From vxwexplo-errs@csg.lbl.gov Tue Jan 1 04:03:09 2002 From: Vxworks Exploder Date: Tue Jan 1 04:03:12 PST 2002 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Tue Jan 1 04:03:05 PST 2002 Subject: Re: which network devices supported by VxWorks? Subject: SBS IP Serial Interface Board Subject: Debugging with Visual SlickEdit Subject: Page fault in "tBoot" Subject: almToggleAlarm Command Help Subject: Re: Using select(), need to monitor the serial port continuously in the thread ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: which network devices supported by VxWorks? Date: Mon, 31 Dec 2001 13:25:24 +0000 (UTC) From: don@mars.dgrc.crc.ca (Donald McLachlan) Organization: The unconfigured xvnews people Message-ID: References: Reply-To: don@mars.dgrc.crc.ca In article , Johan Borkhuis writes: > Amit Raam wrote: > > > Hi, > > Does anyone know of a list of officially supported network devices > > (chips) supported by VxWorks under the XScale architecture? Is there > > such a list on the wrs site? WRS is a pain in the ass about this. It would be nice to be able to go to their web site, look at a list of supported devices, and go and buy something they support. We wanted to do this when transitioning from 68K vme platform to PC. Instead we had to jump through hoops with the pre-sales support people. :-( What a pain. WRS, it will be 2002 tomorrow, so please get with the 90's at least and put this info on your web site. - -- Donald McLachlan E-mail Donald.McLachlan@crc.ca Communications Research Centre / RNS Tel (613) 998-2845 3701 Carling Ave., Fax (613) 998-9648 Ottawa, Ontario K2H 8S2 Canada --------------------------- Newsgroups: comp.os.vxworks Subject: SBS IP Serial Interface Board Date: Mon, 31 Dec 2001 09:49:36 -0500 From: Dick Graham Message-ID: <3C307B00.E5C99747@ll.mit.edu> Hi: Does anyone have a driver for the SBS Greenspring IP-OctalPlus Serial interface board? I have the SBS driver for the PMC-OctalPro, P-Octal, and CP-6201 boards as well as the VxWorks templateSio.c program. These can be used as examples for another serial driver but I'm hoping one already exists. Thanks. Dick Graham --------------------------- Newsgroups: comp.os.vxworks Subject: Debugging with Visual SlickEdit Date: Mon, 31 Dec 2001 16:25:07 GMT From: "spaceman" Organization: Prodigy Internet http://www.prodigy.com Message-ID: Anyone been able to configure Visual SlickEdit as a graphical debugger for Tornado? Jim --------------------------- Newsgroups: comp.os.vxworks Subject: Page fault in "tBoot" Date: 31 Dec 2001 10:43:00 -0800 From: barryrichard@hotmail.com (RichardBarry) Organization: http://groups.google.com/ Message-ID: I have a simple application that has a periodic task that runs each time a timer interrupt frees a semaphore, and two communications tasks (one for Txing and one for Rxing TCP packets). The tasks communicate using message queues with the comms tasks being lower priority. There is also a TFTP client task. For the vast majority of the time everything works well and the system can be left executing for days. However, running the system over an already heavily utilized network (i.e. with lots of trafic unrelated to my app) produces the following error: Page fault program counter: 0x00xxxxxx status register: 0x00010297 error code: 0x00000002 task: 0x72ad4"tBoot" A PCI Intel network card is being used along with a Pentium processor. Any ideas would be gratefully appreciated. Regards, Richard. --------------------------- Newsgroups: comp.os.vxworks Subject: almToggleAlarm Command Help Date: Mon, 31 Dec 2001 21:53:45 GMT From: "magicmoments4" Organization: AT&T Broadband Message-ID: Does anyone know how the command "almToggleAlarm" works in VxWorks? The command wants parameters (slotId, port, ent, defectIdx, Raise). Where would I find these values? Thanks, William magicmoments4@attbi.com --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Using select(), need to monitor the serial port continuously in the thread Date: 1 Jan 2002 00:09:14 -0800 From: shafi4u@rediffmail.com (shafi) Organization: http://groups.google.com/ Message-ID: References: <23de5426.0112302117.12e9fe30@posting.google.com> Hi, First argument in your select seems to be incorrect.. it should be... /* pend, waiting for the file descriptor to become ready */ if( select( nOpenReturn+1, &readFDs, NULL, NULL, NULL ) == ERROR ) return ( ERROR ); shafi manibala@covansys.com (B.Manivannan) wrote in message news:<23de5426.0112302117.12e9fe30@posting.google.com>... > Good Day Viewers, > > 1. I am developing an application which should continuously monitor > the serial port inside the thread for any data arrival. > > 2. In a application, i just created a thread. Inside a thread, i am > opening a port. After that FOREVER loop has started. In side the > forever loop, i have given selecet statement as per the example given > by the help. It goes inside the FOREVER loop and got hanged.. > I Hope the error is in the line, if( FD_ISSET( nOpenReturn, > &readFDs ) ) > I have no idea, why it fails.. > > 3.I am using Tornado II and VxWorks 5.4. > > 4. The Source code, i have generated is as follows > > /*My Source Code Starts Here....*/ > #include "VxWorks.h" > #include "fcntl.h" > #include "selectLib.h" > #include "ioLib.h" > #include "taskLib.h" > #include "sigLib.h" > #include "stdio.h" > #include "string.h" > > #define STACK_SIZE 20000 > #define MAX_FDS 2 > #define MAX_DATA 1024 > > /* Thread Identifier */ > int nReadProcess; > > /* File Descriptor */ > int nOpenReturn; > > /* Thread to monitor and read the serial port continuously */ > void ReadingPort(void); > > /* Application gets initialized here */ > void Read(void) > { > /* Task or thread is being started */ > nReadProcess = taskSpawn ("Reading",200, 0, STACK_SIZE, > (FUNCPTR) ReadingPort,0,0,0,0,0,0,0,0,0,0); > } > void ReadingPort(void) > { > struct fd_set readFDs; > int width, i; > char buffer[MAX_DATA]; > > /* Open File Descriptors */ > nOpenReturn = open("/tyCo/0",2,0); > nStatus1 = ioctl( nOpenReturn, FIOBAUDRATE, 9600 ); > nStatus2 = ioctl( nOpenReturn, FIOSETOPTIONS, OPT_RAW); > > /* Loop for ever for reading data from the serial port */ > FOREVER > { > /* Clear bits in read bit mask */ > FD_ZERO(&readFDs); > > /* Initialise bit masks */ > FD_SET( nOpenReturn, &readFDs); > > /* pend, waiting for the file descriptor to become ready */ > if( select( nOpenReturn, &readFDs, NULL, NULL, NULL ) == ERROR ) > return ( ERROR ); > > /* read from the file descriptor it it gets ready */ > if( FD_ISSET( nOpenReturn, &readFDs ) ) > { > read(nOpenReturn, buffer, MAX_DATA); > printf("The data read from the serial port is %s", buffer); > } > } > > } > /*My Source Code Ends Here....*/ > > 5. Can you suggest me, what is the problem with this code? If it is > not a right way, what is an alternate path to read the message from > the serial port while data occurs at any time. Can you give me some > sample code over here?? > > With Thanks and Regards, > Manivannan.B. --------------------------- End of New-News digest ********************** From vxwexplo-errs@csg.lbl.gov Tue Jan 1 07:04:42 2002 From: "Nachum, Yaron (Yaron)" Date: Tue Jan 1 07:04:44 PST 2002 Subject: dosFsConfigShow causes a failure in writing to the flash device This is a multi-part message in MIME format. ------_=_NextPart_001_01C192D5.847729AB Content-Type: text/plain; charset="windows-1255" Content-Transfer-Encoding: quoted-printable I am running vxWorks 5.4 on a PPC target. I am trying to run dosFs over TFFS. The flash I am using is Intel 28F128J3A starta flash. The problem I have found is - when trying to read the dosFs configuration using dosFsConfigShow while another task is writing to the flash, the write fails. The operation fails on the fflush command - which means the failure occurs when trying to write the data from the buffers to the actual flash. Did anyone encountered with this problem ? ------_=_NextPart_001_01C192D5.847729AB Content-Type: text/html; charset="windows-1255" Content-Transfer-Encoding: quoted-printable dosFsConfigShow causes a failure in writing to the flash = device

I am running vxWorks 5.4 on a = PPC target.

I am trying to run dosFs over = TFFS.

The flash I am using is Intel = 28F128J3A starta flash.

The problem I have found is - = when trying to read the dosFs configuration using dosFsConfigShow while = another task is writing to the flash, the write fails.
The operation fails on the fflush command - which means the failure = occurs when trying to write the data from the buffers to the actual = flash.

Did anyone encountered with = this problem ?

------_=_NextPart_001_01C192D5.847729AB-- From vxwexplo-errs@csg.lbl.gov Tue Jan 1 20:24:49 2002 From: displaystar@netscape.net (L Srinivas Reddy) Date: Tue Jan 1 20:24:51 PST 2002 Subject: VXWORKS:failed in setting the target Dear all, In setting up the target i am using Intel Fast Eth Pro 100 card and enabled FEI driver for the target using TORNADO IDE, after setting up the IP address target is able to download the image from the FTP server, after that it is not comming into network( it is not showing any error in muxload() also). what could be problem. Is it the problem of adding the driver to stack(TCP/IP) or with card. thanks in adv.. for your response regards ls -- __________________________________________________________________ Your favorite stores, helpful shopping tools and great gift ideas. Experience the convenience of buying online with Shop@Netscape! http://shopnow.netscape.com/ Get your own FREE, personal Netscape Mail account today at http://webmail.netscape.com/ From vxwexplo-errs@csg.lbl.gov Tue Jan 1 22:12:13 2002 From: displaystar@netscape.net (L Srinivas Reddy) Date: Tue Jan 1 22:12:16 PST 2002 Subject: VXQWORKS:" failed setting the tartget: " Got It. Hi , Now i am able to get the communiation in the LAN,problem was with "IRQ". Now i would like to use one more same type of card and configure this second card, how to do it , any inputs regarding this are highly appreciated thanks and regards ls Dear all, In setting up the target i am using Intel Fast Eth Pro 100 card and enabled FEI driver for the target using TORNADO IDE, after setting up the IP address target is able to download the image from the FTP server, after that it is not comming into network( it is not showing any error in muxload() also). what could be problem. Is it the problem of adding the driver to stack(TCP/IP) or with card. thanks in adv.. for your response regards ls -- -- __________________________________________________________________ Your favorite stores, helpful shopping tools and great gift ideas. Experience the convenience of buying online with Shop@Netscape! http://shopnow.netscape.com/ Get your own FREE, personal Netscape Mail account today at http://webmail.netscape.com/ From vxwexplo-errs@csg.lbl.gov Wed Jan 2 04:03:10 2002 From: Vxworks Exploder Date: Wed Jan 2 04:03:12 PST 2002 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Wed Jan 2 04:03:06 PST 2002 Subject: Re: which network devices supported by VxWorks? Subject: Re: which network devices supported by VxWorks? Subject: Re: which network devices supported by VxWorks? Subject: Re: Using select(), need to monitor the serial port continuously in the thread Subject: Re: Networking in VxSim Subject: tShell and exception handle Subject: how do microcode Patch for Relocating SMC(uart)_I2C_SPI Parameters Subject: Interrupt Levels - An Intro... Subject: dhcp and bootrom? ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: which network devices supported by VxWorks? Date: Tue, 01 Jan 2002 15:52:09 +0200 From: Amit Raam Organization: Verio Message-ID: References: <3eut2ukj5la3b374ha4fppboifbcnhrd49@4ax.com> I did, to little effect. They insist that I ask about each chip I'm considering. our electronics guys think I'm hiding something.. they can't believe it when I tell them there's no way to tell!. And, naturally, they want to back their design decision with something a little more that an Emailed 'yes' from our local rep.. Even Microsoft has an HCL[1]; why can't WindRiver? [1] Hardware Compatibility List Johan Borkhuis wrote: >Amit Raam wrote: > >> Hi, >> Does anyone know of a list of officially supported network devices >> (chips) supported by VxWorks under the XScale architecture? Is there >> such a list on the wrs site? > >You have to ask WindRiver about this. > >Groeten, > Johan --------------------------- Newsgroups: comp.os.vxworks Subject: Re: which network devices supported by VxWorks? Date: Tue, 01 Jan 2002 19:30:32 GMT From: "cfk@pacbell.net" Organization: Prodigy Internet http://www.prodigy.com Message-ID: References: <3eut2ukj5la3b374ha4fppboifbcnhrd49@4ax.com> Dear Amit: In an analogous fashion to the fact that the hardware engineers have to do their homework to connect a new (to them) peripheral chip to new (to them) hardware, you need to do similar software homework. Basically, there are just a few things to be concerned about with any peripheral chip software in vxWorks and they are: 1) Has the driver accomodated little versus bit endianness 2) Has the driver accomodated macros for inb, outb, inw, outb, inl, outl for I/O versus memory mapped architectures (Intel vs Motorola) 3) Does the driver respond proeprly to DMA and intConnect requests. If all three of these is true then that driver, be it from x86, ebsa285, 68k or xcale BSP should work fine. If it does not, then as a software engineer, you need to figure out if the issue is your hardware engineers ability to connect the pins properly of an actual defect in the driver. I have made both the Intel derived 82559 and AMD97C973 ethernet chips work with the xcale BSP in the last year. The 82559 came from the ebsa285 BSP and the AMD came from the x86 BSP. In both cases, it was merely necessary to ensure that the three items above were accomodated. Charles "Amit Raam" wrote in message news:egf33u4i8e145gqfnpshfpofgvb0ulp2hj@4ax.com... > I did, to little effect. They insist that I ask about each chip I'm considering. > our electronics guys think I'm hiding something.. they can't believe it when I > tell them there's no way to tell!. And, naturally, they want to back their > design decision with something a little more that an Emailed 'yes' from our > local rep.. > Even Microsoft has an HCL[1]; why can't WindRiver? > > > [1] Hardware Compatibility List > > Johan Borkhuis wrote: > > >Amit Raam wrote: > > > >> Hi, > >> Does anyone know of a list of officially supported network devices > >> (chips) supported by VxWorks under the XScale architecture? Is there > >> such a list on the wrs site? > > > >You have to ask WindRiver about this. > > > >Groeten, > > Johan > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: which network devices supported by VxWorks? Date: 1 Jan 2002 14:42:08 -0800 From: john_94501@yahoo.com (John) Organization: http://groups.google.com/ Message-ID: <488e459a.0201011442.6290d564@posting.google.com> References: <3eut2ukj5la3b374ha4fppboifbcnhrd49@4ax.com> Hello Amit, > Even Microsoft has an HCL[1]; why can't WindRiver? Yes, and even with only one architecture to support, their HCL is not always accurate. Unfortunately, even with an architecture as controlled as the PC one, there is still an incredible amount of leeway for the H/W guys to make subtle changes that render a "generic" driver useless. Kudos to MS for trying, but I think that they fighting a losing battle until they own the hardware design too (c.f. Apple). Now, what you can do is find the list of network drivers that are shipped with VxWorks (or, if you don't have it installed anywhere, I can tell you what is installed in my tree). Unless the driver is very CPU-specific (e.g. the PPC860 on-chip network interface driver), it will be compiled for all architectures. That doesn't mean it has been tested on them all though. The person you spoke to at WRS was probably trying to find out the exact chip you were using to get information on whether your combination of CPU and network chip have been seen and/or tested before by them. A quick way of finding out what is normally used on a given architecture is to take a look in the reference BSP at the driver interface stubs. For example, the AE ebsa285 BSP's sysEnd.c file contains a list of cards that have been tested with the board. As somebody else pointed out in a response to your posting, for the most part, the VxWorks drivers try to accomodate all architectures through the use of "magic" macros to deal with endianess, interrupt styles and I/O access styles. When dealing with a custom board though, you must allow some time to iron out the wrinkles, especially when it comes to the interface routines in the BSP (usually in sysEnd.c, though sometimes there are several files with names corresponding to the driver). For XScale, since it is basically an ARM, and I have an ARM installation, here's a list of drivers that I've seen on ARM boards: * DEC 21x4x series [dc] (obsolete chips now I believe) * Intel 82559 [fei] (recent addition to replace dc) * NE2000 [ene] (special CF version for Assabet board) * Olicom [oli] (special PCMCIA version for ARM PID boards - obsolete) One thing to look out for is the presence of the "offset" value in the END load string for the driver you pick. ARM needs this to be set to 2 usually, and some of the older drivers might not support it. Check the reference manual for the driver to see if this is present. I had a quick look at some of the common ones in my AE 1.1 tree, and found that the 3COM 90x & 509, Fujitsu MB86960 and NS DP83932B SONIC drivers don't seem to support it; all the other non-CPU specific ones I looked at did. HTH, John... > > [1] Hardware Compatibility List > > Johan Borkhuis wrote: > > >Amit Raam wrote: > > > >> Hi, > >> Does anyone know of a list of officially supported network devices > >> (chips) supported by VxWorks under the XScale architecture? Is there > >> such a list on the wrs site? > > > >You have to ask WindRiver about this. > > > >Groeten, > > Johan --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Using select(), need to monitor the serial port continuously in the thread Date: 1 Jan 2002 23:12:44 -0800 From: manibala@covansys.com (B.Manivannan) Organization: http://groups.google.com/ Message-ID: <23de5426.0201012312.3e4415fc@posting.google.com> References: <23de5426.0112302117.12e9fe30@posting.google.com> Good Day Mr.Shafi, Thanks for your response. But still the control gets hanged in the FOREVER loop. I made nOpenReturn to nOpenReturn+1. In this select statement only it got hanged.. What i am doing is, I will type the "Read" and enter in the Target Shell..It has to wait till the message arrives to serial port..One the message has come, it has to fall down the flow and should continue to run..Hope you would have understood my requirement. Regards, Manivannan.B. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Networking in VxSim Date: 1 Jan 2002 23:53:39 -0800 From: raj_rashid@hotmail.com (Rashid) Organization: http://groups.google.com/ Message-ID: <2a6118da.0201012353.6c3b987@posting.google.com> References: <15a87607.0112282319.5f582c94@posting.google.com> <2a6118da.0112290814.48658d6e@posting.google.com> <15a87607.0112302127.4e61169a@posting.google.com> > Thanks for the response. But what am in need of is, from the VxWorks > target IP (90.0.0.1) i want to ping another vxworks image ( 90.0.0.2) > in another host. remember i want to ping the other image in another > host from the target shell of 90.0.0.1. My PC ethernet adapter: 192.168.100.69 ( Win NT ) My ULIP adapter: 90.0.0.254 My VxWorks target iamge : 90.0.0.1 In target shell: - -> routeAdd("0.0.0.0","90.0.0.254") value = 0 = 0x0 - -> routeShow ROUTE NET TABLE destination gateway flags Refcnt Use Interface - ---------------------------------------------------------------------------- 0.0.0.0 90.0.0.254 3 0 0 nt0 - ---------------------------------------------------------------------------- ROUTE HOST TABLE destination gateway flags Refcnt Use Interface - ---------------------------------------------------------------------------- 127.0.0.1 127.0.0.1 5 1 4 lo0 - ---------------------------------------------------------------------------- value = 77 = 0x4d = 'M' On this host do, C:\> route ADD 90.0.0.2 192.168.100.144 METRIC 2 Now from target shell: - ->ping "192.168.100.69",4 PING 192.168.100.69: 56 data bytes 64 bytes from 192.168.100.69: icmp_seq=0. time=0. ms 64 bytes from 192.168.100.69: icmp_seq=1. time=0. ms 64 bytes from 192.168.100.69: icmp_seq=2. time=0. ms 64 bytes from 192.168.100.69: icmp_seq=3. time=16. ms - ----192.168.100.69 PING Statistics---- 4 packets transmitted, 4 packets received, 0% packet loss round-trip (ms) min/avg/max = 0/4/16 value = 0 = 0x0 - -> ping "192.168.100.144",4 PING 192.168.100.144: 56 data bytes 64 bytes from 192.168.100.144: icmp_seq=0. time=48. ms 64 bytes from 192.168.100.144: icmp_seq=1. time=32. ms 64 bytes from 192.168.100.144: icmp_seq=2. time=16. ms 64 bytes from 192.168.100.144: icmp_seq=3. time=32. ms - ----192.168.100.144 PING Statistics---- 4 packets transmitted, 4 packets received, 0% packet loss round-trip (ms) min/avg/max = 16/32/48 value = 0 = 0x0 Other end ethernet address : 192.168.100.114 ( Win 2k) Other end ULIP address : 90.0.0.253 Other end VxWorks Target image : 90.0.0.2 Repeat same steps on other host. on this host do, c:\> route ADD 90.0.0.1 192.168.100.69 METRIC 2 from 1st host Target shell : - -> ping "90.0.0.2",4 PING 90.0.0.2: 56 data bytes 64 bytes from 90.0.0.2: icmp_seq=0. time=592. ms 64 bytes from 90.0.0.2: icmp_seq=1. time=400. ms 64 bytes from 90.0.0.2: icmp_seq=2. time=512. ms 64 bytes from 90.0.0.2: icmp_seq=3. time=304. ms - ----90.0.0.2 PING Statistics---- 4 packets transmitted, 4 packets received, 0% packet loss round-trip (ms) min/avg/max = 304/452/592 value = 0 = 0x0 Try these steps, it is working on my machine. Best regards, Rashid --------------------------- Newsgroups: comp.os.vxworks Subject: tShell and exception handle Date: Wed, 2 Jan 2002 18:03:29 +0800 From: "Survey Liu" Organization: Changzhou Telecom Message-ID: Hi,there How will exception be handled while tShell cause it? I have done a lot of test, and result in different ways. 1. When I added a extra hooker via calling excHookAdd, the hooker will not be called if the exc is caused by shell. 2. Using code: pOldVec = intVecGet(); intVecSet( myHandler ); and call pOldVec in myHandler(). In this case, if exception is caused by other tasks than shell, it acts normally. But, if exception is caused by tShell, there seems fall into a nested exc handling. and shell will be locked then. How does tShell processing exception diffs from other tasks? - -- Best regards, Survey Liu - ---------------------------------------------------------------------------- - ---- - -> As you say, you are right. @ @ 4/F Wearnes Technology Bldg. No.10 - --> I do think that you have your idea. | Kefa Rd.,Science & Industry Park - -> Yes, I am really loving you. \|/ Nanshan Shenzhen 518057,PRC - --> Sure, I love you, too. | UTstarcom (China) Shenzhen Branch - -----------------------------------------|---------------------------------- - ---- Tel: +86-755-6683555-4300 | Fax: +86-755-6635550 Email: survey.liu@utstar.com | Web: http://www.utstar.com --------------------------- Newsgroups: comp.os.vxworks Subject: how do microcode Patch for Relocating SMC(uart)_I2C_SPI Parameters Date: Wed, 2 Jan 2002 19:24:32 +0900 From: "it2xpert" Organization: ONSE Telecom Message-ID: Everybody Happy new year !!! I used T2 Vxworks5.4 I desiged the MPC850SR Board.. I want to use the board for 2 ethernet and 1 console, then i desinged SCC2, SCC3 for ethernet and SMC1 for console.... I face to some problem, parameter RAM is overlaied... I downloaded the microcode to the w-motoroal site. how do microcode Patch for Relocating SMC(uart)_I2C_SPI Parameters? thanks jwyu --------------------------- Newsgroups: comp.os.vxworks Subject: Interrupt Levels - An Intro... Date: 2 Jan 2002 03:07:19 -0800 From: arun4vxworks@indiatimes.com (Arun Prasad) Organization: http://groups.google.com/ Message-ID: Hi All, *I have some doubts on ISRs on VxWorks. Infact I need some good introduction on interrupt levels, their usage, limitations. I'm writing NIC driver on VxWorks by using a Linux Driver where there's no interrupt levels. Can anyone throw some torch in this regard? **Also, I'd like to know on IRQ sharing and SMP kernel issues on VxWorks. ***If you can give me some information on good mailing lists and/or news/mailing groups or any good sites or material which explaing these that will be helpful for me to understand interrupt levels, tasks and scheduling in VxWorks. Thank you. Regards Arun --------------------------- Newsgroups: comp.os.vxworks Subject: dhcp and bootrom? Date: Wed, 02 Jan 2002 11:28:53 +0000 From: Owain Phillips Organization: Siemens Inc. Message-ID: <3C32EEF5.AB79C253@siemenscomms.co.uk> This is a multi-part message in MIME format. - --------------002927D1F3244F22F64E5245 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit I'm using Tornado 2.1 for mips and have tryed to turn on DHCP for the bootroms. Does anyone know if this works? I get an exception somewhere in dhcpcBootInit....vxworks library code. Our network interface and bootroms work fine assuming we don't try the dhcp stuff. Regards, Owain - --------------002927D1F3244F22F64E5245 Content-Type: text/x-vcard; charset=us-ascii; name="owain.phillips.vcf" Content-Transfer-Encoding: 7bit Content-Description: Card for Owain Phillips Content-Disposition: attachment; filename="owain.phillips.vcf" begin:vcard n:Phillips;Owain tel;fax:+44 115 943 4969 tel;work:+44 115 943 2167 x-mozilla-html:TRUE url:http://www.siemenscomms.co.uk org:Siemens Comunications Ltd version:2.1 email;internet:owain.phillips@siemenscomms.co.uk title:Software Engineer adr;quoted-printable:;;Technology Drive=0D=0ABeeston;Nottingham;;NG9 1LA;United Kingdom fn:Owain Phillips end:vcard - --------------002927D1F3244F22F64E5245-- --------------------------- End of New-News digest ********************** From vxwexplo-errs@csg.lbl.gov Wed Jan 2 08:16:09 2002 From: "Baron, Ken" Date: Wed Jan 2 08:16:11 PST 2002 Subject: serial port write priority I am using VxWorks 5.4 with Tornado 2. If I have two tasks writing to a serial port and the serial port ring buffer is full, thus causing both tasks to block, which task will unblock first? It seems to me that the tasks are accessing the serial port in a SEM_Q_FIFO manner rather than a SEM_Q_PRIORITY manner. Is there any way to change this? From vxwexplo-errs@csg.lbl.gov Wed Jan 2 20:50:09 2002 From: "Sajjad Ahmed" Date: Wed Jan 2 20:50:12 PST 2002 Subject: Basic Knowledge on Serial communication? Hi VxWorks Gurus, I wanted Some Basic Knowledge on communicating to serial port like: 1)How do i write to serial port from task context? 2)What is Ring Buffer wrt serial prot? 3)example? and more could be appreciated. Thanx in advance. Sajjad. From vxwexplo-errs@csg.lbl.gov Wed Jan 2 23:13:45 2002 From: displaystar@netscape.net (L Srinivas Reddy) Date: Wed Jan 2 23:13:47 PST 2002 Subject: RE: re: VXWORKS:failed in setting the target "Andy Levin" wrote: >>Dear all, >> In setting up the target i am using Intel Fast Eth Pro 100 card and >enabled FEI driver for the target using TORNADO IDE, after setting >>up the IP address target is able to download the image from the FTP >server, after that it is not coming into network( it is not showing any >>error in muxload() also). what could be problem. Is it the problem of >adding the driver to stack(TCP/IP) or with card. >>thanks in adv.. for your response > >How did you enable FEI using the Tornado IDE? When you say "not coming >into network" do you mean the target will not answer a ping? Is it >possible you are using a duplicate IP address? I use the same network >card, but I don't boot manually. I put the following into my >DEFAULT_BOOT_LINE, and all seems to work fine. > >#elif (CPU_VARIANT == PENTIUMPRO) >#define DEFAULT_BOOT_LINE \ > "fei(0,0)(0,0)andyl1:c:\\pumadev\\bootable\\default\\vxWorks >h=192.168.209.6 e=192.168.209.25 u=target pw=ntctntct" >#endif /* (CPU == I80386) */ > >Note that h is the tornado ide host, h is the target, u is the ftp user >name, pw is the ftp password. > >I hope this helps. > >-a > > hi Andy and all, thnaks for your reply , i did the same it is working now with one interface, now i would like to add one more same card so i want to know how to add entry in the config.h, please suggest me something, i am not able to find any documentation regarding this. thanks in adv.. regards LS -- __________________________________________________________________ Your favorite stores, helpful shopping tools and great gift ideas. Experience the convenience of buying online with Shop@Netscape! http://shopnow.netscape.com/ Get your own FREE, personal Netscape Mail account today at http://webmail.netscape.com/ From vxwexplo-errs@csg.lbl.gov Thu Jan 3 03:07:02 2002 From: "Tyler, Paul A (DBY)" Date: Thu Jan 3 03:07:04 PST 2002 Subject: Re: Dumb question about the target-resident shell Try setting sysCplusEnable = 1 in your start script. This allows the shell to mangle the entry name, and it can then find it in the symbol table. Paul Tyler Measurement Systems Specialist Rig Test Systems Team, Facility Engineering Dept. Rolls-Royce plc. e-mail Paul.A.Tyler@Rolls-Royce.com Tel. +44 (1332) 247808 Fax. +44 (1332) 247928 ----- Original Message ----- From: Trevor Ward Newsgroups: comp.os.vxworks Sent: Wednesday, January 02, 2002 4:08 PM Subject: Dumb question about the target-resident shell > Since there aren't any stupid questions (only stupid answers), I thought I'd > provide the exception that proves the rule... > > I've built a bootable image using Tornado 2, extending the default kernel with > some extra .o object modules of my own. The entry point into my (C++) > application is a function called 'fdl' that takes an int parameter. I want to > execute 'fdl(1)' from a startup script whenever the target boots. I have a > target resident symbol table, so I can perform 'lkup "fdl"' on the target shell > and it gives me an entry like: > > fdl(int) 0x0011e3b0 text > > If I type 'fdl(1)' I get the error 'undefined symbol: fdl', but if I type > '0x0011e3b0(1)' then it runs my application as desired. The problem is that the > address of fdl, i.e. 0x0011e3b0, moves around as I change to code and I don't > want to keep having to change the startup script every time I recompile. I'm > sure there must be a really easy way of resolving this problem, but I can't > find it... > > > -- > Posted via Mailgate.ORG Server - http://www.Mailgate.ORG The data contained in, or attached to, this e-mail, may contain confidential information. If you have received it in error you should notify the sender immediately by reply e-mail, delete the message from your system and contact +44(0)117 9797329 (the Rolls-Royce Security Controller) if you need assistance. Please do not copy it for any purpose, or disclose its contents to any other person. An e-mail response to this address may be subject to interception or monitoring for operational reasons or for lawful business practices. (c) 2001 Rolls-Royce plc From vxwexplo-errs@csg.lbl.gov Thu Jan 3 04:03:15 2002 From: Vxworks Exploder Date: Thu Jan 3 04:03:17 PST 2002 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Thu Jan 3 04:03:11 PST 2002 Subject: Re: which network devices supported by VxWorks? Subject: telnet Subject: Re: problem with 82559er driver Subject: Re: Networking in VxSim Subject: Dumb question about the target-resident shell Subject: Re: Detecting unused variables in Diab Data C compiler Subject: VxWorks Routines List Subject: Re: VxWorks Routines List Subject: Re: Dumb question about the target-resident shell Subject: Re: Using select(), need to monitor the serial port continuously in the thread Subject: Re: Dumb question about the target-resident shell Subject: Re: zinc designer Subject: Can't connect to Windows using sockets?? Subject: Re: Debugging with Visual SlickEdit Subject: Re: Interrupt Levels - An Intro... Subject: Re: VxWorks Routines List Subject: Re: Can't connect to Windows using sockets?? Subject: Re: Debugging with Visual SlickEdit Subject: Problem while writing at TFFS (DiskOnChip) Subject: Re: Dumb question about the target-resident shell Subject: Re: Can't connect to Windows using sockets?? Subject: ldModule versus ld < Subject: how do using for SMC2... Subject: Re: Can't connect to Windows using sockets?? Subject: Re: Symbol table errors compiling an image when using STL Subject: Re: what is "makes pointer from integer without a cast"? Subject: Re: malloc in interruption Subject: RAW Socket support ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: which network devices supported by VxWorks? Date: Wed, 2 Jan 2002 13:44:13 +0000 (UTC) From: don@mars.dgrc.crc.ca (Donald McLachlan) Organization: The unconfigured xvnews people Message-ID: References: <488e459a.0201011442.6290d564@posting.google.com> Reply-To: don@mars.dgrc.crc.ca In article <488e459a.0201011442.6290d564@posting.google.com>, john_94501@yahoo.com (John) writes: > Hello Amit, > > > Even Microsoft has an HCL[1]; why can't WindRiver? > > Yes, and even with only one architecture to support, their HCL is not > always accurate. [ snip ] > The person you spoke to at WRS was probably > trying to find out the exact chip you were using to get information on > whether your combination of CPU and network chip have been seen and/or > tested before by them. I'd like to think that is the case but they simply will not tell you what is supported. When I was transitioning from 68K to PC, I asked them what ethernet cards they supported for the PC. The reply was "tell us what card you want to use and we will tell you whether or not it is supported". (one caveat. I wanted a card that supported promiscuous mode.) Basically I wanted them to say "if you buy this card it will work", but they would/could not. :-( --------------------------- Newsgroups: comp.os.vxworks Subject: telnet Date: 2 Jan 2002 15:21:37 CET From: Scanmar AS Organization: Scanmar AS Message-ID: Hi, I've got a problem recently. Sometimes when I try to connect to my target I am locked out, on my target I got the following message "telnetd: someone tried to login." on my PC I got "Sorry, shell is locked.". The strange thing is that the system allow for telnet connection a long time then suddenly this happens, nothing helps except power off. Someone know what's happening? --------------------------- Newsgroups: comp.os.vxworks Subject: Re: problem with 82559er driver Date: Wed, 02 Jan 2002 15:41:55 +0100 From: Michiel Meekes Organization: NLR the Netherlands Message-ID: <3C331C33.7BE19F8B@nlr.nl> References: <3C20C4B3.A9463A@nlr.nl> <488e459a.0112192051.f43b438@posting.google.com> <3c220092$1@brateggebdc5.br-automation.co.at> Werner Schiendl wrote: > "John" wrote in message > news:488e459a.0112192051.f43b438@posting.google.com... > > Hello, > > > > > I'm having a problem with the latest 82559er driver in vxWorks 5.4 for a > > > Pentium target. > > > As long as I don't attach the network cable the driver init's ok. > > > (ifShow "eeE" returns the correct values, all routines are there) > > > > Did you really mean "eeE" as the name? Where did this driver come from? > Normally > > , 82559 devices are called "fei0" (the number will change if there is more > than > > one network card in the target of course). > > eeE is the name of the driver provided from Intel for this network devices > (binary distribution for x86 platform only). > > > > > > After I attach the cable I get a repeating message: > > > > > > " (0x1fa05230 (tNetTask): Panic: where are those rfd buffers " > > > > Sounds like there's a problem with the load string and/or the driver > initialisat > > ion. The number of RFDs is normally specified in the END load string (look > in co > > nfigNet.h for your BSP in most cases). Which arch/BSP are you using? > > > > The documentation with the driver should contain the correct changes > necessary to the BSP in all detail. > OP: Did you setup the driver as explained in the readme? which BSP? using > END driver? or BSD? > > hth > Werner After looking through some old discussions about this subject I found a suggestion that resulted in a solution for the above problem. The suggestion was to modify the BIOS setting "PNP OS Installed" from Yes to No. After this the Intel 82559er MUX driver (using END) worked OK. Thanks to all who helped in this matter. Michiel --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Networking in VxSim Date: 2 Jan 2002 07:15:58 -0800 From: rashanmu@npd.hcltech.com (Ramesh) Organization: http://groups.google.com/ Message-ID: <15a87607.0201020715.270f5712@posting.google.com> References: <15a87607.0112282319.5f582c94@posting.google.com> <2a6118da.0112290814.48658d6e@posting.google.com> <15a87607.0112302127.4e61169a@posting.google.com> <2a6118da.0201012353.6c3b987@posting.google.com> Hi Rashid, Am really thankful to you for the great help. Now it is working fine. Actually i did the the same but forgot to give the metric as 2. THanks for pointing me out. Just i checked out it is working fine. THanks once again. Ramesh raj_rashid@hotmail.com (Rashid) wrote in message news:<2a6118da.0201012353.6c3b987@posting.google.com>... > > Thanks for the response. But what am in need of is, from the VxWorks > > target IP (90.0.0.1) i want to ping another vxworks image ( 90.0.0.2) > > in another host. remember i want to ping the other image in another > > host from the target shell of 90.0.0.1. > --------------------------- Newsgroups: comp.os.vxworks Subject: Dumb question about the target-resident shell Date: Wed, 2 Jan 2002 16:08:32 +0000 (UTC) From: "Trevor Ward" Organization: Mailgate.ORG Server - http://www.Mailgate.ORG Message-ID: <635472238d6165ac734f3c610a2945d0.56035@mygate.mailgate.org> Since there aren't any stupid questions (only stupid answers), I thought I'd provide the exception that proves the rule... I've built a bootable image using Tornado 2, extending the default kernel with some extra .o object modules of my own. The entry point into my (C++) application is a function called 'fdl' that takes an int parameter. I want to execute 'fdl(1)' from a startup script whenever the target boots. I have a target resident symbol table, so I can perform 'lkup "fdl"' on the target shell and it gives me an entry like: fdl(int) 0x0011e3b0 text If I type 'fdl(1)' I get the error 'undefined symbol: fdl', but if I type '0x0011e3b0(1)' then it runs my application as desired. The problem is that the address of fdl, i.e. 0x0011e3b0, moves around as I change to code and I don't want to keep having to change the startup script every time I recompile. I'm sure there must be a really easy way of resolving this problem, but I can't find it... - -- Posted via Mailgate.ORG Server - http://www.Mailgate.ORG --------------------------- Newsgroups: comp.os.psos,comp.os.vxworks Subject: Re: Detecting unused variables in Diab Data C compiler Date: 2 Jan 2002 08:37:21 -0800 From: barryrichard@hotmail.com (RichardBarry) Organization: http://groups.google.com/ Message-ID: References: <3C2ED6C6.336BC905@bigfoot.com> Look up the -Xlint option. This does what you want. --------------------------- Newsgroups: comp.os.vxworks Subject: VxWorks Routines List Date: Wed, 2 Jan 2002 12:20:33 -0500 From: "Lee Cao" Organization: Computer Sciences Corporatioin Message-ID: Hi! I recently was hired to perform some maintenance and analysis on a VxWorks application written for Pentium single board computers. My question is if there is a master list of VxWorks routines that I can reference to see if a routine used is supplied by VxWorks or not, including hardware-specific routines. So far I've looked at the reference manuals provided by VxWorks and the list doesn't seem to be complete. Any thoughts? Thanx for any help. Lee Cao --------------------------- Newsgroups: comp.os.vxworks Subject: Re: VxWorks Routines List Date: Wed, 02 Jan 2002 17:36:37 GMT From: "H. E. Taylor" Organization: MTS Internet Message-ID: <3C336042.7E16@despam.autobahn.mb.ca> References: In article , Lee Cao wrote: > > I recently was hired to perform some maintenance and analysis on a VxWorks > application written for Pentium single board computers. My question is if > there is a master list of VxWorks routines that I can reference to see if a > routine used is supplied by VxWorks or not, including hardware-specific > routines. So far I've looked at the reference manuals provided by VxWorks > and the list doesn't seem to be complete. Any thoughts? > There are a bunch of small & common functions which are part of a VxWorks image, but which are not documented. I assume this is because you have the source code for them anyway. Your best shot might be to hire someone who knows VxWorks. ;-))> Alternatively if you have a development system, you can become friendly with grep. - -het - -- "progress in software has not followed Moore's law." -John Holland Energy Alternatives: http://www.autobahn.mb.ca/~het/energy.html H.E. Taylor http://www.autobahn.mb.ca/~het/ --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Dumb question about the target-resident shell Date: Wed, 2 Jan 2002 12:40:17 -0500 From: "Heath Moore" Message-ID: <7CHY7.184$g17.89478@newsfeed.slurp.net> References: <635472238d6165ac734f3c610a2945d0.56035@mygate.mailgate.org> I've never extended the kernel, so I can provide no help there...but you may try using fdl 1 in your startup script. No parenthesis should be needed. Good luck. "Trevor Ward" wrote in message news:635472238d6165ac734f3c610a2945d0.56035@mygate.mailgate.org... > Since there aren't any stupid questions (only stupid answers), I thought I'd > provide the exception that proves the rule... > > I've built a bootable image using Tornado 2, extending the default kernel with > some extra .o object modules of my own. The entry point into my (C++) > application is a function called 'fdl' that takes an int parameter. I want to > execute 'fdl(1)' from a startup script whenever the target boots. I have a > target resident symbol table, so I can perform 'lkup "fdl"' on the target shell > and it gives me an entry like: > > fdl(int) 0x0011e3b0 text > > If I type 'fdl(1)' I get the error 'undefined symbol: fdl', but if I type > '0x0011e3b0(1)' then it runs my application as desired. The problem is that the > address of fdl, i.e. 0x0011e3b0, moves around as I change to code and I don't > want to keep having to change the startup script every time I recompile. I'm > sure there must be a really easy way of resolving this problem, but I can't > find it... > > > -- > Posted via Mailgate.ORG Server - http://www.Mailgate.ORG --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Using select(), need to monitor the serial port continuously in the thread Date: 02 Jan 2002 12:34:40 -0500 From: Bill Pringlemeir Organization: Bell Sympatico Message-ID: References: <23de5426.0112302117.12e9fe30@posting.google.com> <23de5426.0201012312.3e4415fc@posting.google.com> Sender: bpringlemeir@DeadDuck >>>>> "B" == B Manivannan writes: B> Good Day Mr.Shafi, Thanks for your response. But still the control B> gets hanged in the FOREVER loop. I made nOpenReturn to B> nOpenReturn+1. In this select statement only it got hanged.. B> What i am doing is, B> I will type the "Read" and enter in the Target Shell..It has to B> wait till the message arrives to serial port..One the message has B> come, it has to fall down the flow and should continue to B> run..Hope you would have understood my requirement. I think that a common mistake is found in the hiding of information in the select() macros. These macros call the ioctl function of the drivers and may end up setting available data to be read from the device. I think that the read() should clear the `select bits'. I would suggest trying something like this, /* Clear bits in read bit mask */ FD_ZERO(&readFDs); /* Initialise bit masks */ FD_SET( nOpenReturn, &readFDs); /* Loop for ever for reading data from the serial port */ FOREVER { /* pend, waiting for the file descriptor to become ready */ if( select( nOpenReturn, &readFDs, NULL, NULL, NULL ) == ERROR ) return ( ERROR ); /* read from the file descriptor it it gets ready */ if( FD_ISSET( nOpenReturn, &readFDs ) ) { read(nOpenReturn, buffer, MAX_DATA); printf("The data read from the serial port is %s", buffer); } } This all depends on the implementation of the driver. If the driver does this, case FIOSELECT: if(selWakeupType((SEL_WAKEUP_NODE *) arg) == SELREAD) { selNodeAdd(&Dev->selWakeupList, (SEL_WAKEUP_NODE *)arg); ... OR, case FIOSELECT: if(selWakeupType((SEL_WAKEUP_NODE *) arg) == SELREAD) { selNodeAdd(&Dev->selWakeupList, (SEL_WAKEUP_NODE *)arg); if(dev->DataAvailable) selWakeup(...); ... vxWorks programmer's guide recommends the later behavior. I have found that this can get you into infinite loops due to the setting/reseting of the `DataAvailable' flag, with the hidden ioctl calls in the FD_xxx macros. This is especially true when multiple tasks are pending on the device. I have found weird things on Linux as well when using select()... at least I think it is a tricky mechanism. It will work well when you get it right though! hth, Bill Pringlemeir. - -- Listen to your life insurance tomorrow. Do something nice for your child tomorrow. Your finances will become exposed in the next several years. You'll often be broke. vxWorks FAQ, "http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html" --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Dumb question about the target-resident shell Date: Wed, 2 Jan 2002 12:54:47 -0500 From: "CIND" Organization: Oak Ridge National Lab, Oak Ridge, TN Message-ID: References: <635472238d6165ac734f3c610a2945d0.56035@mygate.mailgate.org> Replace your declaration statement in your header with: extern "C" int fdl(int x); // Your return type and args This will disable name mangling of the function and make it accessable to the shell's lookup function. "Trevor Ward" wrote in message news:635472238d6165ac734f3c610a2945d0.56035@mygate.mailgate.org... > Since there aren't any stupid questions (only stupid answers), I thought I'd > provide the exception that proves the rule... > > I've built a bootable image using Tornado 2, extending the default kernel with > some extra .o object modules of my own. The entry point into my (C++) > application is a function called 'fdl' that takes an int parameter. I want to > execute 'fdl(1)' from a startup script whenever the target boots. I have a > target resident symbol table, so I can perform 'lkup "fdl"' on the target shell > and it gives me an entry like: > > fdl(int) 0x0011e3b0 text > > If I type 'fdl(1)' I get the error 'undefined symbol: fdl', but if I type > '0x0011e3b0(1)' then it runs my application as desired. The problem is that the > address of fdl, i.e. 0x0011e3b0, moves around as I change to code and I don't > want to keep having to change the startup script every time I recompile. I'm > sure there must be a really easy way of resolving this problem, but I can't > find it... > > > -- > Posted via Mailgate.ORG Server - http://www.Mailgate.ORG --------------------------- Newsgroups: comp.os.vxworks Subject: Re: zinc designer Date: 02 Jan 2002 13:56:28 -0500 From: Bill Pringlemeir Organization: Bell Sympatico Message-ID: References: Sender: bpringlemeir@DeadDuck >>>>> "divya" == divya writes: > I have started working in zinc designer.In zinc,there is an option > to use it in embedded systems.But,I find errors while downloading > my file(when zinc is used in tornado environment with > vxWorks).What should I do to remove the error and work > comfortably? Note, Zinc will not work on vxWorks unless you have purchased the WindML and Zinc libraries from WindRiver. You can get Zinc to work with the simulator, but it is probably easier to get it working with a supported BSP (Ie something with WindML video, keyboard and mouse drivers). The free Zinc will not work with the free vxWorks demo kit. Most people will be trying this that ask questions of this sort. Either that or someone who understands the problem has bought the necessary items and given them to you to use... hth, Bill Pringlemeir. - -- DES PLO security [Hello to all my fans in domestic surveillance] counter-intelligence Waco, Texas Honduras spy Rule Psix kibo explosion FBI cryptographic Cocaine nuclear vxWorks FAQ, "http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html" --------------------------- Newsgroups: comp.os.vxworks Subject: Can't connect to Windows using sockets?? Date: Wed, 02 Jan 2002 19:12:01 GMT From: "agin" Organization: Prodigy Internet http://www.prodigy.com Message-ID: <5YIY7.21096$xM1.863790177@newssvr14.news.prodigy.com> I am unsuccessfully trying to connect from a vxWorks client to a win2k server using sockets. I can connect from win2k to vxWorks, vxWorks to vxWorks, win2k to win2k but I am unable to connect from vxWorks to win2k. I am using winsock2. Is there something I am missing? Here is snipet of the vxWorks client code: clientSock = socket (AF_INET, SOCK_STREAM, 0); serverAddr.sin_family = AF_INET; serverAddr.sin_port = htons(SERVER_NUM); if (connect (clientSock, (struct sockaddr *)&serverAddr, sizeof (serverAddr)) == ERROR) { printf ("Connect failed:\n"); printErrno(errnoGet()); close (clientSock); return (ERROR); } Here is a snipet of the Windows Server code: u_long nInterfaceAddr = inet_addr(pcAddress); if (nInterfaceAddr != INADDR_NONE) { SOCKET sd = socket(AF_INET, SOCK_STREAM, 0); if (sd != INVALID_SOCKET) { sockaddr_in sinInterface; sinInterface.sin_family = AF_INET; sinInterface.sin_addr.s_addr = nInterfaceAddr; sinInterface.sin_port = nPort; if (bind(sd, (sockaddr*)&sinInterface, sizeof(sockaddr_in)) != SOCKET_ERROR) { listen(sd, 1); return sd; } } } accept(sd, (sockaddr*)&sinRemote, sizeof(sinRemote)); art --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Debugging with Visual SlickEdit Date: Wed, 02 Jan 2002 19:29:37 +0000 From: molochai Message-ID: <23nv0a.i63.ln@vapour-trail> References: spaceman wrote: > Anyone been able to configure Visual SlickEdit as a graphical debugger for > Tornado? Didn't know it could be used in this way, if it is indeed possible I'd love to know. M --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Interrupt Levels - An Intro... Date: Wed, 02 Jan 2002 19:33:45 +0000 From: molochai Message-ID: References: Arun Prasad wrote: > Hi All, > > *I have some doubts on ISRs on VxWorks. Infact I need some > good introduction on interrupt levels, their usage, limitations. > > I'm writing NIC driver on VxWorks by using a Linux Driver > where there's no interrupt levels. > Can anyone throw some torch in this regard? > > **Also, I'd like to know on IRQ sharing and SMP kernel issues > on VxWorks. > As VxWorks does not do SMP, there are no issues there. As for sharing interrupts, this is entirely possible for most devices, and certainly for any PCI peripherals. > ***If you can give me some information on good mailing lists > and/or > news/mailing groups or any good sites or material which explaing these > that will be helpful for me to understand interrupt levels, tasks and > scheduling in VxWorks. Try the VxWorks Programmers guide for a quick intro. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: VxWorks Routines List Date: Wed, 2 Jan 2002 20:38:45 +0100 From: "Martin Raabe" Organization: T-Online Message-ID: References: Hello Lee Cao, use the Source Code Navigation Software "SNiFF+" to see, what is wher in your source code. SNiFF+ 4.0.x is a WindPowerTool for Tornado. Ciao Martin Raabe PS: Press "F2" in SNiFF+ to get the manuals for the highlighted function. "Lee Cao" schrieb im Newsbeitrag news:a0vfgr$ko6$1@lore.csc.com... > Hi! > > I recently was hired to perform some maintenance and analysis on a VxWorks > application written for Pentium single board computers. My question is if > there is a master list of VxWorks routines that I can reference to see if a > routine used is supplied by VxWorks or not, including hardware-specific > routines. So far I've looked at the reference manuals provided by VxWorks > and the list doesn't seem to be complete. Any thoughts? > > Thanx for any help. > > Lee Cao > > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Can't connect to Windows using sockets?? Date: Wed, 2 Jan 2002 19:37:13 +0000 (UTC) From: don@mars.dgrc.crc.ca (Donald McLachlan) Organization: The unconfigured xvnews people Message-ID: References: <5YIY7.21096$xM1.863790177@newssvr14.news.prodigy.com> Reply-To: don@mars.dgrc.crc.ca Can you ping from vxworks to the win2k server? Can you use something like tcpdump or windump or ethereal to see what is happening to the packets on the wire? What errors/error messages are you getting? Does "SERVER_NUM" == "nPort", and maybe that should be "htons(nPort)". Don In article <5YIY7.21096$xM1.863790177@newssvr14.news.prodigy.com>, "agin" writes: > I am unsuccessfully trying to connect from a vxWorks client to > a win2k server using sockets. > > I can connect from win2k to vxWorks, vxWorks to vxWorks, win2k to win2k but > I am unable to connect from vxWorks to win2k. > > I am using winsock2. > > Is there something I am missing? > > > Here is snipet of the vxWorks client code: > > clientSock = socket (AF_INET, SOCK_STREAM, 0); > > serverAddr.sin_family = AF_INET; > serverAddr.sin_port = htons(SERVER_NUM); > if (connect (clientSock, (struct sockaddr *)&serverAddr, sizeof > (serverAddr)) == ERROR) > { > printf ("Connect failed:\n"); > printErrno(errnoGet()); > > close (clientSock); > return (ERROR); > } > > > Here is a snipet of the Windows Server code: > u_long nInterfaceAddr = inet_addr(pcAddress); > if (nInterfaceAddr != INADDR_NONE) { > SOCKET sd = socket(AF_INET, SOCK_STREAM, 0); > if (sd != INVALID_SOCKET) { > sockaddr_in sinInterface; > sinInterface.sin_family = AF_INET; > sinInterface.sin_addr.s_addr = nInterfaceAddr; > sinInterface.sin_port = nPort; > if (bind(sd, (sockaddr*)&sinInterface, > sizeof(sockaddr_in)) != SOCKET_ERROR) { > listen(sd, 1); > return sd; > } > } > } > > accept(sd, (sockaddr*)&sinRemote, sizeof(sinRemote)); > > > art > > - -- Donald McLachlan E-mail Donald.McLachlan@crc.ca Communications Research Centre / RNS Tel (613) 998-2845 3701 Carling Ave., Fax (613) 998-9648 Ottawa, Ontario K2H 8S2 Canada --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Debugging with Visual SlickEdit Date: Wed, 2 Jan 2002 20:40:13 +0100 From: "Martin Raabe" Organization: T-Online Message-ID: References: Hello Spaceman, no you can't. But using SNiFF+ for Tornado you can in version 4.0.2! Ciao Martin Raabe "spaceman" schrieb im Newsbeitrag news:Dj0Y7.19515$Oj4.4019447993@newssvr21.news.prodigy.com... > Anyone been able to configure Visual SlickEdit as a graphical debugger for > Tornado? > > > Jim > > --------------------------- Newsgroups: comp.os.vxworks Subject: Problem while writing at TFFS (DiskOnChip) Date: Wed, 02 Jan 2002 22:05:10 +0100 From: Tobias Schneider Organization: T-Online Message-ID: <3C337606.74C523A8@web.de> Hi, I have a PC104 board with a DiskOnChip TFFS. Sometimes when I write data to the TFFS the writing process failed(e.g. when using a FTP program hang up) and looking on the TFFS after a reboot a corrupt file is found. The capacity of the DiskOnChip is 4MB and the free space only 1MB. Using the same hardware with the same kernel (only the TFFS driver is replaced by an IDE driver) and a IDE harddisk no problem appears. Does anyone have an idea ? Thanxx Tobias --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Dumb question about the target-resident shell Date: 2 Jan 2002 13:18:52 -0800 From: john_94501@yahoo.com (John) Organization: http://groups.google.com/ Message-ID: <488e459a.0201021318.bbf96f6@posting.google.com> References: <635472238d6165ac734f3c610a2945d0.56035@mygate.mailgate.org> Hello, "CIND" wrote in message news:... > Replace your declaration statement in your header with: > extern "C" int fdl(int x); // Your return type and args > > This will disable name mangling of the function and make it accessable to > the shell's lookup function. The technique will work perfectly, but the explanation is not totally accurate. C++ functions are accessible by the shell's lookup function, the problem is that the shell is a C interpreter and doesn't understand that the types of the parameters need to be included in the name. If you turn off the demangling in the shell (see cplusDemanglerSet() for info on this), then you will get a name with all the mangling visible. That name can be used directly from the shell. Of course, declaring the function as extern "C" is nicer if that's acceptable. HTH, John... > > "Trevor Ward" wrote in message > news:635472238d6165ac734f3c610a2945d0.56035@mygate.mailgate.org... > > Since there aren't any stupid questions (only stupid answers), I thought > I'd > > provide the exception that proves the rule... > > > > I've built a bootable image using Tornado 2, extending the default kernel > with > > some extra .o object modules of my own. The entry point into my (C++) > > application is a function called 'fdl' that takes an int parameter. I want > to > > execute 'fdl(1)' from a startup script whenever the target boots. I have a > > target resident symbol table, so I can perform 'lkup "fdl"' on the target > shell > > and it gives me an entry like: > > > > fdl(int) 0x0011e3b0 text > > > > If I type 'fdl(1)' I get the error 'undefined symbol: fdl', but if I type > > '0x0011e3b0(1)' then it runs my application as desired. The problem is > that the > > address of fdl, i.e. 0x0011e3b0, moves around as I change to code and I > don't > > want to keep having to change the startup script every time I recompile. > I'm > > sure there must be a really easy way of resolving this problem, but I > can't > > find it... > > > > > > -- > > Posted via Mailgate.ORG Server - http://www.Mailgate.ORG --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Can't connect to Windows using sockets?? Date: Wed, 02 Jan 2002 23:51:13 GMT From: "agin" Organization: Prodigy Internet http://www.prodigy.com Message-ID: References: <5YIY7.21096$xM1.863790177@newssvr14.news.prodigy.com> Hey Donald: Thanks for your reply. I can ping the Win2k server from the vxworks. The client gives me a refused connection error. I am using 4242 as my port number for simplicity. art "Donald McLachlan" wrote in message news:a0vnh9$ec7$1@crc-news.crc.ca... > Can you ping from vxworks to the win2k server? > Can you use something like tcpdump or windump or ethereal to see what is happening > to the packets on the wire? > What errors/error messages are you getting? > > Does "SERVER_NUM" == "nPort", and maybe that should be "htons(nPort)". > > Don > > In article <5YIY7.21096$xM1.863790177@newssvr14.news.prodigy.com>, "agin" writes: > > I am unsuccessfully trying to connect from a vxWorks client to > > a win2k server using sockets. > > > > I can connect from win2k to vxWorks, vxWorks to vxWorks, win2k to win2k but > > I am unable to connect from vxWorks to win2k. > > > > I am using winsock2. > > > > Is there something I am missing? > > > > > > Here is snipet of the vxWorks client code: > > > > clientSock = socket (AF_INET, SOCK_STREAM, 0); > > > > serverAddr.sin_family = AF_INET; > > serverAddr.sin_port = htons(SERVER_NUM); > > if (connect (clientSock, (struct sockaddr *)&serverAddr, sizeof > > (serverAddr)) == ERROR) > > { > > printf ("Connect failed:\n"); > > printErrno(errnoGet()); > > > > close (clientSock); > > return (ERROR); > > } > > > > > > Here is a snipet of the Windows Server code: > > u_long nInterfaceAddr = inet_addr(pcAddress); > > if (nInterfaceAddr != INADDR_NONE) { > > SOCKET sd = socket(AF_INET, SOCK_STREAM, 0); > > if (sd != INVALID_SOCKET) { > > sockaddr_in sinInterface; > > sinInterface.sin_family = AF_INET; > > sinInterface.sin_addr.s_addr = nInterfaceAddr; > > sinInterface.sin_port = nPort; > > if (bind(sd, (sockaddr*)&sinInterface, > > sizeof(sockaddr_in)) != SOCKET_ERROR) { > > listen(sd, 1); > > return sd; > > } > > } > > } > > > > accept(sd, (sockaddr*)&sinRemote, sizeof(sinRemote)); > > > > > > art > > > > > > > > -- > Donald McLachlan E-mail Donald.McLachlan@crc.ca > Communications Research Centre / RNS Tel (613) 998-2845 > 3701 Carling Ave., Fax (613) 998-9648 > Ottawa, Ontario > K2H 8S2 > Canada > > --------------------------- Newsgroups: comp.os.vxworks Subject: ldModule versus ld < Date: Thu, 03 Jan 2002 01:45:24 GMT From: "Kelly Hornsby" Organization: Excite@Home - The Leader in Broadband http://home.com/faster Message-ID: hello all I have a PowerPC 750 application that is not overly complex, but it does use C++ and many VxWorks system calls. When I download the program from the WindShell via the download icon and execute it from the Shell, it performs flawlessly. But then: I can take the exact same program (ELF image) and dump it verbatum into flash, and then perform a loadModule from a memdev file so that it is loaded into memory. I can see the symbols I am interested in. If then perform a sysmFindByName, and pick up the start address and start the program that way, the program runs for a while, and then halts. I'm looking for suggestions on the differences between an embedded-loaded application, and a WindShell loaded application. I'm out of ideas here. Any tips, hints, or other comments are welcome. thx --------------------------- Newsgroups: comp.os.vxworks Subject: how do using for SMC2... Date: Thu, 3 Jan 2002 16:54:38 +0900 From: "it2xpert" Organization: ONSE Telecom Message-ID: Hi everybody ! I desinged the MPC850SR board.. I want to use two SCCx(2,3) for ethernet mode... When I used two SCC for ethernet mode, scc3(ethernet mode )'s paramete RAM area is overlaied to the SMC1(console).... I want to use SMC2 for Console.... how do replace SMC1 to SMC2 for console.. this is myself done... - ---------------------------------------------------- config.h - --------------------- #undef CONSOLE_TTY #define CONSOLE_TTY 1 /* console channel */ ( replace 0 to 1) #define N_SIO_CHANNELS 2 /* No. serial I/O channels */ sysSerial.c - -------------------- void sysSerialHwInit (void) { int i; /* an index */ /* intialize the chips device descriptors */ for (i = 1; i < chanNum; i++) { UINT32 regBase; /* BRGCLK freq (Hz) */ ppc860Chan [i].clockRate = BRGCLK_FREQ; /* IMMR reg has base adr */ ppc860Chan [i].regBase = vxImmrGet(); regBase = ppc860Chan [i].regBase; /* use BRG1 for channel 1 and BRG2 for channel 2 */ ppc860Chan [i].bgrNum = (i + 1); /* SMC wired for rs232 */ ppc860Chan [i].uart.smcNum = (i + 1); /* init the number of TBDs */ ppc860Chan [i].uart.txBdNum = ppc860SmcParms[i].smcTbdNum; /* init the number of RBDs */ ppc860Chan [i].uart.rxBdNum = ppc860SmcParms[i].smcRbdNum; /* transmit BD base adrs */ ppc860Chan [i].uart.txBdBase = (SMC_BUF *) (MPC860_REGB_OFFSET + ppc860SmcParms[i].smcTbdOff); /* receive BD base adrs */ ppc860Chan [i].uart.rxBdBase = (SMC_BUF *) (MPC860_REGB_OFFSET + ppc860SmcParms[i].smcRbdOff); /* tx buf base */ ppc860Chan [i].uart.txBufBase = (u_char *) (MPC860_DPRAM_BASE (regBase) + ppc860SmcParms[i].smcTxBufOff); /* rx buf base */ ppc860Chan [i].uart.rxBufBase = (u_char *) (MPC860_DPRAM_BASE (regBase) + ppc860SmcParms[i].smcRxBufOff); /* transmit buffer size */ ppc860Chan [i].uart.txBufSize = ppc860SmcParms[i].smcTxBufSz; /* DPRAM addr of SMC1 params */ ppc860Chan [i].uart.pSmc = (SMC *) ((UINT32) PPC860_DPR_SMC1 (MPC860_DPRAM_BASE (regBase)) + (i * 0x100)); /* SMCMR1 for SMC1 */ ppc860Chan [i].uart.pSmcReg = (SMC_REG *) ((UINT32) MPC860_SMCMR1 (regBase) + (i * 0x10)); /* Mask interrupts */ ppc860Chan [i].uart.pSmcReg->smcm = 0; ppc860Chan [i].pBaud = (UINT32 *) ((UINT32) MPC860_BRGC1 (regBase) + (i * 4)); ppc860Chan [i].channelMode = 0; /* select RS232 pins */ if ( i == 0) *MPC860_PBPAR(regBase) |= 0xC0 ; else { *MPC860_PAPAR( regBase ) |= 0xC0 ; /* select RS232 pins for smc2 */ *MPC860_PADIR( regBase ) &= ~0xC0 ; *MPC860_PAODR( regBase ) &= ~0xC0 ; } /* set it to normal operations */ *MPC860_SDCR(regBase) = SDCR_RAID_BR5; /* reset the chip */ ppc860DevInit(&(ppc860Chan [i])); } } void sysSerialHwInit2 (void) { int i; /* an index */ /* connect serial interrupts */ for (i = 1; i < chanNum; i++) { switch (i) { case 0: (void) intConnect (IV_SMC1, (VOIDFUNCPTR) ppc860Int, (int) &ppc860Chan [i]); break; case 1: (void) intConnect (IV_SMC2_PIP, (VOIDFUNCPTR) ppc860Int, (int) &ppc860Chan [i]); break; default: return; } *CIMR(vxImmrGet()) |= (CIMR_SMC1 >> i); } } bootConfig.c - ------------- #else /* !INCLUDE_TYCODRV_5_2 */ #ifdef INCLUDE_TTY_DEV if (NUM_TTY > 0) { ttyDrv(); /* install console driver */ for (ix = 1; ix < NUM_TTY; ix++) /* create serial devices */ { #if (defined(INCLUDE_WDB) && (WDB_COMM_TYPE == WDB_COMM_SERIAL)) if (ix == WDB_TTY_CHANNEL) /* don't use WDBs channel */ continue; #endif sprintf (tyName, "%s%d", "/tyCo/", ix); (void) ttyDevCreate (tyName, sysSerialChanGet(ix), 512, 512); if (ix == CONSOLE_TTY) /* init the tty console */ { strcpy (consoleName, tyName); consoleFd = open (consoleName, O_RDWR, 0); (void) ioctl (consoleFd, FIOBAUDRATE, CONSOLE_BAUD_RATE); (void) ioctl (consoleFd, FIOSETOPTIONS, OPT_ECHO | OPT_CRMOD | OPT_TANDEM | OPT_7_BIT); } } } #endif /* INCLUDE_TTY_DEV */ - ------- result(console message) - ------- CPU: KCM - PowerPC 850sr Version: 5.4 BSP version: 1.2/1 Creation date: Jan 3 2002, 15:49:47 Press any key to stop auto-boot... 1 [VxWorks KCM]: [VxWorks KCM]: [VxWorks KCM]: [VxWorks KCM]: [VxWorks KCM]: [VxWorks KCM]: [VxWorks KCM]: [VxWorks KCM]: [VxWorks KCM]: [VxWorks KCM why? what's the correct method ?? regards. jwyu --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Can't connect to Windows using sockets?? Date: 2 Jan 2002 23:57:30 -0800 From: pb94087@yahoo.com (P B) Organization: http://groups.google.com/ Message-ID: References: <5YIY7.21096$xM1.863790177@newssvr14.news.prodigy.com> > > > Here is snipet of the vxWorks client code: > > > > > > clientSock = socket (AF_INET, SOCK_STREAM, 0); > > > > > > serverAddr.sin_family = AF_INET; Are you also passing the correct IP address of the server that you are connecting to? > > > serverAddr.sin_port = htons(SERVER_NUM); > > > if (connect (clientSock, (struct sockaddr *)&serverAddr, sizeof > > > (serverAddr)) == ERROR) > > > { > > > printf ("Connect failed:\n"); > > > printErrno(errnoGet()); > > > > > > close (clientSock); > > > return (ERROR); > > > } > > > > > > > > > Here is a snipet of the Windows Server code: > > > u_long nInterfaceAddr = inet_addr(pcAddress); > > > if (nInterfaceAddr != INADDR_NONE) { > > > SOCKET sd = socket(AF_INET, SOCK_STREAM, 0); > > > if (sd != INVALID_SOCKET) { > > > sockaddr_in sinInterface; > > > sinInterface.sin_family = AF_INET; > > > sinInterface.sin_addr.s_addr = nInterfaceAddr; > > > sinInterface.sin_port = nPort; > > > if (bind(sd, (sockaddr*)&sinInterface, > > > sizeof(sockaddr_in)) != SOCKET_ERROR) { > > > listen(sd, 1); perhaps you can try changing it to listen(sd, 5) > > > return sd; > > > } > > > } > > > } > > > do you see the server on the w2K system listening? u can check using netstat. did you try connecting using telnet (telnet IP_ADDR PORT_NUM)? --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Symbol table errors compiling an image when using STL Date: Thu, 3 Jan 2002 10:30:06 -0000 From: "Dave Korn" Organization: Lumber Cartel (tinlc) Members #2234-2237 (owing to browser refresh) Message-ID: References: <3c29bbec$1@news.barak.net.il> Arnon W wrote in message <3c29bbec$1@news.barak.net.il>... >Hi, > >I'm new to this environment so it might be a minor issue I'm having problems >with. >(On the other hand - I couldn't find earlyer posts in this specific matter:) >When I try to integrate it to the image, I get vicious responses at this >point: >ccppc -c -fdollars-in-identifiers -O3 -mstrict-align -nostdinc -DRW_MULTI_T H >READ -D_REENTR >ANT -fvolatile -fno-builtin -fno-for-scope -DCPU=PPC604 -DPRJ_BUILD -DHAL_L I >B >...paths... symTbl.c -o symTbl.o >symTbl.c:2013: syntax error before `.120' >symTbl.c:2014: syntax error before `.139' >symTbl.c:2015: syntax error before `.148' >symTbl.c:8448: `__sn_t24__default_alloc_template2b1i0$__node_allocator_lock' >undeclared he >re (not in a function) >symTbl.c:8448: syntax error before `.120' >symTbl.c:8449: initializer element for `standTbl[1997].value' is not >constant >symTbl.c:8449: warning: initialization from incompatible pointer type >symTbl.c:8449: `__sn_t24__default_alloc_template2b1i0$__node_allocator_lock' >undeclared he >re (not in a function) >symTbl.c:8449: syntax error before `.139' >symTbl.c:8450: initializer element for `standTbl[1998].name' is not constant >symTbl.c:8450: syntax error before string constant >When I took out STL related stuff (I was using ) It compiled well. >Is there a cure? I'm just passing on something I read somewhere else here, STL is really not my field, but ISTR someone saying that to use STL you should only #include - i.e. you use that header and it includes all the other stl headers like and for you. You might like to try this approach and see if it fixes your problem. DaveK - -- Burn your ID card! http://www.optional-identity.org.uk/ Help support the campaign, copy this into your .sig! Proud Member of the Exclusive "I have been plonked by Davee because he thinks I'm interesting" List Member # --------------------------- Newsgroups: comp.os.vxworks Subject: Re: what is "makes pointer from integer without a cast"? Date: Thu, 3 Jan 2002 10:42:44 -0000 From: "Dave Korn" Organization: Lumber Cartel (tinlc) Members #2234-2237 (owing to browser refresh) Message-ID: References: <97c3acac.0112162225.a7dbc13@posting.google.com> <90019cec.0112170439.42b8227a@posting.google.com> Andreas Bachmeier wrote in message <90019cec.0112170439.42b8227a@posting.google.com>... >mimilijie@21cn.com (lijie) wrote in message news:<97c3acac.0112162225.a7dbc13@posting.google.com>... >> during the developing time, i often meet with "makes pointer from >> integer without a cast". > >On the left side, there is a pointer, on the right side an integer. >So You have forgotten the adressoperator. >pPointer = &Integer; > >Andreas It is more likely that he has called a function that returns a pointer, and is assigning the result to a pointer variable, but has not #included the header file that declares the function. If you don't declare a function, the compiler assumes it takes any amount of arguments and returns an int. So the compiler complains about both using the function without declaring it, and also about assigning an integer to a pointer variable. DaveK - -- Burn your ID card! http://www.optional-identity.org.uk/ Help support the campaign, copy this into your .sig! Proud Member of the Exclusive "I have been plonked by Davee because he thinks I'm interesting" List Member # --------------------------- Newsgroups: comp.os.vxworks Subject: Re: malloc in interruption Date: Thu, 3 Jan 2002 11:12:21 -0000 From: "Dave Korn" Organization: Lumber Cartel (tinlc) Members #2234-2237 (owing to browser refresh) Message-ID: References: <3C251DFF.E576635F@home.com> <7pva2uo42dcrgnm2nrab9u7svf9bqttjai@4ax.com> <3C262BE6.B98ED880@home.com> <488e459a.0112231304.48fbd31b@posting.google.com> <3C267804.9BDF6C9D@gv.net> <3C2770B8.B9C@despam.autobahn.mb.ca> <3C293EEE.707CE5CC@home.com> Gwen & Emory Stagmer wrote in message <3C293EEE.707CE5CC@home.com>... >I agree with all of this. Our use of C++ is generally very limited. >This is for spacecraft, so we're generally VERY conservative. I'm only >basically employing C++ for an application framework, and even then it's use >is limited. We mostly use struct's and unions (not classes) [SNIP] You needn't be shy of classes if you're careful: as long as you don't use any virtual functions or virtual base classes there's no overhead in accessing either data or function members, and the data hiding / access control features are very nice to have. The main issue to watch out for in embedded C++ is to make sure there aren't unneccessary temporaries being constructed and destructed all over the place, and to be sure and pass objects around by reference rather than by value so they aren't repeatedly copied. DaveK - -- Burn your ID card! http://www.optional-identity.org.uk/ Help support the campaign, copy this into your .sig! Proud Member of the Exclusive "I have been plonked by Davee because he thinks I'm interesting" List Member # --------------------------- Newsgroups: comp.os.vxworks Subject: RAW Socket support Date: 3 Jan 2002 03:44:32 -0800 From: rashanmu@npd.hcltech.com (Ramesh) Organization: http://groups.google.com/ Message-ID: <15a87607.0201030344.4805f471@posting.google.com> Hi, I want to receive packets of my protocol number ( e.g 140 which i used ) from my VxSIM. IS that possible to do it. Because the Target server waits for IP pkts with this protocol number in IP protocol field, but not receiving the packet ( am sending from VxSIM on another host through the network - remember i can able to send and receive TCP/UDP packets between these two VxSims on n/w ). So when i press CTRL+ Break the target server is crashing. But I can able to build my own IP packet using IP_HDRINCL option, with Protocol number 140 from the VxSIM and sending it to a linux machine. I can able to receive the packets from the Linux machine, generated by the VxSIM. Why the vice versa of above or Getting the raw IP packets generated by another VxSIM ( with proto number 140 ) is not reaching the VxSIM. Is that not possible to receive packets using raw sockets in VxSIM. Kindly Help me. THanks, Ramesh --------------------------- End of New-News digest ********************** From vxwexplo-errs@csg.lbl.gov Fri Jan 4 01:54:08 2002 From: ravi kumar Date: Fri Jan 4 01:54:10 PST 2002 Subject: how to add hook to routeing table hi, i want to maintain my own hook to make an entry into my DB when VxWorks adds route to its routing table. as for as i know we can use enhanced routing table API , but i think this will come only with OSPF product, any other alternative is there or we have use OSPF only to do this. i would like hear some thing about this. regards ravi __________________________________________________ Do You Yahoo!? Send your FREE holiday greetings online! http://greetings.yahoo.com From vxwexplo-errs@csg.lbl.gov Fri Jan 4 04:03:13 2002 From: Vxworks Exploder Date: Fri Jan 4 04:03:15 PST 2002 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Fri Jan 4 04:03:09 PST 2002 Subject: Re: GCC 2.95.4 Subject: Re: problem with 82559er driver Subject: stuck in idleLoop Subject: envoy's snmpd Task crash problem Subject: Re: VxWorks Routines List Subject: Re: VxWorks Routines List Subject: Re: stuck in idleLoop Subject: Re: VxWorks Routines List Subject: vme reset Subject: Re: VxWorks Routines List Subject: Re: smart card reader Subject: SIMNT vxSim Full Simulator on T2CP4 with TMS Subject: I/O ports on PowerPC Subject: Could You Give me a file --> libPPC603gnugcc.a Subject: Interrupt for PPC860 - Download Error occurs for sysBusIntGen function Subject: S_errno_EADDRNOTAVAIL Subject: Slow file accesses with VxSim Subject: Re: Interrupt for PPC860 - Download Error occurs for sysBusIntGen function ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: GCC 2.95.4 Date: Thu, 3 Jan 2002 11:43:24 -0000 From: "Dave Korn" Organization: Lumber Cartel (tinlc) Members #2234-2237 (owing to browser refresh) Message-ID: References: <679c25ff.0112101503.165a710d@posting.google.com> David Ward wrote in message <679c25ff.0112101503.165a710d@posting.google.com>... >I built a GCC 2.95.4 cross tool chain for PPC under Cygwin, following >Dave K's web site for the most part. Very nice - informative and >colorful ! I installed it in a Tornado II tree. But I have some lib >issues when linking my target - these will no doubt look familiar to >some folks - Hi David, sorry it's taken me ages to get round to replying this post, hope it's not too late to be useful to you. Anyway: >undefined ref to __start_cp_handler, __eh_malloc, etc... > >GCC's new mechanism for c++ exception handling. There were lot's of >postings about this back in Feb. The trail goes cold before I find >the best way to resolve it. __start_cp_handler is defined in my new >libgcc - > >/Tornado/host/x86-win32/lib/gcc-lib/powerpc-wrs-vxworks/2.95.4/libgcc.a > >I can get it to link in this by removing -nostdlib or adding -lgcc, >but that lib appears to be in conflict with the old VxWorks run-time >lib - > >/Tornado/target/lib/libPPCEC603gnuvx.a > >which I also need. Do I rebuild this ? The source appears to be >here... or can I just yank out some objects ? If you aren't using C++ exceptions, you can always use -fno-exceptions! Alternatively, try the advice on my website at the 'Info' page, under "What do I need to know to use these compilers?", in the last paragraph beginning "The new compiler comes with a more up-to-date and complete implementation of the C++ STL. To use this, you'll need to configure the kernel to exclude...". Note that you'll want to replace 'libPPC860gnuvx.a' with 'libPPCEC603gnuvx.a" in the arppc command lines. >BTW, here's what my link looks like before playing around with >-nostdlib, -lgcc - >ccppc -nostdlib -r -Wl,-X -o [SNIP] Try and use c++ppc when compiling C++ files if possible as it tend to solve link problems. DaveK - -- Burn your ID card! http://www.optional-identity.org.uk/ Help support the campaign, copy this into your .sig! Proud Member of the Exclusive "I have been plonked by Davee because he thinks I'm interesting" List Member # --------------------------- Newsgroups: comp.os.vxworks Subject: Re: problem with 82559er driver Date: Thu, 3 Jan 2002 15:56:58 +0100 From: "Werner Schiendl" Message-ID: <3c3470da$0$34592$5039e797@newsreader01.highway.telekom.at> References: <3C20C4B3.A9463A@nlr.nl> <488e459a.0112192051.f43b438@posting.google.com> <3c220092$1@brateggebdc5.br-automation.co.at> <3C331C33.7BE19F8B@nlr.nl> "Michiel Meekes" schrieb im Newsbeitrag news:3C331C33.7BE19F8B@nlr.nl... ... > > After looking through some old discussions about this subject I found a > suggestion that > resulted in a solution for the above problem. > The suggestion was to modify the BIOS setting "PNP OS Installed" from Yes to No. > > After this the Intel 82559er MUX driver (using END) worked OK. > > Thanks to all who helped in this matter. > Michiel > The difference this setting usually makes is, that it does not enable Plug&Play devices which are known to be unnecessary at boot time, like sound boards, USB controllers, etc. Sometimes, no resources - like interrupts and memory are assigned either. The OS must than care to assign proper resources and enable the board. It should be possible to do this configuration work in a VxWorks driver if necessary. hth Werner --------------------------- Newsgroups: comp.os.vxworks Subject: stuck in idleLoop Date: 3 Jan 2002 09:29:26 -0800 From: mynamvs@nsc.network.com (Venkat Mynampati) Organization: http://groups.google.com/ Message-ID: <65b2abfc.0201030929.4280fec5@posting.google.com> Hi, i noticed that, my sa110 processor got stuck in idleLoop at: b5a84: e51f247c ldr r2, b5610 <.text> 000b5a88 : b5a88: e5920000 ldr r0, [r2] b5a8c: e3300000 teq r0, #0 b5a90: 1afffffc bne b5a88 ^^^^^^^^^^^^^^^^^^^^^ b5a94: e3a00000 mov r0, #0 b5a98: e5810000 str r0, [r1] b5610: 001375e8 andeqs r7, r3, r8, ror #b Now, [r2] has a value of 1, i.e the address location 0x001375e8 contained a value of 1( this i know because i had means of looking into various addresses), which, i think, made my processor to sit in a tight loop, i.e the value of r0 is always 1 and hence the code always(forever) is spinning on idleLoop. Now,what's happening here? what this idleLoop do? Who would set this bit to zero( i thought any task that is ready-to-run will do this...pls correct me)? and what could be the reasons why this bit has not been set? Has anyone seen this issue before? Where can i learn more about these issues? Thanks and Regards, Venkat --------------------------- Newsgroups: comp.os.vxworks Subject: envoy's snmpd Task crash problem Date: Thu, 3 Jan 2002 20:12:03 +0200 From: "Ofer Goren" Organization: Verio Message-ID: Win2k Sp2 Tornado 2.02 for PPC604 Diab 4.4b Envoy 9.2 - ------------------------ Hi. I'm using MG-SOFT mib browser, version 7.0.0 I compiled a clean vxWork's bootable application for my wrScout BSP. I added the snmp library using the EXTRA_MODULES to my application. Everytime I use SNMP-GET to retrieve something from my ifTable, the snmpd task crashes. Using my debugger I identified that it crashes when returning from A_DecodeOctetStringData(). it crashes because the PC register points to 0x10, and the LR pointer points to 0x0. However, an assembler instruction just before the return, all is OK. It is as if the stack is somehow overrun by someone. Here is the output of "ti "tSnmpd"" called from the shell: NAME ENTRY TID PRI STATUS PC SP ERRNO DELAY - ---------- ------------ -------- --- ---------- -------- -------- ------- -- - --- tSnmpd 106318 1de5050 150 SUSPEND 10 1de4738 0 0 stack: base 0x1de5050 end 0x1dde050 size 28664 high 3872 margin 24792 options: 0x4 VX_DEALLOC_STACK r0 = 0 sp = 1de4738 r2 = 0 r3 = 1ffc898 r4 = 1ffc778 r5 = 7 r6 = 1ffc730 r7 = 1 r8 = 4 r9 = 1e8d38 r10 = 4 r11 = 1de4518 r12 = 0 r13 = 0 r14 = 0 r15 = 0 r16 = 0 r17 = 0 r18 = 33e6 r19 = 0 r20 = 0 r21 = 0 r22 = 0 r23 = 0 r24 = 1ffc778 r25 = 5 r26 = 1ffc90c r27 = 0 r28 = 33ed r29 = 0 r30 = 32ce r31 = 0 msr = 8b022 lr = 0 ctr = 0 pc = 10 cr = 42000088 xer = 0 program Exception current instruction address: 0x000000value = 10 Machine Status Register: 0x00008b022 Condition Register: 0x = 042000088x 0 - -> I'm also attaching my wind.cfg file. Can u see what is the problem? Thanks, - -- Ofer Goren Iamba Technologies begin 666 wind.cfg M.SL["CL[.R!%>&%M<&QE(&-O;F9I9W5R871I;VX@9FEL92!F;W(@8G5I;&1I M;F<@16YV;WD@;VX@=GA7;W)K2]S=&0M;6EB M%=O2!O;F4@;V8@=&AE Organization: Fermi National Accelerator Laboratory Message-ID: <3C34A175.D28A57CF@fnal.gov> References: Hi, What exactly are you looking for? > My question is if > there is a master list of VxWorks routines that I can reference to see if a > routine used is supplied by VxWorks or not, including hardware-specific > routines. If you are just looking for names, build a clean kernel (without your application linked) and run it. Do - -> lkup "" This will give you list of _all_ symbols. That way you will have names of all WRS provided symbols. HTH, Dinker --------------------------- Newsgroups: comp.os.vxworks Subject: Re: VxWorks Routines List Date: Thu, 3 Jan 2002 20:20:42 GMT From: Lee DeRaud Organization: None to speak of Message-ID: <67f93ucpa1rkps39e3vlq8263gg71udrnc@4ax.com> References: <3C34A175.D28A57CF@fnal.gov> Sender: nntp@news.boeing.com (Boeing NNTP News Access) On Thu, 03 Jan 2002 12:22:45 -0600, Dinker Charak wrote: >Hi, > >What exactly are you looking for? > >> My question is if >> there is a master list of VxWorks routines that I can reference to see if a >> routine used is supplied by VxWorks or not, including hardware-specific >> routines. > >If you are just looking for names, build a clean kernel (without your >application linked) and run it. Do > >-> lkup "" > >This will give you list of _all_ symbols. That way you will have names >of all WRS provided symbols. Can't you just save the 'symTbl.c' file from that build? Lee --------------------------- Newsgroups: comp.os.vxworks Subject: Re: stuck in idleLoop Date: 3 Jan 2002 13:51:04 -0800 From: john_94501@yahoo.com (John) Organization: http://groups.google.com/ Message-ID: <488e459a.0201031351.2d10469a@posting.google.com> References: <65b2abfc.0201030929.4280fec5@posting.google.com> Hello, mynamvs@nsc.network.com (Venkat Mynampati) wrote in message news:<65b2abfc.0201030929.4280fec5@posting.google.com>... > Hi, > i noticed that, my sa110 processor got stuck in > idleLoop at: > > b5a84: e51f247c ldr r2, b5610 <.text> > 000b5a88 : > b5a88: e5920000 ldr r0, [r2] > b5a8c: e3300000 teq r0, #0 > b5a90: 1afffffc bne b5a88 > ^^^^^^^^^^^^^^^^^^^^^ > b5a94: e3a00000 mov r0, #0 > b5a98: e5810000 str r0, [r1] > > b5610: 001375e8 andeqs r7, r3, r8, ror #b > > Now, [r2] has a value of 1, i.e the address location > 0x001375e8 contained a value of 1( this i know because > i had means of looking into various addresses), which, > i think, made my processor to sit in a tight loop, i.e > the value of r0 is always 1 and hence the code always(forever) > is spinning on idleLoop. > > Now,what's happening here? what this idleLoop do? Spins all the time there is no task to run; what else would you expect an idle loop to do? The test is against the kernel's work queue being empty. Why? Well, because the only thing that can get you out of the idle loop is an interrupt (think about it for a while if you don't get it immediately). Now, for VxWorks 5.x at least, when in the idle loop the kernel is also in its protected state. That means that when an interrupt occurs any kernel calls, such as semGive(), that the ISR makes are deferred (added to the work queue). When the ISR exits, the idle loop will resume execution, detect that there is work in the work queue and break out of the loop to process it. If that work causes a task to become ready to run, then the scheduler will be invoked and the now ready task will execute until it blocks again. I'm not sure how you are looking at the system though since all the standard tools will cause either the WDB agent or the target shell to be running so you should never see the idle loop. If you are using JTAG or ICE type tools, that might be the cause of your problem, especially if the tool is "handling" all the interrupts in the system. Without at least one interrupt getting through to the kernel, if the system ever becomes idle you will never get back out. HTH, John... > Who would > set this bit to zero( i thought any task that is ready-to-run > will do this...pls correct me)? and what could be the reasons > why this bit has not been set? Has anyone seen this issue before? > Where can i learn more about these issues? > > Thanks and Regards, > Venkat --------------------------- Newsgroups: comp.os.vxworks Subject: Re: VxWorks Routines List Date: Thu, 3 Jan 2002 13:49:20 -0500 From: "Jiandong Wang" Organization: Nortel Message-ID: References: Hi, You can use "lkup" in WindShell. e.g. - -> lkup "pppHookAdd" pppHookAdd 0x0015e8fc text (vxWorks) value = 0 = 0x0 - ->lkup "" ...... (so many routines) Hope it's helpful. - -------------------- "Lee Cao" wrote in message news:a0vfgr$ko6$1@lore.csc.com... > Hi! > > I recently was hired to perform some maintenance and analysis on a VxWorks > application written for Pentium single board computers. My question is if > there is a master list of VxWorks routines that I can reference to see if a > routine used is supplied by VxWorks or not, including hardware-specific > routines. So far I've looked at the reference manuals provided by VxWorks > and the list doesn't seem to be complete. Any thoughts? > > Thanx for any help. > > Lee Cao > > --------------------------- Newsgroups: comp.os.vxworks Subject: vme reset Date: 3 Jan 2002 14:50:21 -0800 From: timothy.j.iskander@lmco.com (Tim Iskander) Organization: http://groups.google.com/ Message-ID: <46e37459.0201031450.20374d8a@posting.google.com> It looks like when I power up our MVME2700 with the vx bootloader selected, it never asserts a VME reset. If I switch over to PPCBug, it does. I cant find anything in the vx docs/etc about it. Has anyone seen this or know anything about it? We have a custom board which needs to see VME RESET at power up. thanks /Tim --------------------------- Newsgroups: comp.os.vxworks Subject: Re: VxWorks Routines List Date: Thu, 03 Jan 2002 15:35:45 -0800 From: "Michael R. Kesti" Organization: MK Associates Message-ID: <3C34EAD1.20E30A2E@gv.net> References: <3C34A175.D28A57CF@fnal.gov> <67f93ucpa1rkps39e3vlq8263gg71udrnc@4ax.com> Reply-To: mkesti@gv.net Lee DeRaud wrote: >On Thu, 03 Jan 2002 12:22:45 -0600, Dinker Charak >wrote: > >>Hi, >> >>What exactly are you looking for? >> >>> My question is if >>> there is a master list of VxWorks routines that I can reference to see if a >>> routine used is supplied by VxWorks or not, including hardware-specific >>> routines. >> >>If you are just looking for names, build a clean kernel (without your >>application linked) and run it. Do >> >>-> lkup "" >> >>This will give you list of _all_ symbols. That way you will have names >>of all WRS provided symbols. > >Can't you just save the 'symTbl.c' file from that build? Using lkup and symTbl.c seem flawed to me, as they will tell you only the symbols available for the libraries that have been included in a particular build. One of vxWork's strengths is its configurability, and it is unlikely that all of the available libraries have been configured into any specific build. The better answer to the OP's question would therefore be that the BSP documentation and Programmer's Reference manual provide such information. What is more, these documents provide not only the names of the available functions, but the information one requires to use them, too. - -- ======================================================================== Michael Kesti | "And like, one and one don't make | two, one and one make one." mkesti@gv.net | - The Who, Bargain --------------------------- Newsgroups: comp.os.vxworks Subject: Re: smart card reader Date: 3 Jan 2002 17:25:19 -0800 From: chrismooremobile@yahoo.com (Chris Moore) Organization: http://groups.google.com/ Message-ID: References: "Zhao Yandong" wrote in message news:... > Our system will use a smart card reader. Now the hardware of smart card > reader has been designed out, and now we need to write the device driver for > it. > > But I don't know how to write the driver code. If I write the driver, which > files should I modify or add ? > > Could anyone give me some guide ? Thanks > > The target os is Vxworks5.4, host is win98, board is ARM Hi, I've designed a h/w + s/w iface to a smart card reader using the ATMEL micros before. Basically you're going to connect the smart card reader to one of the serial ports on your board I assume. So what you need to do is use the open/read/write etc etc for the serial port driver and you will want to design a layer of API's that sit above that driver. For example, (from mem) you need to authenticate yourself towards the card and sometimes for each file on the card so you would design some APIs that the application layer calls to do this, in the API you would place multiple calls towards the hardware interface using the the write(fd, buffer,nbytes) commands... assuming you have already opened the serial device etc. Thats the best way of designing that system. Chris --------------------------- Newsgroups: comp.os.vxworks Subject: SIMNT vxSim Full Simulator on T2CP4 with TMS Date: Fri, 04 Jan 2002 04:48:50 GMT From: "Ricky W. West" Organization: Road Runner - NC Message-ID: <3C353581.8F810156@nc.rr.com> I am working on an application that will be deployed on new hardware using T2CP4 with Tornado for Managed Switches. We have also purchased Tornado 2.0 for All Simulators. It turns out that the full vxSimulator for 2.0 will not work with T2CP4 with the TMS stuff added on top of the T2CP4. The network protocol tool kit has incompatibilities. Now SPR 30894 or some such address the fact that NPT is broken by T2CP4 and you can fix for the "real" target code, a PPC. But, the simulator does not work. Has any one tried to build and use the full vxSim on top of T2CP4 with TMS? My current problem is an exception in a task using the POSIX timer interface. I get a divide by zero exception just after the timer_settime() call. I checked the callback function pointer to verify that it was not zero. Breakpoints at the callback function entry are not being hit. Thanks, Ricky W. West --------------------------- Newsgroups: comp.os.vxworks Subject: I/O ports on PowerPC Date: 3 Jan 2002 23:22:35 -0800 From: arun4vxworks@indiatimes.com (Arun Prasad) Organization: http://groups.google.com/ Message-ID: Hi, I have a question on accessing I/O ports. I need to read/write from/to the network card registers on PowerPC platform. Can anyone tell me the functions provided by VxWorks to read/write Registers directly on PowerPC platforms? Earlier response would be of great help. Thanks in advance. Regards Arun --------------------------- Newsgroups: comp.os.vxworks Subject: Could You Give me a file --> libPPC603gnugcc.a Date: Fri, 4 Jan 2002 07:26:23 +0000 (UTC) From: "Dae-Hee, Choi" Organization: Your Company Message-ID: PLZ... Thank you in advance... --------------------------- Newsgroups: comp.os.vxworks Subject: Interrupt for PPC860 - Download Error occurs for sysBusIntGen function Date: 3 Jan 2002 23:53:01 -0800 From: manibala@covansys.com (B.Manivannan) Organization: http://groups.google.com/ Message-ID: <23de5426.0201032353.566fcb7a@posting.google.com> Good Day viewer, 1. I am using Tornado II and VxWorks 5.4 with the target of PPC 860. 2. I want to generate the sample program for interrupt. 3. I am getting the download error for the function sysBusIntGen. How can i solve that problem? Can you guide me? I am facing lots of problem with the download error. I would like to know, why the download error occurs. I have pasted the sample code here... /*My Sample code starts here */ /* includes */ #include "vxWorks.h" #include "intLib.h" #include "taskLib.h" #include "arch/ppc/ivPpc.h" #include "logLib.h" #include "sysLib.h" /*#include "C:\tornado\target\h\drv\multi\pccchip.h"*/ /* function prototypes */ void interruptHandler(int); void interruptCatcher(void); /*STATUS sysBusIntGen (int, int);*/ /* globals */ #define INTERRUPT_NUM 2 #define INTERRUPT_LEVEL 65 #define ITER1 40 #define LONG_TIME 1000000 #define PRIORITY 100 #define ONE_SECOND 100 void interruptGenerator(void) /* task to generate the SIGINT signal */ { int i, j, taskId, priority; STATUS taskAlive; if((taskId = taskSpawn("interruptCatcher",PRIORITY,0x100,20000,(FUNCPTR)interruptCatcher,0,0,0,0,0,0,0, 0,0,0)) == ERROR) logMsg("taskSpawn interruptCatcher failed\n",0,0,0,0,0,0); for (i=0; i < ITER1; i++) { taskDelay(ONE_SECOND);/* suspend interruptGenerator for one second */ /* check to see if interruptCatcher task is alive! */ if ((taskAlive = taskIdVerify(taskId)) == OK) { logMsg("++++++++++++++++++++++++++Interrupt generated\n",0,0,0,0,0,0); /* generate hardware interrupt 2 */ if((sysBusIntGen(INTERRUPT_NUM,INTERRUPT_LEVEL)) == ERROR) logMsg("Interrupt not generated\n",0,0,0,0,0,0); } else /* interruptCatcher is dead */ break; } logMsg("\n***************interruptGenerator Exited***************\n\n\n\n",0,0,0,0,0,0); } void interruptCatcher(void) /* task to handle the interrupt */ { int i, j; STATUS connected; /* connect the interrupt vector, INTERRUPT_LEVEL, to a specific interrupt handler routine ,interruptHandler, and pass an argument, i */ if((connected = intConnect(INUM_TO_IVEC(INTERRUPT_LEVEL),(VOIDFUNCPTR)interruptHandler,i)) == ERROR) logMsg("intConnect failed\n",0,0,0,0,0,0); for (i=0; i < ITER1; i++) { for (j=0; j < LONG_TIME; j++); logMsg("Normal processing in interruptCatcher\n",0,0,0,0,0,0); } logMsg("\n+++++++++++++++interruptCatcher Exited+++++++++++++++\n",0,0,0,0,0,0); } void interruptHandler(int arg) /* signal handler code */ { int i; logMsg("-------------------------------interrupt caught\n",0,0,0,0,0,0); for (i=0; i < 5; i++) logMsg("interrupt processing\n",0,0,0,0,0,0); } /*My Sample code ends here */ With Thanks and Regards, Manivannan.B. --------------------------- Newsgroups: comp.os.vxworks Subject: S_errno_EADDRNOTAVAIL Date: Fri, 4 Jan 2002 10:36:41 -0000 From: <@logica.com> Organization: Logica UK Limited Message-ID: <1010140603.575609@ernani.logica.co.uk> Hi all, Anyone know what the errorno 0x31 (S_errno_EADDRNOTAVAIL) means? Its the only return I get from the call Status = setsockopt( ClientSocket, IPPROTO_IP, IP_ADD_MEMBERSHIP, (char *)&MulticastReq, sizeof(MulticastReq) ); I can't find any definition of it in the manuals anywhere so I'm unsure what to do in order to fix it!!! thanks in advance, Mark. --------------------------- Newsgroups: comp.os.vxworks Subject: Slow file accesses with VxSim Date: Thu, 3 Jan 2002 21:29:32 +0000 (UTC) From: "martin brook" Organization: BT Openworld Message-ID: I'm using TSFS device as I could not get passfs to work. Any tips for speeding up accesses ? TTFN martin - --- Outgoing mail is certified Virus Free. Checked by AVG anti-virus system (http://www.grisoft.com). Version: 6.0.313 / Virus Database: 174 - Release Date: 1/2/02 --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Interrupt for PPC860 - Download Error occurs for sysBusIntGen function Date: 4 Jan 2002 03:09:11 -0800 From: manibala@covansys.com (B.Manivannan) Organization: http://groups.google.com/ Message-ID: <23de5426.0201040309.7127349b@posting.google.com> References: <23de5426.0201032353.566fcb7a@posting.google.com> For Simulator using SIMNT it is working fine. For bootable image application, for simpc, it is compiling successfuly..but how can it be downloaded in the ppc860 target. With thanks and regards, Manivannan.B. --------------------------- End of New-News digest ********************** From vxwexplo-errs@csg.lbl.gov Fri Jan 4 06:45:28 2002 From: "John W Cosgrove" Date: Fri Jan 4 06:45:30 PST 2002 Subject: Re: comp.os.vxworks newsdigest Tim, The operation needed is to write to the Universe's MISC_CTL register (Offset 404), setting bit designated SW_SYSRST (bit 22). In our config (VME_EXTEND) this is accomplished by a long word write to address 0xfd050404 of 0x4000. I added the following to my sysLib.c to encapsulate the operation: /****************************************************************************** * * sysVmeReset - Causes a reset of the VMEbus to occur * * This routine will write a one to the Bus Reset bit of the Universe chip * register #404 of the Universe chip. * * From a C point of view, the routine is defined as follows: * * void sysVmeReset() * * RETURNS: Nothing, bus resets * * RSC Add on, {jwc:991029} */ void sysVmeReset(void) { *UNIVERSE_MISC_CTL |= LONGSWAP(MISC_CTL_SW_SRST); } John Cosgrove (401)842-4167 jwc@ssd.ray.com To: vxworks_users@csg.lbl.gov Subject: comp.os.vxworks newsdigest Newsgroups: comp.os.vxworks Subject: vme reset Date: 3 Jan 2002 14:50:21 -0800 From: timothy.j.iskander@lmco.com (Tim Iskander) Organization: http://groups.google.com/ Message-ID: <46e37459.0201031450.20374d8a@posting.google.com> It looks like when I power up our MVME2700 with the vx bootloader selected, it never asserts a VME reset. If I switch over to PPCBug, it does. I cant find anything in the vx docs/etc about it. Has anyone seen this or know anything about it? We have a custom board which needs to see VME RESET at power up. thanks /Tim From vxwexplo-errs@csg.lbl.gov Fri Jan 4 06:54:35 2002 From: "Cigdem TURAN" Date: Fri Jan 4 06:54:37 PST 2002 Subject: Ramix PMC237 expansion board Hi all, Is there anybody who have used Ramix's PMC237 PMC expansion board? There is no Vxworks driver for the board. There is problem with slot PMC0. Please share your experience with the board. Thank you very much. Cigdem Turan ASELSAN, TURKEY From vxwexplo-errs@csg.lbl.gov Sat Jan 5 04:03:10 2002 From: Vxworks Exploder Date: Sat Jan 5 04:03:13 PST 2002 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Sat Jan 5 04:03:07 PST 2002 Subject: Re: QSPAN with PowerQUICC Subject: Re: Interrupt Levels - An Intro... Subject: test, please ignore Subject: Re: HW benchmark under VxWorks Subject: Question about END/NPT driver. Thanks in advance! Subject: Re: vme reset Subject: Re: QSPAN with PowerQUICC Subject: vxWorks Flash File System Subject: Re: S_errno_EADDRNOTAVAIL Subject: vxworks, hard disk crashes, and NBX-100 Subject: Can't attach unknown device Subject: 0xd003 status error displayed by bootrom ------------------------------------------------------- Newsgroups: comp.sys.powerpc.tech,comp.os.vxworks Subject: Re: QSPAN with PowerQUICC Date: 04 Jan 2002 16:58:31 GMT From: pdesnoyers@sourcetree.chinook.com (Peter Desnoyers) Organization: Concentric Internet Services Message-ID: References: <3C3095E7.AB53450D@mindspring.com> <3C319464.BE59B19E@varndellengineering.com> On Tue, 01 Jan 2002 05:50:12 -0500, George Varndell wrote: >Mike Moran wrote: >> >> I have an MDP8xxPro development board from ESTC. >> It has a Tundra QSPAN PCI Host Bridge, and I have >> a USB OHCI PCI card in the single PCI slot of the >> development board. > >The EST (actually WindRiver) board does not have external >address multiplexing, which is required when DRAM is >accessed on the QBUS. Try using the SRAM on the Pro board >instead of the DRAM. It works more better. ;) In addition, make sure that the addresses you are giving the USB controller are proper PCI bus addresses. E.g. on our in-house board, sort-of-copied from the MBX860, PCI bus address 0x80000000 is mapped to QBus address 0. Check the QBus target configuration registers (PCI target config? I forget) Email me if you want more help - I read this group infrequently... - -- ..................................................................... Peter Desnoyers (781) 457-1165 pdesnoyers@chinook.com Chinook Communications (617) 661-1979 pjd@fred.cambridge.ma.us 100 Hayden Ave, Lexington MA 02421 --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Interrupt Levels - An Intro... Date: 4 Jan 2002 09:07:45 -0800 From: arun4vxworks@indiatimes.com (Arun Prasad) Organization: http://groups.google.com/ Message-ID: References: Hai, Thanks for your follow-up. <>I have gone through the relevant portions of this guide. But I am not satisfied with it. Now I need more information on interrupt levels and its significance in writing device drivers on VxWorks. Can anyone throw some torch on these topics? Regards Arun molochai wrote in message news:... > Arun Prasad wrote: > > > Hi All, > > > > *I have some doubts on ISRs on VxWorks. Infact I need some > > good introduction on interrupt levels, their usage, limitations. > > > > I'm writing NIC driver on VxWorks by using a Linux Driver > > where there's no interrupt levels. > > Can anyone throw some torch in this regard? > > > > **Also, I'd like to know on IRQ sharing and SMP kernel issues > > on VxWorks. > > > > As VxWorks does not do SMP, there are no issues there. As for sharing > interrupts, this is entirely possible for most devices, and certainly for > any PCI peripherals. > > > > ***If you can give me some information on good mailing lists > > and/or > > news/mailing groups or any good sites or material which explaing these > > that will be helpful for me to understand interrupt levels, tasks and > > scheduling in VxWorks. > > Try the VxWorks Programmers guide for a quick intro. --------------------------- Newsgroups: comp.os.vxworks Subject: test, please ignore Date: Fri, 4 Jan 2002 09:53:23 -0800 From: "Lee Hao" Message-ID: Reply-To: "Lee Hao" ignore, please --------------------------- Newsgroups: comp.os.vxworks,comp.benchmarks Subject: Re: HW benchmark under VxWorks Date: 4 Jan 2002 11:03:03 -0800 From: akarpov@my-deja.com (Alex) Organization: http://groups.google.com/ Message-ID: <4f09bb68.0201041103.4c1626bd@posting.google.com> References: <4f09bb68.0112211217.407f2053@posting.google.com> Hi folks, Welcome back all of you who just returned from the holidays. I'd like to give it another try & post my message again: news:<4f09bb68.0112211217.407f2053@posting.google.com>... > Hi, > > I'd like to examine performance of my hardware platform (PPC) through > SPEC CPU95 tests. Unfortunately I found out the SPEC95 suite doesn't support > VxWorks. I was wondering if you could share any information on running SPEC > or other third party benchmarking kits under VxWorks. > > Thanks, Alex --------------------------- Newsgroups: comp.os.vxworks Subject: Question about END/NPT driver. Thanks in advance! Date: Fri, 4 Jan 2002 14:13:17 -0500 From: "Jiandong Wang" Organization: Nortel Message-ID: Question about END/NPT driver. Thanks in advance! I write an END/NPT driver (a .c file). But when I try to include it into vxWorks image, the project can't create the image. There are 3 files: mydriver.c lowlayerbufmanager.h --- define a class ClassLowlayerbufmanager lowlayerbufmanager.cpp In mydriver.c , ClassLowlayerbufmanager::instance()->memberfuncion() should be called to receive/send data. So, "lowlayerbufmanager.h" is included in mydriver.c . The compiler reports the error at the keyword "class". (I think C compiler doesn't know it.) Then, I add another 2 files: 1. temp.h callObjectMemberFunction(); /* prototype declaration */ 2. temp.cpp #include "lowlayerbufmanager.h" callObjectMemberFunction() { ClassLowlayerbufmanager::instance()->memberfuncion() } Now, in mydriver.c, only temp.h should be included. But after compiling all files, the ldppc reports "undefined reference to callObjectMemberFunction() " I don't know how to let a function in the .c file can call another function(or instance's memberfunction) in other .cpp file. Anybody can give me some advices? I really appeciate your help! Regards & Thanks, Jiandong Wang --------------------------- Newsgroups: comp.os.vxworks Subject: Re: vme reset Date: Fri, 04 Jan 2002 13:15:18 -0600 From: Andrew Johnson Organization: APS, Argonne National Laboratory Message-ID: <3C35FF46.45A29536@aps.anl.gov> References: <46e37459.0201031450.20374d8a@posting.google.com> Tim Iskander wrote: > > It looks like when I power up our MVME2700 with the vx bootloader > selected, > it never asserts a VME reset. If I switch over to PPCBug, it does. I > cant find > anything in the vx docs/etc about it. Has anyone seen this or know > anything about it? We have a custom board which needs to see VME RESET > at power up. There's a bit in the Universe chip's MISC_CTL register which you can set to initiate a bus reset. That register gets initialized in universe.c:sysUniverseInit() so you could just insert MISC_CTL_SW_SRST | in the lines that are ORed into temp_data. However IIRC the VME backplane circuitry should be holding SYSRESET* low for 200ms after +5V becomes stable, so your board should be seeing a reset pulse at powerup anyway. Are you sure your backplane is wired properly? - - Andrew - -- Perfection is reached, not when there is no longer anything to add, but when there is no longer anything to take away. - - Antoine de Saint-Exupery --------------------------- Newsgroups: comp.sys.powerpc.tech,comp.os.vxworks Subject: Re: QSPAN with PowerQUICC Date: Fri, 04 Jan 2002 16:02:12 -0500 From: Mike Moran Organization: MindSpring Enterprises Message-ID: <3C361854.3D33FE92@mindspring.com> References: <3C3095E7.AB53450D@mindspring.com> <3C319464.BE59B19E@varndellengineering.com> Reply-To: moran@mindspring.com George Varndell wrote: > > Mike Moran wrote: > > > > I have an MDP8xxPro development board from ESTC. > > It has a Tundra QSPAN PCI Host Bridge, and I have > > a USB OHCI PCI card in the single PCI slot of the > > development board. > > The EST (actually WindRiver) board does not have external > address multiplexing, which is required when DRAM is > accessed on the QBUS. Try using the SRAM on the Pro board > instead of the DRAM. It works more better. ;) Indeed Mr. Varndell! You are *the* man! It sucks that I have to use a "specialized" memory space, but it works just like you said. Now I can get back to the actual task at hand, which is an embedded USB OHCI HCD. thanks, mike - -- Michael N. Moran 5009 Old Field Ct. Kennesaw, GA 30144 (h) 770 516 7918 (c) 678 521 5460 The Beatles were wrong: 1 & 1 & 1 is 1 --------------------------- Newsgroups: comp.os.vxworks Subject: vxWorks Flash File System Date: 4 Jan 2002 16:43:33 -0800 From: e-che@ntl30.com (eche) Organization: http://groups.google.com/ Message-ID: I made flash file system initialized on vxWorks 5.4. but I need some help for completing. I loaded vxWorks image from network via ftp method. I want this loaded image to be copied on flash .(Purpose) It doesn't work. So I did like this. In bootConfig.c, first I made ramDrv. ramDrv(); pBlkDev = ramDevCreate(0, 512, 400, 3000, 0);/* ramAddr, bytesperblk, blkspertrack, nBlocks, blkOffset */ dosFsMkfs("/ram", pBlkDev); and In bootCmdLoop, case '$': /* net load ,copy it and flashLoad */ if (netBootLoad (pLine, &entry)==OK) { flashFInit(); copy("/ram/vxWorks","/rfa2/vxWorks"); /* flashFsBlkFlush2(0); flashFsBlkWrite2(pBlkDev,0, 32, (char *)&entry); */ printf("copied from ram to flash!\n"); } else { printf("net boot load failed!\n"); } if (flashLoad ("/rfa2/vxWorks", &entry) == OK) { printf ("entry = 0x%x\n", (int) entry); /*go(entry);*/ } else { printf("flashLoad failed!\n"); taskDelay (sysClkRateGet ()); /* pause a second */ /*reboot (BOOT_NO_AUTOBOOT);*/ /* something is awry */ } break; Regards, eche --------------------------- Newsgroups: comp.os.vxworks Subject: Re: S_errno_EADDRNOTAVAIL Date: Sat, 05 Jan 2002 01:27:26 GMT From: "cfk@pacbell.net" Organization: Prodigy Internet http://www.prodigy.com Message-ID: <2EsZ7.23764$_Z2.925686963@newssvr21.news.prodigy.com> References: <1010140603.575609@ernani.logica.co.uk> Dear Mark: I believe the setsockopt is telling you that it is unable to add your routine as a member to the multicast group you are passing into it. Probably because this multicast group does not exist on your network. Remember that error returns from vxWorks have the lower 8 bits set as the classic errno with the upper 8 bits from vxWorks itself. This merely means here that the setsockop call failed because of some underlying network configuration not setup properly. Charles <@logica.com> wrote in message news:1010140603.575609@ernani.logica.co.uk... > Hi all, > > Anyone know what the errorno 0x31 (S_errno_EADDRNOTAVAIL) means? Its the > only return I get from the call > > Status = setsockopt( ClientSocket, > IPPROTO_IP, > IP_ADD_MEMBERSHIP, > (char *)&MulticastReq, > sizeof(MulticastReq) ); > > I can't find any definition of it in the manuals anywhere so I'm unsure > what to do in order to fix it!!! > > thanks in advance, > Mark. > > > > > > --------------------------- Newsgroups: comp.os.vxworks Subject: vxworks, hard disk crashes, and NBX-100 Date: Fri, 04 Jan 2002 20:23:42 -0500 From: "Yan Seiner" Organization: http://extra.newsguy.com Message-ID: <20020104.202339.928894674.2005@oberon.cardinal.lan> All right, I'm a complete vxworks newbie. I have an NBX-100 phone system, which runs vxworks. We've had this system for two months, in which time we have had 2 complete hard disk crashes. The system uses Seagate IDE drives, which from experience are rock reliable. They just don't crash that often. The only indication of anything wrong is that we get an error (going from memory) error on socket 39, errno = 32 on reinstalling the OS. 3Com ( the vendor) is basically saying keep replacing the disks. I can't stand to have more downtime. From looking at the OS, it appears to be a *nix variant, which would make error 32 a broken pipe. I have no idea what socket 39 might be. Any guidance or ideas? 3Com's support basically is a read-from-the-menu type of operation - and their idea is that if it doesn't work, it's someone else's problem. I am somewhat frustrated with them at this point. - --Yan --------------------------- Newsgroups: comp.os.vxworks Subject: Can't attach unknown device Date: 5 Jan 2002 00:19:27 -0800 From: vividh@hotmail.com (rasayani) Organization: http://groups.google.com/ Message-ID: <28809481.0201050019.2a1cbdb9@posting.google.com> I am trying to ipAttach to (1,"fei") but it gives a error Can't attach unknown fei(unit 1). I can attach to fei0 though. Viv --------------------------- Newsgroups: comp.os.vxworks Subject: 0xd003 status error displayed by bootrom Date: 5 Jan 2002 02:38:55 -0800 From: rsramkee@mailcity.com (Ramki) Organization: http://groups.google.com/ Message-ID: Hi, I have an x86 based target with an ne2000 ISA ethernet card (brand accton). I tried building a bootrom wherein I would like to download the VxWorks image using this card over FTP from an FTP server. I enabled INCLUDE_ENE in config.h under my BSP directory (pcpentium). The comments in the header file indicated that this was the flag for including a ne2000 driver. This actually is supposed to enable BSD type driver, but effectively it in turn defines another macro, INCLUDE_ENE_END, including an END compatible ne2000 driver. When I am booting thru the bootrom, the card is identified fine and I get the message, "Attaching to interface ene0..." "Attaching to interface lo0... Done" etc... but when it tries to download and load the image, the target hangs for sometime at the "loading..." prompt and then gives me the status error 0xd0003... Going through the programmers documentation, I found that this error code relates to the iosLibrary and the error code is, S_ioslib_INVALID_FILE_DESCRIPTOR (M_ioslib | 3). I am stuck here as I am not able to relate this error code to whats supposedly happening in the target. I also dont see any "muxload Failed" message. So, the ne2000card it seems has been identified fine. When I try the same process with another 3COM PCI card in the same target with the built in driver for the 3COM card, with the same FTP server, everything works fine. Can someone help me out with this. Am I missing something or is there an issue with ISA/EISA? Thanks for your time. Ramakrishnan --------------------------- End of New-News digest ********************** From vxwexplo-errs@csg.lbl.gov Sun Jan 6 04:03:07 2002 From: Vxworks Exploder Date: Sun Jan 6 04:03:09 PST 2002 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Sun Jan 6 04:03:04 PST 2002 Subject: Re: pci error in IDT Rc32332 Subject: Re: pci error in IDT Rc32332 Subject: Re: 0xd003 status error displayed by bootrom Subject: Re: Can't attach unknown device ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: pci error in IDT Rc32332 Date: 5 Jan 2002 04:20:03 -0800 From: sczhang@esteemnetworks.com (ShiChao Zhang) Organization: http://groups.google.com/ Message-ID: <4ceedd43.0201050420.7ee19f02@posting.google.com> References: <4ceedd43.0112282323.368465f8@posting.google.com> Johan Borkhuis wrote in message news:... > sczhang@esteemnetworks.com (ShiChao Zhang) wrote: > > > 1. When I try to use the PCI Configuration Data Register at > > 1800_2cfc,Address Register at 1800_2cf8 to access the PCI > > configuration space, any write operation is OK(each operation comprise > > of a write access to address Reg and a write/read operation to data > > Reg). But If I perform a read operation, it will generate a Bus Error > > Exception. Then I tried to locate why the exception take place. And I > > found that if I put a access to Address Reg followed by a printf > > function call, then begin the normal read operation, it will be OK, no > > exception will occur. > > I can't understand it. > > Did you try to use the memory mapped version of the PCI access? You can > also access the I/O space using address 0x188xxxxx, and you can access the > memory space using addresses 0x40000000 .. 0x5fffffff and 0x60000000 .. > 0x7fffffff. You can use different endianess for these areas, but they can > point to the same PCI space. > > > Groeten, > Johan I have tried it, but has no effect. And if I disable the exception, all the value I got is the address of the PCI configuration Address Register:0x18002cfc. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: pci error in IDT Rc32332 Date: 5 Jan 2002 20:24:32 GMT From: Johan Borkhuis Organization: Agere Systems Message-ID: References: <4ceedd43.0112282323.368465f8@posting.google.com> <4ceedd43.0201050420.7ee19f02@posting.google.com> sczhang@esteemnetworks.com (ShiChao Zhang) wrote: > I have tried it, but has no effect. And if I disable the exception, > all the value I got is the address of the PCI configuration Address > Register:0x18002cfc. Did you try the command "pciDeviceShow" and/or "pciHeaderShow"? Groeten, Johan - -- o o o o o o o . . . _____________________________ o _____ || Johan Borkhuis | .][__n_n_|DD[ ====_____ | borkhuis@agere.com | >(________|__|_[_________]_|__________________________| _/oo OOOOO oo` ooo ooo 'o!o!o o!o!o` === VxWorks FAQ: http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html === --------------------------- Newsgroups: comp.os.vxworks Subject: Re: 0xd003 status error displayed by bootrom Date: 5 Jan 2002 12:50:39 -0800 From: john_94501@yahoo.com (John) Organization: http://groups.google.com/ Message-ID: <488e459a.0201051250.f7fd0fd@posting.google.com> References: Hello, Normally this problem is caused by one of: 1) The boot file name is invalid (typo perhaps) 2) The ftp server is not configured correctly (you don't say whether it was a Windoze box or a Unix machine that was serving) 3) The user name and/or password in the boot info are not correct 4) Network problems (either physical, or things like the gateway and netmask settings being wrong). Is your ftp server on the same subnet as the target board? Did you set the netmask explicitly in the boot parameters, or just rely on the mask being determined automatically from the address? The easiest way to check 1 through 3 above is to simply ftp to the server from another machine, say a windoze box. Log in as the user you put in the bootline, and then try to get the file using the exact pathname from the bootline. If that works, then you can concentrate on the network side of the problem. See if you can snoop/sniff packets from the network to see what is happening. Try pinging the target while it tries to download the image. Double check all the address settings. HTH, John... rsramkee@mailcity.com (Ramki) wrote in message news:... > Hi, > I have an x86 based target with an ne2000 ISA ethernet card (brand > accton). I tried building a bootrom wherein I would like to download > the VxWorks image using this card over FTP from an FTP server. I > enabled INCLUDE_ENE in config.h under my BSP directory (pcpentium). > The comments in the header file indicated that this was the flag for > including a ne2000 driver. This actually is supposed to enable BSD > type driver, but effectively it in turn defines another macro, > INCLUDE_ENE_END, including an END compatible ne2000 driver. > > When I am booting thru the bootrom, the card is identified fine and I > get the message, > "Attaching to interface ene0..." > "Attaching to interface lo0... Done" > etc... but when it tries to download and load the image, the target > hangs for sometime at the "loading..." prompt and then gives me the > status error 0xd0003... > > Going through the programmers documentation, I found that this error > code relates to the iosLibrary and the error code is, > S_ioslib_INVALID_FILE_DESCRIPTOR (M_ioslib | 3). > > I am stuck here as I am not able to relate this error code to whats > supposedly happening in the target. > > I also dont see any "muxload Failed" message. So, the ne2000card it > seems has been identified fine. > > When I try the same process with another 3COM PCI card in the same > target with the built in driver for the 3COM card, with the same FTP > server, everything works fine. > > Can someone help me out with this. Am I missing something or is there > an issue with ISA/EISA? > > Thanks for your time. > Ramakrishnan --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Can't attach unknown device Date: 5 Jan 2002 12:54:18 -0800 From: john_94501@yahoo.com (John) Organization: http://groups.google.com/ Message-ID: <488e459a.0201051254.32fe7920@posting.google.com> References: <28809481.0201050019.2a1cbdb9@posting.google.com> Hello, Do you have an entry for the second fei in the endDevTbl for your BSP? Check in configNet.h to make sure that you have two entries for fei (or you have at least two dummy entries, and your BSP auto-detects multiple cards and will create suitable entries in this table automatically - some BSPs do this, but not all, and since you omitted to tell us anything about the system you are working on, I can't be more precise). HTH, John... vividh@hotmail.com (rasayani) wrote in message news:<28809481.0201050019.2a1cbdb9@posting.google.com>... > I am trying to ipAttach to (1,"fei") but it gives a error > Can't attach unknown fei(unit 1). I can attach to fei0 though. > > Viv --------------------------- End of New-News digest ********************** From vxwexplo-errs@csg.lbl.gov Mon Jan 7 04:03:10 2002 From: Vxworks Exploder Date: Mon Jan 7 04:03:12 PST 2002 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Mon Jan 7 04:03:06 PST 2002 Subject: Re: vxworks, hard disk crashes, and NBX-100 Subject: SiS630E/SiS900 Ehternet support for VxWorks? Subject: Re: SiS630E/SiS900 Ehternet support for VxWorks? Subject: Re: vxworks, hard disk crashes, and NBX-100 Subject: Re: Symbol table errors compiling an image when using STL Subject: Re: vxworks, hard disk crashes, and NBX-100 Subject: Re: Could You Give me a file --> libPPC603gnugcc.a Subject: Re: 0xd003 status error displayed by bootrom Subject: make a boot disk Subject: tNetTask suspend while trying to load two END's Subject: Re: 0xd003 status error displayed by bootrom Subject: Re: 0xd003 status error displayed by bootrom ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: vxworks, hard disk crashes, and NBX-100 Date: Sun, 6 Jan 2002 16:19:11 +0200 From: "Leonid Rosenboim" Organization: Verio Message-ID: References: <20020104.202339.928894674.2005@oberon.cardinal.lan> Yan, 3COM is not the onlly company getting dummies into support pisitions, it's probably with most companies the same story, WindRiver is probably not exceptional in this respect either. Regarding the socket 39, it's meaningless to me, and I have been around VxWorks for well over a decade. Regarding hard disk crashes, this is where I might be helpful, as I got over 15 years of experience with hard disks (and other storage devices): Reliability of a standard HD is not only a function of the HD itself, but also of its environment, including heat dissapation, mechanical vibration and power supply. Some disks are more sensitive then others to each environmental parameter, but they all are sensitive to this to some extent, hence your experience with a particular HD model on your rack mount servers, can grealty differ from the statistics of the same disk drive in desktop systems. I would recommend that you try to replace this HD of yours with a different brand, but more then anything, 3COM should be able to provide a better sulution. If you threat to cancel yior order of the PBX, or even make a threat to sue 3COM for all lost work hours due to their faulty products, you will get attention of some more senior people, and they will try to solve this somehow. Meanwhile, replace the HD. - - Leonid "Yan Seiner" wrote in message news:20020104.202339.928894674.2005@oberon.cardinal.lan... > All right, I'm a complete vxworks newbie. > > I have an NBX-100 phone system, which runs vxworks. We've had this system > for two months, in which time we have had 2 complete hard disk crashes. > > The system uses Seagate IDE drives, which from experience are rock > reliable. They just don't crash that often. > > The only indication of anything wrong is that we get an error (going from > memory) error on socket 39, errno = 32 on reinstalling the OS. > > 3Com ( the vendor) is basically saying keep replacing the disks. I can't > stand to have more downtime. From looking at the OS, it appears to be a > *nix variant, which would make error 32 a broken pipe. I have no idea > what socket 39 might be. > > Any guidance or ideas? > > 3Com's support basically is a read-from-the-menu type of operation - and > their idea is that if it doesn't work, it's someone else's problem. I am > somewhat frustrated with them at this point. > > --Yan --------------------------- Newsgroups: comp.os.vxworks Subject: SiS630E/SiS900 Ehternet support for VxWorks? Date: Mon, 07 Jan 2002 00:18:01 GMT From: "Mike Anderson" Organization: RoadRunner - Cox Message-ID: Greetings! I've seen a couple of posts over the last couple of years regarding this chipset, but never a response. So, I'll ask this one again. Has anyone seen or heard of an Ethernet driver for VxWorks that works on the SiS630E (SiS900) chipset? TIA, Mike Anderson - -- - -- ======================================================= T h e P T R G r o u p, I n c. ======================================================= Embedded, Real-Time Solutions and Training Mike Anderson mailto:mike@ThePTRGroup.com Chief Scientist http://www.ThePTRGroup.com C: (703) 585-9384 V: (703) 430-3748 ======================================================= "Software development is like making a baby... You can't make a baby in one month by impregnating nine women. Some things just take time." --------------------------- Newsgroups: comp.os.vxworks Subject: Re: SiS630E/SiS900 Ehternet support for VxWorks? Date: Mon, 07 Jan 2002 00:29:52 GMT From: "cfk@pacbell.net" Organization: Prodigy Internet http://www.prodigy.com Message-ID: <4_5_7.25372$nv4.2121265170@newssvr14.news.prodigy.com> References: Dear Mike: I had an occaision a few months ago to study the Sis900 and it appears to be basically a 2nd source of an NE2000. It looks register compatible and all of that. I would imagine the generic ne driver should work. Have not tested it, just studied the chip spec for half a day trying to convince myself it mapped in as something. Charles "Mike Anderson" wrote in message news:ZO5_7.58922$mp3.27263625@typhoon.southeast.rr.com... > Greetings! > > I've seen a couple of posts over the last couple of years regarding this > chipset, > but never a response. So, I'll ask this one again. Has anyone seen or > heard of > an Ethernet driver for VxWorks that works on the SiS630E (SiS900) chipset? > > TIA, > > Mike Anderson > > -- > -- > ======================================================= > T h e P T R G r o u p, I n c. > ======================================================= > Embedded, Real-Time Solutions and Training > > Mike Anderson mailto:mike@ThePTRGroup.com > Chief Scientist http://www.ThePTRGroup.com > C: (703) 585-9384 V: (703) 430-3748 > ======================================================= > "Software development is like making a baby... > You can't make a baby in one month by impregnating > nine women. Some things just take time." > > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: vxworks, hard disk crashes, and NBX-100 Date: Sun, 06 Jan 2002 19:34:34 -0500 From: "Yan Seiner" Organization: http://extra.newsguy.com Message-ID: <20020106.193426.920271899.13512@oberon.cardinal.lan> References: <20020104.202339.928894674.2005@oberon.cardinal.lan> In article , "Leonid Rosenboim" wrote: > Yan, 3COM is not the onlly company getting dummies into support > pisitions, it's probably with most companies the same story, WindRiver > is probably not exceptional in this respect either. > > Regarding the socket 39, it's meaningless to me, and I have been around > VxWorks > for well over a decade. > > Regarding hard disk crashes, this is where I might be helpful, as I got > over 15 years > of experience with hard disks (and other storage devices): Reliability > of a standard HD is not only a function of the HD itself, but also of > its > environment, including heat dissapation, mechanical vibration and power > supply. > Some disks are more sensitive then others to each environmental > parameter, but > they all are sensitive to this to some extent, hence your experience > with a particular > HD model on your rack mount servers, can grealty differ from the > statistics of the > same disk drive in desktop systems. 3Com has already brought up those issues - trying to "blame the customer". The NBX100 is rack mounted; the rack is bolted to the floor and ceiling and there is no vibration to speak of. The rack has 9 other hard drives - - all but two Seagates - and none of the others has crashed. The power is clean - this is a new building, the rack is fed by a dedicated outlet, and is filtered with a surge/lighting arrestor, a line conditioner, and a UPS, all name brand, all running at less tha 50% capacity. > > I would recommend that you try to replace this HD of yours with a > different brand, but more then anything, 3COM should be able to provide > a better sulution. > If you threat to cancel yior order of the PBX, or even make a threat to > sue 3COM > for all lost work hours due to their faulty products, you will get > attention of some > more senior people, and they will try to solve this somehow. Meanwhile, > replace the HD. > > - Leonid I may scream and yell at 3Com; unfortunately 3Com does not talk the end users but works only through dealers. The dealer has been excellent, but having worked with open source for a bunch of eyars, closed systems really leave me frustrated :-( If it crashed again, I'll find someone at 3COm to explain the facts of life to :-) - --Yan --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Symbol table errors compiling an image when using STL Date: 7 Jan 2002 00:00:34 -0600 From: "Arnon W" Message-ID: <3c3833dd$1@news.barak.net.il> References: <3c29bbec$1@news.barak.net.il> Hi, Didn't work. I added only these 3 lines to one of my cpp's: #include (or #include ) using namespace std; list test; The funny thing is that the module containing the STL part compiles with no problems, and the code runs well as a downloadable module. The minute I try to integrate it into the image - boom! Arnon "Dave Korn" wrote in message news:W4XY7.206$qS6.464044@newsr2.u-net.net... > Arnon W wrote in message <3c29bbec$1@news.barak.net.il>... > >Hi, > > > >I'm new to this environment so it might be a minor issue I'm having > problems > >with. > >(On the other hand - I couldn't find earlyer posts in this specific > matter:) > > >When I try to integrate it to the image, I get vicious responses at this > >point: > > >ccppc -c -fdollars-in-identifiers -O3 -mstrict-align -nostdinc -DRW_MULTI_T > H > >READ -D_REENTR > >ANT -fvolatile -fno-builtin -fno-for-scope -DCPU=PPC604 -DPRJ_BUILD -DHAL_L > I > >B > >...paths... symTbl.c -o symTbl.o > >symTbl.c:2013: syntax error before `.120' > >symTbl.c:2014: syntax error before `.139' > >symTbl.c:2015: syntax error before `.148' > >symTbl.c:8448: > `__sn_t24__default_alloc_template2b1i0$__node_allocator_lock' > >undeclared he > >re (not in a function) > >symTbl.c:8448: syntax error before `.120' > >symTbl.c:8449: initializer element for `standTbl[1997].value' is not > >constant > >symTbl.c:8449: warning: initialization from incompatible pointer type > >symTbl.c:8449: > `__sn_t24__default_alloc_template2b1i0$__node_allocator_lock' > >undeclared he > >re (not in a function) > >symTbl.c:8449: syntax error before `.139' > >symTbl.c:8450: initializer element for `standTbl[1998].name' is not > constant > >symTbl.c:8450: syntax error before string constant > > >When I took out STL related stuff (I was using ) It compiled well. > >Is there a cure? > > I'm just passing on something I read somewhere else here, STL is really > not my field, but ISTR someone saying that to use STL you should only > #include - i.e. you use that header and it includes all the other > stl headers like and for you. You might like to try this > approach and see if it fixes your problem. > > DaveK > -- > Burn your ID card! http://www.optional-identity.org.uk/ > Help support the campaign, copy this into your .sig! > Proud Member of the Exclusive "I have been plonked by Davee because he > thinks I'm interesting" List > Member # > > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: vxworks, hard disk crashes, and NBX-100 Date: Mon, 7 Jan 2002 09:08:17 +0200 From: "Leonid Rosenboim" Organization: Verio Message-ID: References: <20020104.202339.928894674.2005@oberon.cardinal.lan> <20020106.193426.920271899.13512@oberon.cardinal.lan> Yan, Both in terms of vibration and power source I meant the HD's environment, while you are talking about the externa environment of the box. In rack mounted boxes, hard rack attachment can actually be the cause for problems, transmitting vibrations instead of absorbing them. Also, fans can cause quite a lot of internal vibration. The system's internal power supply can be rather unstable if some devices have a highly dynamic consumption pattern, sometimes the disk drive itself can cause the 12V voltage to fluctuate beyond specified tolerance, if measured on the HD power connector. HD's internal low voltage protection only works when power goes down for a long time, but strong ripple can cause irreparable damage to HD. I am not suggesting you to modify the 3COM system, just offering some insight to help you understand the fenomena. - - Leonid "Yan Seiner" wrote in message news:20020106.193426.920271899.13512@oberon.cardinal.lan... > In article , "Leonid Rosenboim" > wrote: > > > Yan, 3COM is not the onlly company getting dummies into support > > pisitions, it's probably with most companies the same story, WindRiver > > is probably not exceptional in this respect either. > > > > Regarding the socket 39, it's meaningless to me, and I have been around > > VxWorks > > for well over a decade. > > > > Regarding hard disk crashes, this is where I might be helpful, as I got > > over 15 years > > of experience with hard disks (and other storage devices): Reliability > > of a standard HD is not only a function of the HD itself, but also of > > its > > environment, including heat dissapation, mechanical vibration and power > > supply. > > Some disks are more sensitive then others to each environmental > > parameter, but > > they all are sensitive to this to some extent, hence your experience > > with a particular > > HD model on your rack mount servers, can grealty differ from the > > statistics of the > > same disk drive in desktop systems. > > 3Com has already brought up those issues - trying to "blame the > customer". The NBX100 is rack mounted; the rack is bolted to the floor > and ceiling and there is no vibration to speak of. The rack has 9 other hard drives > - all but two Seagates - and none of the others has crashed. > > The power is clean - this is a new building, the rack is fed by a > dedicated outlet, and is filtered with a surge/lighting arrestor, a line > conditioner, and a UPS, all name brand, all running at less tha 50% > capacity. > > > > > I would recommend that you try to replace this HD of yours with a > > different brand, but more then anything, 3COM should be able to provide > > a better sulution. > > If you threat to cancel yior order of the PBX, or even make a threat to > > sue 3COM > > for all lost work hours due to their faulty products, you will get > > attention of some > > more senior people, and they will try to solve this somehow. Meanwhile, > > replace the HD. > > > > - Leonid > > I may scream and yell at 3Com; unfortunately 3Com does not talk the end > users but works only through dealers. The dealer has been excellent, but > having worked with open source for a bunch of eyars, closed systems > really leave me frustrated :-( If it crashed again, I'll find someone at > 3COm to explain the facts of life to :-) > > --Yan --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Could You Give me a file --> libPPC603gnugcc.a Date: 7 Jan 2002 07:50:10 GMT From: Johan Borkhuis Organization: Agere Systems Message-ID: References: This file is part of the normal Tornado distribution, so you have to get a license for this. Groeten, Johan - -- o o o o o o o . . . _____________________________ o _____ || Johan Borkhuis | .][__n_n_|DD[ ====_____ | borkhuis@agere.com | >(________|__|_[_________]_|__________________________| _/oo OOOOO oo` ooo ooo 'o!o!o o!o!o` === VxWorks FAQ: http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html === --------------------------- Newsgroups: comp.os.vxworks Subject: Re: 0xd003 status error displayed by bootrom Date: 7 Jan 2002 07:57:28 GMT From: Johan Borkhuis Organization: Agere Systems Message-ID: References: <488e459a.0201051250.f7fd0fd@posting.google.com> john_94501@yahoo.com (John) wrote: > The easiest way to check 1 through 3 above is to simply ftp to the > server from another machine, say a windoze box. Log in as the user you > put in the bootline, and then try to get the file using the exact > pathname from the bootline. Also keep a close watch at the log of the FTP server, this will show a lot, for example invalid login or invalid path names. Groeten, Johan - -- o o o o o o o . . . _____________________________ o _____ || Johan Borkhuis | .][__n_n_|DD[ ====_____ | borkhuis@agere.com | >(________|__|_[_________]_|__________________________| _/oo OOOOO oo` ooo ooo 'o!o!o o!o!o` === VxWorks FAQ: http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html === --------------------------- Newsgroups: comp.os.vxworks Subject: make a boot disk Date: Mon, 7 Jan 2002 16:11:39 +0800 From: "Zhao Yandong" Organization: Bentium Ltd. (CN99) Message-ID: I want to use my pc as a target. And now I want to make a boot disk I know I should edit the DEFAULT_BOOT_LINE in config.h, but I don't know how to write it. My net adapter is realtek RTL 8139 Should I change other settings in config.h? Thank you Zhao Yandong --------------------------- Newsgroups: comp.os.vxworks Subject: tNetTask suspend while trying to load two END's Date: 7 Jan 2002 02:55:19 -0800 From: odedn@shiron.com (oj) Organization: http://groups.google.com/ Message-ID: <9d60aef6.0201070255.3c105617@posting.google.com> hi ya all vxWorkers!! i have encountered a problem while trying to load two ends. the first is an HDLC driver with the unit id=2, the second is a software END with the unit id=1, i also have an Ethernet driver wich loads at boot time with the unit id=0. the software End Driver is loaded with initialization look like this: END_TBL_ENTRY rcv; rcv.unit = 1; rcv.endLoadFunc = ReceiverEndLoad; rcv.endLoan = TRUE; rcv.endLoadString = "1:0:1:0"; rcv.pBSP = NULL; rcv.processed = FALSE; the HDLC End Driver is loaded with initialization look like this: END_TBL_ENTRY rg; rg.unit = 2; rg.endLoadFunc = Sens82xxEndLoad; rg.endLoan = TRUE; rg.endLoadString = ""; rg.pBSP = NULL; rg.processed = FALSE; when loading each of the drivers without trying to load the other every thing seems to work ok!! if i load first the HDLC everything seems ok , and than try to load the "software " END i get the message (the same goes if i first try to load the "software" End): Exception next instruction address: 0x0007c6f0 Machine Status Register: 0x00049000 Condition Register: 0x44020040 Task: 0x3db9928 "tNetTask" The tNetTask is suspend and i lose connection with target server. when running muxShow from Hyper Terminal i get: Device: motfcc Unit: 0 Description: Motorola FCC Ethernet Enhanced Network Driver Protocol: IP 4.4 ARP Type: 2054 Recv 0x7d334 Shutdown 0x7d5f8 Protocol: IP 4.4 TCP/IP Type: 2048 Recv 0x7d334 Shutdown 0x7d530 Device: Hdlc Unit: 2 Description: MPC82xx Network Driver Protocol: IP 4.4 ARP Type: 2054 Recv 0x7d334 Shutdown 0x7d5f8 Protocol: IP 4.4 TCP/IP Type: 2048 Recv 0x7d334 Shutdown 0x7d530 Device: IS-Rcvr Unit: 1 Description: InterSKY Receiver Enhanced Network Driver. Protocol: IP 4.4 ARP Type: 2054 Recv 0x7d334 Shutdown 0x7d5f8 Protocol: IP 4.4 TCP/IP Type: 2048 Recv 0x7d334 Shutdown 0x7d530 value = 0 = 0x0 which seems OK!! when running ifShow i get: motfcc (unit number 0): Flags: (0x8063) UP BROADCAST MULTICAST ARP RUNNING Type: ETHERNET_CSMACD Internet address: 10.150.11.54 Broadcast address: 10.255.255.255 Netmask 0xff000000 Subnetmask 0xff000000 Ethernet address is 08:00:3e:03:02:01 Metric is 0 Maximum Transfer Unit size is 1500 755 packets received; 130 packets sent 631 multicast packets received 1 multicast packets sent 0 input errors; 0 output errors 0 collisions; 0 dropped lo (unit number 0): Flags: (0x8069) UP LOOPBACK MULTICAST ARP RUNNING Type: SOFTWARE_LOOPBACK Internet address: 127.0.0.1 Netmask 0xff000000 Subnetmask 0xff000000 Metric is 0 Maximum Transfer Unit size is 32768 4 packets received; 4 packets sent 0 multicast packets received 0 multicast packets sent 0 input errors; 0 output errors 0 collisions; 0 dropped Hdlc (unit number 2): Flags: (0x8063) UP BROADCAST MULTICAST ARP RUNNING Type: ETHERNET_CSMACD Internet address: 200.20.1.1 Broadcast address: 200.20.1.255 Netmask 0xffffff00 Subnetmask 0xffffff00 Ethernet address is 61:66:61:66:61:66 Metric is 0 Maximum Transfer Unit size is 1500 0 packets received; 1 packets sent 0 multicast packets received 0 multicast packets sent 0 input errors; 0 output errors 0 collisions; 0 dropped IS-Rcvr (unit number 1): Flags: (0x8063) UP BROADCAST MULTICAST ARP RUNNING Type: ETHERNET_CSMACD Internet address: 192.168.1.1 Broadcast address: 192.168.1.255 Netmask 0xffffff00 Subnetmask 0xffffff00 Ethernet address is 0a:0b:0c:0d:0e:0f Metric is 0 Maximum Transfer Unit size is 1520 0 packets received; 3 packets sent 0 multicast packets received 0 multicast packets sent 0 input errors; 489500 output errors 0 collisions; 0 dropped value = 29 = 0x1d with the 489500 output errors on the IS-Rcvr "software" End. i beleive that somehow there is collisions between these two drivers but i dont have a clue what is the reason for this. some more useful information: 1. when loading each driver alone no broblems occur. 2. to initialize the mBlk memarea i use memaligne. 3. to initialize the cBlk memarea i use cacheDmaMalloc. 4. i have separate net pools for each driver. 5. i use the MPC8260 on tornado 2.0 youre help will be most blessed thanx. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: 0xd003 status error displayed by bootrom Date: 7 Jan 2002 03:21:08 -0800 From: rsramkee@mailcity.com (Ramki) Organization: http://groups.google.com/ Message-ID: References: <488e459a.0201051250.f7fd0fd@posting.google.com> Hi John, Thanks for the reply. > Normally this problem is caused by one of: > > 1) The boot file name is invalid (typo perhaps) > > 2) The ftp server is not configured correctly (you don't say whether > it was a Windoze box or a Unix machine that was serving) Windows, the FTP server that comes along with Tornado. > 3) The user name and/or password in the boot info are not correct > > 4) Network problems (either physical, or things like the gateway and > netmask settings being wrong). Is your ftp server on the same subnet > as the target board? Did you set the netmask explicitly in the boot > parameters, or just rely on the mask being determined automatically > from the address? Yes, I checked all these. It was to rule out all these problems that I repeated the same experiment with a 3COM PCI ethernet card in the same board under the same environment. I let the subnet to be determined automatically from the address. All the boot parameters, like the IP Address, the subnet, the FTP server's address, user name, password were all the same. Only the device name changed, from elPci (for 3COM) to ene (for the ne2000 card). The 3COM card works like a charm, downloads the file and runs the image without any problems. > If that works, then you can concentrate on the network side of the > problem. See if you can snoop/sniff packets from the network to see > what is happening. Try pinging the target while it tries to download > the image. Double check all the address settings. I tried this using a packet sniffer(NetXRay). Absolutely no packets coming out:( I am really out of clues as to what could be it. Thanks for your time. It helped. Regards Ramki --------------------------- Newsgroups: comp.os.vxworks Subject: Re: 0xd003 status error displayed by bootrom Date: 7 Jan 2002 03:23:49 -0800 From: rsramkee@mailcity.com (Ramki) Organization: http://groups.google.com/ Message-ID: References: <488e459a.0201051250.f7fd0fd@posting.google.com> Hi, Thanks for your reply. > > The easiest way to check 1 through 3 above is to simply ftp to the > > server from another machine, say a windoze box. Log in as the user you > > put in the bootline, and then try to get the file using the exact > > pathname from the bootline. > > Also keep a close watch at the log of the FTP server, this will show a lot, > for example invalid login or invalid path names. I did this. The log showed no signs of any activity, confirming that no packets were coming out of the card. Even a packet sniffer showed the same results. Please let me know if there could be any other possible problems I may have missed. Thanks for your time. regards Ramki --------------------------- End of New-News digest ********************** From vxwexplo-errs@csg.lbl.gov Mon Jan 7 10:45:20 2002 From: "Patrick T. Pinkowski" Date: Mon Jan 7 10:45:23 PST 2002 Subject: PPP and VxWorks This is a multi-part message in MIME format. --------------55EF95F0DEEBC5D2256BB505 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit I am trying to get PPP to operate over a Modem connection. I am able to dial out and get a good modem connection. Then, I perform a pppInit using the serial port which is connected to the modem. I set the PPP debug option On, and PPP indicated that it was sending out LCP packets, and that's what it is supposed to do. I tooled up the serial device driver associated with the "/tyCo/1" serial interface, and found that PPP was not actually sending out any bytes. Has anyone encountered a similar problem? Any help would be greatly appreciated. Thanks in advance, Pat -- Patrick T. Pinkowski SPARTA, Inc. 5875 Trinity Parkway, Suite 300 Centreville, Va 20120-1921 phone: (703) 988-8500 x385 fax: (703) 988-8599 --------------55EF95F0DEEBC5D2256BB505 Content-Type: text/x-vcard; charset=us-ascii; name="ptp.vcf" Content-Transfer-Encoding: 7bit Content-Description: Card for Patrick T. Pinkowski Content-Disposition: attachment; filename="ptp.vcf" begin:vcard n:Pinkowski;Patrick tel;cell:(703) 597-7795 tel;fax:(703) 988-8599 tel;home:(703) 858-0810 tel;work:(703) 988-8500 x385 x-mozilla-html:FALSE url:www.mclean.sparta.com org:SPARTA, Inc. adr:;;5875 Trinity Parkway, Suite 300;Centreville;Va;20120-1921;USA version:2.1 email;internet:ptp@sparta.com title:Senior Principle Computer Engineer fn:Patrick T. Pinkowski end:vcard --------------55EF95F0DEEBC5D2256BB505-- From vxwexplo-errs@csg.lbl.gov Mon Jan 7 11:38:32 2002 From: Donald.McLachlan@crc.ca Date: Mon Jan 7 11:38:34 PST 2002 Subject: Re: lspci I've been corresponding with SeaLevel but I think maybe this is more of a vxWorks question so I'll post it here. Background info follows, but my questions are: - Why does vxWorks say the card's I/O addresses are at 7001 and 7401 when Linux says they are at 7000 and 7401? - Also, why 2 addresses 700X and 740X; begin and end, or aliasing? - Both report memory address f9000000. SeaLevel says the registers are only available via I/O ... so what's this memory? Thanks, Don ----- I'm running booting vxWorks from a floppy on a PC that was used for linux ... so if I boot from the hard disk I have linux. Using the Linux lspci command I get: > lspci -v -d 0x135e:0x5102 00:0d.0 Serial controller: Sealevel Systems Inc: Unknown device 5102 (rev 01) (prog-if 02 [16550]) Subsystem: Sealevel Systems Inc: Unknown device 5102 Flags: medium devsel, IRQ 12 Memory at f9000000 (32-bit, non-prefetchable) I/O ports at 7000 I/O ports at 7400 ----- Using vxWorks command from the target shell I get: -> pciDeviceShow(0) Scanning function 0 of each PCI device on bus 0 Using configuration mechanism 1 bus device function vendorID deviceID class 00000000 00000000 00000000 00001106 00000691 00000600 00000000 00000001 00000000 00001106 00008598 00040600 00000000 00000007 00000000 00001106 00000596 00010600 00000000 0000000a 00000000 000010b7 00009055 00000200 00000000 0000000d 00000000 0000135e 00005102 02000700 value = 0 = 0x0 -> pciFindDeviceShow(0x0000135e, 0x00005102, 0) deviceId = 0x00005102 vendorId = 0x0000135e index = 0x00000000 busNo = 0x00000000 deviceNo = 0x0000000d funcNo = 0x00000000 value = 0 = 0x0 -> pciHeaderShow(0, 0x0000000d, 0x00000000) vendor ID = 0x135e device ID = 0x5102 command register = 0x0003 status register = 0x0280 revision ID = 0x01 class code = 0x07 sub class code = 0x00 programming interface = 0x02 cache line = 0x08 latency time = 0x00 header type = 0x00 BIST = 0x00 base address 0 = 0xf9000000 base address 1 = 0x00007001 base address 2 = 0x00007401 base address 3 = 0x00000000 base address 4 = 0x00000000 base address 5 = 0x00000000 cardBus CIS pointer = 0x00000000 sub system vendor ID = 0x135e sub system ID = 0x5102 expansion ROM base address = 0x00000000 interrupt line = 0x0c interrupt pin = 0x01 min Grant = 0x00 max Latency = 0x00 value = 0 = 0x0 From vxwexplo-errs@csg.lbl.gov Mon Jan 7 11:50:41 2002 From: "Whiteis, Peter G" Date: Mon Jan 7 11:50:43 PST 2002 Subject: VxWorks - unresolved "new" operator for C++ All, We are attempting to dynamically link in an object file produced from C++ source code. Our file is very simple ... it just allocates some memory via the "new" operator. When we attempt to download the object module to our target via the "ld" command, we get the following unresolved symbols: -> ld < test1.o Undefined symbol: __builtin_vec_new (binding 1 type 0) Undefined symbol: __builtin_delete (binding 1 type 0) value = 8378304 = 0x7fd7c0 = myTest(void) + 0x3f0 -> We are building using Tornado2 for a PPC860 family, and are including all the c++ components available as well as c++ runtime support. The symbol table is builtin with the image. We are able to lamely fix the problem when we directly build our test c++ file directly into the vxWorks images. Then we are able to successfully invoke the function call. Has anyone had this problem or found a work around for this? Thanks, -Pete Whiteis- Corning Inc From vxwexplo-errs@csg.lbl.gov Mon Jan 7 14:17:28 2002 From: "Mahesh Rajmane" Date: Mon Jan 7 14:17:31 PST 2002 Subject: Loading Vxworks Hi, I'm new to Vxworks and GNU. Until now our tool chain has been Metaware compilers and Beacon development tools. I have two questions: 1) What is file format of final VxWork image (bootable as well as downloadable)? 2) Is it possible to get OMF format for Vxworks image? Please help me out if anybody has idea of this. Thanks and Regards, Mahesh From vxwexplo-errs@csg.lbl.gov Tue Jan 8 03:29:46 2002 From: Shrishail Date: Tue Jan 8 03:29:48 PST 2002 Subject: fixed address loading Hi List, How can I load "select functions" from a object module at a known address. Using the #pragma use_section, in the source file helps in doing so. But what do I do in case I have only the object file? Typically some functions in the n/w stack of vxworks. Hope to get some input. Regards, Shrishail From vxwexplo-errs@csg.lbl.gov Tue Jan 8 04:03:17 2002 From: Vxworks Exploder Date: Tue Jan 8 04:03:19 PST 2002 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Tue Jan 8 04:03:13 PST 2002 Subject: Re: Question about END/NPT driver. Thanks in advance! Subject: powerQUICK-II evaluation board Subject: Re: powerQUICK-II evaluation board Subject: Re: VxWorks Routines List Subject: Re: VxWorks Routines List Subject: Re: tNetTask suspend while trying to load two END's Subject: broadcast UDP Receive and send problem Subject: Re: fei82559 hangs on tx Subject: new image building problem Subject: Re: broadcast UDP Receive and send problem Subject: Re: vxWorks Flash File System Subject: Re: Symbol table errors compiling an image when using STL Subject: Re: new image building problem Subject: Does Tornado require write access to download links/libraries? Subject: why does taskDelay(1) take so long? Subject: Re: 0xd003 status error displayed by bootrom Subject: Re: I/O ports on PowerPC Subject: readyQHead - how to read it? Subject: Re: VxWorks Routines List Subject: Target Shell does not support multi-users... Subject: serial port initialization when serial device is missing Subject: LC82C169 Driver... Subject: Re: new image building problem Subject: Sun RPC or microsoft RPC in VxDCOM ? Subject: Re: Target Shell does not support multi-users... Subject: Caching, I/O Space memory mapping and DMA Subject: any other debug method? Subject: What version of NPT am I using now ? Subject: Re: readyQHead - how to read it? Subject: powerQUICK-II evaluation boards question Subject: Re: why does taskDelay(1) take so long? Subject: Optimizing VxWorks boot time ? Subject: Re: any other debug method? ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: Question about END/NPT driver. Thanks in advance! Date: 7 Jan 2002 04:18:58 -0800 From: houcheng@ms17.hinet.net (houcheng) Organization: http://groups.google.com/ Message-ID: References: "Jiandong Wang" wrote in message news:... > > I don't know how to let a function in the .c file can call another > function(or instance's memberfunction) in other .cpp file. Anybody can give > me some advices? I really appeciate your help! > > Regards & Thanks, > > Jiandong Wang This is more like a OOP or a C/C++ programming question, not a question about NPT. To call a function in C++ source file, you can define a extern "C" function in the C++ source file, like this: extern "C" foo_c_functoin(int arg) { // any thing you want to do .... } Then you can call this function on .C source file. By declare a function as "extern "C" ", the C++ compiler will treat it as a C function, not C++ function, and this function's name will not be renamed by C++ rules, and will be inserted into symbol table directly. I am developing a IP over ATM, NPT driver now. Currently I am using SIMNT as my NPT template driver platform, and I will porting it to the ads860 platform. Maybe we can disscuss the NPT driver issues :) MY EMAIL: houcheng@turbonetworks.com.tw_NO_SPAM --------------------------- Newsgroups: comp.os.vxworks Subject: powerQUICK-II evaluation board Date: 7 Jan 2002 06:48:38 -0600 From: "Ilya Krasavin" Message-ID: <3c3996d1$1@news.barak.net.il> Reply-To: "Ilya Krasavin" Hi gurus. May anybody to recommnd me the evaluation board for powerQUICK-II. I found 3 (ACTIS, IPnexus and ADS), but I don't know which one is better. Thank you for a help. Ilya --------------------------- Newsgroups: comp.os.vxworks Subject: Re: powerQUICK-II evaluation board Date: Mon, 07 Jan 2002 09:31:02 -0500 From: George Varndell Organization: Posted Via Binaries.net = SPEED+RETENTION+COMPLETION = http://www.binaries.net Message-ID: <3C39B126.959408E7@varndellengineering.com> References: <3c3996d1$1@news.barak.net.il> Ilya Krasavin wrote: > > Hi gurus. > May anybody to recommnd me the evaluation board for powerQUICK-II. I found 3 > (ACTIS, IPnexus and ADS), but I don't know which one is better. > Thank you for a help. > Ilya Go to http://www.embeddedplanet.com/products/ep8260.asp In my opinion, this board is the best value available. Regards, George Varndell - -- Varndell Engineering, LLC. PPC and XScale Expertise The VE 82559 END driver demo version is now available. Compare the performance difference -- free, no obligation. Send email request to demo@varndellengineering.com. ______________________________________________________________________________ Posted Via Binaries.net = SPEED+RETENTION+COMPLETION = http://www.binaries.net --------------------------- Newsgroups: comp.os.vxworks Subject: Re: VxWorks Routines List Date: Mon, 7 Jan 2002 14:53:36 GMT From: Lee DeRaud Organization: None to speak of Message-ID: References: <3C34A175.D28A57CF@fnal.gov> <67f93ucpa1rkps39e3vlq8263gg71udrnc@4ax.com> <3C34EAD1.20E30A2E@gv.net> Sender: nntp@news.boeing.com (Boeing NNTP News Access) On Thu, 03 Jan 2002 15:35:45 -0800, "Michael R. Kesti" wrote: >Lee DeRaud wrote: > >>On Thu, 03 Jan 2002 12:22:45 -0600, Dinker Charak >>wrote: >> >>>Hi, >>> >>>What exactly are you looking for? >>> >>>> My question is if >>>> there is a master list of VxWorks routines that I can reference to see if a >>>> routine used is supplied by VxWorks or not, including hardware-specific >>>> routines. >>> >>>If you are just looking for names, build a clean kernel (without your >>>application linked) and run it. Do >>> >>>-> lkup "" >>> >>>This will give you list of _all_ symbols. That way you will have names >>>of all WRS provided symbols. >> >>Can't you just save the 'symTbl.c' file from that build? > >Using lkup and symTbl.c seem flawed to me, as they will tell you only >the symbols available for the libraries that have been included in a >particular build. One of vxWork's strengths is its configurability, >and it is unlikely that all of the available libraries have been configured >into any specific build. I think we both assumed he would do a special build with *all* of the 'INCLUDE_*' switches defined...it might take more than one pass (depending on what he's looking for), as some of the switches are incompatible. >The better answer to the OP's question would >therefore be that the BSP documentation and Programmer's Reference manual >provide such information. What is more, these documents provide not only >the names of the available functions, but the information one requires to >use them, too. I should point out that the quality and completeness of BSP documentation varies wildly between vendors. Not to mention that if the WRS documents were as good as you imply, the volume on this newsgroup would be a *lot* lower. Lee --------------------------- Newsgroups: comp.os.vxworks Subject: Re: VxWorks Routines List Date: Mon, 07 Jan 2002 07:44:40 -0800 From: "Michael R. Kesti" Organization: MK Associates Message-ID: <3C39C268.F01B7B3@gv.net> References: <3C34A175.D28A57CF@fnal.gov> <67f93ucpa1rkps39e3vlq8263gg71udrnc@4ax.com> <3C34EAD1.20E30A2E@gv.net> Reply-To: mkesti@gv.net Lee DeRaud wrote: >On Thu, 03 Jan 2002 15:35:45 -0800, "Michael R. Kesti" wrote: > >>Using lkup and symTbl.c seem flawed to me, as they will tell you only >>the symbols available for the libraries that have been included in a >>particular build. One of vxWork's strengths is its configurability, >>and it is unlikely that all of the available libraries have been configured >>into any specific build. > >I think we both assumed he would do a special build with *all* of the >'INCLUDE_*' switches defined...it might take more than one pass >(depending on what he's looking for), as some of the switches are >incompatible. This still only generate lists of function names with no documentation of their calling sequences and is therefore a waste of time, IMO. >>The better answer to the OP's question would >>therefore be that the BSP documentation and Programmer's Reference manual >>provide such information. What is more, these documents provide not only >>the names of the available functions, but the information one requires to >>use them, too. > >I should point out that the quality and completeness of BSP >documentation varies wildly between vendors. I don't doubt that. > Not to mention that if >the WRS documents were as good as you imply, the volume on this >newsgroup would be a *lot* lower. I find the Programmer's Reference to be fairly complete and accurate and that the effect to which you refer is due largely to the inadequacies of the VxWorks User's Guide. - -- ======================================================================== Michael Kesti | "And like, one and one don't make | two, one and one make one." mkesti@gv.net | - The Who, Bargain --------------------------- Newsgroups: comp.os.vxworks Subject: Re: tNetTask suspend while trying to load two END's Date: 07 Jan 2002 11:00:25 -0500 From: Bill Pringlemeir Organization: Bell Sympatico Message-ID: References: <9d60aef6.0201070255.3c105617@posting.google.com> Sender: bpringlemeir@DeadDuck >>>>> "oj" == oj writes: oj> hi ya all vxWorkers!! i have encountered a problem while trying oj> to load two ends. the first is an HDLC driver with the unit oj> id=2, the second is a software END with the unit id=1, i also oj> have an Ethernet driver wich loads at boot time with the unit oj> id=0. Your network names are motfcc0, lo0, Hdlc2, IS-Rcvr1. The maximum network name size is eight [I don't know if that includes a null]. You might try making "IS-Rcvr1" shorter. Also, you don't need to increment the unit number if the name is different (afaik). The unit number can always be zero. I think that the maximum number of END devices is eight, so you should be fine there. I have a system with two PPP, one ethernet and the loop back interface. You might wish to check over your software END (and possibly the HDLC if it is custom). Could there be some sort of interrupt conflict with the HDLC and the software driver? You will have to tell us more about the drivers before we could help. It could quite possibly be something related a conflict of 8260 resources (baud rate generators, etc). regards, Bill Pringlemeir. - -- Loaning money to your sister will bring dividends. Don't forget to see your psychologist. Listen to your chicken tomorrow. Your finances will become positive in the next several seconds. You'll often be unhappy. 12 is your lucky number. vxWorks FAQ, "http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html" --------------------------- Newsgroups: comp.os.vxworks Subject: broadcast UDP Receive and send problem Date: Mon, 07 Jan 2002 16:25:28 +0000 From: Jeremy Coker Organization: Land & Sea Systems Message-ID: <3C39CBF8.9B93CD17@bae.co.uk> Hi, the configuration I have is: mvme2400 (although I need this to work on a mvm5101; there seems to be a bug in windrivers bsp for the 5101 that causes it not to send broadcast messages). windriver BSP Problem: Whenever I send broadcast udp messages from one 2400 to another 2400, I do not appear to get any data loss on the first run of the program. However after running this program a second time I get a data loss on the very last datagram/message (any ideas why this is happening); I have found a poor solution that is not of much use to me, which is to reboot the cards and run the program again, this then sends and receives all the packets o.k. Why would this be ? I've tried all the suggestions of increasing the buffer size and so on, but this does not appear to have any affect. This happens at any time of the day (so cannot be due to network traffic etc...) Regards Jeremy --------------------------- Newsgroups: comp.os.vxworks Subject: Re: fei82559 hangs on tx Date: 7 Jan 2002 09:44:00 -0800 From: ios71@hotmail.com (Iosif Harutyunov) Organization: http://groups.google.com/ Message-ID: References: I posted recently regarding this issue. This is second try. I have problem with FEI BSD driver (82559ER on x86 platform), when network driver stop transmitting frames because of lack of free CFD buffers. Under some condition, presumably heavy load, something happens during feiTxStartup executions and as a result one of CFD buffers remains marked as busy (its status is not CFD_S_COMPLETE). After while number of "busy" CFD grows and at some point transmission stalls. It never recovers from this state even after watchdog calls, so I assuming something happening with CFD ring so CU cannot complete CFD list processing. Has anybody come across with something similar? Leonid, as it looks like from your post (see below) we have similar problem that you had with fei BSD driver. Could you please give more details about it. Thank you in advance, Iosif,_ > > > I had a similar problem with the FEI BSD driver two years ago. > > > If I remember correctly, you dont touch the transmit chain while the Tx > > is active, but you queue up the new packets in a new Command list, which > > is made active once the MAC is done with the current list. > > > > I can probably dig up my diffs, but I am not sure how applicable > > these are since I did this on teh BSD and you are uding the END drivers. > > > > - Leonid > > > "BT" wrote in message > > news:3c0cdc78.605875953@news.eunet.no... > > > We have an PPC750 with two Intel82559ER network controllers. The > > > driver is fei82559End. > > > > > > With high load on one of the networks the driver stops transmitting. > > > The other network is OK. I can use WindView etc. for debugging. > > > > > > I have found that this happens when the SCB status word indicates > > > CU status = LPQ active. If I ensures that I do NOT write anything > > > to the CFD's while the SCB status/CU status is not in suspend the > > > driver does not hang on tx anymore. > > > > > > I do not like this solution as I have to trhow away network packets > > > before they are given to the controller. And since I have free CFDs > > > I should be able to buffer up more packets. > > > > > > Has anybody come across this problem? Or even better give me a fix? > > > > > > BT --------------------------- Newsgroups: comp.os.vxworks Subject: new image building problem Date: 7 Jan 2002 10:54:28 -0800 From: dragonli@yahoo.com (June) Organization: http://groups.google.com/ Message-ID: i want to include resolvLib to the image. I modified configAll.h. and started a new project "create a bootable vxworks image" in Tornado II based on a BSP (..\IXP1200\BoardSupport\VxWorks\IXP1200EB). I got an error saying the project creation failed. Details: invalid command name "::toolFindByFile". So what is wrong then? anybody knows? thanks --------------------------- Newsgroups: comp.os.vxworks Subject: Re: broadcast UDP Receive and send problem Date: Mon, 7 Jan 2002 21:59:04 +0200 From: "Leonid Rosenboim" Organization: Verio Message-ID: References: <3C39CBF8.9B93CD17@bae.co.uk> Although this looks like a network driver bug, did you check if the loss happens in transmission or in reception ? I.e. if you hook a packet snoffer (or a tcpdump on Linux) do you see al packets go out or do you see some already lost ? "Jeremy Coker" wrote in message news:3C39CBF8.9B93CD17@bae.co.uk... > Hi, the configuration I have is: > > mvme2400 (although I need this to work on a mvm5101; there seems to > be a bug in > windrivers bsp for the 5101 that causes it not to send broadcast > messages). > > windriver BSP > > > Problem: > Whenever I send broadcast udp messages from one 2400 to another 2400, I > do not > appear to get any data loss on the first run of the program. > > However after running this program a second time I get a data loss on > the very last > datagram/message (any ideas why this is happening); > > I have found a poor solution that is not of much use to me, which is to > reboot the > cards and run the program again, this then sends and receives all the > packets o.k. > > Why would this be ? > > I've tried all the suggestions of increasing the buffer size and so on, > but this does not > appear to have any affect. > > This happens at any time of the day (so cannot be due to network traffic > > etc...) > > > Regards > > Jeremy > > > > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: vxWorks Flash File System Date: Mon, 7 Jan 2002 22:03:38 +0200 From: "Leonid Rosenboim" Organization: Verio Message-ID: References: Which Flash file system are you using ? How do you know your FFS initialization is correct, and the Flash is functional ? The boot image is a large file, did you try the FFS with smaller file ? In the example below, I dont see how the image file gets onto the RAM disk. Overall it seems you need to do some reading of the fine manuals, dude. "eche" wrote in message news:c74dde4e.0201041643.2d6a6997@posting.google.com... > I made flash file system initialized on vxWorks 5.4. > but I need some help for completing. > > I loaded vxWorks image from network via ftp method. > > I want this loaded image to be copied on flash .(Purpose) > It doesn't work. > > So I did like this. > > In bootConfig.c, > > first I made ramDrv. > > ramDrv(); > pBlkDev = ramDevCreate(0, 512, 400, 3000, 0);/* ramAddr, > bytesperblk, blkspertrack, nBlocks, blkOffset */ > dosFsMkfs("/ram", pBlkDev); > > and In bootCmdLoop, > > > case '$': /* net load ,copy it and flashLoad */ > > if (netBootLoad (pLine, &entry)==OK) > { > > flashFInit(); > copy("/ram/vxWorks","/rfa2/vxWorks"); > /* > flashFsBlkFlush2(0); > flashFsBlkWrite2(pBlkDev,0, 32, (char *)&entry); > */ > > printf("copied from ram to flash!\n"); > } > else > { > printf("net boot load failed!\n"); > } > > > if (flashLoad ("/rfa2/vxWorks", &entry) == OK) > { > printf ("entry = 0x%x\n", (int) entry); > /*go(entry);*/ > } > else > { > printf("flashLoad failed!\n"); > > taskDelay (sysClkRateGet ()); /* pause a second */ > /*reboot (BOOT_NO_AUTOBOOT);*/ /* something is awry */ > } > break; > > > Regards, > > eche --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Symbol table errors compiling an image when using STL Date: 7 Jan 2002 13:46:35 -0800 From: john_94501@yahoo.com (John) Organization: http://groups.google.com/ Message-ID: <488e459a.0201071346.16dad6f2@posting.google.com> References: <3c29bbec$1@news.barak.net.il> <3c3833dd$1@news.barak.net.il> Hello, Your problem is probably the use of namespaces - I don't believe that they are supported by any version of Tornado/VxWorks at the moment. HTH, John... "Arnon W" wrote in message news:<3c3833dd$1@news.barak.net.il>... > Hi, > > Didn't work. > I added only these 3 lines to one of my cpp's: > > #include (or #include ) > using namespace std; > list test; > > The funny thing is that the module containing the STL part compiles with no > problems, > and the code runs well as a downloadable module. > The minute I try to integrate it into the image - boom! > > > Arnon > > "Dave Korn" wrote in message > news:W4XY7.206$qS6.464044@newsr2.u-net.net... > > Arnon W wrote in message <3c29bbec$1@news.barak.net.il>... > > >Hi, > > > > > >I'm new to this environment so it might be a minor issue I'm having > problems > > >with. > > >(On the other hand - I couldn't find earlyer posts in this specific > > matter:) > > > > >When I try to integrate it to the image, I get vicious responses at this > > >point: > > > > > >ccppc -c -fdollars-in-identifiers -O3 -mstrict-align -nostdinc -DRW_MULTI_T > > H > > >READ -D_REENTR > > > >ANT -fvolatile -fno-builtin -fno-for-scope -DCPU=PPC604 -DPRJ_BUILD -DHAL_L > > I > > >B > > >...paths... symTbl.c -o symTbl.o > > >symTbl.c:2013: syntax error before `.120' > > >symTbl.c:2014: syntax error before `.139' > > >symTbl.c:2015: syntax error before `.148' > > >symTbl.c:8448: > `__sn_t24__default_alloc_template2b1i0$__node_allocator_lock' > > >undeclared he > > >re (not in a function) > > >symTbl.c:8448: syntax error before `.120' > > >symTbl.c:8449: initializer element for `standTbl[1997].value' is not > > >constant > > >symTbl.c:8449: warning: initialization from incompatible pointer type > > >symTbl.c:8449: > `__sn_t24__default_alloc_template2b1i0$__node_allocator_lock' > > >undeclared he > > >re (not in a function) > > >symTbl.c:8449: syntax error before `.139' > > >symTbl.c:8450: initializer element for `standTbl[1998].name' is not > constant > > >symTbl.c:8450: syntax error before string constant > > > >When I took out STL related stuff (I was using ) It compiled well. > > >Is there a cure? > > > > I'm just passing on something I read somewhere else here, STL is really > > not my field, but ISTR someone saying that to use STL you should only > > #include - i.e. you use that header and it includes all the other > > stl headers like and for you. You might like to try this > > approach and see if it fixes your problem. > > > > DaveK > > -- > > Burn your ID card! http://www.optional-identity.org.uk/ > > Help support the campaign, copy this into your .sig! > > Proud Member of the Exclusive "I have been plonked by Davee because he > > thinks I'm interesting" List > > Member # > > > > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: new image building problem Date: 7 Jan 2002 13:52:09 -0800 From: john_94501@yahoo.com (John) Organization: http://groups.google.com/ Message-ID: <488e459a.0201071352.79519543@posting.google.com> References: Hello, Make sure you have added values to the following parameters: RESOLVER_DOMAIN RESOLVER_DOMAIN_SERVER They are not given defaults by the project facility (after all, what would they default to?), so you need to define them. Unfortunately, it seems that the project facility does not give a useful error in this case. HTH, John... dragonli@yahoo.com (June) wrote in message news:... > i want to include resolvLib to the image. I modified configAll.h. and > started a new project "create a bootable vxworks image" in Tornado II > based on a BSP (..\IXP1200\BoardSupport\VxWorks\IXP1200EB). I got an > error saying the project creation failed. Details: invalid command > name "::toolFindByFile". So what is wrong then? anybody knows? thanks --------------------------- Newsgroups: comp.os.vxworks Subject: Does Tornado require write access to download links/libraries? Date: 7 Jan 2002 14:16:03 -0800 From: matthew.becker@usa.xerox.com (Matthew Becker) Organization: http://groups.google.com/ Message-ID: <94e55b6.0201071416.23f691ff@posting.google.com> Hello, I've noticed that if I don't have write access to a library/link, that I am unable to download it to the target. I select "Project->Download" from the Tornado main menu. I then select the appropriate file from the "Download objects" GUI. I then receive the following message, > F:\Runtime.Phoenix8260\MainApp\obj.iphoenixPlus\appiphoenixPlus > This file exists with Read Only attributes. > Please use a different file name. I have read-execute access to this file, but not write access. Does Tornado require write access in order to download links/libraries (I am using Tornado 2.0.2)? If so, then could someone please explain why this is the case. Thanks in advance. - -Matt - - Matthew Becker Xerox Corporation matthew.becker@usa.xerox.com --------------------------- Newsgroups: comp.os.vxworks Subject: why does taskDelay(1) take so long? Date: 7 Jan 2002 14:18:03 -0800 From: eaglejump@yahoo.com (Bob Johnson) Organization: http://groups.google.com/ Message-ID: <67868445.0201071418.5c211c82@posting.google.com> For our system, we use a high system clock rate of approximately 2000 ticks per second (shown by sysClkRateGet). If I call tickGet() at random times, I get numbers with no specific pattern. However, if I call taskDelay(1), and then tickGet, I ALWAYS get a value which a is a multiple of 20. It seems that the taskDelay is taking much longer than 1 tick. I tried running my test at priority 3 and priority 100 and neither priority made a difference. Any ideas? Thanks! --------------------------- Newsgroups: comp.os.vxworks Subject: Re: 0xd003 status error displayed by bootrom Date: 7 Jan 2002 14:20:32 -0800 From: john_94501@yahoo.com (John) Organization: http://groups.google.com/ Message-ID: <488e459a.0201071420.338716fa@posting.google.com> References: <488e459a.0201051250.f7fd0fd@posting.google.com> Hello, > Yes, I checked all these. It was to rule out all these problems that I > repeated the same experiment with a 3COM PCI ethernet card in the same > board under the same environment. I let the subnet to be determined > automatically from the address. All the boot parameters, like the IP > Address, the subnet, the FTP server's address, user name, password > were all the same. Only the device name changed, from elPci (for 3COM) > to ene (for the ne2000 card). The 3COM card works like a charm, > downloads the file and runs the image without any problems. OK, so that at least points firmly at the card. Next... 1) Does the card work in a Windows box (or other OS)? That will help check that the card itself is correct, and that the jumpers are all set correctly for the network you're connecting to. 2) Is there a light on the hub/switch you're connecting to that will show that the link is live? 3) Have you managed to verify the interrupt and address range settings for the card? Not transmitting anything could be be caused by these being incorrect. Also, check for packets from the card's MAC (ethernet) address, rather than from its IP address. I suspect that you will get one packet, probably an ARP, when the board is switched on. If that is the case, check the IRQ settings very carefully. > I am really out of clues as to what could be it. Next thing to try, since you have a card that lets you boot, is to add both cards to the system (and both drivers to the kernel). Make the 3COM card your boot device, and then try shut down the 3COM card and manually bring up the NE2000 from the target based shell (I suggest the target shell since that will avoid need for a network connection to use the tools - you could set it up for serial WDB if you prefer the host based tools). Some functions you might find useful here: a) To detach the 3COM, ipDetach (0, "elPci") b) To attach the device to the IP stack, ipAttach (1, "ene") [this might need to be (0, "ene") - try it if 1 doesn't work] c) To determine the existence of a device/unit pair, say ene0, use ifunit("ene0") d) Don't know how useful this will be, but there is an eneShow() routine you could try. HTH, John... > Thanks for your time. It helped. > > Regards > Ramki --------------------------- Newsgroups: comp.os.vxworks Subject: Re: I/O ports on PowerPC Date: Mon, 7 Jan 2002 23:51:54 +0100 From: "JJS" Message-ID: References: Arun Prasad a écrit dans le message : e38d912c.0201032322.2998a475@posting.google.com... > Hi, > > I have a question on accessing I/O ports. > I need to read/write from/to the network card registers on PowerPC > platform. > > Can anyone tell me the functions provided by VxWorks to read/write > Registers directly on PowerPC platforms? > > Earlier response would be of great help. > > Thanks in advance. > Regards > Arun Have you tried this : char * ptreg = (char*) 0xXXXX (adress of a byte reg) *ptreg = val; Jack --------------------------- Newsgroups: comp.os.vxworks Subject: readyQHead - how to read it? Date: 7 Jan 2002 15:52:23 -0800 From: mynamvs@nsc.network.com (Venkat Mynampati) Organization: http://groups.google.com/ Message-ID: <65b2abfc.0201071552.49376cb@posting.google.com> Hi, I want to go thru readyQHead and find out what tasks are there in readyQ, can you pls help me out here? in kernelLibP.h readyQHead is defined as Q_HEAD readyQHead; and in qLib.h i found that struct Q_HEAD { Q_NODE *pFirstNode; ... ... } and struct Q_NODE { UINT qPriv1;UINT qPriv2; ...; ... } ; My doubt is how to read ReadyQ, by using readyQHead..i mean i can get pFirstNode, but what this pFirstNode contains, i mean what this Q_NODE contains? address of the a TCB? in qLib.h, there's qInfo() function, what it does? Thanks Venkat --------------------------- Newsgroups: comp.os.vxworks Subject: Re: VxWorks Routines List Date: Mon, 07 Jan 2002 18:02:31 -0600 From: Dinker Charak Organization: Fermi National Accelerator Laboratory Message-ID: <3C3A3717.9BECBC6D@fnal.gov> References: <3C34A175.D28A57CF@fnal.gov> <67f93ucpa1rkps39e3vlq8263gg71udrnc@4ax.com> <3C34EAD1.20E30A2E@gv.net> <3C39C268.F01B7B3@gv.net> > >>Using lkup and symTbl.c seem flawed to me, as they will tell you only > >>the symbols available for the libraries that have been included in a > >>particular build. One of vxWork's strengths is its configurability, > >>and it is unlikely that all of the available libraries have been configured > >>into any specific build. > > > >I think we both assumed he would do a special build with *all* of the > >'INCLUDE_*' switches defined...it might take more than one pass > >(depending on what he's looking for), as some of the switches are > >incompatible. > > This still only generate lists of function names with no documentation of > their calling sequences and is therefore a waste of time, IMO. > >>>My question is >>>if >>>there is a master list of VxWorks routines that I can reference to see if >>>a >>>routine used is supplied by VxWorks or not, including hardware-specific >>>routines. From the original query it _did_ seem he wanted a list of names of routines provided by WRS and to which I guess we all responded and the suggested methods works well. Guess, Lee Cao need to come back and tell us more. In any case, what he is trying to do is interesting and it will be worthwile watching him solve the problem. Regards, Dinker --------------------------- Newsgroups: comp.os.vxworks Subject: Target Shell does not support multi-users... Date: Tue, 08 Jan 2002 01:19:32 GMT From: "Jae-Hwa Lee" Organization: Dacom Message-ID: Dear everyone. Is it possible that multil users access a target shell simultaneously? For example, I want to access the target system via both a system console and a telnet. =========================================================== Jay Lee --------------------------- Newsgroups: comp.os.vxworks Subject: serial port initialization when serial device is missing Date: Tue, 08 Jan 2002 02:05:56 GMT From: "Kelly Hornsby" Organization: Excite@Home - The Leader in Broadband http://home.com/faster Message-ID: <8us_7.31904$LQ1.10302865@news2.nash1.tn.home.com> Hello All I have a PowerPC design that uses a PMC-based serial device that has as NS16552 on it. I added a serial driver to the bsp (my card has a PMC genesis). the serial device works great, and my application runs well with it. I have a need to remove the serial PMC card when the PowerPC card is actually installed in its holder. However, my application acts very weird. What are the steps to make VxWorks not "see" a serial device so that printf and logMsgs go into the bit bucket. I realize that I can redirect processing flow based on the pci signature of the card, and I have all that, but I need a way to redirect the serial outputs to where the OS thinks the serial output is being performed. Any ideas? Anythig is welcomed. --------------------------- Newsgroups: comp.os.vxworks Subject: LC82C169 Driver... Date: 7 Jan 2002 20:23:35 -0800 From: hemu_u@hotmail.com (Hemant) Organization: http://groups.google.com/ Message-ID: <21595172.0201072023.67f7fc78@posting.google.com> Hi All, I am supposed to write a network driver for NETGEAR card with LC82C169 (FA310Tx) for IDT Evaluation Board. But I don't know how to start as there is no BSP template for the above driver. I am trying to use the dec21x40End.c but it is not working at all. Can anybody suggest me about the same? Thanks for the help which you may provide, Hemant. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: new image building problem Date: 7 Jan 2002 20:45:54 -0800 From: dragonli@yahoo.com (June) Organization: http://groups.google.com/ Message-ID: References: <488e459a.0201071352.79519543@posting.google.com> yes, i added values to those parameters. the problem is i even can not create the project. john_94501@yahoo.com (John) wrote in message news:<488e459a.0201071352.79519543@posting.google.com>... > Hello, > > Make sure you have added values to the following parameters: > > RESOLVER_DOMAIN > RESOLVER_DOMAIN_SERVER > > They are not given defaults by the project facility (after all, what > would they default to?), so you need to define them. Unfortunately, it > seems that the project facility does not give a useful error in this > case. > > HTH, > > John... > --------------------------- Newsgroups: comp.os.vxworks Subject: Sun RPC or microsoft RPC in VxDCOM ? Date: 7 Jan 2002 20:46:34 -0800 From: ishwara_a@yahoo.com (Ishwar) Organization: http://groups.google.com/ Message-ID: <5286f56c.0201072046.ed8d7a1@posting.google.com> Hi, I have some confusion over the RPC implemented in VxWorks. VxWorks documentation says that it uses sun RPC. But WRS product VxDCOM's brochures say that VxDCOM can work with NT workstations seamlessly. i.e. it should be having microsoft RPC. I understand that there are some slight differences between these two RPC implementations. When we have VxDCOM along with traditional VxWorks, do we have choice of RPC implementations ??? Thanks, Ishwar --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Target Shell does not support multi-users... Date: Tue, 08 Jan 2002 12:16:19 +0530 From: Pavan Kumar T V Organization: Bell Labs Development Centre, Lucent Technologies Message-ID: <3C3A95BB.6A9EB6B0@lucent.com> References: Jay Lee, The shell is not reentrant, since yacc does not generate a reentrant parser. Therefore, there can be only a single shell executing at one time Hope this answers your question. - - pavan Jae-Hwa Lee wrote: > > Dear everyone. > > Is it possible that multil users access a target shell simultaneously? > For example, I want to access the target system via both a system console > and a telnet. > > =========================================================== > Jay Lee --------------------------- Newsgroups: comp.os.vxworks Subject: Caching, I/O Space memory mapping and DMA Date: 7 Jan 2002 23:17:37 -0800 From: arun4vxworks@indiatimes.com (Arun Prasad) Organization: http://groups.google.com/ Message-ID: Hi, I'm having some serious of doubts on Caching, I/O Space memory mapping and DMA data transfer on IBM PPC405 platform. Can anyone clarify these doubts? Do we need to invalidate the cache irrespective of our Card is whether memory mapped or i/o mapped while accessing device registers? And one more doubt. Whether the memory space which we are getting after configuring (pciAutoConfig) the PCI card (for memory mapped devices ) for I/O space is a cacheable or non-cacheable one? I think, it will not be a cacheable one as this memory is shared by the Host CPU as well as by the Card. Please do clarify me on this regard. If we directly access the device registers (sysInbyte() sysOutByte()), do we need to invalidate the cache as we are accessing HW registers? And when we need to do cache flush? (within ISR??? or in some other places?) One more thing, Do we need to take care of Cache validate/invalidate if we do not handle DMA data transfer??? Thanks in advance and have a nice time. Regards Arun Note : I'm writing driver for a ethernet card on IBM PPC405 --------------------------- Newsgroups: comp.os.vxworks Subject: any other debug method? Date: 7 Jan 2002 23:52:19 -0800 From: feng_w@hotmail.com (wang feng) Organization: http://groups.google.com/ Message-ID: <6ccda887.0201072352.64f62e3e@posting.google.com> hi, friends, we have softswitch project on ppc system, and control system are hot-backup with two cards, problem is they use same ip, the master use it to handle outgoing message and the slave only can recieve info from master. we use nt2000 as the host to debug the master through local network, but the slave cann't be attched. how to debug the slave through other ways just like com port or paralel port? thanks in advance Wangfeng --------------------------- Newsgroups: comp.os.vxworks Subject: What version of NPT am I using now ? Date: 8 Jan 2002 00:10:58 -0800 From: houcheng@ms17.hinet.net (houcheng) Organization: http://groups.google.com/ Message-ID: My Tornado version is Tornado 2.02 PowerPC (AP) Tornado 2.02 PowerPC ads860 (BSP) I encounter some errors when compiling my NPT template driver, but this driver works fine in Tornado Prototyper Plus. The Prototyper Plus is using NPT version 4. Who can tell me what version of NPT is inside the Tornado 2.02? Or Tonado 2.02 does not contain NPT? I download the NPT source from WinSurf, and installing on my host, but still doesn't work! Plese help me, Thanks in advance. houcheng@turbonetworks.com_NO_SPAM. Lin Houcheng --------------------------- Newsgroups: comp.os.vxworks Subject: Re: readyQHead - how to read it? Date: Tue, 08 Jan 2002 08:57:08 GMT From: Ephraim Gadsby Organization: - minimal Message-ID: References: <65b2abfc.0201071552.49376cb@posting.google.com> On 7 Jan 2002 15:52:23 -0800, mynamvs@nsc.network.com (Venkat Mynampati) wrote: >Hi, >I want to go thru readyQHead and find out what tasks >are there in readyQ, can you pls help me out here? > >in kernelLibP.h readyQHead is defined as > >Q_HEAD readyQHead; > >and in qLib.h i found that >struct Q_HEAD { > Q_NODE *pFirstNode; > ... > ... >} >and > struct Q_NODE > { > UINT qPriv1;UINT qPriv2; > ...; ... > } ; > >My doubt is how to read ReadyQ, by using readyQHead..i mean i can >get pFirstNode, but what this pFirstNode contains, i mean what this >Q_NODE contains? address of the a TCB? Whatever it is you are try to do, it is probably a bad idea. You might be better-off, posting your problem and asking for advice. If you are just trying to understand how the ready queue works then take a look at how it is used, following through taskDelay(0) is quite instructive, and fairly straightforward. --------------------------- Newsgroups: comp.os.vxworks Subject: powerQUICK-II evaluation boards question Date: 8 Jan 2002 02:21:33 -0800 From: ilkras@yahoo.com (Ilya Krasavin) Organization: http://groups.google.com/ Message-ID: <9ca84f0f.0201080221.1fb27112@posting.google.com> Hi gurus. May anybody to recommnd me the evaluation board for powerQUICK-II. I found 3 (ACTIS, IPnexus and ADS), but I don't know which one is better. Thank you for a help. Ilya --------------------------- Newsgroups: comp.os.vxworks Subject: Re: why does taskDelay(1) take so long? Date: 8 Jan 2002 02:32:16 -0800 From: ilkras@yahoo.com (Ilya Krasavin) Organization: http://groups.google.com/ Message-ID: <9ca84f0f.0201080232.e09498d@posting.google.com> References: <67868445.0201071418.5c211c82@posting.google.com> eaglejump@yahoo.com (Bob Johnson) wrote in message news:<67868445.0201071418.5c211c82@posting.google.com>... > For our system, we use a high system clock rate of approximately 2000 > ticks per second (shown by sysClkRateGet). If I call tickGet() at > random times, I get numbers with no specific pattern. > > However, if I call taskDelay(1), and then tickGet, I ALWAYS get a > value which a is a multiple of 20. It seems that the taskDelay is > taking much longer than 1 tick. > > I tried running my test at priority 3 and priority 100 and neither > priority made a difference. > > Any ideas? > Thanks! May be the timeslice that your scheduler uses is multiple of the 20 ms. Every time you call to the taskDelay() - you actually calls your scheduler and signal it that another task may be run now, so scheduler allocates the timeslice to it. After the timeout of original task is expired it will get control immidiately only if it runs on highest priority among other tasks. Try to rise priority of your task. Ilya --------------------------- Newsgroups: comp.os.vxworks Subject: Optimizing VxWorks boot time ? Date: Tue, 8 Jan 2002 11:54:32 +0100 From: "Søren Abildgaard" Organization: TDC Internet Message-ID: <3c3acff3$0$5452$edfadb0f@dspool01.news.tele.dk> Reply-To: "Søren Abildgaard" Hi *.*, We are running a project, where we are using a x86 platform and a vxworks 5.2. I am fairly new to VxWorks, but my problem is the boottime we are getting. It takes about 13 seconds from the BIOS POST ends until VXworks starts to load our application. The kernel I have built I have left out all components, which we don´t need. What parameters do I have left to tweak on? And whats dertermines this boottime, which seems fairly long for an embedded system? Yours sincerly Søren Abildgaard Terma A/S Dk Denmark --------------------------- Newsgroups: comp.os.vxworks Subject: Re: any other debug method? Date: Tue, 8 Jan 2002 11:57:42 +0100 From: "Werner Schiendl" Message-ID: <3c3ad0a9@brateggebdc5.br-automation.co.at> References: <6ccda887.0201072352.64f62e3e@posting.google.com> Hi, Select the proper WDB target agent in the target server configuration and configure the bootable VxWorks image with the matching WDB target agent. Be sure to also verify the target agent's parameters like baudrate and interface to use. hth Werner "wang feng" wrote in message news:6ccda887.0201072352.64f62e3e@posting.google.com... > hi, friends, > we have softswitch project on ppc system, and control system are > hot-backup with two cards, problem is they use same ip, the master use > it to handle outgoing message and the slave only can recieve info > from master. we use nt2000 as the host to debug the master through > local network, but the slave cann't be attched. how to debug the slave > through other ways just like com port or paralel port? > > thanks in advance > > Wangfeng --------------------------- End of New-News digest ********************** From vxwexplo-errs@csg.lbl.gov Tue Jan 8 04:57:04 2002 From: ulf.fiederling@dornier.eads.net Date: Tue Jan 8 04:57:06 PST 2002 Subject: Hi We are using Tornado II on a Pentium platform with the UGL graphics package. Due to our very small monitor we need a font size greater than 24 (up to 48). The graphics package is only delivered with fonts up to 24. Is ther a solution to get / include bigger fonts in our BSP configuration. Best Regards Ulf Fiederling From vxwexplo-errs@csg.lbl.gov Tue Jan 8 05:29:17 2002 From: lig Date: Tue Jan 8 05:29:19 PST 2002 Subject: about IP forword... Hi,all i'm new to vxworks, and i want to know what need to do to let vxworks forwarding IP packets? i remember there is a macro looks likes "IP_FORWARDING" which need to be seted . where i can find it? thank in front! BR stonefish From vxwexplo-errs@csg.lbl.gov Wed Jan 9 04:03:19 2002 From: Vxworks Exploder Date: Wed Jan 9 04:03:21 PST 2002 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Wed Jan 9 04:03:14 PST 2002 Subject: Re: why does taskDelay(1) take so long? Subject: Re: Optimizing VxWorks boot time ? Subject: Re: Optimizing VxWorks boot time ? Subject: Re: why does taskDelay(1) take so long? Subject: shared memory network question Subject: Re: tNetTask suspend while trying to load two END's Subject: Re: why does taskDelay(1) take so long? Subject: Re: Caching, I/O Space memory mapping and DMA Subject: Re: why does taskDelay(1) take so long? Subject: Re: readyQHead - how to read it? Subject: Re: shared memory network question Subject: Re: serial port initialization when serial device is missing Subject: VME bus interrupts Subject: Re: VME bus interrupts Subject: Re: VME bus interrupts Subject: BootRom and MMU Subject: Re: why does taskDelay(1) take so long? Subject: Re: any other debug method? Subject: Re: why does taskDelay(1) take so long? Subject: Re: why does taskDelay(1) take so long? Subject: Re: memcpy problem Subject: Re: readyQHead - how to read it? Subject: Re: VME bus interrupts Subject: bus error exception in ppc8260 Subject: Include Paths Subject: Re: bus error exception in ppc8260 Subject: Re: bus error exception in ppc8260 Subject: Re: Optimizing VxWorks boot time ? Subject: Re: Include Paths Subject: Re: bus error exception in ppc8260 Subject: Bug Report? Loading more than one END ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: why does taskDelay(1) take so long? Date: Tue, 8 Jan 2002 13:23:23 +0100 From: "Martin Raabe" Organization: T-Online Message-ID: References: <67868445.0201071418.5c211c82@posting.google.com> Hello Bob, taskDelay(0) is giving up the time slice, taskDelay(1) is waiting one or two ticks! After that the task gets to the state ready again and it is up to the priority scheme, when the task next gets the CPU! Ciao Martin Raabe "Bob Johnson" schrieb im Newsbeitrag news:67868445.0201071418.5c211c82@posting.google.com... > For our system, we use a high system clock rate of approximately 2000 > ticks per second (shown by sysClkRateGet). If I call tickGet() at > random times, I get numbers with no specific pattern. > > However, if I call taskDelay(1), and then tickGet, I ALWAYS get a > value which a is a multiple of 20. It seems that the taskDelay is > taking much longer than 1 tick. > > I tried running my test at priority 3 and priority 100 and neither > priority made a difference. > > Any ideas? > Thanks! --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Optimizing VxWorks boot time ? Date: Tue, 8 Jan 2002 12:54:35 -0000 From: "Martin Dowie" Message-ID: <3c3aebda$1@pull.gecm.com> References: <3c3acff3$0$5452$edfadb0f@dspool01.news.tele.dk> "Søren Abildgaard" wrote in message news:3c3acff3$0$5452$edfadb0f@dspool01.news.tele.dk... > Hi *.*, > > We are running a project, where we are using a x86 platform > and a vxworks 5.2. > > I am fairly new to VxWorks, but my problem is the boottime we > are getting. It takes about 13 seconds from the BIOS POST ends > until VXworks starts to load our application. > The kernel I have built I have left out all components, which we > don´t need. > > What parameters do I have left to tweak on? And whats dertermines this > boottime, which seems fairly long for an embedded system? How much memory have you told VxWorks you have and how does that compare with how much you actually use? We have found that reducing the amount you tell VxWorks you have can make a considerable difference to the time. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Optimizing VxWorks boot time ? Date: Tue, 08 Jan 2002 12:59:11 +0000 From: Graham Baxter Organization: Graham Baxter (Software) Limited Message-ID: <3C3AED1F.E28D2EE4@NOSPAM.bcs.org.uk> References: <3c3acff3$0$5452$edfadb0f@dspool01.news.tele.dk> Soren, My BSP is PowerPC but this could apply to X86 too. VxWorks copies the ROM image into RAM and then fills the RAM before and after the image with zeros (bootInit.c). When there is a lot of RAM this filling can take a long time. Regards, Graham Baxter (VxWorks and pSOS BSP'S) Freelance Software Engineer - AVAILABLE FOR NEW ASSIGNMENT SOON gbaxter@NOSPAM.bcs.org.uk --------------------------- Newsgroups: comp.os.vxworks Subject: Re: why does taskDelay(1) take so long? Date: 8 Jan 2002 06:57:00 -0800 From: eaglejump@yahoo.com (Bob Johnson) Organization: http://groups.google.com/ Message-ID: <67868445.0201080656.7edd0fe7@posting.google.com> References: <67868445.0201071418.5c211c82@posting.google.com> <9ca84f0f.0201080232.e09498d@posting.google.com> ilkras@yahoo.com (Ilya Krasavin) wrote in message news:<9ca84f0f.0201080232.e09498d@posting.google.com>... > eaglejump@yahoo.com (Bob Johnson) wrote in message news:<67868445.0201071418.5c211c82@posting.google.com>... > > For our system, we use a high system clock rate of approximately 2000 > > ticks per second (shown by sysClkRateGet). If I call tickGet() at > > random times, I get numbers with no specific pattern. > > > > However, if I call taskDelay(1), and then tickGet, I ALWAYS get a > > value which a is a multiple of 20. It seems that the taskDelay is > > taking much longer than 1 tick. > > > > I tried running my test at priority 3 and priority 100 and neither > > priority made a difference. > > > > Any ideas? > > Thanks! > > May be the timeslice that your scheduler uses is multiple of the 20 > ms. Every time you call to the taskDelay() - you actually calls your > scheduler and signal it that another task may be run now, so scheduler > allocates the timeslice to it. After the timeout of original task is > expired it will get control immidiately only if it runs on highest > priority among other tasks. > Try to rise priority of your task. > Ilya Thanks for the suggestions. Unfortunately.. To test how often the scheduler runs, I used the function taskSwitchHookAdd. This will call a function during every task switch. In that function, I incremented a counter. During every taskDelay(1), the counter increments 20-30 times. That means that task switching is occurring as expected. As far as the priority is concerned, I am already running at priority 3, and that doesn't help. I increased the priority to 1, just to see what would happen. That didn't make a difference either. Any more suggestions? --------------------------- Newsgroups: comp.os.vxworks Subject: shared memory network question Date: Tue, 08 Jan 2002 14:51:42 +0000 From: Support Engineer 4 Organization: Spider Software Limited Message-ID: <3C3B077D.E4BEA62A@spider.com> Hi, Anyone seen this error message before? What does it mean? Backplane anchor at 0x4100... Attaching network interface sm0... done. Error setting inet address of sm0 to 194.217.109.171, errno = 0x11 Thanks, Roland. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: tNetTask suspend while trying to load two END's Date: 8 Jan 2002 07:06:32 -0800 From: akarpov@my-deja.com (Alex) Organization: http://groups.google.com/ Message-ID: <4f09bb68.0201080706.74c79d7e@posting.google.com> References: <9d60aef6.0201070255.3c105617@posting.google.com> Oded, I was wondering if disabling caches would improve the situation. Alex odedn@shiron.com (oj) wrote in message news:<9d60aef6.0201070255.3c105617@posting.google.com>... > hi ya all vxWorkers!! > > i have encountered a problem while trying to load two ends. > ..... > Exception next instruction address: 0x0007c6f0 > Machine Status Register: 0x00049000 > Condition Register: 0x44020040 > Task: 0x3db9928 "tNetTask" > ...... > some more useful information: > 1. when loading each driver alone no broblems occur. > 2. to initialize the mBlk memarea i use memaligne. > 3. to initialize the cBlk memarea i use cacheDmaMalloc. > 4. i have separate net pools for each driver. > 5. i use the MPC8260 on tornado 2.0 > > youre help will be most blessed > thanx. VxWorks --------------------------- Newsgroups: comp.os.vxworks Subject: Re: why does taskDelay(1) take so long? Date: 8 Jan 2002 15:27:47 GMT From: Johan Borkhuis Organization: Agere Systems Message-ID: References: <67868445.0201071418.5c211c82@posting.google.com> <9ca84f0f.0201080232.e09498d@posting.google.com> <67868445.0201080656.7edd0fe7@posting.google.com> eaglejump@yahoo.com (Bob Johnson) wrote: > To test how often the scheduler runs, I used the function > taskSwitchHookAdd. This will call a function during every task > switch. In that function, I incremented a counter. During every > taskDelay(1), the counter increments 20-30 times. That means that > task switching is occurring as expected. Can you also add a counter to sysClkInt, and see if this routine is called? Groeten, Johan - -- o o o o o o o . . . _____________________________ o _____ || Johan Borkhuis | .][__n_n_|DD[ ====_____ | borkhuis@agere.com | >(________|__|_[_________]_|__________________________| _/oo OOOOO oo` ooo ooo 'o!o!o o!o!o` === VxWorks FAQ: http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html === --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Caching, I/O Space memory mapping and DMA Date: Tue, 8 Jan 2002 17:24:19 +0100 From: "Michael Lawnick" Organization: Buergernetz Dillingen Message-ID: References: Reply-To: "Michael Lawnick" Sender: mlawnick@p5088f893.dip.t-dialin.net "Arun Prasad" schrieb im Newsbeitrag news:e38d912c.0201072317.7b0bef7b@posting.google.com... > Hi, > > I'm having some serious of doubts on Caching, I/O Space memory > mapping and DMA data transfer on IBM PPC405 platform. Can anyone > clarify these doubts? > > Do we need to invalidate the cache irrespective of our Card is > whether memory mapped or i/o mapped while accessing device registers? > > And one more doubt. > Whether the memory space which we are getting after configuring > (pciAutoConfig) the PCI card (for memory mapped devices ) for I/O > space is a cacheable or non-cacheable one? > I think, it will not be a cacheable one as this memory is shared by > the Host > CPU as well as by the Card. Please do clarify me on this regard. > > If we directly access the device registers (sysInbyte() > sysOutByte()), do we > need to invalidate the cache as we are accessing HW registers? > > And when we need to do cache flush? (within ISR??? or in some other > places?) > One more thing, > Do we need to take care of Cache validate/invalidate if we > do not handle DMA data transfer??? > > Thanks in advance and have a nice time. > > Regards > Arun > Note : I'm writing driver for a ethernet card on IBM PPC405 Hi Arun, Caching is independand from I/O- or memory mapped PCI space. If you haven't set an address space as uncachable that is used by two masters, you need flush/invalidate. Nevertheless I recommend to set all PCI-space to uncachable. For CPU-owned memory, accesses by external devices (aka DMA) there are two ways: - - enable MMU and do a cacheDmaMalloc. This sets the allocated memory uncachable. - - do cacheFlush before notifying the device of buffer changes (send) and a cacheInvalidate before access after notification (aka txInterrupt) by device. CAVE AT: buffers may not overlap in cache on concurrent accesses . If your NIC has local memory, where it buffers data and CPU must actively fetch it, you shouldn't have any cache troubles when disabling cache for PCI address space of your NIC. HTH - -- Mit freundlichen Grüßen, Michael Lawnick ============================================== SOFTEC GmbH Tel +49-731-96600-0 Promenade 17 Fax +49-731-96600-23 D-89073 Ulm Michael Lawnick Germany lawnick@softec.de ============================================== --------------------------- Newsgroups: comp.os.vxworks Subject: Re: why does taskDelay(1) take so long? Date: Tue, 8 Jan 2002 16:29:12 +0000 (UTC) From: don@mars.dgrc.crc.ca (Donald McLachlan) Organization: The unconfigured xvnews people Message-ID: References: <67868445.0201080656.7edd0fe7@posting.google.com> Reply-To: don@mars.dgrc.crc.ca Does your system have a slow ISR somewhere? Does your system have a slow peice of code inside a taskLock/unlock() or intLock/unlock() block? In article <67868445.0201080656.7edd0fe7@posting.google.com>, eaglejump@yahoo.com (Bob Johnson) writes: > ilkras@yahoo.com (Ilya Krasavin) wrote in message news:<9ca84f0f.0201080232.e09498d@posting.google.com>... > > eaglejump@yahoo.com (Bob Johnson) wrote in message news:<67868445.0201071418.5c211c82@posting.google.com>... > > > For our system, we use a high system clock rate of approximately 2000 > > > ticks per second (shown by sysClkRateGet). If I call tickGet() at > > > random times, I get numbers with no specific pattern. > > > > > > However, if I call taskDelay(1), and then tickGet, I ALWAYS get a > > > value which a is a multiple of 20. It seems that the taskDelay is > > > taking much longer than 1 tick. > > > > > > I tried running my test at priority 3 and priority 100 and neither > > > priority made a difference. > > > > > > Any ideas? > > > Thanks! > > > > May be the timeslice that your scheduler uses is multiple of the 20 > > ms. Every time you call to the taskDelay() - you actually calls your > > scheduler and signal it that another task may be run now, so scheduler > > allocates the timeslice to it. After the timeout of original task is > > expired it will get control immidiately only if it runs on highest > > priority among other tasks. > > Try to rise priority of your task. > > Ilya > > Thanks for the suggestions. Unfortunately.. > > To test how often the scheduler runs, I used the function > taskSwitchHookAdd. This will call a function during every task > switch. In that function, I incremented a counter. During every > taskDelay(1), the counter increments 20-30 times. That means that > task switching is occurring as expected. > > As far as the priority is concerned, I am already running at priority > 3, and that doesn't help. I increased the priority to 1, just to see > what would happen. That didn't make a difference either. > > Any more suggestions? - -- Donald McLachlan E-mail Donald.McLachlan@crc.ca Communications Research Centre / RNS Tel (613) 998-2845 3701 Carling Ave., Fax (613) 998-9648 Ottawa, Ontario K2H 8S2 Canada --------------------------- Newsgroups: comp.os.vxworks Subject: Re: readyQHead - how to read it? Date: 8 Jan 2002 08:31:34 -0800 From: mynamvs@nsc.network.com (Venkat Mynampati) Organization: http://groups.google.com/ Message-ID: <65b2abfc.0201080831.6404a2d5@posting.google.com> References: <65b2abfc.0201071552.49376cb@posting.google.com> Ephraim Gadsby wrote in message news:... > On 7 Jan 2002 15:52:23 -0800, mynamvs@nsc.network.com (Venkat > Mynampati) wrote: > > >Hi, > >I want to go thru readyQHead and find out what tasks > >are there in readyQ, can you pls help me out here? > > > >in kernelLibP.h readyQHead is defined as > > > >Q_HEAD readyQHead; > > > >and in qLib.h i found that > >struct Q_HEAD { > > Q_NODE *pFirstNode; > > ... > > ... > >} > >and > > struct Q_NODE > > { > > UINT qPriv1;UINT qPriv2; > > ...; ... > > } ; > > > >My doubt is how to read ReadyQ, by using readyQHead..i mean i can > >get pFirstNode, but what this pFirstNode contains, i mean what this > >Q_NODE contains? address of the a TCB? > > Whatever it is you are try to do, it is probably a bad idea. You might > be better-off, posting your problem and asking for advice. > > If you are just trying to understand how the ready queue works then > take a look at how it is used, following through taskDelay(0) is quite > instructive, and fairly straightforward. Well, my problem is that, i think i am stuck in idle loop and i know for sure that there are user tasks.Also, my board allows me to peek into the memory and what i found out was that workQIsEmpty is 1 and taskIdCurrent is tNetTask.When my processor is in this state,i.e not responding or doing user work, i have another processor, which will send a NMI. Now, what i want to do is to gather all the state info this processor( the one that's hanging) and as part of such info, i want to read the readyQ and pendQ and collect the tasks and their state's. Any help is appreciated. Venkat --------------------------- Newsgroups: comp.os.vxworks Subject: Re: shared memory network question Date: Tue, 8 Jan 2002 17:30:46 +0100 From: "Michael Lawnick" Organization: Buergernetz Dillingen Message-ID: References: <3C3B077D.E4BEA62A@spider.com> Reply-To: "Michael Lawnick" Sender: mlawnick@p5088f893.dip.t-dialin.net Hi Support Engineer 4 errno.h : #define EEXIST 17 /* File exists */ seems that this address is already in use !? Remember that your masters IP-address should be in an other subnet than the backplane address (easiest to configure). - -- Mit freundlichen Grüßen, Michael Lawnick ============================================== SOFTEC GmbH Tel +49-731-96600-0 Promenade 17 Fax +49-731-96600-23 D-89073 Ulm Michael Lawnick Germany lawnick@softec.de ============================================== "Support Engineer 4" schrieb im Newsbeitrag news:3C3B077D.E4BEA62A@spider.com... > > Hi, > > Anyone seen this error message before? What does it mean? > > Backplane anchor at 0x4100... Attaching network interface sm0... done. > Error setting inet address of sm0 to 194.217.109.171, errno = 0x11 > > Thanks, > > Roland. > > > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: serial port initialization when serial device is missing Date: Tue, 08 Jan 2002 17:06:53 +0000 From: Graham Baxter Organization: Graham Baxter (Software) Limited Message-ID: <3C3B272D.E4A46B40@NOSPAM.bcs.org.uk> References: <8us_7.31904$LQ1.10302865@news2.nash1.tn.home.com> This is how you might open the null device: int nullFd = open("/null",O_RDWR,0); if(nullFd == ERROR) { /* ERROR */ } This is how you might re-direct logMsg: logFdSet(nullFd); This is how you might re-redirect stdio: ioGlobalStdSet (STD_IN,nullFd); ioGlobalStdSet (STD_OUT,nullFd); ioGlobalStdSet (STD_ERR,nullFd); Regards, Graham Baxter (VxWorks and pSOS BSP'S) Freelance Software Engineer - AVAILABLE FOR NEW ASSIGNMENT SOON gbaxter@NOSPAM.bcs.org.uk --------------------------- Newsgroups: comp.os.vxworks Subject: VME bus interrupts Date: Tue, 08 Jan 2002 12:26:08 -0600 From: James Toland Organization: Raytheon Company Message-ID: <3C3B39C0.DBED8355@raytheon.com> Hello,

I'm having trouble testing my ISR routines.  When I generate a VME bus interrupt using the vxWorks supplied function "sysBusIntGen (level, vector)", my ISR behaves as expected.  However, when I generate a VME bus interrupt using the VMetro Bus Analyzer for the same level and vector, vxWorks announces: "workQpanic: kernel work queue overflow... autoboot in progress"?!!!  Has anyone out there used a VMetro Bus analyzer to generate interrupts?  Anyone know what I'm doing incorrectly?  Some simple code below demonstrates this problem.

Environment: MVME2700 SBC with VxWorks 5.4 and a VMetro Bus Analyzer to capture VME bus activity and generate interrupts.

#include "vxWorks.h"
#include "logLib.h"
#include "sysLib.h"
#include "intLib.h"
#include "iv.h"

void my_Isr (int vector)
{
  logMsg ("Received: %x \n", vector, 0,0,0,0,0);
}

void int_setup (int level, int vector)
{
  intConnect( (VOIDFUNCPTR *)INUM_TO_IVEC(vector), my_Isr, vector);
  sysIntEnable(level);
}

Tornado Shell prompt -> int_setup( 1, 0x64 )
Tornado Shell prompt -> sysBusIntGen ( 1, 0x64)
ISR prints "Received: 64" to console.

-OR-

Tornado Shell prompt -> int_setup( 1, 0x64 )
initiate VMetro interrupt level 1 vector 0x64
ERROR: workQpanic and reboot!

Thanks
 
  --------------------------- Newsgroups: comp.os.vxworks Subject: Re: VME bus interrupts Date: Tue, 8 Jan 2002 22:40:13 +0200 From: "Leonid Rosenboim" Organization: Verio Message-ID: References: <3C3B39C0.DBED8355@raytheon.com> This is a multi-part message in MIME format. - ------=_NextPart_000_003E_01C19895.6FB27540 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable Apparently the VMetro issues interrupts at a too fast rate, too fast for the CPU to handle, ehnce the Panic message, or alternatively, your ISR is doing too much stuff so it can not handle too many inpteerupts per second. I used VMetro, but it was ages ago and not for generating interrupts but to trace various event, it's a hell of a tool, real life saver! See where you control the mode (one-shot or repetitive) and rate of the interrupts the VMetro generates. - - Leonid "James Toland" wrote in message = news:3C3B39C0.DBED8355@raytheon.com... Hello,=20 I'm having trouble testing my ISR routines. When I generate a VME bus = interrupt using the vxWorks supplied function "sysBusIntGen (level, = vector)", my ISR behaves as expected. However, when I generate a VME = bus interrupt using the VMetro Bus Analyzer for the same level and = vector, vxWorks announces: "workQpanic: kernel work queue overflow... = autoboot in progress"?!!! Has anyone out there used a VMetro Bus = analyzer to generate interrupts? Anyone know what I'm doing = incorrectly? Some simple code below demonstrates this problem.=20 Environment: MVME2700 SBC with VxWorks 5.4 and a VMetro Bus Analyzer = to capture VME bus activity and generate interrupts.=20 #include "vxWorks.h"=20 #include "logLib.h"=20 #include "sysLib.h"=20 #include "intLib.h"=20 #include "iv.h"=20 void my_Isr (int vector)=20 {=20 logMsg ("Received: %x \n", vector, 0,0,0,0,0);=20 }=20 void int_setup (int level, int vector)=20 {=20 intConnect( (VOIDFUNCPTR *)INUM_TO_IVEC(vector), my_Isr, vector);=20 sysIntEnable(level);=20 }=20 Tornado Shell prompt -> int_setup( 1, 0x64 )=20 Tornado Shell prompt -> sysBusIntGen ( 1, 0x64)=20 ISR prints "Received: 64" to console.=20 -OR-=20 Tornado Shell prompt -> int_setup( 1, 0x64 )=20 initiate VMetro interrupt level 1 vector 0x64=20 ERROR: workQpanic and reboot!=20 Thanks=20 =20 =20 - ------=_NextPart_000_003E_01C19895.6FB27540 Content-Type: text/html; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable

Apparently the VMetro issues = interrupts at a too=20 fast rate,
too fast for the CPU to handle, ehnce the = Panic=20 message,
or alternatively, your ISR is doing too much = stuff so=20 it
can not handle too many inpteerupts per=20 second.
I used VMetro, but it was ages ago and not for = generating
interrupts but to trace various event, it's a = hell of a=20 tool,
real life saver!
See where you control the mode (one-shot or = repetitive)=20 and
rate of the interrupts the VMetro=20 generates.
 
- Leonid
"James Toland" <j-toland@raytheon.com> = wrote in=20 message news:3C3B39C0.DBED8355@rayth= eon.com...
Hello,=20

I'm having trouble testing my ISR routines.  When I generate a = VME bus=20 interrupt using the vxWorks supplied function "sysBusIntGen (level, = vector)",=20 my ISR behaves as expected.  However, when I generate a VME bus = interrupt=20 using the VMetro Bus Analyzer for the same level and vector, vxWorks=20 announces: "workQpanic: kernel work queue overflow... autoboot in=20 progress"?!!!  Has anyone out there used a VMetro Bus analyzer to = generate interrupts?  Anyone know what I'm doing = incorrectly?  Some=20 simple code below demonstrates this problem.=20

Environment: MVME2700 SBC with VxWorks 5.4 and a VMetro Bus = Analyzer to=20 capture VME bus activity and generate interrupts.=20

#include "vxWorks.h"
#include "logLib.h"
#include = "sysLib.h"=20
#include "intLib.h"
#include "iv.h"=20

void my_Isr (int vector)
{
  logMsg ("Received: %x = \n",=20 vector, 0,0,0,0,0);
}=20

void int_setup (int level, int vector)
{
  intConnect( = (VOIDFUNCPTR *)INUM_TO_IVEC(vector), my_Isr, vector);
 =20 sysIntEnable(level);
}=20

Tornado Shell prompt -> int_setup( 1, 0x64 )
Tornado Shell = prompt=20 -> sysBusIntGen ( 1, 0x64)
ISR prints "Received: 64" to = console.=20

-OR-=20

Tornado Shell prompt -> int_setup( 1, 0x64 )
initiate VMetro = interrupt level 1 vector 0x64
ERROR: workQpanic and reboot!=20

Thanks
 
 

- ------=_NextPart_000_003E_01C19895.6FB27540-- --------------------------- Newsgroups: comp.os.vxworks Subject: Re: VME bus interrupts Date: Tue, 8 Jan 2002 20:56:41 +0000 (UTC) From: don@mars.dgrc.crc.ca (Donald McLachlan) Organization: The unconfigured xvnews people Message-ID: References: Reply-To: don@mars.dgrc.crc.ca a couple years ago someone ported a driver I had written. The signature we had was if during startup (before everything was ready to run) we were receiving data on the sync serial port we would get the same "workQpanic: kernel work queue overflow" error message. While looking at another issue I noticed that when they ported the driver they made some changes (read bugs). It turned out that if the RX ring buffer was full, they did not ack the interrupt, thus the interrupt Q was overflowing. You might have a similiar premature (incorrect) exit from and ISR somewhere. Don - -- Donald McLachlan E-mail Donald.McLachlan@crc.ca Communications Research Centre / RNS Tel (613) 998-2845 3701 Carling Ave., Fax (613) 998-9648 Ottawa, Ontario K2H 8S2 Canada --------------------------- Newsgroups: comp.os.vxworks Subject: BootRom and MMU Date: 8 Jan 2002 13:33:11 -0800 From: vloscomp@yahoo.com (Vinh) Organization: http://groups.google.com/ Message-ID: Hi all, I am trying to debug our VxWorks boot for PPC, and i can't seem to used VisionClick to step over an assembly instruction (tlbie) in the '../target/templatePpc/sysALib.s'. This is the sniped of code. sysALoop: tlbie p0 addi p0,p0,0x1000 /* increment bits 15-19 */ bdnz sysALoop /* decrement CTR, branch if CTR != 0 */ sync /* sync instr req'd after tlbie */ Can any one explain to me why we need to invalidates the TLB in _sysInit? I though that when VxWorks comes up, it will used it's MMU lib to initialized the PPC MMU. Some how the instruction caused the VisionProbe to disconnect. Any help is greatly appreciated. Thank you, Vinh Lam --------------------------- Newsgroups: comp.os.vxworks Subject: Re: why does taskDelay(1) take so long? Date: Tue, 8 Jan 2002 18:04:30 -0500 From: "Jiandong Wang" Organization: Nortel Message-ID: References: <67868445.0201071418.5c211c82@posting.google.com> Hi, Bob, sysClkRateSet(nTicks) is a BSP-specific routine. Did you check the BSP document if you can use 2000 as parameter?(Alough sysClkRateGet shows it has been set correctly.) In our system, I normally use 1000 to set the highest time-resolution. I suggest you set 1000 ticks per second firstly to see if this works correctly. Then, try 2000. - ---------------------- "Bob Johnson" wrote in message news:67868445.0201071418.5c211c82@posting.google.com... > For our system, we use a high system clock rate of approximately 2000 > ticks per second (shown by sysClkRateGet). If I call tickGet() at > random times, I get numbers with no specific pattern. > > However, if I call taskDelay(1), and then tickGet, I ALWAYS get a > value which a is a multiple of 20. It seems that the taskDelay is > taking much longer than 1 tick. > > I tried running my test at priority 3 and priority 100 and neither > priority made a difference. > > Any ideas? > Thanks! --------------------------- Newsgroups: comp.os.vxworks Subject: Re: any other debug method? Date: 8 Jan 2002 17:20:32 -0800 From: feng_w@hotmail.com (wang feng) Organization: http://groups.google.com/ Message-ID: <6ccda887.0201081720.50754b1b@posting.google.com> References: <6ccda887.0201072352.64f62e3e@posting.google.com> <3c3ad0a9@brateggebdc5.br-automation.co.at> hi, ok, thanks a lot for your advice. hth Wangfeng "Werner Schiendl" wrote in message news:<3c3ad0a9@brateggebdc5.br-automation.co.at>... > Hi, > > Select the proper WDB target agent in the target server configuration and > configure the bootable VxWorks image with the matching WDB target agent. > Be sure to also verify the target agent's parameters like baudrate and > interface to use. > > hth > Werner > > "wang feng" wrote in message > news:6ccda887.0201072352.64f62e3e@posting.google.com... > > hi, friends, > > we have softswitch project on ppc system, and control system are > > hot-backup with two cards, problem is they use same ip, the master use > > it to handle outgoing message and the slave only can recieve info > > from master. we use nt2000 as the host to debug the master through > > local network, but the slave cann't be attched. how to debug the slave > > through other ways just like com port or paralel port? > > > > thanks in advance > > > > Wangfeng --------------------------- Newsgroups: comp.os.vxworks Subject: Re: why does taskDelay(1) take so long? Date: Wed, 09 Jan 2002 02:19:08 GMT From: Ephraim Gadsby Organization: - minimal Message-ID: References: <67868445.0201071418.5c211c82@posting.google.com> <9ca84f0f.0201080232.e09498d@posting.google.com> <67868445.0201080656.7edd0fe7@posting.google.com> On 8 Jan 2002 06:57:00 -0800, eaglejump@yahoo.com (Bob Johnson) wrote: >ilkras@yahoo.com (Ilya Krasavin) wrote in message news:<9ca84f0f.0201080232.e09498d@posting.google.com>... >> eaglejump@yahoo.com (Bob Johnson) wrote in message news:<67868445.0201071418.5c211c82@posting.google.com>... >> > For our system, we use a high system clock rate of approximately 2000 >> > ticks per second (shown by sysClkRateGet). If I call tickGet() at >> > random times, I get numbers with no specific pattern. >> > >> > However, if I call taskDelay(1), and then tickGet, I ALWAYS get a >> > value which a is a multiple of 20. It seems that the taskDelay is >> > taking much longer than 1 tick. >> > >> > I tried running my test at priority 3 and priority 100 and neither >> > priority made a difference. >> > >> > Any ideas? >> > Thanks! >> >> May be the timeslice that your scheduler uses is multiple of the 20 >> ms. Every time you call to the taskDelay() - you actually calls your >> scheduler and signal it that another task may be run now, so scheduler >> allocates the timeslice to it. After the timeout of original task is >> expired it will get control immidiately only if it runs on highest >> priority among other tasks. >> Try to rise priority of your task. >> Ilya > >Thanks for the suggestions. Unfortunately.. > >To test how often the scheduler runs, I used the function >taskSwitchHookAdd. This will call a function during every task >switch. In that function, I incremented a counter. During every >taskDelay(1), the counter increments 20-30 times. That means that >task switching is occurring as expected. > >As far as the priority is concerned, I am already running at priority >3, and that doesn't help. I increased the priority to 1, just to see >what would happen. That didn't make a difference either. There is nothing intrinsically wrong about 20-30 context switches during one taskDelay(1) (or even during one tick) but it might be useful to see what those tasks are. The task-switch handler gets the TID of the old and new tasks as arguments. I'd dump them into an array during the delay period. In VxWork the sequence taskx -> Idle state -> taskx does not involve any context switches, so presumably there must be two or more highly active tasks with a real or inherited priority 0. Also in msg <9tbnst$8l8$1@uranium.btinternet.com> David Laight refered to some bugs showing-up above 2k ticks/s. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: why does taskDelay(1) take so long? Date: 8 Jan 2002 18:38:23 -0800 From: eaglejump@yahoo.com (Bob Johnson) Organization: http://groups.google.com/ Message-ID: <67868445.0201081838.39d07205@posting.google.com> References: <67868445.0201071418.5c211c82@posting.google.com> Thanks to Donald and everybody else for their help! I am fairly certain that a low priority task is locking interrupts. It only unlocks the interrupts on 20 tick boundaries. Bob eaglejump@yahoo.com (Bob Johnson) wrote in message news:<67868445.0201071418.5c211c82@posting.google.com>... > For our system, we use a high system clock rate of approximately 2000 > ticks per second (shown by sysClkRateGet). If I call tickGet() at > random times, I get numbers with no specific pattern. > > However, if I call taskDelay(1), and then tickGet, I ALWAYS get a > value which a is a multiple of 20. It seems that the taskDelay is > taking much longer than 1 tick. > > I tried running my test at priority 3 and priority 100 and neither > priority made a difference. > > Any ideas? > Thanks! --------------------------- Newsgroups: comp.os.vxworks Subject: Re: memcpy problem Date: 8 Jan 2002 19:51:15 -0800 From: dragonli@yahoo.com (June) Organization: http://groups.google.com/ Message-ID: References: <3C210DEE.6537F8F6@digidata.com> My memcpy problem is solved. I just messed up with those pointers and address. thank you, everyone! --------------------------- Newsgroups: comp.os.vxworks Subject: Re: readyQHead - how to read it? Date: Wed, 09 Jan 2002 05:16:57 GMT From: Ephraim Gadsby Organization: - minimal Message-ID: References: <65b2abfc.0201071552.49376cb@posting.google.com> <65b2abfc.0201080831.6404a2d5@posting.google.com> On 8 Jan 2002 08:31:34 -0800, mynamvs@nsc.network.com (Venkat Mynampati) wrote: >Ephraim Gadsby wrote in message news:... >> On 7 Jan 2002 15:52:23 -0800, mynamvs@nsc.network.com (Venkat >> Mynampati) wrote: >> >> >Hi, >> >I want to go thru readyQHead and find out what tasks >> >are there in readyQ, can you pls help me out here? >> > >> >in kernelLibP.h readyQHead is defined as >> > >> >Q_HEAD readyQHead; >> > >> >and in qLib.h i found that >> >struct Q_HEAD { >> > Q_NODE *pFirstNode; >> > ... >> > ... >> >} >> >and >> > struct Q_NODE >> > { >> > UINT qPriv1;UINT qPriv2; >> > ...; ... >> > } ; >> > >> >My doubt is how to read ReadyQ, by using readyQHead..i mean i can >> >get pFirstNode, but what this pFirstNode contains, i mean what this >> >Q_NODE contains? address of the a TCB? >> >> Whatever it is you are try to do, it is probably a bad idea. You might >> be better-off, posting your problem and asking for advice. >> >> If you are just trying to understand how the ready queue works then >> take a look at how it is used, following through taskDelay(0) is quite >> instructive, and fairly straightforward. >Well, my problem is that, i think i am stuck in idle loop and i know >for sure that there are user tasks.Also, my board allows me to peek into >the memory and what i found out was that workQIsEmpty is 1 and taskIdCurrent >is tNetTask.When my processor is in this state,i.e not responding or doing >user work, i have another processor, which will send a NMI. >Now, what i want to do is to gather all the state info this processor( the one >that's hanging) and as part of such info, i want to read the readyQ and pendQ >and collect the tasks and their state's. Sound to me as if some hardware problem has hung the processor, or a bug has caused some serious corruption to the kernel's data structures. I'd do the following - - check the ISRs (and other handlers with similar restrictions like watchdogs and context switch hooks) to see that they only use allowed library call e.g. no selects, semTakes, mallocs, printfs, or C++, that might conceal these functions. I've seen this kind of thing hang VxWorks several times. - - check the cpu manufacturers errata documentation for this kind of behavior (particularly if it is new or exotic) - - check that all the tasks have adequate stack memory, when a stack overflows it often corrupts a TCB. - - try to track down the point at which it locks-up e.g keep a history of ISR entries and exits. If its a hardware probelm you may find it this way If you want to follow the ready queue the current default priority queue implemention is in qPriBMapLib.c. One thing to watch out for is a trick that is commonly used in the kernel, where some important member structure is delibrately put at the start of a larger structure so that the same pointer can be used for both. For example the first member of a TCB is the ready/pend queue node, so that a pointer to the node, and a pointer to the TCB are interchangeable. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: VME bus interrupts Date: Wed, 9 Jan 2002 11:02:59 +0200 From: "Leonid Rosenboim" Organization: Verio Message-ID: References: <3C3B39C0.DBED8355@raytheon.com> This is a multi-part message in MIME format. - ------=_NextPart_000_0020_01C198FD.3332CE40 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable Another point you could look at is VME interrupt acknowledge - some boards do it in hardware, others need the software to do this from the ISR, e.g. call sysBusIntAck(level), otherwise the interrupt = source will be kept on forever which will definitely cause the panic = message. Ints generated with sysBusIntGen may be "pulsed" so you could get away = without acking them, ... - - Leonid P.S. Seems the years are taking their toll, the ack thing would have = been obvious in the years when VME was our daily bread. "James Toland" wrote in message = news:3C3B39C0.DBED8355@raytheon.com... Hello,=20 I'm having trouble testing my ISR routines. When I generate a VME bus = interrupt using the vxWorks supplied function "sysBusIntGen (level, = vector)", my ISR behaves as expected. However, when I generate a VME = bus interrupt using the VMetro Bus Analyzer for the same level and = vector, vxWorks announces: "workQpanic: kernel work queue overflow... = autoboot in progress"?!!! Has anyone out there used a VMetro Bus = analyzer to generate interrupts? Anyone know what I'm doing = incorrectly? Some simple code below demonstrates this problem.=20 Environment: MVME2700 SBC with VxWorks 5.4 and a VMetro Bus Analyzer = to capture VME bus activity and generate interrupts.=20 #include "vxWorks.h"=20 #include "logLib.h"=20 #include "sysLib.h"=20 #include "intLib.h"=20 #include "iv.h"=20 void my_Isr (int vector)=20 {=20 logMsg ("Received: %x \n", vector, 0,0,0,0,0);=20 }=20 void int_setup (int level, int vector)=20 {=20 intConnect( (VOIDFUNCPTR *)INUM_TO_IVEC(vector), my_Isr, vector);=20 sysIntEnable(level);=20 }=20 Tornado Shell prompt -> int_setup( 1, 0x64 )=20 Tornado Shell prompt -> sysBusIntGen ( 1, 0x64)=20 ISR prints "Received: 64" to console.=20 -OR-=20 Tornado Shell prompt -> int_setup( 1, 0x64 )=20 initiate VMetro interrupt level 1 vector 0x64=20 ERROR: workQpanic and reboot!=20 Thanks=20 =20 =20 - ------=_NextPart_000_0020_01C198FD.3332CE40 Content-Type: text/html; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable
Another point you could look at is = VME interrupt=20 acknowledge -
some boards do it in hardware, others need the = software=20 to do this
from the ISR, e.g. call sysBusIntAck(level), = otherwise=20 the interrupt source will be kept on forever which will definitely cause = the=20 panic message.
Ints generated with sysBusIntGen may be = "pulsed" so you=20 could get away without acking them, ...
 
- Leonid
P.S. Seems the years are taking their toll, = the ack=20 thing would have been obvious in the years when VME was our daily=20 bread.
"James Toland" <j-toland@raytheon.com> = wrote in=20 message news:3C3B39C0.DBED8355@rayth= eon.com...
Hello,=20

I'm having trouble testing my ISR routines.  When I generate a = VME bus=20 interrupt using the vxWorks supplied function "sysBusIntGen (level, = vector)",=20 my ISR behaves as expected.  However, when I generate a VME bus = interrupt=20 using the VMetro Bus Analyzer for the same level and vector, vxWorks=20 announces: "workQpanic: kernel work queue overflow... autoboot in=20 progress"?!!!  Has anyone out there used a VMetro Bus analyzer to = generate interrupts?  Anyone know what I'm doing = incorrectly?  Some=20 simple code below demonstrates this problem.=20

Environment: MVME2700 SBC with VxWorks 5.4 and a VMetro Bus = Analyzer to=20 capture VME bus activity and generate interrupts.=20

#include "vxWorks.h"
#include "logLib.h"
#include = "sysLib.h"=20
#include "intLib.h"
#include "iv.h"=20

void my_Isr (int vector)
{
  logMsg ("Received: %x = \n",=20 vector, 0,0,0,0,0);
}=20

void int_setup (int level, int vector)
{
  intConnect( = (VOIDFUNCPTR *)INUM_TO_IVEC(vector), my_Isr, vector);
 =20 sysIntEnable(level);
}=20

Tornado Shell prompt -> int_setup( 1, 0x64 )
Tornado Shell = prompt=20 -> sysBusIntGen ( 1, 0x64)
ISR prints "Received: 64" to = console.=20

-OR-=20

Tornado Shell prompt -> int_setup( 1, 0x64 )
initiate VMetro = interrupt level 1 vector 0x64
ERROR: workQpanic and reboot!=20

Thanks
 
 

- ------=_NextPart_000_0020_01C198FD.3332CE40-- --------------------------- Newsgroups: comp.os.vxworks Subject: bus error exception in ppc8260 Date: 9 Jan 2002 01:16:52 -0800 From: ruchama@septier.com (Ruchama Goldshmidt) Organization: http://groups.google.com/ Message-ID: <5df340e0.0201090116.70d0958d@posting.google.com> Hello vxWorkers!! does anyone know how can I know exactly what was the address that the access to him cause the bus error exception? I know that the are some register that can tell me that but I don't what are them. (I try to develop a new exception handler and I really need this detail) TIA Ruchama Goldshmidt Septier Communication Ltd. --------------------------- Newsgroups: comp.os.vxworks Subject: Include Paths Date: Wed, 09 Jan 2002 11:02:53 +0000 From: Paul Caswell Organization: Land & Sea Systems Message-ID: <3C3C235D.A2669DFF@baesystems.com> Hello vxWorks people, I'm using T2.0 on NT4. I have a downloadable project created in C:\TornadoDev\myProject and I want to add files to it from another location (ie C:\TornadoDev\Common). It's easy enough to add the files to the project workspace but when I then try and build the compiler complains that it can't find the one of the required headers (it's in the common directory). I tried adding -I to the project compiler options with no luck. In VC++ and UNIX this is easy so I get the feeling that I'm just missing something simple. Any help would be appreciated. Paul --------------------------- Newsgroups: comp.os.vxworks Subject: Re: bus error exception in ppc8260 Date: Wed, 09 Jan 2002 06:34:43 -0500 From: joe durusau Organization: BELLSOUTH.net Message-ID: <3C3C2AD3.42E7E3FE@bellsouth.net> References: <5df340e0.0201090116.70d0958d@posting.google.com> For most ppc implementations, you cannot recover from a bus error, because the processor has multiple instructions in progress at a given time, and the machine state is non-recoverable. Check the docs on your specific machine before possibly wasting time on an exception handler that will not be able to work. Most BSPs support vxMemProbe which will allow you to check for bus errors before creating one. Speaking only for myself, Joe Durusau Ruchama Goldshmidt wrote: > Hello vxWorkers!! > > does anyone know how can I know exactly what was the address that the > access to him cause the bus error exception? > I know that the are some register that can tell me that but I don't > what are them. > > (I try to develop a new exception handler and I really need this > detail) > > TIA > > Ruchama Goldshmidt > Septier Communication Ltd. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: bus error exception in ppc8260 Date: Wed, 9 Jan 2002 10:33:45 -0000 From: "Dave Korn" Organization: Lumber Cartel (tinlc) Members #2234-2237 (owing to browser refresh) Message-ID: <20W_7.337$qS6.871834@newsr2.u-net.net> References: <5df340e0.0201090116.70d0958d@posting.google.com> Ruchama Goldshmidt wrote in message <5df340e0.0201090116.70d0958d@posting.google.com>... >Hello vxWorkers!! > >does anyone know how can I know exactly what was the address that the >access to him cause the bus error exception? >I know that the are some register that can tell me that but I don't >what are them. > >(I try to develop a new exception handler and I really need this >detail) It's in the DAR, stored in the exception stack frame that is passed to the handler. Here's an example of a DSI exception handler that I use: void SetupDSIExceptionHandler (void) { excConnect ((VOIDFUNCPTR *)_EXC_OFF_DATA, (VOIDFUNCPTR)DSIExceptionHandler); } // These can be adjusted from the shell. int bspTraceDSIExceptions =1; int bspContinueDSIExceptions = 0; void DSIExceptionHandler (ESFPPC *esf) { // We are in task context, not ISR, so printf etc. is ok to use here. printf ("\n\nDSI exception: PC $%08x DAR $%08x DSISR $%08x\n", esf->regSet.pc, esf->dar, esf->dsisr); if (bspTraceDSIExceptions) bspHowDidIGetHere (); // If we're continuing, go back now; otherwise suspend here // and let the user decide what to do about it. if (!bspContinueDSIExceptions) taskSuspend (0); } static int bspTraceWrapper (SEM_ID sema, int tidtotrace, ...) { if (taskIdVerify (tidtotrace) == ERROR) { printf ("tid not valid $%08x\n", tidtotrace); i (0); } else { printf ("\n**** Application stack trace ***\n"); tt (tidtotrace); printf ("\n\n"); } semGive(sema); return(TRUE); } // You can't call tt directly from the code, as the // stack trace function doesn't work on its own task. void bspHowDidIGetHere (void) { SEM_ID sema; STATUS err; sema = semBCreate(SEM_Q_FIFO, SEM_EMPTY); // the priority here puts the task below shell, network, // exctask and logtask, and most drivers. err = taskSpawn("tBspTraceback", 70, VX_FP_TASK, 24*1024, (FUNCPTR)bspTraceWrapper, (int)sema, taskIdSelf (), 0, 0, 0, 0, 0, 0, 0, 0); if (err == ERROR) printf("Sorry, dunno how i got here\n"); else semTake(sema, WAIT_FOREVER); semDelete(sema); } DaveK - -- moderator of alt.talk.rec.soc.biz.news.comp.humanities.meow.misc.moderated.meow Burn your ID card! http://www.optional-identity.org.uk/ Help support the campaign, copy this into your .sig! Proud Member of the Exclusive "I have been plonked by Davee because he thinks I'm interesting" List Member # --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Optimizing VxWorks boot time ? Date: Wed, 9 Jan 2002 12:43:49 +0100 From: "Søren Abildgaard" Organization: TDC Internet Message-ID: <3c3c2d00$0$89102$edfadb0f@dspool01.news.tele.dk> References: <3c3acff3$0$5452$edfadb0f@dspool01.news.tele.dk> <3c3aebda$1@pull.gecm.com> Reply-To: "Søren Abildgaard" We have 32 MB of RAM on-board and all is currently reported to VxWorks. I could proberly reduce this to half the size. I will test out asap. Any other hints of how to attack the boot time problem?? "Martin Dowie" wrote in message news:3c3aebda$1@pull.gecm.com... > "Søren Abildgaard" wrote in message > news:3c3acff3$0$5452$edfadb0f@dspool01.news.tele.dk... > > Hi *.*, > > > > We are running a project, where we are using a x86 platform > > and a vxworks 5.2. > > > > I am fairly new to VxWorks, but my problem is the boottime we > > are getting. It takes about 13 seconds from the BIOS POST ends > > until VXworks starts to load our application. > > The kernel I have built I have left out all components, which we > > don´t need. > > > > What parameters do I have left to tweak on? And whats dertermines this > > boottime, which seems fairly long for an embedded system? > > How much memory have you told VxWorks you have and how does that > compare with how much you actually use? We have found that reducing > the amount you tell VxWorks you have can make a considerable > difference to the time. > > > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Include Paths Date: Wed, 09 Jan 2002 11:42:56 +0000 From: Graham Baxter Organization: Graham Baxter (Software) Limited Message-ID: <3C3C2CC0.ECB904AF@NOSPAM.bcs.org.uk> References: <3C3C235D.A2669DFF@baesystems.com> Paul, Try replacing the backward slashes with forward slashes. (back slash has a special meaning in make) I hope this helps. Regards, Graham Baxter (VxWorks and pSOS BSP'S) Freelance Software Engineer - AVAILABLE FOR NEW ASSIGNMENT SOON gbaxter@NOSPAM.bcs.org.uk --------------------------- Newsgroups: comp.os.vxworks Subject: Re: bus error exception in ppc8260 Date: Wed, 09 Jan 2002 11:44:17 +0000 From: Graham Baxter Organization: Graham Baxter (Software) Limited Message-ID: <3C3C2D11.1326A76A@NOSPAM.bcs.org.uk> References: <5df340e0.0201090116.70d0958d@posting.google.com> Ruchama, Try the following: write your own handler like: #include void myHandler(register int task,register int vecNum,register ESFPPC *pEsf) __attribute__ ((noreturn)); void myHandler(register int task,register int vecNum,register ESFPPC *pEsf) { /* your handler here */ } Install it with: excHookAdd((FUNCPTR)myHandler); The handler can access the registers: pEsf->regSet.msr pEsf->regSet.lr pEsf->regSet.ctr pEsf->regSet.pc pEsf->regSet.cr pEsf->regSet.xer ETC. see excPpcLib.h I hope this helps. Regards, Graham Baxter (VxWorks and pSOS BSP'S) Freelance Software Engineer - AVAILABLE FOR NEW ASSIGNMENT SOON gbaxter@NOSPAM.bcs.org.uk --------------------------- Newsgroups: comp.os.vxworks Subject: Bug Report? Loading more than one END Date: 9 Jan 2002 03:58:22 -0800 From: odedn@shiron.com (oj) Organization: http://groups.google.com/ Message-ID: <9d60aef6.0201090358.1d8143cc@posting.google.com> hi ya all vxWorkers i have been struggling for a week with this problem: i was trying to load two END's , the image i am using has its IP_MAX_UNITS param configure to 1, when i tried to load the first END from application level i got the message to increase my IP_MAX_UNITS , i increase it from my application before loading the END and everything went just fine. the same goes when i tried to load the second END but this time when i increase the IP_MAX_UNITS from application level i got some weird exception check error. i was sure that the problem is somehow relate to the second END implementation... but it didn't. only when i change the IP_MAX_UNITS in my image things worked out!!! why the hell changing the IP_MAX_UNITS from application level solved the problem for the first END and not for the second ? i have no idea. if any of u will encounters such a problem this maybe a solution for ya, if someone can explain this strange vxWorks behavior please let me know. thanx. --------------------------- End of New-News digest ********************** From vxwexplo-errs@csg.lbl.gov Wed Jan 9 07:04:37 2002 From: "Shah, Ranjana" Date: Wed Jan 9 07:04:39 PST 2002 Subject: Re: why does taskDelay(1) take so long? Bob, This may or may not be your problem, but may be worth a try. VxWorks has its own internal "software" system clock, based on the hardware system clock. The hardware real time clock is separate from VxWorks' clock and hardware system clock. When the BSP is initialized, it sets the system clock to generate an interrupt, usually 60 times a second using sysClkRateSet(). When the VxWorks time library (defined in time.h) is initialized, it calibrates its granuality to this value. When you override this with another call to sysClkRateSet(), it does not re-initilaize the VxWorks' time library. The function clock_setres() must be called to resync VxWorks every time the system timer granuality is changed. Ranjana Shah Staff Engineer Telephonics Corp. Voice: (631) 549-6104 FAX: (631) 755-7642 "Bob Johnson" wrote in message news:67868445.0201071418.5c211c82@posting.google.com... > For our system, we use a high system clock rate of approximately 2000 > ticks per second (shown by sysClkRateGet). If I call tickGet() at > random times, I get numbers with no specific pattern. > > However, if I call taskDelay(1), and then tickGet, I ALWAYS get a > value which a is a multiple of 20. It seems that the taskDelay is > taking much longer than 1 tick. > > I tried running my test at priority 3 and priority 100 and neither > priority made a difference. > > Any ideas? > Thanks! From vxwexplo-errs@csg.lbl.gov Wed Jan 9 07:30:59 2002 From: "Sinn, Larry" Date: Wed Jan 9 07:31:01 PST 2002 Subject: Re: vme reset > > Tim Iskander wrote: > > > > > > It looks like when I power up our MVME2700 with the vx bootloader > > > selected, > > > it never asserts a VME reset. If I switch over to PPCBug, it does. I > > > cant find > > > anything in the vx docs/etc about it. Has anyone seen this or know > > > anything about it? We have a custom board which needs to see VME RESET > > > at power up. > > > > Andrew wrote: > > > > There's a bit in the Universe chip's MISC_CTL register which you can set > > to initiate a bus reset. That register gets initialized in > > universe.c:sysUniverseInit() so you could just insert > > MISC_CTL_SW_SRST | > > in the lines that are ORed into temp_data. > > > However IIRC the VME backplane circuitry should be holding SYSRESET* low > > for 200ms after +5V becomes stable, so your board should be seeing a > reset > > pulse at powerup anyway. Are you sure your backplane is wired properly? > > > On the mvme2300, a VME Reset would also cause a VxWorks reboot. > I solved it by using 2 memory locations just above the vector addresses > (I think it's 0x400) and setting very unique values to determine if this > was a cold or warm reboot. A cold reboot I would do a VME Reset. > I also had to modify the memory clear routine so it would not > clear these 2 locations. > > Larry. > > > Larry Sinn > KLA-Tencor MS I-1009 408 875 0247 voice > 160 Rio Robles 408 875 6179 fax > San Jose, Ca 95134-1809 larry.sinn@kla-tencor.com > From vxwexplo-errs@csg.lbl.gov Wed Jan 9 17:25:31 2002 From: "Jun Wu" Date: Wed Jan 9 17:25:33 PST 2002 Subject: which case etherOutput fail Hello experts, I use ethernet Input Hook and etherOutput to implement a simple gateway (Torndao2.0/Vxworks 5.4, intel SA1110 assabet board ). In low speed traffic case it works well. but when the traffic speed is higher than 384kbps (both inbound and oubound ), the hook can still capture ethernet packets, whilst call etherOutput fails (return ERROR). Does anybody know the inner world of etherOutput function ? Which cases cause it fail ? Or is there any method debugging into it ? Best regards Jun Wu From vxwexplo-errs@csg.lbl.gov Wed Jan 9 19:26:50 2002 From: nnnzx@263.net Date: Wed Jan 9 19:26:52 PST 2002 Subject: =?gb2312?B?aW8gc3BhY2UgcmVx?= --Boundary-=_JGYNrqwbeNRltuUfdtXDbZTBIrHz Content-Type: text/plain; charset="gb2312" Content-Transfer-Encoding: 8bit Hello, I'm working on a vxWorks driver for a PCI board which uses the PLX-9052 chip.I am using the pcPentium BSP . I am able to read the PCI configuration space without problems and I can see the BARS. I am able to read the PLX registers through I/O space.The board's IO Space mapped from 0xa000~0xafff ,and i can access all space in 0xa000~0xafff use windriver,but I am unable to accesses the all IO space 0xa000~0xafff under vxworks.I can only access IO space 0xa000~0xa0ff and 0xa400~0xa4ff.Could you tell me what reason about this and how can i do? Hope, your helps ZhangXin __________________________________________ È«³¡ÌؼۣºÊýÂëÏà»ú¡¢ÕÆÉϵçÄÔ¡¢MP3 http://shopping.263.net/category21.htm ÌìÓ®³´¹É,ÌìÉúÎÒÓ®! http://stock19.263.net/download/borntowin.htm --Boundary-=_JGYNrqwbeNRltuUfdtXDbZTBIrHz Content-Type: text/html; charset="gb2312" Content-Transfer-Encoding: 8bit Hello,
I'm working on a vxWorks driver for a PCI board which uses the PLX-9052 chip.I am using the pcPentium BSP . I am able to read the PCI configuration space without problems and I can see the BARS. I am able to read the PLX registers through I/O space.The board's IO Space mapped from 0xa000~0xafff ,and i can access all space in 0xa000~0xafff use windriver,but I am unable to accesses the all IO space 0xa000~0xafff under vxworks.I can only access IO space 0xa000~0xa0ff and 0xa400~0xa4ff.Could you tell me what reason about this and how can i do?

Hope, your helps

ZhangXin




__________________________________________

È«³¡ÌؼۣºÊýÂëÏà»ú¡¢ÕÆÉϵçÄÔ¡¢MP3 http://shopping.263.net/category21.htm
ÌìÓ®³´¹É,ÌìÉúÎÒÓ®! http://stock19.263.net/download/borntowin.htm
--Boundary-=_JGYNrqwbeNRltuUfdtXDbZTBIrHz-- From vxwexplo-errs@csg.lbl.gov Wed Jan 9 20:46:11 2002 From: Jitendra kanitkar Date: Wed Jan 9 20:46:14 PST 2002 Subject: RE: comp.os.vxworks newsdigest Hi, I don't think you need to change the IP_MAX_UNITS in applcation for each and every card. Change it in configNet.h file in your BSP and if you are using tornado project then there also you need to change the parameter. Some times you need to undef IP_MAX_UNITS in configNet.h before you make it 2, (that depends on situation). I hope this helps you. Jitendra. Newsgroups: comp.os.vxworks Subject: Bug Report? Loading more than one END Date: 9 Jan 2002 03:58:22 -0800 From: odedn@shiron.com (oj) Organization: http://groups.google.com/ Message-ID: <9d60aef6.0201090358.1d8143cc@posting.google.com> hi ya all vxWorkers i have been struggling for a week with this problem: i was trying to load two END's , the image i am using has its IP_MAX_UNITS param configure to 1, when i tried to load the first END from application level i got the message to increase my IP_MAX_UNITS , i increase it from my application before loading the END and everything went just fine. the same goes when i tried to load the second END but this time when i increase the IP_MAX_UNITS from application level i got some weird exception check error. i was sure that the problem is somehow relate to the second END implementation... but it didn't. only when i change the IP_MAX_UNITS in my image things worked out!!! why the hell changing the IP_MAX_UNITS from application level solved the problem for the first END and not for the second ? i have no idea. if any of u will encounters such a problem this maybe a solution for ya, if someone can explain this strange vxWorks behavior please let me know. thanx. From vxwexplo-errs@csg.lbl.gov Thu Jan 10 04:03:15 2002 From: Vxworks Exploder Date: Thu Jan 10 04:03:17 PST 2002 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Thu Jan 10 04:03:11 PST 2002 Subject: Re: Include Paths Subject: Re: Optimizing VxWorks boot time ? Subject: Re: Optimizing VxWorks boot time ? Subject: Re: vxWorks Flash File System Subject: Re: Optimizing VxWorks boot time ? Subject: Re: lspci Subject: Re: Optimizing VxWorks boot time ? Subject: Re: Include Paths Subject: Re: lspci Subject: Re: Optimizing VxWorks boot time ? Subject: Re: vxworks, hard disk crashes, and NBX-100 Subject: Re: BootRom and MMU Subject: Re: Using select(), need to monitor the serial port continuously in the thread Subject: Socket back end and fd table of the I/O system Subject: Re: Bug Report? Loading more than one END Subject: Re: VME bus interrupts Subject: Re: why does taskDelay(1) take so long? Subject: Re: Socket back end and fd table of the I/O system Subject: Re: why does taskDelay(1) take so long? Subject: "end.h" included in C++ code in Tornada 2.0.2 Subject: Re: why does taskDelay(1) take so long? Subject: Co-processer unuseable exception Subject: Problem with __FiPc with ipAttach() Subject: [HELP]UDP TX/RX packet mismatch problem... Subject: Re: Bug Report? Loading more than one END Subject: Re: [HELP]UDP TX/RX packet mismatch problem... ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: Include Paths Date: 09 Jan 2002 13:29:39 +0200 From: Reinhold Wirth Organization: WEB.DE Message-ID: <3c3c37b3$1@netnews.web.de> References: <3C3C235D.A2669DFF@baesystems.com> Paul Caswell wrote: >Hello vxWorks people, >I'm using T2.0 on NT4. I have a downloadable project created in >C:\TornadoDev\myProject and I want to add files to it from another >location (ie C:\TornadoDev\Common). It's easy enough to add the files >to the project workspace but when I then try and build the compiler >complains that it can't find the one of the required headers (it's in >the common directory). I tried adding -I to the project compiler >options with no luck. In VC++ and UNIX this is easy so I get the >feeling that I'm just missing something simple. Any help would be >appreciated. Hi Paul, try it with add the macro VPATH (value:) - -- __________________________________________________________ News suchen, lesen, schreiben mit http://newsgroups.web.de --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Optimizing VxWorks boot time ? Date: Wed, 9 Jan 2002 13:24:09 +0100 From: "Frank Wolf" Message-ID: <0jch1a.q4t.ln@gum.de> References: <3c3acff3$0$5452$edfadb0f@dspool01.news.tele.dk> <3c3aebda$1@pull.gecm.com> <3c3c2d00$0$89102$edfadb0f@dspool01.news.tele.dk> Hi there! We've modified the routines LOCAL void copyLongs (source, destination, nlongs); LOCAL void fillLongs (buf, nlongs, val); to use the DMA controller instead of CPU to perform this task. This helped us to get far lower startup times. Frank "Søren Abildgaard" wrote in message news:3c3c2d00$0$89102$edfadb0f@dspool01.news.tele.dk... > We have 32 MB of RAM on-board and all is currently > reported to VxWorks. I could proberly reduce this to > half the size. I will test out asap. Any other hints of how to > attack the boot time problem?? > > "Martin Dowie" wrote in message > news:3c3aebda$1@pull.gecm.com... > > "Søren Abildgaard" wrote in message > > news:3c3acff3$0$5452$edfadb0f@dspool01.news.tele.dk... > > > Hi *.*, > > > > > > We are running a project, where we are using a x86 platform > > > and a vxworks 5.2. > > > > > > I am fairly new to VxWorks, but my problem is the boottime we > > > are getting. It takes about 13 seconds from the BIOS POST ends > > > until VXworks starts to load our application. > > > The kernel I have built I have left out all components, which we > > > don´t need. > > > > > > What parameters do I have left to tweak on? And whats dertermines this > > > boottime, which seems fairly long for an embedded system? > > > > How much memory have you told VxWorks you have and how does that > > compare with how much you actually use? We have found that reducing > > the amount you tell VxWorks you have can make a considerable > > difference to the time. > > > > > > > > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Optimizing VxWorks boot time ? Date: Wed, 9 Jan 2002 12:58:07 -0000 From: "Martin Dowie" Message-ID: <3c3c3e2c$1@pull.gecm.com> References: <3c3acff3$0$5452$edfadb0f@dspool01.news.tele.dk> <3c3aebda$1@pull.gecm.com> <3c3c2d00$0$89102$edfadb0f@dspool01.news.tele.dk> "Søren Abildgaard" wrote in message news:3c3c2d00$0$89102$edfadb0f@dspool01.news.tele.dk... > We have 32 MB of RAM on-board and all is currently > reported to VxWorks. I could proberly reduce this to > half the size. I will test out asap. Any other hints of how to > attack the boot time problem?? Hmmm, avoid initialising devices you don't need. I've heard stories (though never seen this myself) that some Ethernet chips can take up to (maybe more than?!) 1 second to initialise. If you can still meet any (speed) performance requirements, you could trying running from EEPROM (saves the loading of the SDRAM and the uncompressing, assuming your image is compressed to start with). --------------------------- Newsgroups: comp.os.vxworks Subject: Re: vxWorks Flash File System Date: 9 Jan 2002 04:59:44 -0800 From: x-xz@263.net (Xie Xiaozhuo) Organization: http://groups.google.com/ Message-ID: <9ed09ad5.0201090459.bcb2d91@posting.google.com> References: I didn't see you load your vxWorks image to your ram device. If your function netBootload() loads vxWorks from out-of-band port, vxWorks now should be in your ram beginning at a certain position (in my switch, that is RAM_LOW_ADDR = 0x80100000 ) just open your /rfa2/vxWorks and copy it to your RAM_LOW_ADDR. But I think a better way is to copy data directly from the data socket of ftp connection to flash. below is for referrence: char *flashBuf = (char *)RAM_LOW_ADDR; if ((fd = open("/rfa2/vxWorks", O_WRONLY | O_CREAT, 0644)) == ERROR) { printErr("Cannot open \"%s\".\n", fileName); goto fail; } printf("Writing data ...\n"); i = write(fd, flashBuf, size); /*size is size of your vxWorks image*/ close(fd); printf("done %d\n", i); > I made flash file system initialized on vxWorks 5.4. > but I need some help for completing. > > I loaded vxWorks image from network via ftp method. > > I want this loaded image to be copied on flash .(Purpose) > It doesn't work. > > So I did like this. > > In bootConfig.c, > > first I made ramDrv. > > ramDrv(); > pBlkDev = ramDevCreate(0, 512, 400, 3000, 0);/* ramAddr, > bytesperblk, blkspertrack, nBlocks, blkOffset */ > dosFsMkfs("/ram", pBlkDev); > > and In bootCmdLoop, > > > case '$': /* net load ,copy it and flashLoad */ > > if (netBootLoad (pLine, &entry)==OK) > { > > flashFInit(); > copy("/ram/vxWorks","/rfa2/vxWorks"); > /* > flashFsBlkFlush2(0); > flashFsBlkWrite2(pBlkDev,0, 32, (char *)&entry); > */ > > printf("copied from ram to flash!\n"); > } > else > { > printf("net boot load failed!\n"); > } > > > if (flashLoad ("/rfa2/vxWorks", &entry) == OK) > { > printf ("entry = 0x%x\n", (int) entry); > /*go(entry);*/ > } > else > { > printf("flashLoad failed!\n"); > > taskDelay (sysClkRateGet ()); /* pause a second */ > /*reboot (BOOT_NO_AUTOBOOT);*/ /* something is awry */ > } > break; > > > Regards, > > eche --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Optimizing VxWorks boot time ? Date: Wed, 9 Jan 2002 14:46:02 +0100 From: "Søren Abildgaard" Organization: TDC Internet Message-ID: <3c3c499d$0$35462$edfadb0f@dspool01.news.tele.dk> References: <3c3acff3$0$5452$edfadb0f@dspool01.news.tele.dk> <3c3aebda$1@pull.gecm.com> <3c3c2d00$0$89102$edfadb0f@dspool01.news.tele.dk> <0jch1a.q4t.ln@gum.de> Reply-To: "Søren Abildgaard" I am question mark here. How should I use the DMA controller ? Regards, Søren "Frank Wolf" wrote in message news:0jch1a.q4t.ln@gum.de... > Hi there! > > We've modified the routines > > LOCAL void copyLongs (source, destination, nlongs); > LOCAL void fillLongs (buf, nlongs, val); > > to use the DMA controller instead of CPU to perform this task. > This helped us to get far lower startup times. > > Frank > > > "Søren Abildgaard" wrote in message > news:3c3c2d00$0$89102$edfadb0f@dspool01.news.tele.dk... > > We have 32 MB of RAM on-board and all is currently > > reported to VxWorks. I could proberly reduce this to > > half the size. I will test out asap. Any other hints of how to > > attack the boot time problem?? > > > > "Martin Dowie" wrote in message > > news:3c3aebda$1@pull.gecm.com... > > > "Søren Abildgaard" wrote in message > > > news:3c3acff3$0$5452$edfadb0f@dspool01.news.tele.dk... > > > > Hi *.*, > > > > > > > > We are running a project, where we are using a x86 platform > > > > and a vxworks 5.2. > > > > > > > > I am fairly new to VxWorks, but my problem is the boottime we > > > > are getting. It takes about 13 seconds from the BIOS POST ends > > > > until VXworks starts to load our application. > > > > The kernel I have built I have left out all components, which we > > > > don´t need. > > > > > > > > What parameters do I have left to tweak on? And whats dertermines this > > > > boottime, which seems fairly long for an embedded system? > > > > > > How much memory have you told VxWorks you have and how does that > > > compare with how much you actually use? We have found that reducing > > > the amount you tell VxWorks you have can make a considerable > > > difference to the time. > > > > > > > > > > > > > > > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: lspci Date: Wed, 9 Jan 2002 15:05:35 +0000 (UTC) From: don@mars.dgrc.crc.ca (Donald McLachlan) Organization: The unconfigured xvnews people Message-ID: Followup-To: comp.os.vxworks Reply-To: don@mars.dgrc.crc.ca Background info follows, but my questions are: - - Why does vxWorks say the card's I/O addresses are at 7001 and 7401 when Linux says they are at 7000 and 7401? - - Also, why 2 addresses 700X and 740X; begin and end, or aliasing? - - Both report memory address f9000000. SeaLevel says the registers are only available via I/O ... so what's this memory? Thanks, Don - ----- I'm running booting vxWorks from a floppy on a PC that was used for linux ... so if I boot from the hard disk I have linux. Using the Linux lspci command I get: > lspci -v -d 0x135e:0x5102 00:0d.0 Serial controller: Sealevel Systems Inc: Unknown device 5102 (rev 01) (prog-if 02 [16550]) Subsystem: Sealevel Systems Inc: Unknown device 5102 Flags: medium devsel, IRQ 12 Memory at f9000000 (32-bit, non-prefetchable) I/O ports at 7000 I/O ports at 7400 - ----- Using vxWorks command from the target shell I get: -> pciDeviceShow(0) Scanning function 0 of each PCI device on bus 0 Using configuration mechanism 1 bus device function vendorID deviceID class 00000000 00000000 00000000 00001106 00000691 00000600 00000000 00000001 00000000 00001106 00008598 00040600 00000000 00000007 00000000 00001106 00000596 00010600 00000000 0000000a 00000000 000010b7 00009055 00000200 00000000 0000000d 00000000 0000135e 00005102 02000700 value = 0 = 0x0 -> pciFindDeviceShow(0x0000135e, 0x00005102, 0) deviceId = 0x00005102 vendorId = 0x0000135e index = 0x00000000 busNo = 0x00000000 deviceNo = 0x0000000d funcNo = 0x00000000 value = 0 = 0x0 -> pciHeaderShow(0, 0x0000000d, 0x00000000) vendor ID = 0x135e device ID = 0x5102 command register = 0x0003 status register = 0x0280 revision ID = 0x01 class code = 0x07 sub class code = 0x00 programming interface = 0x02 cache line = 0x08 latency time = 0x00 header type = 0x00 BIST = 0x00 base address 0 = 0xf9000000 base address 1 = 0x00007001 base address 2 = 0x00007401 base address 3 = 0x00000000 base address 4 = 0x00000000 base address 5 = 0x00000000 cardBus CIS pointer = 0x00000000 sub system vendor ID = 0x135e sub system ID = 0x5102 expansion ROM base address = 0x00000000 interrupt line = 0x0c interrupt pin = 0x01 min Grant = 0x00 max Latency = 0x00 value = 0 = 0x0 - -- Donald McLachlan E-mail Donald.McLachlan@crc.ca Communications Research Centre / RNS Tel (613) 998-2845 3701 Carling Ave., Fax (613) 998-9648 Ottawa, Ontario K2H 8S2 Canada --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Optimizing VxWorks boot time ? Date: Wed, 9 Jan 2002 15:40:07 -0000 From: "Martin Dowie" Message-ID: <3c3c6424$1@pull.gecm.com> References: <3c3acff3$0$5452$edfadb0f@dspool01.news.tele.dk> <3c3aebda$1@pull.gecm.com> <3c3c2d00$0$89102$edfadb0f@dspool01.news.tele.dk> <0jch1a.q4t.ln@gum.de> <3c3c499d$0$35462$edfadb0f@dspool01.news.tele.dk> "Søren Abildgaard" wrote in message news:3c3c499d$0$35462$edfadb0f@dspool01.news.tele.dk... > I am question mark here. How should I use the > DMA controller ? As it may do this in the "background", freeing up your CPU to be getting on with other tasks in parallel. It may even be faster at doing it also. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Include Paths Date: Wed, 09 Jan 2002 16:01:21 +0000 From: Paul Caswell Organization: Land & Sea Systems Message-ID: <3C3C6951.8FF4BB94@baesystems.com> References: <3C3C235D.A2669DFF@baesystems.com> Thanks Both, I changed to forward slashes '/' and it worked with -I --------------------------- Newsgroups: comp.os.vxworks Subject: Re: lspci Date: Wed, 9 Jan 2002 17:46:36 +0100 From: "Michael Lawnick" Organization: Buergernetz Dillingen Message-ID: References: Reply-To: "Michael Lawnick" Sender: mlawnick@pd9e65c63.dip.t-dialin.net Hi Donald, your vxWorks' output says the same as linux. The 1 at the end of 7001 and 7401 is the indicator of I/O-space. Could the memory area at f9.. be a buffer ? Write 0xFFFFFFFF to it and read back again. Look at the unset bits from low to high and you'll have the width of your device's address decoder (which _might_ be the size of device's buffer). - -- Mit freundlichen Grüßen, Michael Lawnick ============================================== SOFTEC GmbH Tel +49-731-96600-0 Promenade 17 Fax +49-731-96600-23 D-89073 Ulm Michael Lawnick Germany lawnick@softec.de ============================================== "Donald McLachlan" schrieb im Newsbeitrag news:a1hm7v$ecv$1@crc-news.crc.ca... > > Background info follows, but my questions are: > > - Why does vxWorks say the card's I/O addresses are at 7001 and 7401 when > Linux says they are at 7000 and 7401? > - Also, why 2 addresses 700X and 740X; begin and end, or aliasing? > - Both report memory address f9000000. SeaLevel says the registers > are only available via I/O ... so what's this memory? > > Thanks, > Don > > ----- > > I'm running booting vxWorks from a floppy on a PC that was used for linux ... > so if I boot from the hard disk I have linux. > > Using the Linux lspci command I get: > > > lspci -v -d 0x135e:0x5102 > 00:0d.0 Serial controller: Sealevel Systems Inc: Unknown device 5102 (rev 01) (prog-if 02 [16550]) > Subsystem: Sealevel Systems Inc: Unknown device 5102 > Flags: medium devsel, IRQ 12 > Memory at f9000000 (32-bit, non-prefetchable) > I/O ports at 7000 > I/O ports at 7400 > > ----- > > Using vxWorks command from the target shell I get: > > -> pciDeviceShow(0) > Scanning function 0 of each PCI device on bus 0 > Using configuration mechanism 1 > bus device function vendorID deviceID class > 00000000 00000000 00000000 00001106 00000691 00000600 > 00000000 00000001 00000000 00001106 00008598 00040600 > 00000000 00000007 00000000 00001106 00000596 00010600 > 00000000 0000000a 00000000 000010b7 00009055 00000200 > 00000000 0000000d 00000000 0000135e 00005102 02000700 > value = 0 = 0x0 > -> pciFindDeviceShow(0x0000135e, 0x00005102, 0) > deviceId = 0x00005102 > vendorId = 0x0000135e > index = 0x00000000 > busNo = 0x00000000 > deviceNo = 0x0000000d > funcNo = 0x00000000 > value = 0 = 0x0 > -> pciHeaderShow(0, 0x0000000d, 0x00000000) > vendor ID = 0x135e > device ID = 0x5102 > command register = 0x0003 > status register = 0x0280 > revision ID = 0x01 > class code = 0x07 > sub class code = 0x00 > programming interface = 0x02 > cache line = 0x08 > latency time = 0x00 > header type = 0x00 > BIST = 0x00 > base address 0 = 0xf9000000 > base address 1 = 0x00007001 > base address 2 = 0x00007401 > base address 3 = 0x00000000 > base address 4 = 0x00000000 > base address 5 = 0x00000000 > cardBus CIS pointer = 0x00000000 > sub system vendor ID = 0x135e > sub system ID = 0x5102 > expansion ROM base address = 0x00000000 > interrupt line = 0x0c > interrupt pin = 0x01 > min Grant = 0x00 > max Latency = 0x00 > value = 0 = 0x0 > > > -- > Donald McLachlan E-mail Donald.McLachlan@crc.ca > Communications Research Centre / RNS Tel (613) 998-2845 > 3701 Carling Ave., Fax (613) 998-9648 > Ottawa, Ontario > K2H 8S2 > Canada > > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Optimizing VxWorks boot time ? Date: Wed, 09 Jan 2002 09:27:48 -0800 From: "Michael R. Kesti" Organization: MK Associates Message-ID: <3C3C7D94.B47C2FCC@gv.net> References: <3c3acff3$0$5452$edfadb0f@dspool01.news.tele.dk> Reply-To: mkesti@gv.net "Søren Abildgaard" wrote: >We are running a project, where we are using a x86 platform >and a vxworks 5.2. > >I am fairly new to VxWorks, but my problem is the boottime we >are getting. It takes about 13 seconds from the BIOS POST ends >until VXworks starts to load our application. >The kernel I have built I have left out all components, which we >don´t need. > >What parameters do I have left to tweak on? And whats dertermines this >boottime, which seems fairly long for an embedded system? I'm not certain how this would apply to your environment, but in a PPC860 project in which I was involed, we GREATLY reduced boot time by enabling the instruction cache toward the end of romInit.s. The BSP from which we derived our BSP had deferred cache enabling to the projects' bootable vxWorks images that are loaded from the host or, in deployed systems, from ROM. - -- ======================================================================== Michael Kesti | "And like, one and one don't make | two, one and one make one." mkesti@gv.net | - The Who, Bargain --------------------------- Newsgroups: comp.os.vxworks Subject: Re: vxworks, hard disk crashes, and NBX-100 Date: Wed, 9 Jan 2002 17:45:18 -0000 From: "Dave Korn" Organization: Lumber Cartel (tinlc) Members #2234-2237 (owing to browser refresh) Message-ID: <9f%_7.372$qS6.886248@newsr2.u-net.net> References: <20020104.202339.928894674.2005@oberon.cardinal.lan> Yan Seiner wrote in message <20020104.202339.928894674.2005@oberon.cardinal.lan>... >All right, I'm a complete vxworks newbie. > >I have an NBX-100 phone system, which runs vxworks. We've had this system >for two months, in which time we have had 2 complete hard disk crashes. > >The system uses Seagate IDE drives, which from experience are rock >reliable. They just don't crash that often. > >The only indication of anything wrong is that we get an error (going from >memory) error on socket 39, errno = 32 on reinstalling the OS. > >3Com ( the vendor) is basically saying keep replacing the disks. I can't >stand to have more downtime. From looking at the OS, it appears to be a >*nix variant, which would make error 32 a broken pipe. I have no idea >what socket 39 might be. > >Any guidance or ideas? Just one thought: that 32 might be a hex number rather than decimal, in which case the error would actually be #define ENOTSOCK 50 /* Socket operation on non-socket */ which might make sense in the context of the socket number - somehow the software is using a normal file descriptor number as if it were a socket, or vice-versa. That sounds to me like something's gone very badly wrong internally, and there's insufficient error-handling in the software to take care of it. Is there any NVRAM in the thing that you can pull the battery out of to get a factory reset? Perhaps it's got some internal configuration info that is badly out-of-kilter. Don't do this without consulting the supplier though, since it could just as easily render the thing unworkable as fix it, it depends ! DaveK - -- moderator of alt.talk.rec.soc.biz.news.comp.humanities.meow.misc.moderated.meow Burn your ID card! http://www.optional-identity.org.uk/ Help support the campaign, copy this into your .sig! Proud Member of the Exclusive "I have been plonked by Davee because he thinks I'm interesting" List Member # --------------------------- Newsgroups: comp.os.vxworks Subject: Re: BootRom and MMU Date: Wed, 09 Jan 2002 17:59:24 GMT From: "Jim" Organization: Magma Communications Ltd. Message-ID: <0y%_7.14882$8e.753501@news> References: Here's a couple things I have noticed using the ICE and visionClcik ... hope it helps ... 1.) In the source window when your stepping over stuff with vision click ... it doesn't always look like the pc moves ... but if the top corner of your source window says your at address 0x02800100 and you decide to change this to 0x02800104 (by simply typing it in and press enter in the dialogue box) then step (using the single step button) you will find it takes you to the correct line ... its really misleading and TERRIBLE that the software behaves this way ... 2.) As well when your stepping realize that the line being pointed to will no execute until you hit enter ... 3.) As well be sure to be careful that you run commands that are using the correct IMMR value ... if you boot your board the bootloader is normally written like you are using the default IMMR value from the hardware reset word ... so becareful if you init the ICE with the "in" command ... but if you initialize the ICE with the command "inn; sr sypcr ff83" you can avoid these conflicts ... Cheers, Jim "Vinh" wrote in message news:c129947c.0201081333.5d35caae@posting.google.com... > Hi all, > I am trying to debug our VxWorks boot for PPC, and i can't seem to > used VisionClick to step over an assembly instruction (tlbie) in the > '../target/templatePpc/sysALib.s'. > This is the sniped of code. > > sysALoop: > tlbie p0 > addi p0,p0,0x1000 /* increment bits 15-19 */ > bdnz sysALoop /* decrement CTR, branch if CTR != 0 */ > sync /* sync instr req'd after tlbie */ > > > Can any one explain to me why we need to invalidates the TLB in > _sysInit? > I though that when VxWorks comes up, it will used it's MMU lib to > initialized the PPC MMU. Some how the instruction caused the > VisionProbe to disconnect. > > > Any help is greatly appreciated. > > Thank you, > Vinh Lam --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Using select(), need to monitor the serial port continuously in the thread Date: Wed, 9 Jan 2002 18:12:26 -0000 From: "Dave Korn" Organization: Lumber Cartel (tinlc) Members #2234-2237 (owing to browser refresh) Message-ID: References: <23de5426.0112302117.12e9fe30@posting.google.com> B.Manivannan wrote in message <23de5426.0112302117.12e9fe30@posting.google.com>... >Good Day Viewers, > >1. I am developing an application which should continuously monitor >the serial port inside the thread for any data arrival. > >2. In a application, i just created a thread. Inside a thread, i am >opening a port. After that FOREVER loop has started. In side the >forever loop, i have given selecet statement as per the example given >by the help. It goes inside the FOREVER loop and got hanged.. > /* Open File Descriptors */ > nOpenReturn = open("/tyCo/0",2,0); Here you open a channel which is already opened by tShell. And you do not check for a return error code. If the open call fails, nOpenReturn will get -1 and all the code which uses it to select() will fail very badly, since file descriptor -1 will never be ready! Also > nReadProcess = taskSpawn ("Reading",200, 0, STACK_SIZE, > (FUNCPTR) ReadingPort,0,0,0,0,0,0,0,0,0,0); your task is running at priority 200, but the tShell task has priority 1. So it is very likely that the shell will intercept all the characters from the serial port and your task will not get them. You could make sure the shell does not interfere by disabling it or by lowering its priority, but even better would be not trying to use the same serial port for two different programs at the same time... DaveK - -- moderator of alt.talk.rec.soc.biz.news.comp.humanities.meow.misc.moderated.meow Burn your ID card! http://www.optional-identity.org.uk/ Help support the campaign, copy this into your .sig! Proud Member of the Exclusive "I have been plonked by Davee because he thinks I'm interesting" List Member # --------------------------- Newsgroups: comp.os.vxworks Subject: Socket back end and fd table of the I/O system Date: 9 Jan 2002 19:07:22 CET From: "Roger Bodén" Message-ID: <3c3c86da@news.wineasy.se> Hi all, I'm implementing a so-called socket back end to a stack of our own. The socket API functions seem to work fine, but if I try to use the functions of the I/O system (e.g. read(), write(), etc) on a socket descriptor, they fail with errno set to S_iosLib_INVALID_FILE_DESCRIPTOR (0xD0003). The starting point of my problems seem to be that iosFdValue() fails for the socket descriptor. My guess is that this is due to that the file descriptor returned by a socket back-end function (e.g. usrSockAccept()) is not registered in the file descriptor table of the I/O system. How can I register my file descriptors with the file descriptor table of the I/O system? - -- Roger Bodén INTERPEAK Box 7498 SE-103 92 Stockholm Sweden Office: +46 8 545 275 80 Fax +46 8 545 275 99 www.interpeak.com --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Bug Report? Loading more than one END Date: Wed, 9 Jan 2002 18:07:13 +0000 (UTC) From: "David Laight" Organization: BT Openworld Message-ID: References: <9d60aef6.0201090358.1d8143cc@posting.google.com> I'm fairly sure that IP_MAX_UNITS is used to size a malloc()ed array during network startup. So you need to have built an image with a larger value (ar at least arrange to increase the corresponding variable before the network code initialises). Just writing in a larget value will cause the code to overwrite things :-) David oj wrote in message news:9d60aef6.0201090358.1d8143cc@posting.google.com... > hi ya all vxWorkers > > i have been struggling for a week with this problem: > i was trying to load two END's , the image i am using has its > IP_MAX_UNITS param > configure to 1, when i tried to load the first END from application > level i got the message to increase my IP_MAX_UNITS , i increase it > from my application before loading the END and everything went just > fine. > the same goes when i tried to load the second END but this time when i > increase the IP_MAX_UNITS from application level i got some weird > exception check error. > i was sure that the problem is somehow relate to the second END > implementation... but it didn't. > only when i change the IP_MAX_UNITS in my image things worked out!!! > > why the hell changing the IP_MAX_UNITS from application level solved > the problem for the first END and not for the second ? i have no > idea. > > if any of u will encounters such a problem this maybe a solution for > ya, > if someone can explain this strange vxWorks behavior please let me > know. > > thanx. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: VME bus interrupts Date: Wed, 09 Jan 2002 11:52:07 -0600 From: James Toland Organization: Raytheon Company Message-ID: <3C3C8347.35726ECB@raytheon.com> References: <3C3B39C0.DBED8355@raytheon.com> The VMetro CCA was not deasserting the interrupt request after the ISR routine was executed.  Reseated the VMetro piggyback module and the interrupts started working properly.  -James

Leonid Rosenboim wrote:

Apparently the VMetro issues interrupts at a too fast rate,too fast for the CPU to handle, ehnce the Panic message,or alternatively, your ISR is doing too much stuff so itcan not handle too many inpteerupts per second.I used VMetro, but it was ages ago and not for generatinginterrupts but to trace various event, it's a hell of a tool,real life saver!See where you control the mode (one-shot or repetitive) andrate of the interrupts the VMetro generates. - Leonid
"James Toland" <j-toland@raytheon.com> wrote in message news:3C3B39C0.DBED8355@raytheon.com...Hello,

I'm having trouble testing my ISR routines.  When I generate a VME bus interrupt using the vxWorks supplied function "sysBusIntGen (level, vector)", my ISR behaves as expected.  However, when I generate a VME bus interrupt using the VMetro Bus Analyzer for the same level and vector, vxWorks announces: "workQpanic: kernel work queue overflow... autoboot in progress"?!!!  Has anyone out there used a VMetro Bus analyzer to generate interrupts?  Anyone know what I'm doing incorrectly?  Some simple code below demonstrates this problem.

Environment: MVME2700 SBC with VxWorks 5.4 and a VMetro Bus Analyzer to capture VME bus activity and generate interrupts.

#include "vxWorks.h"
#include "logLib.h"
#include "sysLib.h"
#include "intLib.h"
#include "iv.h"

void my_Isr (int vector)
{
  logMsg ("Received: %x \n", vector, 0,0,0,0,0);
}

void int_setup (int level, int vector)
{
  intConnect( (VOIDFUNCPTR *)INUM_TO_IVEC(vector), my_Isr, vector);
  sysIntEnable(level);
}

Tornado Shell prompt -> int_setup( 1, 0x64 )
Tornado Shell prompt -> sysBusIntGen ( 1, 0x64)
ISR prints "Received: 64" to console.

-OR-

Tornado Shell prompt -> int_setup( 1, 0x64 )
initiate VMetro interrupt level 1 vector 0x64
ERROR: workQpanic and reboot!

Thanks
 
 

--------------------------- Newsgroups: comp.os.vxworks Subject: Re: why does taskDelay(1) take so long? Date: Wed, 09 Jan 2002 20:37:10 GMT From: Ephraim Gadsby Organization: - minimal Message-ID: References: <67868445.0201071418.5c211c82@posting.google.com> <67868445.0201081838.39d07205@posting.google.com> On 8 Jan 2002 18:38:23 -0800, eaglejump@yahoo.com (Bob Johnson) wrote: >Thanks to Donald and everybody else for their help! I am fairly >certain that a low priority task is locking interrupts. It only >unlocks the interrupts on 20 tick boundaries. I'd be interested to know what is going on there. If this low-priority task locks-out the clock interrupts, how does it know when the 20 ticks boundary is reached? It could spin on a counter that is synchronized to the system clock, but in that case wouldn't most of the clock interrupts be lost completely, so that the tick-count would not increment by twenty. Even if the clock can queue-up 20 interrupts, then the fact that taskDelay(1) always returns on 20 tick boundaries suggests that the task is spinning with interrupts locked-out for almost all the time. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Socket back end and fd table of the I/O system Date: Wed, 9 Jan 2002 23:19:04 +0200 From: "Leonid Rosenboim" Organization: Verio Message-ID: References: <3c3c86da@news.wineasy.se> I beleive that to write this back end properly, you need access to the source code of the iosLib socket binding functions, or at least hire someone who has access to or remembers these sources by heart ;-) - - Leonid "Roger Bodén" wrote in message news:3c3c86da@news.wineasy.se... > Hi all, > > I'm implementing a so-called socket back end to a stack of our own. The > socket API functions seem to work fine, but if I try to use the functions of > the I/O system (e.g. read(), write(), etc) on a socket descriptor, they fail > with errno set to S_iosLib_INVALID_FILE_DESCRIPTOR (0xD0003). The starting > point of my problems seem to be that iosFdValue() fails for the socket > descriptor. My guess is that this is due to that the file descriptor > returned by a socket back-end function (e.g. usrSockAccept()) is not > registered in the file descriptor table of the I/O system. > > How can I register my file descriptors with the file descriptor table of the > I/O system? > > -- > Roger Bodén > INTERPEAK > Box 7498 > SE-103 92 Stockholm > Sweden > Office: +46 8 545 275 80 > Fax +46 8 545 275 99 > www.interpeak.com > > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: why does taskDelay(1) take so long? Date: Wed, 09 Jan 2002 22:09:20 GMT From: "Frank" Message-ID: References: <67868445.0201071418.5c211c82@posting.google.com> <67868445.0201081838.39d07205@posting.google.com> Hi, VxWorkser I got another question: If I set sysClkRate=2000, Does that take lots of overhead to OS kernel ? As I know from document, normally the sysClkRate should be set "<600",while default is "60". Is it good to set "2000" so high in real product? Frank - -- "Ephraim Gadsby" wrote in message news:a92p3uc8e8os3ean6cesvjrcssu5g9oiuc@4ax.com... > On 8 Jan 2002 18:38:23 -0800, eaglejump@yahoo.com (Bob Johnson) wrote: > > >Thanks to Donald and everybody else for their help! I am fairly > >certain that a low priority task is locking interrupts. It only > >unlocks the interrupts on 20 tick boundaries. > > I'd be interested to know what is going on there. > > If this low-priority task locks-out the clock interrupts, how does it > know when the 20 ticks boundary is reached? > > It could spin on a counter that is synchronized to the system > clock, but in that case wouldn't most of the clock interrupts be lost > completely, so that the tick-count would not increment by twenty. > > Even if the clock can queue-up 20 interrupts, then the fact that > taskDelay(1) always returns on 20 tick boundaries suggests that the > task is spinning with interrupts locked-out for almost all the time. --------------------------- Newsgroups: comp.os.vxworks Subject: "end.h" included in C++ code in Tornada 2.0.2 Date: Wed, 9 Jan 2002 17:00:40 -0500 From: "Jiandong Wang" Organization: Nortel Message-ID: Hi, Guys, I meet a problem about compiling C++ code in which the "end.h" is included in Tornado 2.0.2 . Compiler always tells there are syntax errors in m2Lib.h . ( No error in Tornado 2.0.1) Anybody met such a problem? Thanks in advance for help! Here are 5 short files to describle this problem. - ---- test11.c ----- #include "vxWorks.h" #include "end.h" #include "test12.h" int test11() { int i; i = callObjectMemberFunction(); return i; } - ---- test12.h ---- int callObjectMemberFunction(); - ---- test12.cpp ---- #include "vxWorks.h" #include "end.h" #include "testcpp.h" #include "test12.h" extern "C" int callObjectMemberFunction() { CTemp aaa; return aaa.func1(); } - ---- testcpp.h---- #ifndef TEST_CPP_INCLUDED #define TEST_CPP_INCLUDED class CTemp { public: CTemp(){}; ~CTemp(){}; public: int func1(); }; #endif - ---- testcpp.cpp ---- #include "vxWorks.h" #include "end.h" #include "testcpp.h" int CTemp::func1() { return 5; } - -------- end of 5 short files --------- --------------------------- Newsgroups: comp.os.vxworks Subject: Re: why does taskDelay(1) take so long? Date: Wed, 9 Jan 2002 22:16:01 +0000 (UTC) From: don@mars.dgrc.crc.ca (Donald McLachlan) Organization: The unconfigured xvnews people Message-ID: References: Reply-To: don@mars.dgrc.crc.ca In article , "Frank" writes: > Hi, VxWorkser > I got another question: > If I set sysClkRate=2000, Does that take lots of overhead to OS kernel ? > As I know from document, normally the sysClkRate should be set "<600",while > default is "60". > Is it good to set "2000" so high in real product? > > Frank Depends on processor speed and interupt efficiency. Seat of the pants answer: - - run spy to find out % time spent in kernel. - - call sysClkRateSet() - - call clock_setres() - - run spy to find out % time spent in kernel. adjust as you see fit. - -- Donald McLachlan E-mail Donald.McLachlan@crc.ca Communications Research Centre / RNS Tel (613) 998-2845 3701 Carling Ave., Fax (613) 998-9648 Ottawa, Ontario K2H 8S2 Canada --------------------------- Newsgroups: comp.os.vxworks Subject: Co-processer unuseable exception Date: Wed, 09 Jan 2002 17:28:53 -0500 From: Stephen Roderick Organization: University of Maryland College Park Message-ID: <3C3CC423.DC65B2E6@ssl.umd.edu> Hi We're intermittently getting one of these on a MIPS R4000 based board. The function it occurs in is very simple, and is simply doing some unit conversion from floats to integers. We've got overflow checking/protection built in there, though this shouldn't be the issue. We've also looked at the MIPS manuals for some clues but they're not particularly helpful (whatever happened to RTFM working for you? :-) ) Is anyone able to shed some light on a possible reason? TIA Stephen --------------------------- Newsgroups: comp.os.vxworks Subject: Problem with __FiPc with ipAttach() Date: Thu, 10 Jan 2002 06:19:48 GMT From: "Nejhdeh Ghevondian" Organization: Integrated Spectronics Message-ID: <8oa%7.12718$9u6.50804@news-server.bigpond.net.au> Reply-To: "Nejhdeh Ghevondian" Can anyone help me!! When I include the following routine in my application source: #include { ipAttach(0,"fei"); } the Tornado environment adds __FiPc to the end of ipAttach, which ends up being: ipAttach__FiPc This becomes a problem since there is a symbol conflict with VxWorks image file, since it expects ipAttach(). Why does this happen? Rergards Nejhdeh Ghevondian --------------------------- Newsgroups: comp.os.vxworks Subject: [HELP]UDP TX/RX packet mismatch problem... Date: Thu, 10 Jan 2002 17:02:16 +0900 From: "chihwankim" Organization: ELIMnet Internet News Message-ID: I have two MPC8260 boards A and B. I open UDP socket both of them and try to transmit and receive UDP packets for each size of datagram. I assume that A is transmitting UDP datagram to B every each 10ms and B is receiving datagram from A. For 256/512/1024 bytes of datagram, there is no error in the receiving side. But 64/128 bytes of datagram, there occurs missed packets in the receiving side. So, I added some debugging code to check every ethernet packet to count the number of packets with specified UDP port in BSP. But results is just the same with above. Anyone has some ideas about this problem? Thanks. chkim in korea. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Bug Report? Loading more than one END Date: 10 Jan 2002 09:44:55 GMT From: Johan Borkhuis Organization: Agere Systems Message-ID: References: <9d60aef6.0201090358.1d8143cc@posting.google.com> odedn@shiron.com (oj) wrote: > i have been struggling for a week with this problem: > i was trying to load two END's , the image i am using has its > IP_MAX_UNITS param > configure to 1, when i tried to load the first END from application > level i got the message to increase my IP_MAX_UNITS , i increase it > from my application before loading the END and everything went just > fine. IP_MAX_UNITS is a macro, you have to set it compile time. Groeten, Johan - -- o o o o o o o . . . _____________________________ o _____ || Johan Borkhuis | .][__n_n_|DD[ ====_____ | borkhuis@agere.com | >(________|__|_[_________]_|__________________________| _/oo OOOOO oo` ooo ooo 'o!o!o o!o!o` === VxWorks FAQ: http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html === --------------------------- Newsgroups: comp.os.vxworks Subject: Re: [HELP]UDP TX/RX packet mismatch problem... Date: Thu, 10 Jan 2002 06:39:18 -0500 From: joe durusau Organization: Bellsouth.Net Message-ID: <3C3D7D66.2D999429@bellsouth.net> References: Sounds like you might be having collisions. Note that UDP is dy definition not guaranteed to work. You need either an absolutely quiet net into which to transmit, or some higher level has to check for errors and arrange for re-transmission if required. Speaking only for myself, Joe Durusau chihwankim wrote: > I have two MPC8260 boards A and B. I open UDP socket both of them and try to > transmit and receive UDP packets for each size of datagram. I assume that A > is transmitting UDP datagram to B every each 10ms and B is receiving > datagram from A. > > For 256/512/1024 bytes of datagram, there is no error in the receiving side. > But 64/128 bytes of datagram, there occurs missed packets in the receiving > side. > > So, I added some debugging code to check every ethernet packet to count the > number of packets with specified UDP port in BSP. But results is just the > same with above. > > Anyone has some ideas about this problem? > Thanks. > chkim in korea. --------------------------- End of New-News digest ********************** From vxwexplo-errs@csg.lbl.gov Thu Jan 10 06:50:56 2002 From: "Lisa P. Rego" Date: Thu Jan 10 06:50:58 PST 2002 Subject: Memory Loss using Green Hills Tornado Tools We are running VxWorks 5.3.1 under Tornado 1.0.1, and are using the Green Hills 1.8.9 Tornado PowerPC Development Environment. Our target is a PPC 2700 board. We boot from and do all File I/O via a SCSI hard-drive using vxWorks dosFs1 library calls. We have an embedded Fortran application which is spawned from a C wrapper as a vxWorks task. This application was originally written to run in batch mode as a Unix process. Because it is non-reentrant, we unload the module after each invocation, and reload it upon the next invocation. One problem we have is that each time we run this application (which does File I/O of some pretty large files, among other things), we exhibit a memory loss of about 5 Kbytes each time. We had been theorizing that our old Fortran task may not have been real good at file i/o maintenance, and was possibly leaving files open without closing them. From what we understand, under vxWorks, each time a file is opened, a buffer is allocated which is (num Sectors per cluster) * (number of bytes per sector) size. With our partitionning, this would be about 72 Kbytes, which doesn't jive with the memory loss we see. Our C wrapper calls a Green Hills routine, _ghexit, to clean up after the Fortran application. We can find little documentation about this routine, and I was wondering if it really did clean up File I/O as the comments imply. Has any one had any experience with the Green Hills development environment that could help me out? At this point we are trying to narrow down where the memory leak is coming, but it seems quite likely that it could be either our Fortran application or the Fortran libraries. Thanks very much! Lisa ------------------------------------------------------------------------------- Lisa P. Rego Raytheon Electronic Systems Division 1847 W. Main Rd Portsmouth, RI 02871 (401) 842-4212 (office) (401) 842-2911 (lab) lap1@ssd.ray.com From vxwexplo-errs@csg.lbl.gov Thu Jan 10 22:47:19 2002 From: pranjan@hss.hns.com Date: Thu Jan 10 22:47:21 PST 2002 Subject: Timestamp resolution on Vx-works Hi, I have a query regarding the resolution of the timestamp on Vx-works. To get a feel of the execution time of one of our utility functions, we try to get the time- stamp before the function is called and time- stamp after the function has been called. For example, t1 = clock_gettime() function() t2= clock_gettime() But (t2-t1) always comes out to be zero. We tried changing the resolutions both programatically and compile time. We had set the resolution to 10 ns but again the time difference was 0. We had enabled HighResolution Time stamp in the vxWorks image also. I am not sure if we have missed out something. Can someone please throw some light on what needs to be done. with regards, ranjan. Hughes Software Systems From vxwexplo-errs@csg.lbl.gov Fri Jan 11 01:28:13 2002 From: "Pradeep G" Date: Fri Jan 11 01:28:15 PST 2002 Subject: Controlling FTP process. Hi, Though this query is not directly related to VxWorks, I think the problem is relevant to VxWorks also. We are using OSE Delta 4.1.1 for our project. We are using MPC 860 processor with 32 MB MMC. It crashes when I try to put some big file into it when there is no sufficient space left in MMC (for obvious reasons). I need to control the download of such a big file to MMC when there is less space in MMC. Is there any way I can monitor the FTP process. Thanks in Advance pradeep Hughes Software Systems From vxwexplo-errs@csg.lbl.gov Fri Jan 11 02:36:00 2002 From: NAVEEN BHATIA Date: Fri Jan 11 02:36:03 PST 2002 Subject: Segmentation violation Error in VxWorks hi, i am trying the following code in VxWorks: #include void main (void) { char *data_PTR; data_PTR = (char *) malloc (100); /* assign a 64-bit value */ *((unsigned long long *)data_PTR) = 1; } When I run this snippet i get the following error. What could be wrong?( i have taken the memory allocated to be big enough to not let the pointer data_PTR fall off it.) -> main Segmentation violation program counter: 0x00016b88 next program counter: 0x00016b8c processor status: 0xfe001004 access address: 0x00000000 162a8 vxTaskEntry +c : shell (1, 0, 0, 0, 0, 0) 79034 shell +154: 7904c (0, 3, 7f, 0, 0, 0) 79240 shell +360: execute (3b1198, 0, 80, 0, 3b1198, 3b0f80) 7937c execute +c4 : yyparse (40e0f0, 10c800, ffffffff, 3b0f80, 3b0f78, 40f688) dcc14 yyparse +74c: db06c (40e0c0, 40e0e0, 10, 2bb, 4, 0) db1b8 yystart +728: 0 (0, 0, 0, 0, 0, 0) shell restarted. Please help. Naveen Bhatia, Logic Eastern, NOIDA, INDIA __________________________________________________ Do You Yahoo!? Send FREE video emails in Yahoo! Mail! http://promo.yahoo.com/videomail/ From vxwexplo-errs@csg.lbl.gov Fri Jan 11 04:03:19 2002 From: Vxworks Exploder Date: Fri Jan 11 04:03:22 PST 2002 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Fri Jan 11 04:03:14 PST 2002 Subject: Re: Optimizing VxWorks boot time ? Subject: Re: Optimizing VxWorks boot time ? Subject: Change module name Subject: Re: Problem with __FiPc with ipAttach() Subject: Re: Change module name Subject: Re: Optimizing VxWorks boot time ? Subject: Re: [HELP]UDP TX/RX packet mismatch problem... Subject: Re: DosFs 2.0 Product Information Subject: Re: Problem with __FiPc with ipAttach() Subject: Re: Optimizing VxWorks boot time ? Subject: Re: Optimizing VxWorks boot time ? Subject: Re: [HELP]UDP TX/RX packet mismatch problem... Subject: gigabit PMCs, SBCs with integrated RACEway++ questions Subject: Visual Studio Integration Subject: VxWorks Boot on MVME5100 (PPC 7400) and Tornado 2.1 Subject: HELP finding a newer LIBGCC.A with source code and/or building instructions please Subject: Re: [HELP]UDP TX/RX packet mismatch problem... Subject: Re: I/O ports on PowerPC Subject: Re: Caching, I/O Space memory mapping and DMA Subject: Re: LC82C169 Driver... Subject: supplement Subject: Re: _protoSwIndex and inetsw[ ] used in our OSPF implementation Subject: Re: bus error exception in ppc8260 Subject: _protoSwIndex and inetsw[ ] used in our OSPF implementation Subject: Re: pci error in IDT Rc32332 Subject: Re: bus error exception in ppc8260 Subject: mpc850 upmx program question Subject: DosFS2 or the device driver Subject: Re: LC82C169 Driver... Subject: Re: supplement Subject: WIND_BASE vs TGT_DIR Subject: Help with Interrupts and Console Subject: flash and MMU Subject: [Q] Problem when making VxWorks bootrom and boot image Subject: Re: Caching, I/O Space memory mapping and DMA Subject: Re: VxWorks Boot on MVME5100 (PPC 7400) and Tornado 2.1 Subject: Tornado project launcher not working ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: Optimizing VxWorks boot time ? Date: Thu, 10 Jan 2002 13:26:22 +0100 From: "Romano Signorelli" Organization: Centro Servizi Interbusiness Message-ID: References: <3c3acff3$0$5452$edfadb0f@dspool01.news.tele.dk> <3c3aebda$1@pull.gecm.com> <3c3c2d00$0$89102$edfadb0f@dspool01.news.tele.dk> <0jch1a.q4t.ln@gum.de> If not private, is it possibile to take a look at the source code? Thanks in advance, bye Romano Signorelli "Frank Wolf" ha scritto nel messaggio news:0jch1a.q4t.ln@gum.de... > Hi there! > > We've modified the routines > > LOCAL void copyLongs (source, destination, nlongs); > LOCAL void fillLongs (buf, nlongs, val); > > to use the DMA controller instead of CPU to perform this task. > This helped us to get far lower startup times. > > Frank --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Optimizing VxWorks boot time ? Date: Thu, 10 Jan 2002 13:28:50 +0100 From: "Romano Signorelli" Organization: Centro Servizi Interbusiness Message-ID: References: <3c3acff3$0$5452$edfadb0f@dspool01.news.tele.dk> <3c3aebda$1@pull.gecm.com> <3c3c2d00$0$89102$edfadb0f@dspool01.news.tele.dk> <0jch1a.q4t.ln@gum.de> <3c3c499d$0$35462$edfadb0f@dspool01.news.tele.dk> <3c3c6424$1@pull.gecm.com> If not private, is it possibile to take a look at the source code? Thanks in advance, bye Romano Signorelli P.S.: sorry for multi-post message "Frank Wolf" ha scritto nel messaggio news:0jch1a.q4t.ln@gum.de... > Hi there! > > We've modified the routines > > LOCAL void copyLongs (source, destination, nlongs); > LOCAL void fillLongs (buf, nlongs, val); > > to use the DMA controller instead of CPU to perform this task. > This helped us to get far lower startup times. > > Frank --------------------------- Newsgroups: comp.os.vxworks Subject: Change module name Date: Thu, 10 Jan 2002 14:26:48 +0100 From: "Bostjan Mulej" Organization: Slovenija OnLine - SiOL Message-ID: Hi! Is it possible to change the name of already loaded module? I searched through documentation, but I didn't found anything useful. Regards, Bostjan --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Problem with __FiPc with ipAttach() Date: Thu, 10 Jan 2002 15:25:13 +0100 From: "Werner Schiendl" Message-ID: <3c3da44b$1@brateggebdc5.br-automation.co.at> References: <8oa%7.12718$9u6.50804@news-server.bigpond.net.au> Hi, you are compiling C++ source? Seems the header file does not care for this properly, so the compiler thinks ipAttach uses C++ bindings. Try to use extern "C" { #include } to overcome this problem hth Werner "Nejhdeh Ghevondian" wrote in message news:8oa%7.12718$9u6.50804@news-server.bigpond.net.au... > Can anyone help me!! > When I include the following routine in my application source: > > #include > { > ipAttach(0,"fei"); > } > > the Tornado environment adds __FiPc to the end of ipAttach, which ends up > being: > ipAttach__FiPc > > This becomes a problem since there is a symbol conflict with VxWorks image > file, since it expects > ipAttach(). > > Why does this happen? > > Rergards > > Nejhdeh Ghevondian > > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Change module name Date: Thu, 10 Jan 2002 16:45:11 +0200 From: "Leonid Rosenboim" Organization: Verio Message-ID: References: It's one of those questions, first reaction, oops I dont know that, second reaction, what the heck for? Module names dont mean much, they are only used for correlating with sources by the source code debugger, no more. FUnction names dont matter either, as the external address resolution happens during loading. If you can figure out a darn good reason to dive into this mud, you can do this using symTblLib low level functions to mess with the module names as stored in a specific symbol table with low level symLib functions. HTH, - Leonid "Bostjan Mulej" wrote in message news:rEg%7.671$a6.35740@news.siol.net... > Hi! > > Is it possible to change the name of already loaded module? > > I searched through documentation, but I didn't found anything > useful. > > Regards, Bostjan > > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Optimizing VxWorks boot time ? Date: Thu, 10 Jan 2002 15:23:36 +0000 (UTC) From: David Laight Organization: BT Openworld Message-ID: <3C3DB247.96CCF127@l8s.co.uk> References: <3c3acff3$0$5452$edfadb0f@dspool01.news.tele.dk> <3C3C7D94.B47C2FCC@gv.net> "Michael R. Kesti" wrote: > > "Søren Abildgaard" wrote: > > >We are running a project, where we are using a x86 platform > >and a vxworks 5.2. > > > >I am fairly new to VxWorks, but my problem is the boottime we > >are getting. It takes about 13 seconds from the BIOS POST ends > >until VXworks starts to load our application. > >The kernel I have built I have left out all components, which we > >don´t need. > > > >What parameters do I have left to tweak on? And whats dertermines this > >boottime, which seems fairly long for an embedded system? > > I'm not certain how this would apply to your environment, but in a > PPC860 project in which I was involed, we GREATLY reduced boot time > by enabling the instruction cache toward the end of romInit.s. The > BSP from which we derived our BSP had deferred cache enabling to the > projects' bootable vxWorks images that are loaded from the host or, > in deployed systems, from ROM. Things will speed up no end if you enable the I cache and D cache - - especially for the ROM space. Beware that anything that writes to the EEPROMS needs uncached access, probably best to map them twice! You probably need to flush the D cache before you try to execute anything that is still in it. Ideally this should be done after the uncompress calls - but you can't (easily) add code there. The start of usrInit() will probbaly do (all/usrConfig.c and all/bootConfig.c). This problem may be an ARM one though, I added code to flush the mini-cache. David --------------------------- Newsgroups: comp.os.vxworks Subject: Re: [HELP]UDP TX/RX packet mismatch problem... Date: Thu, 10 Jan 2002 15:32:19 +0000 (UTC) From: David Laight Organization: BT Openworld Message-ID: <3C3DB452.4DFBFCBF@l8s.co.uk> References: <3C3D7D66.2D999429@bellsouth.net> joe durusau wrote: > > Sounds like you might be having collisions. Note that UDP is dy > definition > not guaranteed to work. You need either an absolutely quiet net into which > to transmit, or some higher level has to check for errors and arrange for > re-transmission if required. Not collisions - they are completely benign. The ethernet MAC engine will automatically retransmit them. > > Speaking only for myself, > > Joe Durusau > > chihwankim wrote: > > > I have two MPC8260 boards A and B. I open UDP socket both of them and try to > > transmit and receive UDP packets for each size of datagram. I assume that A > > is transmitting UDP datagram to B every each 10ms and B is receiving > > datagram from A. Are you sure you are sending one every 10ms? That is only 100/sec nothing should fail at that data rate! > > > > For 256/512/1024 bytes of datagram, there is no error in the receiving side. > > But 64/128 bytes of datagram, there occurs missed packets in the receiving > > side. So big packets are ok, but you get errors with small ones. Mmmmmm.... If you really do have the tx rate capped properly then something is definitely awol in the receiving system. If you don't have the tx rate capped, then you are probably seeing the effect of marginally different costs in transmitting and receiving packets. (cost/byte + cost/packet) Check with some LAN monitor software that the packets are all txed. David --------------------------- Newsgroups: comp.os.vxworks Subject: Re: DosFs 2.0 Product Information Date: Thu, 10 Jan 2002 18:18:54 +0200 From: "Leonid Rosenboim" Organization: Verio Message-ID: References: <52f21cc5.0112211105.7e146316@posting.google.com> Guys, just for summary: DosFs 2.0 has a problem with Tornado 2.x on PowerPC 603/604 architecture (or other derivatives using the 603/604 compule flag), which is due to PowerPC compiler improper code generation for 64-bit integer operations used throughout DosFs 2.0. This bug could is tricky to catch, it will bite you only if you have some tasks doing Floating Point calculation, and task switch taking place between these and other tasks doing DosFs I/O in midst of 64-bit arithmetics. So do not be fooled to beleive you eint got it! For those who got DosFs 2.0 source code, this can be solved by either compiling the library with -msoft-float compile flag, or by disabling 64-bit math (not recommended). For binary only licensees, you need a patch which is available from WINDSURF. References: SPR # 71089 Additional SPR # 33221,33702,33684,34704,62415,28428 "Bob Waisnor" wrote in message news:52f21cc5.0112211105.7e146316@posting.google.com... > Looked at this DosFs 2.0 file corruption (file size -1)problem for a > LONG time. Both Windriver and M-systems did the "Gee, it shouldn't do > that". Come to find out that the problem is in the dosfs 2.0 binary. > Here we re-compiled and did not use 2 Gig support and the thing works. > My guess it's a compile issue.... > > "Don Dewar" wrote in message news:... > > I was also shocked to find this out. I called my FAE who said I had to pay > > for it. When I pointed out I heard it was free with a support contract he > > relented and sent me the binaries. They charge an outrageous amount of > > money for the sources. > > > > The integration went ok, but we suddenly found some of our files > > disappearing or getting corrupt. We tried to debug it for months with very > > little support from WRS. Finally, we got the sources and still could not > > figure it out. When we compiled with less than 2 MB support all our > > problems all vanished. > > > > Anyway they don't seem to support it very well. > > > > Good luck. > > -- Don > > "Leonid Rosenboim" wrote in message > > news:newscache$w61gog$hug$1@lnews.actcom.co.il... > > > I tried to find the technical specifications of DosFs 2.0 on the WindRiver > > > Web site to show someone, and to my surprise, I found no mention of it. > > > This seems shocking to me that this optional component is not advertised, > > > and I remain puzzled as to the reasons why. > > > > > > Surely I don't expect any official answer from WindRiver (although that > > > might > > > be nice to chat again with my old pals), but I would appreciate if anyone > > > can > > > shed some light on this enigma. > > > > > > Of course this is mere curiosity. > > > - Leonid > > > > > > > > > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Problem with __FiPc with ipAttach() Date: Thu, 10 Jan 2002 14:31:41 -0000 From: "Dave Korn" Organization: Lumber Cartel (tinlc) Members #2234-2237 (owing to browser refresh) Message-ID: References: <8oa%7.12718$9u6.50804@news-server.bigpond.net.au> Nejhdeh Ghevondian wrote in message <8oa%7.12718$9u6.50804@news-server.bigpond.net.au>... >Can anyone help me!! >When I include the following routine in my application source: > >#include >{ >ipAttach(0,"fei"); >} > >the Tornado environment adds __FiPc to the end of ipAttach, which ends up >being: >ipAttach__FiPc > >This becomes a problem since there is a symbol conflict with VxWorks image >file, since it expects >ipAttach(). > >Why does this happen? It happens because the file with your code making the call is in C++, and C++ uses name-mangling to make sure that you can't call a function passing the wrong arguments. However the O/S functions are all in plain C, so they don't have mangled names. WRS have taken care of this in their header files by wrapping the contents with #ifdef __cplusplus extern "C" { #endif at the start and #ifdef __cplusplus } #endif at the end; this tells the C++ compiler that the .h file refers to functions in a .C file, so the compiler knows not to mangle their names. Unfortunately, WRS have forgotten to include these parts in the ipProto.h header file, so you'll have to do it yourself, by wrapping the #include in your program in extern "C", like this: extern "C" { #include } DaveK - -- moderator of alt.talk.rec.soc.biz.news.comp.humanities.meow.misc.moderated.meow Burn your ID card! http://www.optional-identity.org.uk/ Help support the campaign, copy this into your .sig! Proud Member of the Exclusive "I have been plonked by Davee because he thinks I'm interesting" List Member # --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Optimizing VxWorks boot time ? Date: Thu, 10 Jan 2002 11:08:11 -0600 From: "Tuan Pham" Organization: Motorola NSS Message-ID: References: <3c3acff3$0$5452$edfadb0f@dspool01.news.tele.dk> <3c3aebda$1@pull.gecm.com> <3c3c2d00$0$89102$edfadb0f@dspool01.news.tele.dk> <0jch1a.q4t.ln@gum.de> Hello all, I just wanted some more specifics of using the DMA controller to perform this task? Is there a built in library? Thanks in advance. - -Tuan Pham "Frank Wolf" wrote in message news:0jch1a.q4t.ln@gum.de... > Hi there! > > We've modified the routines > > LOCAL void copyLongs (source, destination, nlongs); > LOCAL void fillLongs (buf, nlongs, val); > > to use the DMA controller instead of CPU to perform this task. > This helped us to get far lower startup times. > > Frank > > > "Søren Abildgaard" wrote in message > news:3c3c2d00$0$89102$edfadb0f@dspool01.news.tele.dk... > > We have 32 MB of RAM on-board and all is currently > > reported to VxWorks. I could proberly reduce this to > > half the size. I will test out asap. Any other hints of how to > > attack the boot time problem?? > > > > "Martin Dowie" wrote in message > > news:3c3aebda$1@pull.gecm.com... > > > "Søren Abildgaard" wrote in message > > > news:3c3acff3$0$5452$edfadb0f@dspool01.news.tele.dk... > > > > Hi *.*, > > > > > > > > We are running a project, where we are using a x86 platform > > > > and a vxworks 5.2. > > > > > > > > I am fairly new to VxWorks, but my problem is the boottime we > > > > are getting. It takes about 13 seconds from the BIOS POST ends > > > > until VXworks starts to load our application. > > > > The kernel I have built I have left out all components, which we > > > > don´t need. > > > > > > > > What parameters do I have left to tweak on? And whats dertermines this > > > > boottime, which seems fairly long for an embedded system? > > > > > > How much memory have you told VxWorks you have and how does that > > > compare with how much you actually use? We have found that reducing > > > the amount you tell VxWorks you have can make a considerable > > > difference to the time. > > > > > > > > > > > > > > > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Optimizing VxWorks boot time ? Date: 10 Jan 2002 12:42:52 -0500 From: Bill Pringlemeir Organization: Bell Sympatico Message-ID: References: <3c3acff3$0$5452$edfadb0f@dspool01.news.tele.dk> <3c3aebda$1@pull.gecm.com> <3c3c2d00$0$89102$edfadb0f@dspool01.news.tele.dk> <0jch1a.q4t.ln@gum.de> Sender: bpringlemeir@DeadDuck >>> "Frank Wolf" wrote: >> We've modified the routines >> >> LOCAL void copyLongs (source, destination, nlongs); LOCAL void >> fillLongs (buf, nlongs, val); >> >> to use the DMA controller instead of CPU to perform this task. >> This helped us to get far lower startup times. >>>>> "Tuan" == Tuan Pham writes: Tuan> Hello all, I just wanted some more specifics of using the DMA Tuan> controller to perform this task? Is there a built in library? Tuan> Thanks in advance. There are no vxWorks standard DMA routines. There might be for your BSP. However, often DMA is used to move data between hardware and the main memory. To zero or copy a block you often have to do memory to memory transfers. I believe that you are using a PC/x86 based system so I will make no comment on what your hardware can do [as I know nothing about x86's]. Anyways, there are no standard vxWorks routines and you might have to do some work to use DMA. regards, Bill Pringlemeir. - -- TORPEDOED BY NETSCAPE. SINKING. U-181. vxWorks FAQ, "http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html" --------------------------- Newsgroups: comp.os.vxworks Subject: Re: [HELP]UDP TX/RX packet mismatch problem... Date: Thu, 10 Jan 2002 17:59:58 GMT From: Dan Gold Organization: http://www.newsranger.com Message-ID: References: <3C3D7D66.2D999429@bellsouth.net> <3C3DB452.4DFBFCBF@l8s.co.uk> In article <3C3DB452.4DFBFCBF@l8s.co.uk>, David Laight says... > >>> chihwankim wrote: >> >> > I have two MPC8260 boards A and B. I open UDP socket both of them and try to >> > transmit and receive UDP packets for each size of datagram. I assume that A >> > is transmitting UDP datagram to B every each 10ms and B is receiving >> > datagram from A. > Are these custom 8260 designs or evaluation boards ? And the software (i.e. the drivers) ? Did you do the port, or are they stock ? Are you sure you have the latest drivers ? I know WRS has fixed at least 1 or 2 bugs in the FCC driver... I'm assuming this is Ethernet on the FCC, not SCC. And do you only have 1 FCC active ? >> > >> > For 256/512/1024 bytes of datagram, there is no error in the receiving side. >> > But 64/128 bytes of datagram, there occurs missed packets in the receiving >> > side. > >So big packets are ok, but you get errors with small ones. >Mmmmmm.... > >If you really do have the tx rate capped properly then something is >definitely awol in the receiving system. > >If you don't have the tx rate capped, then you are probably seeing >the effect of marginally different costs in transmitting and >receiving packets. (cost/byte + cost/packet) > >Check with some LAN monitor software that the packets are all txed. The original poster said he added some code in the BSP to count packets with the correct UDP port - was that on the Tx or Rx side ? If it was the Rx, you might want to add code to the Tx side to make sure all packets are getting Tx'd. Also, do you see any MAC layer Tx or Rx errors ? If you're sure the packets are being Tx'd, you might {ip,udp}StatShow and see if they tell you anything. See what statistics are available to you from the driver on both Tx and Rx sides, that may give some clue. hth, Dan Gold gold@ensemble.com --------------------------- Newsgroups: comp.os.vxworks Subject: gigabit PMCs, SBCs with integrated RACEway++ questions Date: Thu, 10 Jan 2002 19:30:58 GMT From: Randy Ryan Organization: Our-Kids Message-ID: <3C3DEBF2.956A0F15@pacbell.net> 1: Anybody using Gigabit PMC modules and achieving good throughput rates. Trying to narrow down the fantasy from the real :) 2: Is anybody using an SBC with integrated RACEWay++ chip, such as Transtech-DSP/DNA ? The diagrams imply that data can flow directly from the PMC site to the RACEWay chip. Can you also simultaneously move data into the memory through the bridge too? How problematic is it to move data from that PMC bus side into memory for processing and back out to the RACEWay++ chip (while still acquiring from the PMC). Thanks much! Randy NovaSol, http://www.nova-sol.com Looking for Sr. Engineers who want to live and work in Hawaii. --------------------------- Newsgroups: comp.os.vxworks Subject: Visual Studio Integration Date: Thu, 10 Jan 2002 19:38:26 GMT From: "Chad" Organization: Excite@Home - The Leader in Broadband http://home.com/faster Message-ID: Hi, I got a sample makefile from the TornadoII FAQ I'm working with to try and allow me to browse and compile my ppc project source from visual studio. Does anybody know where I can get more info on doing this? --------------------------- Newsgroups: comp.os.vxworks Subject: VxWorks Boot on MVME5100 (PPC 7400) and Tornado 2.1 Date: Thu, 10 Jan 2002 20:25:13 GMT From: "Kyle Rutledge" Organization: Prodigy Internet http://www.prodigy.com Message-ID: I've flashed VxWorks Boot onto a MVME5100 card and it boots to that point fine, but when VxWorks Boot goes to the host to download VxWorks(OS) I get the following: >>>>> Attached TCP/IP interface to fei0. Attaching network interface lo0... done. Loading... Error code 2: Access violation tftpGet: Error ogcurred while transgerring the file. Erroneous header read Error loading file: errno = 0x0. Can't load boot file!! >>> On the host side, ftp reports: >>>> Connection received from XXX.XXX.XXX.XXX on port XXXX Read request for file Error: EACCESS. Ext error The operation completed successfully. >>> I don't believe that there is anything physically wrong with the comm link as the vxWorks Boot binary was download by ftp via the PPC6bug command "niop", before it was flashed to the 5100. The ftp user permissions on the host are sufficient to allow downloading of the file, as I have done it from other computers attached to the host. Any ideas would be appreciated. Kyle --------------------------- Newsgroups: comp.os.vxworks Subject: HELP finding a newer LIBGCC.A with source code and/or building instructions please Date: Thu, 10 Jan 2002 15:22:44 -0600 From: "Tim" Organization: Nicolet Technologies Message-ID: Reply-To: "Tim" Sender: moeller@niti.com Running Tornado 2.0 on a Windows 2000 host, targeting the PowerPC 405GP (without FPU.) The libgcc.a that I was supplied with seems to be at least 5 years old: C:\Tornado\host\x86-win32\lib\gcc-lib\powerpc-wrs-vxworks\cygnus-2.7.2-96012 6\soft-float\libgcc.a I would a newer version and source code and instructions for building the same. Thanks in advance for any help/pointers, -- Tim --------------------------- Newsgroups: comp.os.vxworks Subject: Re: [HELP]UDP TX/RX packet mismatch problem... Date: Fri, 11 Jan 2002 12:38:04 +0900 From: "chihwankim" Organization: ELIMnet Internet News Message-ID: References: <3C3D7D66.2D999429@bellsouth.net> <3C3DB452.4DFBFCBF@l8s.co.uk> Well, thanks for your comments. To make sure that TX side successfully transmitted its packet, I made another test. 8260 A(TX side) transmits its packet periodically(10 ms) to PC and then PC sends to 8260 B(RX side) the same data whenever it receives from 8260 A. PC is based on window 2000 operating system. The result shows that PC received all the datagram from 8260 A without any missing, but 8260 B still has missed packets. The ratio is just the same with 8260 to 8260 case. And I added a code in BSP ethernet driver to check TX packets with specified UDP port, and the count is just the same with application-transmitted-packets. For understanding, below is the result of udpstatShow() when TX count=5328, RX count=5324(4 missing) - ---------------------------------------------------------------- TX side -> udpstatShow UDP: 5343 total packets 15 input packets 5328 output packets 0 incomplete header 0 bad data length field 0 bad checksum 12 broadcasts received with no ports 0 full socket 3 pcb cache lookups failed 3 pcb hash lookups failed RX side - -> udpstatShow UDP: 5324 total packets 5324 input packets 0 output packets 0 incomplete header 0 bad data length field 0 bad checksum 1 broadcasts received with no ports 0 full socket 5323 pcb cache lookups failed 3 pcb hash lookups failed - ---------------------------------------------------------------- What can I do further? thanks. "Dan Gold" wrote in message news:yEk%7.11864$cD4.25231@www.newsranger.com... > In article <3C3DB452.4DFBFCBF@l8s.co.uk>, David Laight says... > > > >>> chihwankim wrote: > >> > >> > I have two MPC8260 boards A and B. I open UDP socket both of them and try to > >> > transmit and receive UDP packets for each size of datagram. I assume that A > >> > is transmitting UDP datagram to B every each 10ms and B is receiving > >> > datagram from A. > > > > Are these custom 8260 designs or evaluation boards ? And the software (i.e. the > drivers) ? > Did you do the port, or are they stock ? Are you sure you have the latest > drivers ? I know WRS > has fixed at least 1 or 2 bugs in the FCC driver... > > I'm assuming this is Ethernet on the FCC, not SCC. And do you only have 1 FCC > active ? > >> > > >> > For 256/512/1024 bytes of datagram, there is no error in the receiving side. > >> > But 64/128 bytes of datagram, there occurs missed packets in the receiving > >> > side. > > > >So big packets are ok, but you get errors with small ones. > >Mmmmmm.... > > > >If you really do have the tx rate capped properly then something is > >definitely awol in the receiving system. > > > >If you don't have the tx rate capped, then you are probably seeing > >the effect of marginally different costs in transmitting and > >receiving packets. (cost/byte + cost/packet) > > > >Check with some LAN monitor software that the packets are all txed. > > The original poster said he added some code in the BSP to > count packets with the correct UDP port - was that on the Tx or Rx side ? > If it was the Rx, you might want to add code to the Tx side to make > sure all packets are getting Tx'd. Also, do you see any MAC layer > Tx or Rx errors ? If you're sure the packets are being Tx'd, you might > {ip,udp}StatShow and see if they tell you anything. > > See what statistics are available to you from the driver on both Tx and > Rx sides, that may give some clue. > > hth, > > Dan Gold gold@ensemble.com --------------------------- Newsgroups: comp.os.vxworks Subject: Re: I/O ports on PowerPC Date: 9 Jan 2002 10:44:56 -0800 From: arun4vxworks@indiatimes.com (Arun Prasad) Organization: http://groups.google.com/ Message-ID: References: Hi, Thank you for your reply. <>Nope. I haven't tried. I'll give a try and get back to you. Thanks again. Regards run "JJS" wrote in message news:... > Arun Prasad a écrit dans le message : > e38d912c.0201032322.2998a475@posting.google.com... > > Hi, > > > > I have a question on accessing I/O ports. > > I need to read/write from/to the network card registers on PowerPC > > platform. > > > > Can anyone tell me the functions provided by VxWorks to read/write > > Registers directly on PowerPC platforms? > > > > Earlier response would be of great help. > > > > Thanks in advance. > > Regards > > Arun > > Have you tried this : > char * ptreg = (char*) 0xXXXX (adress of a byte reg) > *ptreg = val; > Jack --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Caching, I/O Space memory mapping and DMA Date: 9 Jan 2002 11:54:06 -0800 From: arun4vxworks@indiatimes.com (Arun Prasad) Organization: http://groups.google.com/ Message-ID: References: <8bG_7.327$qS6.799090@newsr2.u-net.net> Dear Dave, Thank you for your detailed reply/clarifications. <>Yes. You are absolutely right. I DO NOT HAVE EXPERIENCE other than x86, though I READ a bit on Z80 and Motorola 68000 series chips. <>Is it so? I thought it's device/controller/card specific. If my memory is correct, I've read from Goankar's Microprocessor book that it is device-specific. But again that book dealt with 8085 and 8086 processors. And previously I worked on Linux Driver for SCSI Raid Controllers with i960 processors in them and I found I/O mapped and memory mapped cards separately and accordingly all the device access functions varied. It raises one more question from me. Hope you can/will answer that too. What happens if I have a ethernet controller which has a processor like intel's i960 (Hope this supports I/O mapped I/O) which supports I/O mapped I/O and if that controller is fixed on PPC system? Will PPC still map the device in to it's memory? << PCI space should indeed always be non-cacheable. But the system cannot do this for you. You must set up the BATs or PTEs (see manual re: sysBatDesc and sysPhysMemDesc) that cover the address range used by the PCI bus to have the G (guarded) and I (cache-inhibited) flags. Once you have done that, every card will be non-cacheable no matter where in PCI space the system decides to auto config it. >>Oh. Thanks for this information. But I haven't found sysBatDesc. But I was able to find sysPhysMemDesc. Here I'm giving the snap shot of it. (void *) PCI_MEMORY_START, (void *) PCI_MEMORY_START, PCI_MEMORY_MAP_END - PCI_MEMORY_START + 1, VM_STATE_MASK_VALID | VM_STATE_MASK_WRITABLE | VM_STATE_MASK_CACHEABLE | VM_STATE_MASK_GUARDED, VM_STATE_VALID | VM_STATE_WRITABLE | VM_STATE_CACHEABLE_NOT | VM_STATE_GUARDED Can you write me on how to interpret this? Because there are two entries are there two records are there. In this example, there are two PCI_MEMORY_START are there and correspondingly two different sets of flags are associated with them, one with Guarded and and non cacheable and the other one with cacheable and unguarded. I don't know why there are two different entries are there. Can you let me know this? <<> > If we directly access the device registers (sysInbyte() sysOutByte()), do we need to invalidate the cache as we are accessing HW registers? <>Oops... This is confusing again. From the file /* sysLib.c - IBM 405GP Walnut system-dependent routines */ I could able to find sysInByte() and sysOutByte() functions and similar PCI access routines to read/write one byte/Word/Long. But within the definitions sysInByte() and sysOutByte() everything treated as memory mapped only.I could able to see simple assignment statements and EIEIO_SYNC macros only. I guess these are only wrapper functions which simulates I/O mapped IO. Please do clarify me if I'm wong. << However I am very surprised to hear that you have an ethernet card that has no DMA. >>I'm sorry for the confusion. Our Card has its own DMA controller and its own Buffer/memory. <>I've made a note of it. I have one more question. This is nothing to do with VxWorks or PPC. Just having a curiosity to know whether the e-mail ID you have given () is a valid one? If I send a mail, will it reach you? Thank you so much for your reply. I would also like to thank Michael Lawnick (Lawnick@softec.de) for his reply mail. <>Oh. I. See. Yes. This is the scenario with our controller (It has its own memory and it buffers.) Regards Arun - -- Arun Prasad Software Engineer Adaptec India Pvt. Limited (A wholly owned subsidiary of Adaptec Inc., USA) 6-3-1086, IV Floor, Vista Grand Towers, Rajbhavan Road, Somajiguda, Hydereabad - 500082 Phone: (O):91-40-6661555/56/57/58 (Ext):234 Email: arun_prasad@adaptecin.com "Dave Korn" wrote in message news:<8bG_7.327$qS6.799090@newsr2.u-net.net>... > Arun Prasad wrote in message ... > >Hi, > > > > I'm having some serious of doubts on Caching, I/O Space memory > >mapping and DMA data transfer on IBM PPC405 platform. Can anyone > >clarify these doubts? > > Hi Arun, > > Your main confusion is because you are thinking of the PPC as if it was an > x86 chip. It isn't, and one big difference is that there is no separate I/O > space and memory space on > PPC: everything is mapped to memory space, and there is no I/O port space. > > > Do we need to invalidate the cache irrespective of our Card is > >whether memory mapped or i/o mapped while accessing device registers? > > Every card is memory mapped in PPC. So if the registers are mapped to > cacheable address space, you would have to flush the cache when writing to > them, and invalidate it before reading from them. This is tricky, which is > why it is much preferable to map device registers into non-cacheable address > space. > > > Whether the memory space which we are getting after configuring > >(pciAutoConfig) the PCI card (for memory mapped devices ) for I/O > >space is a cacheable or non-cacheable one? > > I think, it will not be a cacheable one as this memory is shared by > >the Host > >CPU as well as by the Card. Please do clarify me on this regard. > > PCI space should indeed always be non-cacheable. But the system cannot do > this for you. You must set up the BATs or PTEs (see manual re: sysBatDesc > and sysPhysMemDesc) that cover the address range used by the PCI bus to have > the G (guarded) and I (cache-inhibited) flags. Once you have done that, > every card will be non-cacheable no matter where in PCI space the system > decides to auto config it. > > > If we directly access the device registers (sysInbyte() > >sysOutByte()), do we > >need to invalidate the cache as we are accessing HW registers? > > These functions don't exist on ppc! *Everything* is memory mapped, so you > just set a pointer to point to it and read/write by using "*ptr", the same > as any other memory location. The location itself should be non-cacheable > for the device registers, but if for some reason you went crazy and wanted > to cache the entire PCI bus address space, then yes, you would need to flush > the cache before a write or invalidate it before a read. > > > One more thing, > > Do we need to take care of Cache validate/invalidate if we > >do not handle DMA data transfer??? > > No: assuming the PCI bus address space is properly set up as > non-cacheable, then reading and writing the registers will be ok; and if the > device does not alter system RAM, then there is no need to worry about the > caches. > > However I am very surprised to hear that you have an ethernet card that > has no DMA. How does the data get into the CPU's memory? Do you have to > read the data packet from the ethernet chip itself using the cpu? > > > And when we need to do cache flush? (within ISR??? or in some other > >places?) > > You can flush the cache any time you like - see the documentation for > cacheFlush - but the best time to do so is immediately before beginning to > write the commands to the device registers that would cause it to access the > system RAM. > > DaveK --------------------------- Newsgroups: comp.os.vxworks Subject: Re: LC82C169 Driver... Date: 9 Jan 2002 14:47:08 -0800 From: brian_doe@hotmail.com (brian) Organization: http://groups.google.com/ Message-ID: References: <21595172.0201072023.67f7fc78@posting.google.com> <3C3AF6D0.9F54E13F@varndellengineering.com> I would imagine he is looking for something helpful and constructive. Not some pompous ass like you pointing out useless things to him. Maybe he's working for a small shop. Maybe they're broke. Maybe he's gung ho and looking for a starting point. Who knows? Does Varndell Engineering get a lot of business with this shitty attitude? I hope not. Brian George Varndell wrote in message news:<3C3AF6D0.9F54E13F@varndellengineering.com>... > Hemant wrote: > > > > Hi All, > > I am supposed to write a network driver for NETGEAR card with LC82C169 > > (FA310Tx) for IDT Evaluation Board. > > But I don't know how to start as there is no BSP template for the > > above driver. > > I am trying to use the dec21x40End.c but it is not working at all. > > Can anybody suggest me about the same? > > I'm not sure what kind of reply you're hoping for. > Did you think someone would take pity on you and send you a > working driver? Did you think someone would start at first > principles and spell out everything you need to do -- step > by step? I really don't want to beat you up, but your employer > needs to hire a consultant -- you're clearly not qualified for > the task you've been assigned. > > Regards, > George Varndell --------------------------- Newsgroups: comp.os.vxworks Subject: supplement Date: 9 Jan 2002 23:08:18 -0800 From: lihuapeng@21cn.com (DragonSpring) Organization: http://groups.google.com/ Message-ID: <1d411918.0201092308.3f947de4@posting.google.com> Is there any other method to get the OSPF packet from all END-drivered network interfaces besides _protoSwIndex and inetsw[]? --------------------------- Newsgroups: comp.os.vxworks Subject: Re: _protoSwIndex and inetsw[ ] used in our OSPF implementation Date: 10 Jan 2002 20:55:00 -0800 From: lihuapeng@21cn.com (DragonSpring) Organization: http://groups.google.com/ Message-ID: <1d411918.0201102054.4fea409@posting.google.com> References: <1d411918.0201091833.3205a89f@posting.google.com> Help! Why I can not get OSPF packet via inetsw[IP_PROTO_NUM_MAX]? lihuapeng@21cn.com (DragonSpring) wrote in message news:<1d411918.0201091833.3205a89f@posting.google.com>... > Hi, > > A question about _protoSwIndex and > inetsw[IP_PROTO_NUM_MAX].IP_PROTO_NUM_MAX is defined as 8 in VxWorks. > > Environment: target: PPC860 rev. D4 > Host: WinNT 4.0, Tornado 2.0/VxWorks 5.4 for PPC, Tornado > 2.0 Cumulative Patch 2 (SPR T2CP2), NPT 5.4(TDK-13274), NPT patch(SPR > 30894) > > We are developing our OSPF protocol (rfc2328). We want to get all OSPF > packet from IP layer. So we use the code as following: > > int ospf_init() > { > FAST struct protosw *p_ospf_protocol_switch; > > if (_protoSwIndex >= ((sizeof(inetsw))/(sizeof(inetsw[0])))) > return (-1) ; > > p_ospf_protocol_switch = &inetsw [_protoSwIndex]; > > if (p_ospf_protocol_switch == NULL) > return (-1); > > if (p_ospf_protocol_switch->pr_domain != NULL) > return (1); /* already initialized */ > > p_ospf_protocol_switch->pr_type = SOCK_RAW; > p_ospf_protocol_switch->pr_domain = &inetdomain; > p_ospf_protocol_switch->pr_protocol = IPPROTO_OSPF; > p_ospf_protocol_switch->pr_flags = PR_ATOMIC|PR_ADDR; > p_ospf_protocol_switch->pr_input = ospf_receive; > p_ospf_protocol_switch->pr_output = ospf_send; > p_ospf_protocol_switch->pr_ctlinput = raw_ctlinput; > p_ospf_protocol_switch->pr_ctloutput = rip_ctloutput; > p_ospf_protocol_switch->pr_usrreq = rip_usrreq; > p_ospf_protocol_switch->pr_init = 0 ; > p_ospf_protocol_switch->pr_fasttimo = 0; > p_ospf_protocol_switch->pr_slowtimo = 0; > p_ospf_protocol_switch->pr_drain = 0; > p_ospf_protocol_switch->pr_sysctl = 0; > > _protoSwIndex++; > > ... > } > > void ospf_receive(struct mbuf *m) > { > /* determine which interface this OSPF packet is received from */ > ... > } > > void ospf_send(struct mbuf *m) > { > ... > } > > We invoke ospf_init() in usrAppInit() (which is in usrAppInit.c) to > start the OSPF process. > > Now my problem is that I can never get any OSPF packet in > ospf_recieve(). It seems that the IP protocol stack never call > ospf_receive(). OSPF takes inetsw[7]. inetsw[IP_PROTO_NUM_MAX] seems > take no effect. Why? Does the IP protocol discard the OSPF packet? Is > there any problem in the VxWorks configurtion? BTW, we have observed > the motFecEnd.c (END driver) has already passed the OSPF packet to MUX > layer. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: bus error exception in ppc8260 Date: 9 Jan 2002 23:11:53 -0800 From: ruchama@septier.com (Ruchama Goldshmidt) Organization: http://groups.google.com/ Message-ID: <5df340e0.0201092311.471518d2@posting.google.com> References: <5df340e0.0201090116.70d0958d@posting.google.com> <20W_7.337$qS6.871834@newsr2.u-net.net> "Dave Korn" wrote in message news:<20W_7.337$qS6.871834@newsr2.u-net.net>... > It's in the DAR, stored in the exception stack frame that is passed to the > handler. Here's an example of a DSI exception handler that I use: thanks for the answer , but still I have problem ,and when I try to look in the DAR register I see 0. what this is mean? TIA Ruchama --------------------------- Newsgroups: comp.os.vxworks Subject: _protoSwIndex and inetsw[ ] used in our OSPF implementation Date: 9 Jan 2002 18:33:03 -0800 From: lihuapeng@21cn.com (DragonSpring) Organization: http://groups.google.com/ Message-ID: <1d411918.0201091833.3205a89f@posting.google.com> Hi, A question about _protoSwIndex and inetsw[IP_PROTO_NUM_MAX].IP_PROTO_NUM_MAX is defined as 8 in VxWorks. Environment: target: PPC860 rev. D4 Host: WinNT 4.0, Tornado 2.0/VxWorks 5.4 for PPC, Tornado 2.0 Cumulative Patch 2 (SPR T2CP2), NPT 5.4(TDK-13274), NPT patch(SPR 30894) We are developing our OSPF protocol (rfc2328). We want to get all OSPF packet from IP layer. So we use the code as following: int ospf_init() { FAST struct protosw *p_ospf_protocol_switch; if (_protoSwIndex >= ((sizeof(inetsw))/(sizeof(inetsw[0])))) return (-1) ; p_ospf_protocol_switch = &inetsw [_protoSwIndex]; if (p_ospf_protocol_switch == NULL) return (-1); if (p_ospf_protocol_switch->pr_domain != NULL) return (1); /* already initialized */ p_ospf_protocol_switch->pr_type = SOCK_RAW; p_ospf_protocol_switch->pr_domain = &inetdomain; p_ospf_protocol_switch->pr_protocol = IPPROTO_OSPF; p_ospf_protocol_switch->pr_flags = PR_ATOMIC|PR_ADDR; p_ospf_protocol_switch->pr_input = ospf_receive; p_ospf_protocol_switch->pr_output = ospf_send; p_ospf_protocol_switch->pr_ctlinput = raw_ctlinput; p_ospf_protocol_switch->pr_ctloutput = rip_ctloutput; p_ospf_protocol_switch->pr_usrreq = rip_usrreq; p_ospf_protocol_switch->pr_init = 0 ; p_ospf_protocol_switch->pr_fasttimo = 0; p_ospf_protocol_switch->pr_slowtimo = 0; p_ospf_protocol_switch->pr_drain = 0; p_ospf_protocol_switch->pr_sysctl = 0; _protoSwIndex++; ... } void ospf_receive(struct mbuf *m) { /* determine which interface this OSPF packet is received from */ ... } void ospf_send(struct mbuf *m) { ... } We invoke ospf_init() in usrAppInit() (which is in usrAppInit.c) to start the OSPF process. Now my problem is that I can never get any OSPF packet in ospf_recieve(). It seems that the IP protocol stack never call ospf_receive(). OSPF takes inetsw[7]. inetsw[IP_PROTO_NUM_MAX] seems take no effect. Why? Does the IP protocol discard the OSPF packet? Is there any problem in the VxWorks configurtion? BTW, we have observed the motFecEnd.c (END driver) has already passed the OSPF packet to MUX layer. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: pci error in IDT Rc32332 Date: 10 Jan 2002 05:04:12 -0800 From: sczhang@esteemnetworks.com (ShiChao Zhang) Organization: http://groups.google.com/ Message-ID: <4ceedd43.0201100504.2690427a@posting.google.com> References: <4ceedd43.0112282323.368465f8@posting.google.com> <4ceedd43.0201050420.7ee19f02@posting.google.com> Johan Borkhuis wrote in message news:... > sczhang@esteemnetworks.com (ShiChao Zhang) wrote: > > > I have tried it, but has no effect. And if I disable the exception, > > all the value I got is the address of the PCI configuration Address > > Register:0x18002cfc. > > Did you try the command "pciDeviceShow" and/or "pciHeaderShow"? > > Groeten, > Johan Oh, I solved this problem. The reason is using the -O2 compiler option, which reorder my c code(convert a write,read operation to read,write operation). Now the pciDeviceShow is OK! Thanks! --------------------------- Newsgroups: comp.os.vxworks Subject: Re: bus error exception in ppc8260 Date: 10 Jan 2002 03:33:50 -0800 From: ruchama@septier.com (Ruchama Goldshmidt) Organization: http://groups.google.com/ Message-ID: <5df340e0.0201100333.f4d097f@posting.google.com> References: <5df340e0.0201090116.70d0958d@posting.google.com> <20W_7.337$qS6.871834@newsr2.u-net.net> <5df340e0.0201092311.471518d2@posting.google.com> ruchama@septier.com (Ruchama Goldshmidt) wrote in message news:<5df340e0.0201092311.471518d2@posting.google.com>... > "Dave Korn" wrote in message news:<20W_7.337$qS6.871834@newsr2.u-net.net>... > > > It's in the DAR, stored in the exception stack frame that is passed to the > > handler. Here's an example of a DSI exception handler that I use: > > thanks for the answer , but still I have problem ,and when I try to > look in the DAR register I see 0. > > what this is mean? > > TIA > Ruchama The exception I want to handle is the machine check exception maybe there is differnt between the DSI exception in the example (vector 0x300) and the Machine check vector (number 0x200)? TIA Ruchama --------------------------- Newsgroups: comp.os.vxworks Subject: mpc850 upmx program question Date: 10 Jan 2002 02:50:43 -0800 From: jlddotcn@hotmail.com (juliedan) Organization: http://groups.google.com/ Message-ID: Could anybody tell me how to use SDRAM timing diagram to make upmTableSdram? I hope you can help me in details(especially the process). thank you in advance. Liedan Ju mail: jlddotcn@hotmail.com --------------------------- Newsgroups: comp.os.vxworks Subject: DosFS2 or the device driver Date: 10 Jan 2002 07:13:10 -0800 From: scott_b_klein@md.northgrum.com (Scott Klein) Organization: http://groups.google.com/ Message-ID: <69517bb8.0201100713.63d1cc56@posting.google.com> Our application consists of many tasks, all disk requests are submitted from the same task: tFileProcessor, which issues write requests as fast as possible to 4 files (not in a rotational manner, most i/o's are to 1 file). Our cluster size on the disk is 32KB and our writes are either 64KB or 1MB (the cache bypass thershold is 64KB). We found that, during normal operation, our 1MB write requests take about 15-20 ms to complete and that tFileProcessor uses about 6% of the CPU time with the CPU being idle about 75% of the time. We discovered that we will get a 1MB write request that takes 5 seconds before returning successfully. When this happens, tFileProcessor uses about 75% of the CPU time and we get messages from the driver that the disk has a problem. Subsequent writes to this file take about 170 ms, but other files continue at 17ms. We are trying to figure out if this is a Wind River problem or a device driver problem. Because it is only 1 file affected, we think that the file system has a problem. Wind River software would also be suspect if device driver functions do not count toward task CPU usage. Thanks for any clarification you can provide. Scott B. Klein Senior Engineer Electronics Sensors and Systems Sector Northrop Grumman Corporation Post Office Box 746 - Mail Stop 285A Baltimore, Maryland 21203-0746 Telephone (410) 765-0910 Fax: (410) 981-2045 scott_b_klein@md.northgrum.com --------------------------- Newsgroups: comp.os.vxworks Subject: Re: LC82C169 Driver... Date: 9 Jan 2002 21:48:35 -0800 From: hemu_u@hotmail.com (Hemant) Organization: http://groups.google.com/ Message-ID: <21595172.0201092148.6e2677a4@posting.google.com> References: <21595172.0201072023.67f7fc78@posting.google.com> <3C3AF6D0.9F54E13F@varndellengineering.com> That was so nice of you Brian...Thanx for that. That was the first message I had posted on this group and after seeing that ---reply I decided that it would be my last messsage.But it seems there are good ppl also here. :) By the way...the driver is almost complete, only thing is that the chip sometimes hangs (don't know why!!!) I will now rely on traditional FIFO ring than using hash to give a try! brian_doe@hotmail.com (brian) wrote in message news:... > I would imagine he is looking for something helpful and constructive. > Not some pompous ass like you pointing out useless things to him. > Maybe he's working for a small shop. Maybe they're broke. Maybe he's > gung ho and looking for a starting point. Who knows? > > Does Varndell Engineering get a lot of business with this shitty > attitude? I hope not. > > Brian > > George Varndell wrote in message news:<3C3AF6D0.9F54E13F@varndellengineering.com>... > > Hemant wrote: > > > > > > Hi All, > > > I am supposed to write a network driver for NETGEAR card with LC82C169 > > > (FA310Tx) for IDT Evaluation Board. > > > But I don't know how to start as there is no BSP template for the > > > above driver. > > > I am trying to use the dec21x40End.c but it is not working at all. > > > Can anybody suggest me about the same? > > > > I'm not sure what kind of reply you're hoping for. > > Did you think someone would take pity on you and send you a > > working driver? Did you think someone would start at first > > principles and spell out everything you need to do -- step > > by step? I really don't want to beat you up, but your employer > > needs to hire a consultant -- you're clearly not qualified for > > the task you've been assigned. > > > > Regards, > > George Varndell --------------------------- Newsgroups: comp.os.vxworks Subject: Re: supplement Date: 10 Jan 2002 05:27:12 -0800 From: lihuapeng@21cn.com (DragonSpring) Organization: http://groups.google.com/ Message-ID: <1d411918.0201100527.511764bd@posting.google.com> References: <1d411918.0201092308.3f947de4@posting.google.com> lihuapeng@21cn.com (DragonSpring) wrote in message news:<1d411918.0201092308.3f947de4@posting.google.com>... > Is there any other method to get the OSPF packet from all END-drivered > network interfaces besides _protoSwIndex and inetsw[]? This message is a supplement to early message "_protoSwIndex and inetsw[ ] used in our OSPF implementation". --------------------------- Newsgroups: comp.os.vxworks Subject: WIND_BASE vs TGT_DIR Date: 10 Jan 2002 08:50:12 -0800 From: rwest@hatterasnetworks.com (RickyW) Organization: http://groups.google.com/ Message-ID: <834f1c2f.0201100850.680cb428@posting.google.com> One thing that has bothered me for quite a while is the reliance of Tornado on the WIND_BASE variable for both tools and target. I can understand the desire for simplicity but WindRivers package deployment mechanism does seem to support multiple targets with different package configurations. Moreover, when patches or packages are applied, it is a hope and a prayer that you current configuration doesn't get screwed up. Does anyone else share this frustration? I've often thought about a TARGET_BASE variable to use when setting TGT_DIR whenever it is defined rather than WIND_BASE. Has anyone done this? --------------------------- Newsgroups: comp.os.vxworks Subject: Help with Interrupts and Console Date: 10 Jan 2002 12:13:04 -0800 From: karahanduman@hotmail.com (Kara) Organization: http://groups.google.com/ Message-ID: Greetings, I installed an interrupt (vector 65) on my custom platfrom. This interrupt is planned for Scc but in the hardware it is used for my custom peripheral. My interrupt handler gets called only if I press a key on the console (over serial port), if I don't press a key I don't get interrupts. Do you have any idea what is the reason? Regards, Kara- --------------------------- Newsgroups: comp.os.vxworks Subject: flash and MMU Date: 10 Jan 2002 18:32:16 -0800 From: phil_chee@sina.com (ruready) Organization: http://groups.google.com/ Message-ID: Why I cannot read or write flash, when I enable MMU in BSP? When reading flash, I can only get the data of 0xff. But if I disable the MMU, everything is OK. That's what I write in sysPhysMemDesc[] PHYS_MEM_DESC sysPhysMemDesc [] = { { /* Vector Table and Interrupt Stack */ (void *) LOCAL_MEM_LOCAL_ADRS, (void *) LOCAL_MEM_LOCAL_ADRS, RAM_LOW_ADRS, VM_STATE_MASK_VALID | VM_STATE_MASK_WRITABLE | VM_STATE_MASK_CACHEABLE, VM_STATE_VALID | VM_STATE_WRITABLE | VM_STATE_CACHEABLE_NOT }, { /* RAM on 60x bus - Must be second entry for Auto Sizing */ (void *) RAM_LOW_ADRS, (void *) RAM_LOW_ADRS, LOCAL_MEM_SIZE - RAM_LOW_ADRS, VM_STATE_MASK_VALID | VM_STATE_MASK_WRITABLE | VM_STATE_MASK_CACHEABLE | VM_STATE_MASK_MEM_COHERENCY, VM_STATE_VALID | VM_STATE_WRITABLE | VM_STATE_CACHEABLE | VM_STATE_MEM_COHERENCY }, { /* Board Control and Status Registers */ (void *) BCSR_BASE_ADRS, (void *) BCSR_BASE_ADRS, BCSRS_SIZE, VM_STATE_MASK_VALID | VM_STATE_MASK_WRITABLE | VM_STATE_MASK_CACHEABLE | VM_STATE_MASK_GUARDED, VM_STATE_VALID | VM_STATE_WRITABLE | VM_STATE_CACHEABLE_NOT | VM_STATE_GUARDED }, { /* MPC8260 Internal Memory Map */ (void *) DEFAULT_IMM_ADRS, (void *) DEFAULT_IMM_ADRS, IMM_SIZE, VM_STATE_MASK_VALID | VM_STATE_MASK_WRITABLE | VM_STATE_MASK_CACHEABLE | VM_STATE_MASK_GUARDED, VM_STATE_VALID | VM_STATE_WRITABLE | VM_STATE_CACHEABLE_NOT | VM_STATE_GUARDED }, { /* Flash */ (void *) ROM_BASE_ADRS, (void *) ROM_BASE_ADRS, ROM_SIZE, VM_STATE_MASK_VALID | VM_STATE_MASK_WRITABLE | VM_STATE_MASK_CACHEABLE | VM_STATE_MASK_GUARDED, VM_STATE_VALID | VM_STATE_WRITABLE | VM_STATE_CACHEABLE_NOT | VM_STATE_GUARDED } }; Could someone do me a faver helping me to solve this problem? --------------------------- Newsgroups: comp.os.vxworks Subject: [Q] Problem when making VxWorks bootrom and boot image Date: Fri, 11 Jan 2002 15:04:38 +0900 From: "kimks" Organization: Korea Telecom Message-ID: Hi all... I have 2 questions about VxWorks bootrom & boot image. I believe that you can help me to solve my problem. [1] I'm trying to build VxWorks boot image (vxworks_rom) using Tornado 2 project tool, but it does not work. When I traced the boot sequence, I found that the jump to absEntry crashes system. When I'm trying to build a compressed boot image (vxworks_romCompress), inflate() doesn't work; control enters inflate() and never returns; How can I fix my problem ? [2] When making a bootrom, I'd like to configure RAM_HIGH_ADRS more than 0x01f00000. This is why my vxwork image is more than 35Mbs. But when I configure RAM_HIGH_ADRS more than 0x01f00000, bootrom does't work. It works when RAM_HIGH_ADRS is configured less than 0x01f00000. RAM_HIGH_ADRS has any limitation in value ? Any help would be appreciated. kimks knuth73@dreamwiz.com --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Caching, I/O Space memory mapping and DMA Date: Fri, 11 Jan 2002 09:01:57 +0100 From: "Michael Lawnick" Organization: Buergernetz Dillingen Message-ID: References: <8bG_7.327$qS6.799090@newsr2.u-net.net> Reply-To: "Michael Lawnick" Sender: mlawnick@p5088fa06.dip.t-dialin.net Hi Arund, as you posted your direct EMails with Dave here, I assume you want different/additional opinions to your question. Here my 2 Cents: As mentioned correctly by Dave, PPC always see's PCI as memory mapped, but this does not mean, that a PPC can't do PCI-I/O. It means that there is a memory space, where the PCI-bridge of PPC will translate all accesses into I/O-accesses and there are other spaces, that are interpreted in other ways. Your excerpt from sysPhysMemDesc is just _one_ entry and its correct: VM_STATE_MASK_VALID | VM_STATE_MASK_WRITABLE |VM_STATE_MASK_CACHEABLE |VM_STATE_MASK_GUARDED tells which flags you want to change (_MASK_) VM_STATE_VALID | VM_STATE_WRITABLE | VM_STATE_CACHEABLE_NOT |VM_STATE_GUARDED enables the entry and sets it writable, uncached and guarded - just as you need it. You write: > >>I'm sorry for the confusion. Our Card has its own DMA controller and > its own Buffer/memory. This seems to be suspicious - any confusion possible ? Either DMA or local buffer, both would be redundant (surely still possible, but who wants to pay?) On our PPC405, our scndNIC 's registers have to be set up through PCI, data is written to CPU-memory via DMA (by the NIC). As we have shut off MMU, I must always do invalidate before data reading and flush before telling the NIC that there's a new packet to transmit. If our MMU would be enabled, I could allocate memory with cacheDmaMalloc and wouldn't need to care. To complete your confusion about sysInByte, sysOutByte, ..., there are more commands: sysPciInByte, sysPciOutByte,... The trick is: the sysInXXX variant will write data as is, typically found on Intel arch. sysPciInXXX will make a swap before access. This is needed, as PPC is big-endian, but PCI is little-endian. Second issue are the instruction pipes of PPC. As you might do two successive accesses to different PCI-registers, but the result of the second one might be influenced by the first (e.g. first enable device via control register, then read from status register), it must be ensured, that the first instruction completed before the second. This is done by a sync instruction as you might see in sysALib.s - -- Mit freundlichen Grüßen, Michael Lawnick ============================================== SOFTEC GmbH Tel +49-731-96600-0 Promenade 17 Fax +49-731-96600-23 D-89073 Ulm Michael Lawnick Germany lawnick@softec.de ============================================== "Arun Prasad" schrieb im Newsbeitrag news:e38d912c.0201091154.76631ff5@posting.google.com... > Dear Dave, > > Thank you for your detailed reply/clarifications. > > < was an x86 chip. > >>Yes. You are absolutely right. I DO NOT HAVE EXPERIENCE other than > x86, though I READ a bit on Z80 and Motorola 68000 series chips. > > < space and memory space on PPC: everything is mapped to memory space, > and there is no I/O port space. > < >>Is it so? I thought it's device/controller/card specific. If my > memory is correct, I've read from Goankar's Microprocessor book that > it is device-specific. But again that book dealt with 8085 and 8086 > processors. And previously I worked on Linux Driver for SCSI Raid > Controllers with i960 processors in them and I found I/O mapped and > memory mapped cards separately and accordingly all the device access > functions varied. > It raises one more question from me. Hope you can/will answer that > too. > What happens if I have a ethernet controller which has a processor > like intel's i960 (Hope this supports I/O mapped I/O) which supports > I/O mapped I/O and if that controller is fixed on PPC system? > Will PPC still map the device in to it's memory? > > << PCI space should indeed always be non-cacheable. But the system > cannot do this for you. You must set up the BATs or PTEs (see manual > re: sysBatDesc and sysPhysMemDesc) that cover the address range used > by the PCI bus to have the G (guarded) and I (cache-inhibited) flags. > Once you have done that, every card will be non-cacheable no matter > where in PCI space the system decides to auto config it. > >>Oh. Thanks for this information. But I haven't found sysBatDesc. But > I was able to find sysPhysMemDesc. Here I'm giving the snap shot of > it. > > (void *) PCI_MEMORY_START, > (void *) PCI_MEMORY_START, > PCI_MEMORY_MAP_END - PCI_MEMORY_START + 1, > VM_STATE_MASK_VALID | VM_STATE_MASK_WRITABLE | > VM_STATE_MASK_CACHEABLE | > VM_STATE_MASK_GUARDED, > VM_STATE_VALID | VM_STATE_WRITABLE | > VM_STATE_CACHEABLE_NOT | > VM_STATE_GUARDED > > Can you write me on how to interpret this? Because there are two > entries are there two records are there. In this example, there are > two PCI_MEMORY_START are there and correspondingly two different sets > of flags are associated with them, one with Guarded and and non > cacheable and the other one with cacheable and unguarded. I don't know > why there are two different entries are there. Can you let me know > this? > > <<> > If we directly access the device registers (sysInbyte() > sysOutByte()), do we need to invalidate the cache as we are accessing > HW registers? > < > >>Oops... This is confusing again. > From the file /* sysLib.c - IBM 405GP Walnut system-dependent routines > */ > I could able to find sysInByte() and sysOutByte() functions and > similar PCI access routines to read/write one byte/Word/Long. > But within the definitions sysInByte() and sysOutByte() everything > treated as memory mapped only.I could able to see simple assignment > statements and EIEIO_SYNC macros only. > I guess these are only wrapper functions which simulates I/O mapped > IO. Please do clarify me if I'm wong. > > << However I am very surprised to hear that you have an ethernet card > that has no DMA. > >>I'm sorry for the confusion. Our Card has its own DMA controller and > its own Buffer/memory. > > < the commands to the device registers that would cause it to access the > system RAM. > >>I've made a note of it. > > I have one more question. This is nothing to do with VxWorks or PPC. > Just having a curiosity to know whether the e-mail ID you have given > () is a valid one? If I send a mail, will > it reach you? > > Thank you so much for your reply. > > I would also like to thank Michael Lawnick (Lawnick@softec.de) for > his reply mail. > > < actively > fetch it, you shouldn't have any cache troubles when disabling cache > for PCI > address space of your NIC. > >>Oh. I. See. Yes. This is the scenario with our controller (It has > its own memory and it buffers.) > > Regards > Arun > -- > Arun Prasad > Software Engineer > Adaptec India Pvt. Limited > (A wholly owned subsidiary of Adaptec Inc., USA) > 6-3-1086, IV Floor, Vista Grand Towers, > Rajbhavan Road, Somajiguda, > Hydereabad - 500082 > Phone: (O):91-40-6661555/56/57/58 > (Ext):234 > Email: arun_prasad@adaptecin.com > > "Dave Korn" wrote in message news:<8bG_7.327$qS6.799090@newsr2.u-net.net>... > > Arun Prasad wrote in message ... > > >Hi, > > > > > > I'm having some serious of doubts on Caching, I/O Space memory > > >mapping and DMA data transfer on IBM PPC405 platform. Can anyone > > >clarify these doubts? > > > > Hi Arun, > > > > Your main confusion is because you are thinking of the PPC as if it was an > > x86 chip. It isn't, and one big difference is that there is no separate I/O > > space and memory space on > > PPC: everything is mapped to memory space, and there is no I/O port space. > > > > > Do we need to invalidate the cache irrespective of our Card is > > >whether memory mapped or i/o mapped while accessing device registers? > > > > Every card is memory mapped in PPC. So if the registers are mapped to > > cacheable address space, you would have to flush the cache when writing to > > them, and invalidate it before reading from them. This is tricky, which is > > why it is much preferable to map device registers into non-cacheable address > > space. > > > > > Whether the memory space which we are getting after configuring > > >(pciAutoConfig) the PCI card (for memory mapped devices ) for I/O > > >space is a cacheable or non-cacheable one? > > > I think, it will not be a cacheable one as this memory is shared by > > >the Host > > >CPU as well as by the Card. Please do clarify me on this regard. > > > > PCI space should indeed always be non-cacheable. But the system cannot do > > this for you. You must set up the BATs or PTEs (see manual re: sysBatDesc > > and sysPhysMemDesc) that cover the address range used by the PCI bus to have > > the G (guarded) and I (cache-inhibited) flags. Once you have done that, > > every card will be non-cacheable no matter where in PCI space the system > > decides to auto config it. > > > > > If we directly access the device registers (sysInbyte() > > >sysOutByte()), do we > > >need to invalidate the cache as we are accessing HW registers? > > > > These functions don't exist on ppc! *Everything* is memory mapped, so you > > just set a pointer to point to it and read/write by using "*ptr", the same > > as any other memory location. The location itself should be non-cacheable > > for the device registers, but if for some reason you went crazy and wanted > > to cache the entire PCI bus address space, then yes, you would need to flush > > the cache before a write or invalidate it before a read. > > > > > One more thing, > > > Do we need to take care of Cache validate/invalidate if we > > >do not handle DMA data transfer??? > > > > No: assuming the PCI bus address space is properly set up as > > non-cacheable, then reading and writing the registers will be ok; and if the > > device does not alter system RAM, then there is no need to worry about the > > caches. > > > > However I am very surprised to hear that you have an ethernet card that > > has no DMA. How does the data get into the CPU's memory? Do you have to > > read the data packet from the ethernet chip itself using the cpu? > > > > > And when we need to do cache flush? (within ISR??? or in some other > > >places?) > > > > You can flush the cache any time you like - see the documentation for > > cacheFlush - but the best time to do so is immediately before beginning to > > write the commands to the device registers that would cause it to access the > > system RAM. > > > > DaveK --------------------------- Newsgroups: comp.os.vxworks Subject: Re: VxWorks Boot on MVME5100 (PPC 7400) and Tornado 2.1 Date: Fri, 11 Jan 2002 09:46:49 +0100 From: "Michael Lawnick" Organization: Buergernetz Dillingen Message-ID: References: Reply-To: "Michael Lawnick" Sender: mlawnick@p5088fa06.dip.t-dialin.net Hi Kyle, is your swapping between ftp and tftp one of your typos or do you really try to boot via tftp from a ftp-server ? > tftpGet: Error ogcurred while transgerring the file. <-> > On the host side, ftp reports: Assuming it's a typo: Don't use stripppc, objcopyppc or anything else to remove symbol information from your code! AFAIK, all host-tools for PPC-targets that manipulate symbol information are buggy. Some don't work at all, others not in conjunction with debugger information,... . Use only images provided by the IDE. We had a similiar problem when having stripped an image by objcopyppc that contained debug info. Recompiled without debug-info and all worked. - -- Mit freundlichen Grüßen, Michael Lawnick ============================================== SOFTEC GmbH Tel +49-731-96600-0 Promenade 17 Fax +49-731-96600-23 D-89073 Ulm Michael Lawnick Germany lawnick@softec.de ============================================== "Kyle Rutledge" schrieb im Newsbeitrag news:JMm%7.1777$AL3.348689537@newssvr17.news.prodigy.com... > I've flashed VxWorks Boot onto a MVME5100 card and it boots to that point > fine, but when VxWorks Boot goes to the host to download VxWorks(OS) I get > the following: > > >>>>> > Attached TCP/IP interface to fei0. > Attaching network interface lo0... done. > Loading... Error code 2: Access violation > tftpGet: Error ogcurred while transgerring the file. > Erroneous header read > > Error loading file: errno = 0x0. > Can't load boot file!! > >>> > > On the host side, ftp reports: > > >>>> > Connection received from XXX.XXX.XXX.XXX on port XXXX > Read request for file > Error: EACCESS. Ext error The operation completed successfully. > >>> > > I don't believe that there is anything physically wrong with the comm link > as the vxWorks Boot binary was download by ftp via the PPC6bug command > "niop", before it was flashed to the 5100. The ftp user permissions on the > host are sufficient to allow downloading of the file, as I have done it from > other computers attached to the host. > > Any ideas would be appreciated. > > Kyle > > > --------------------------- Newsgroups: comp.os.vxworks Subject: Tornado project launcher not working Date: Fri, 11 Jan 2002 09:54:15 +0000 (UTC) From: "Arne Henrichsen" Organization: Mailgate.ORG Server - http://www.Mailgate.ORG Message-ID: <71cd02f888c82c8631d4d31fbdb94d0e.8627@mygate.mailgate.org> I have installed Tornadoon a sun machine. When I type 'launch', the Tornado launcher comes up. But when pressing the 'Project' or 'WindView' button, the following happens: Tornado 2.0 Copyright (C) 1998, Wind River Systems, Inc. All Rights Reserved But no project window gets loaded... I am using tornado2.0.2 on sun-solaris. Does anybody have similar problems? Thanks Arne Henrichsen - -- Posted via Mailgate.ORG Server - http://www.Mailgate.ORG --------------------------- End of New-News digest ********************** From vxwexplo-errs@csg.lbl.gov Fri Jan 11 05:38:23 2002 From: Donald McLachlan Date: Fri Jan 11 05:38:26 PST 2002 Subject: Re: Timestamp resolution on Vx-works > Subject: Timestamp resolution on Vx-works > Submitted-by: pranjan@hss.hns.com > > Hi, > > I have a query regarding the resolution of the timestamp on Vx-works. To get a > feel of the execution time of one of our utility functions, we try to get the > time- stamp before the function is called and time- stamp after the function has > been called. > > For example, > > t1 = clock_gettime() > function() > t2= clock_gettime() > > But (t2-t1) always comes out to be zero. From the shell, try: timexN function For example my system has a tick granularity of 16.67 milliseconds: -> sysClkRateGet value = 60 = 0x3c = '<' thus the following measures 0 time taken by sysDelay(): -> tickGet ; sysDelay ; tickGet value = 3753812 = 0x394754 value = 0 = 0x0 value = 3753812 = 0x394754 but timexN shows: -> timexN sysDelay timex: 125000 reps, time per rep = 1 +/- 0 (0%) microsecs value = 58 = 0x3a = ':' [ Didn't someone out there say the target shell was the best feature of vxWorks? :-) ] Don From vxwexplo-errs@csg.lbl.gov Fri Jan 11 05:49:27 2002 From: Donald.McLachlan@crc.ca Date: Fri Jan 11 05:49:29 PST 2002 Subject: Re: Timestamp resolution on Vx-works > Subject: Timestamp resolution on Vx-works > Submitted-by: pranjan@hss.hns.com > > Hi, > > I have a query regarding the resolution of the timestamp on Vx-works. To get a > feel of the execution time of one of our utility functions, we try to get the > time- stamp before the function is called and time- stamp after the function has > been called. > > For example, > > t1 = clock_gettime() > function() > t2= clock_gettime() > > But (t2-t1) always comes out to be zero. From the shell, try: timexN function For example my system has a tick granularity of 16.67 milliseconds: -> sysClkRateGet value = 60 = 0x3c = '<' thus the following measures 0 time taken by sysDelay(): -> tickGet ; sysDelay ; tickGet value = 3753812 = 0x394754 value = 0 = 0x0 value = 3753812 = 0x394754 but timexN shows: -> timexN sysDelay timex: 125000 reps, time per rep = 1 +/- 0 (0%) microsecs value = 58 = 0x3a = ':' [ Didn't someone out there say the target shell was the best feature of vxWorks? :-) ] Don From vxwexplo-errs@csg.lbl.gov Fri Jan 11 07:08:10 2002 From: "Shah, Ranjana" Date: Fri Jan 11 07:08:12 PST 2002 Subject: Re: Timestamp resolution on Vx-works VxWorks Ranjan, It appears that you are looking for the result in the wrong place. The call t1 = clock_gettime() would give OK(0) in t1. Actually the timestamp comes in one of the parameters. Here is an example of the proper usage: ULONG PosixSec, PosixNs; timespec timespec_now; clock_gettime(CLOCK_REALTIME, ×pec_now); PosixSec = timespec_now.tv_sec; //seconds PosixNs = timespec_now.tv_nsec; //nano seconds Note: timespec is defined in time.h as follows: struct timespec { /* interval = tv_sec*10**9 + tv_nsec */ time_t tv_sec; /* seconds */ long tv_nsec; /* nanoseconds (0 - 1,000,000,000) */ }; and #define CLOCK_REALTIME 0x0 /* system wide realtime clock */ I hope this helps. Ranjana Shah Staff Engineer Telephonics Corp. Voice: (631) 549-6104 FAX: (631) 755-7642 From vxwexplo-errs@csg.lbl.gov Fri Jan 11 09:20:20 2002 From: Bruce Wilson Date: Fri Jan 11 09:20:23 PST 2002 Subject: Re: Tornado project launcher not working I have also experienced exactly the same Tornado 2 behavior. Tech support offered the following suggestions: (1) Make sure the installing user has read/write privileges to the system, and is installing under CDE (common desktop environment). [NOTE: I use OpenWindows with twm window manager rather than the CDE, and everything works after I did steps 4 and 5 below] (2) Check the font path with "xset q" If it looks pretty long and the tornado install isn't working, then run "xset fp default" followed by "xset q" again. Make sure it is now the system default. See if this helps. (3) Make sure the MWHOME environment variable is not defined. Rename/remove the .wind/mw directory in the user's home area. Check if there is another MainWin (MW)-enabled application on the host by looking at the LD_LIBRARY_PATH (e.g. SDS debugger has caused problems) (4) Use the truss command on the host to get clues about the failure: "truss -o outfilename misbehaving_program_name" (e.g. truss -o projfail.out tornado -or- truss -o wvfail windview) [NOTE: this was useful for me. I found some files that needed executable permission for everyone. There's a mountain of output from truss] (5) Mainwin uses System V interprocess communications entities like shared memory and semaphores on the host. If an application using Mainwin is killed, it may not have a chance to clean up these resources and they are left behind, preventing their use when invoked later. Run "ipcs -a" to see what's allocated. Then run ipcrm as shown below to delete all but the ones owned by root. Check the result by running "ipcs -a" again, to verify only root-owned items exist. Then reboot the host [NOTE: I'm not positive the reboot is necessary]. EXAMPLE SESSION: ipcs -a IPC status from as of Thu Apr 6 14:49:08 2000 Message Queue facility not in system. T ID KEY MODE OWNER GROUP CREATOR CGROUP NATTCH SEGSZ CPID LPID ATIME DTIME CTIME Shared Memory: m 0 0x50000993 --rw-r--r-- root root root root 1 68 486 486 6:21:13 6:21:13 6:21:13 m 1 0x31000c90 --rw-rw-r-- will wrs will wrs 1 8192 930 20677 14:34:47 14:35:02 9:17:20 m 2 0x31000c91 --rw-rw-r-- will wrs will wrs 1 1048576 930 20677 14:34:47 14:35:02 9:17:20 m 2203 0x87654321 --rw-rw-rw- will wrs will wrs 1 1048576 9920 9920 16:50:09 no-entry 16:50:09 m 2704 00000000 --rw-rw-rw- will wrs will wrs 0 16384 17092 0 no-entry no-entry 9:54:25 T ID KEY MODE OWNER GROUP CREATOR CGROUP NSEMS OTIME CTIME Semaphores: s 0 0x31000c90 --ra-ra-r-- will wrs will wrs 3 14:35:02 9:17:20 If you see something like this, then there could be problems with shared memory or semaphores that have already been allocated. Ask the user to run: ipcrm > ipcrm -m 1 -m 2 -m 2203 -m 2704 -s 0x31000c90 The command above will clean out the shared memory and semphores allocated by Mainwin. A subsequent ipcs -a will show that only the shared memory created by root remains. > ipcs -a IPC status from as of Thu Apr 6 14:52:28 2000 Message Queue facility not in system. T ID KEY MODE OWNER GROUP CREATOR CGROUP NATTCH SEGSZ CPID LPID ATIME DTIME CTIME Shared Memory: m 0 0x50000993 --rw-r--r-- root root root root 1 68 486 486 6:21:13 6:21:13 6:21:13 T ID KEY MODE OWNER GROUP CREATOR CGROUP NSEMS OTIME CTIME Semaphores: Good luck solving this problem. Another thing to know about is that the first time you select the project or windview tools, a font cache gets set up on the host, and the process is filled with sleeps that make it take 2 to 4 minutes! So, after trying all the suggestions, reboot your host, bring up the launcher, select the project or windview tool and go take a coffee break. The window might actually appear while you're gone. Fun, huh? --Bruce Wilson University of California Lawrence Livermore National Lab >Subject: Tornado project launcher not working >Date: Fri, 11 Jan 2002 09:54:15 +0000 (UTC) >From: "Arne Henrichsen" >Organization: Mailgate.ORG Server - http://www.Mailgate.ORG >Message-ID: <71cd02f888c82c8631d4d31fbdb94d0e.8627@mygate.mailgate.org> > >I have installed Tornadoon a sun machine. When I type 'launch', the Tornado >launcher comes up. But when pressing the 'Project' or 'WindView' button, the >following happens: > >Tornado 2.0 >Copyright (C) 1998, Wind River Systems, Inc. >All Rights Reserved > >But no project window gets loaded... > >I am using tornado2.0.2 on sun-solaris. > >Does anybody have similar problems? > >Thanks >Arne Henrichsen From vxwexplo-errs@csg.lbl.gov Sat Jan 12 04:03:12 2002 From: Vxworks Exploder Date: Sat Jan 12 04:03:14 PST 2002 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Sat Jan 12 04:03:08 PST 2002 Subject: Why the target server can't connect with target agent use wdb serial connect? Subject: Re: HELP finding a newer LIBGCC.A with source code and/or building instructions please Subject: Re: VxWorks Boot on MVME5100 (PPC 7400) and Tornado 2.1 Subject: Re: [HELP]UDP TX/RX packet mismatch problem... Subject: Re: DosFS2 or the device driver Subject: Datacell Snapper on Synergy VGM5/VSS4 Subject: Re: Libraries for MPC8260 Subject: Re: BootRom and MMU Subject: How to enable CACHE on MPC 8260? Subject: Re: How to enable CACHE on MPC 8260? Subject: Re: Co-processer unuseable exception Subject: source code from W. Richard Stevens book Subject: Re: source code from W. Richard Stevens book Subject: ipAttach problem ?? ( two end driver) Subject: Re: Tornado project launcher not working ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Why the target server can't connect with target agent use wdb serial connect? Date: 11 Jan 2002 05:24:03 -0800 From: jlddotcn@hotmail.com (juliedan) Organization: http://groups.google.com/ Message-ID: I use a bootrom program as the bsp to generate a bootable vxworks image project. I did follows: (1) In bootrom's Makefile, I comment follows: #BOOTINIT = bootinit.c #BOOTCONFIG = bootconfig.c (2) the project was created sccussfully. (3) select wdb serial connection component for debug, and other components. select ping client component. select telnet server compnent. ...other needed components. (4) Rebuild all, scceeded. got vxworks.bin (5) download the vxworks.bin to flash by bootrom program. (6) reboot the system (7) the Hyperterminal display follows: ... Attaching interface lsi0...done. Adding 2407 symbols for standalone. Vxworks. Copyright 1984-1998 Wind River Systems, Inc. Cpu: ARM PID - ARM7TMMI (ARM) Bsp Version: 1.2/0 Creation data: Dec 29, 2001 WDB: Ready - -> (8) - ->i (the task informations, such as _tennetd, _shell, and so on) (9) config target server: (10) close Hyperterminal to free COM2. (11) start the target server. can't connect target agent, errors information as follows: Error: clnttty_rcv read timed out. (12) in pc windows dos console: ping 192.168.0.10 (the target ip address) result: request timed out (13) question1: Why the target server can't connect with target agent use wdb serial connect. question2: why the pc can't get target ping response message? thanks in advance. Liedan Ju --------------------------- Newsgroups: comp.os.vxworks Subject: Re: HELP finding a newer LIBGCC.A with source code and/or building instructions please Date: Fri, 11 Jan 2002 14:00:48 -0000 From: "Dave Korn" Organization: Lumber Cartel (tinlc) Members #2234-2237 (owing to browser refresh) Message-ID: References: Tim wrote in message ... >Running Tornado 2.0 on a Windows 2000 host, targeting the PowerPC 405GP >(without FPU.) > >The libgcc.a that I was supplied with seems to be at least 5 years old: >C:\Tornado\host\x86-win32\lib\gcc-lib\powerpc-wrs-vxworks\cygnus-2.7.2-9601 2 >6\soft-float\libgcc.a > >I would a newer version and source code and instructions for building the >same. http://www.newgcc4vxworks4ppc.cjb.net/ hth, DaveK - -- moderator of alt.talk.rec.soc.biz.news.comp.humanities.meow.misc.moderated.meow Burn your ID card! http://www.optional-identity.org.uk/ Help support the campaign, copy this into your .sig! Proud Member of the Exclusive "I have been plonked by Davee because he thinks I'm interesting" List Member # --------------------------- Newsgroups: comp.os.vxworks Subject: Re: VxWorks Boot on MVME5100 (PPC 7400) and Tornado 2.1 Date: Fri, 11 Jan 2002 14:37:13 GMT From: "Kyle Rutledge" Organization: Prodigy Internet http://www.prodigy.com Message-ID: References: Michael, I wish it was a typo, then I wouldn't feel as stupid. I misread the docmentation with regard to the "flags" field in the VxWorks Boot parameters. I mistook flag 0x80 to force ftp transfer as opposed to RSH transfer. It, of course, forces tftp transfers. So yeah, I was trying to boot via tftp from an ftp server. Thanks for your help, I probably would have looked at that for a very long time without recognizing my mistake. Kyle "Michael Lawnick" wrote in message news:a1m8ps$2de$1@snoopy.bndlg.de... > Hi Kyle, > > is your swapping between ftp and tftp one of your typos or do you really try > to boot via tftp from a ftp-server ? > > tftpGet: Error ogcurred while transgerring the file. > <-> > > On the host side, ftp reports: > > Assuming it's a typo: > Don't use stripppc, objcopyppc or anything else to remove symbol information > from your code! > AFAIK, all host-tools for PPC-targets that manipulate symbol information are > buggy. Some don't work at all, others not in conjunction with debugger > information,... . Use only images provided by the IDE. > We had a similiar problem when having stripped an image by objcopyppc that > contained debug info. Recompiled without debug-info and all worked. > -- > Mit freundlichen Grüßen, > Michael Lawnick > ============================================== > SOFTEC GmbH Tel +49-731-96600-0 > Promenade 17 Fax +49-731-96600-23 > D-89073 Ulm Michael Lawnick > Germany lawnick@softec.de > ============================================== > > "Kyle Rutledge" schrieb im Newsbeitrag > news:JMm%7.1777$AL3.348689537@newssvr17.news.prodigy.com... > > I've flashed VxWorks Boot onto a MVME5100 card and it boots to that point > > fine, but when VxWorks Boot goes to the host to download VxWorks(OS) I get > > the following: > > > > >>>>> > > Attached TCP/IP interface to fei0. > > Attaching network interface lo0... done. > > Loading... Error code 2: Access violation > > tftpGet: Error ogcurred while transgerring the file. > > Erroneous header read > > > > Error loading file: errno = 0x0. > > Can't load boot file!! > > >>> > > > > On the host side, ftp reports: > > > > >>>> > > Connection received from XXX.XXX.XXX.XXX on port XXXX > > Read request for file > > Error: EACCESS. Ext error The operation completed successfully. > > >>> > > > > I don't believe that there is anything physically wrong with the comm link > > as the vxWorks Boot binary was download by ftp via the PPC6bug command > > "niop", before it was flashed to the 5100. The ftp user permissions on > the > > host are sufficient to allow downloading of the file, as I have done it > from > > other computers attached to the host. > > > > Any ideas would be appreciated. > > > > Kyle > > > > > > > > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: [HELP]UDP TX/RX packet mismatch problem... Date: Fri, 11 Jan 2002 17:00:50 +0000 From: Graham Baxter Organization: Graham Baxter (Software) Limited Message-ID: <3C3F1A42.C2014EE2@NOSPAM.bcs.org.uk> References: <3C3D7D66.2D999429@bellsouth.net> <3C3DB452.4DFBFCBF@l8s.co.uk> I had a similar problem when I configured the TX and RX clocks the wrong way around. They were close enough to each other to work most of the time! Packet loss was 3-4%. - -- Regards, Graham Baxter (VxWorks and pSOS BSP'S) Freelance Software Engineer gbaxter@NOSPAM.bcs.org.uk --------------------------- Newsgroups: comp.os.vxworks Subject: Re: DosFS2 or the device driver Date: Fri, 11 Jan 2002 19:25:54 +0200 From: "Leonid Rosenboim" Organization: Verio Message-ID: References: <69517bb8.0201100713.63d1cc56@posting.google.com> In this case, I bet this is a driver issue, as I seen similar issue in an older version of the WRS driver for IDE/ATA - the driver waits for an interrupt for most things, but there are wait loops in the driver which use CPU polling. This might be acceptable if polling for up to few microsecs, but what if the bit doesn't flip ? ... exactly, the driver spits an error message, and eats up CPU time in tons before it does so. ALso, you are lucky you get the error message, which is a clue as to where in the driver code you should look for the error. If the message comes from disk cache or file system code, that means the driver eint printing nothing, just returning an error, which is a clue in a lesser degree, Good luck, - - Leonid "Scott Klein" wrote in message news:69517bb8.0201100713.63d1cc56@posting.google.com... > Our application consists of many tasks, all disk requests are > submitted from the same task: tFileProcessor, which issues write > requests as fast as possible to 4 files (not in a rotational manner, > most i/o's are to 1 file). Our cluster size on the disk is 32KB and > our writes are either 64KB or 1MB (the cache bypass thershold is > 64KB). We found that, during normal operation, our 1MB write requests > take about 15-20 ms to complete and that tFileProcessor uses about 6% > of the CPU time with the CPU being idle about 75% of the time. > > We discovered that we will get a 1MB write request that takes 5 > seconds before returning successfully. When this happens, > tFileProcessor uses about 75% of the CPU time and we get messages from > the driver that the disk has a problem. Subsequent writes to this file > take about 170 ms, but other files continue at 17ms. > > We are trying to figure out if this is a Wind River problem or a > device driver problem. Because it is only 1 file affected, we think > that the file system has a problem. Wind River software would also be > suspect if device driver functions do not count toward task CPU usage. > > Thanks for any clarification you can provide. > > > Scott B. Klein > Senior Engineer > Electronics Sensors and Systems Sector > Northrop Grumman Corporation > Post Office Box 746 - Mail Stop 285A > Baltimore, Maryland 21203-0746 > Telephone (410) 765-0910 > Fax: (410) 981-2045 > scott_b_klein@md.northgrum.com --------------------------- Newsgroups: comp.os.vxworks Subject: Datacell Snapper on Synergy VGM5/VSS4 Date: 11 Jan 2002 18:40:17 GMT From: Paolo Bellutta Organization: Jet Propulsion Laboratory - Pasadena CA Message-ID: Help! I'm trying to find out the correct configuration to use for a Datacell Snapper24 PMC module (frame grabber) to be used on a Synergy VGM5 or a VSS4. I currently have the configuration for a Motorola MVME2400 and a Cetia (Thales) VMPC6C-Dual. Reading the BSP and hardware manual didn't help in figuring out a setup that is working. My hope is that somebody on this forum is using a similar configuration: Snapper24 and a Synergy board. If you do, and are willing to share some info, please drop me a line. I would be extremely grateful. Thanks. Paolo Bellutta --------------------------- Newsgroups: comp.os.vxworks,comp.sys.powerpc.tech Subject: Re: Libraries for MPC8260 Date: 11 Jan 2002 11:24:22 -0800 From: jd3work@yahoo.com (jnd3) Organization: http://groups.google.com/ Message-ID: <3fcd02dd.0201111124.23b9a838@posting.google.com> References: <2a6118da.0112302226.83b6acd@posting.google.com> Hi Rashid, You should use the PPCEC603gnu library for the MPC8260 (the 8260 is built on the 603 core). You'll have to use libraries that are specific to the 8260 (603), as the 860 libraries are not compatible (voice of experience: I spent about 8 months porting an embedded application from an 860 to an 8260). Hope this helps... Jim raj_rashid@hotmail.com (Rashid) wrote in message news:<2a6118da.0112302226.83b6acd@posting.google.com>... > Hi all, > > I am working on Tornado-II for PowerPC on Solaris. Which GNU > tool should i use for MPC8260? In my Tornado we are having PPC860gnu > but no PPC8260gnu. Also the communication interface libraries like > motFccEnd.o are for MPC860, can i use these libraries for MPC8260 as > well? > > Thanks, > > Best Regards, > Rashid --------------------------- Newsgroups: comp.os.vxworks Subject: Re: BootRom and MMU Date: 11 Jan 2002 14:56:33 -0800 From: vloscomp@yahoo.com (Vinh) Organization: http://groups.google.com/ Message-ID: References: <0y%_7.14882$8e.753501@news> Hi Jim, Thank you for the pointer. I had been able to step into VxWorks code, e.g. romInit.s, for PPC750 and it seem to be fined. However, when i tried to debug the VxWorks 'sysALib.s' it fail to step over one instruction 'tlbie'. I think this involved the MMU. If you or anyone have any other suggestion, it will be greatly appreciated. Thank you, Vinh "Jim" wrote in message news:<0y%_7.14882$8e.753501@news>... > Here's a couple things I have noticed using the ICE and visionClcik ... hope > it helps ... > > 1.) In the source window when your stepping over stuff with vision click ... > it doesn't always look like the pc moves ... but if the top corner of your > source window says your at address 0x02800100 and you decide to change this > to 0x02800104 (by simply typing it in and press enter in the dialogue box) > then step (using the single step button) you will find it takes you to the > correct line ... its really misleading and TERRIBLE that the software > behaves this way ... > I don't have any problem here, i always look to the left column for addresses. > 2.) As well when your stepping realize that the line being pointed to will > no execute until you hit enter ... > Agree, but it VisionProbe should not disconect from the CPU. > 3.) As well be sure to be careful that you run commands that are using the > correct IMMR value ... if you boot your board the bootloader is normally > written like you are using the default IMMR value from the hardware reset > word ... so becareful if you init the ICE with the "in" command ... but if > you initialize the ICE with the command "inn; sr sypcr ff83" you can avoid > these conflicts ... > Agree, i don't have any problem here. Actually, i always download a reset register file to ensured the CPU is in the correct state. > Cheers, > > Jim > > "Vinh" wrote in message > news:c129947c.0201081333.5d35caae@posting.google.com... > > Hi all, > > I am trying to debug our VxWorks boot for PPC, and i can't seem to > > used VisionClick to step over an assembly instruction (tlbie) in the > > '../target/templatePpc/sysALib.s'. > > This is the sniped of code. > > > > sysALoop: > > tlbie p0 > > addi p0,p0,0x1000 /* increment bits 15-19 */ > > bdnz sysALoop /* decrement CTR, branch if CTR != 0 */ > > sync /* sync instr req'd after tlbie */ > > > > > > Can any one explain to me why we need to invalidates the TLB in > > _sysInit? > > I though that when VxWorks comes up, it will used it's MMU lib to > > initialized the PPC MMU. Some how the instruction caused the > > VisionProbe to disconnect. > > > > > > Any help is greatly appreciated. > > > > Thank you, > > Vinh Lam --------------------------- Newsgroups: comp.os.vxworks Subject: How to enable CACHE on MPC 8260? Date: Fri, 11 Jan 2002 18:24:37 -0500 From: Babu Yama Organization: Lucent Technologies Message-ID: <3C3F7435.9FC82931@lucent.com> Hi, I want to know how I can enable data and instruction cache on MPC8260. The core standard separate 16 KB each DCACHE and ICACHE. I am using WindRiver's wrSbc8260 BSP33 as the reference code. In this bsp, in config.h file, I have defined these 3 macros. #define INCLUDE_CACHE_SUPPORT #define USER_I_CACHE_ENABLE #define USER_D_CACHE_ENABLE As a result, cacheLibInit(), cacheEnable() are being called in usrConfig.c. However, in HID0 register [ICE = Instruction Cache Enable, bit 16; DCE = Data Cache Enable, bit 17] are not set and Caching is not enabled. If I set these bits 16,17 manually from visionICE emulator and then let the code run, it crashes with "unknown reason error". Any help and ideas will be greatly appreciated, Thanks, Babu --------------------------- Newsgroups: comp.os.vxworks Subject: Re: How to enable CACHE on MPC 8260? Date: Fri, 11 Jan 2002 23:55:49 GMT From: Dan Gold Organization: http://www.newsranger.com Message-ID: <9YK%7.1599$TS6.3953@www.newsranger.com> References: <3C3F7435.9FC82931@lucent.com> In article <3C3F7435.9FC82931@lucent.com>, Babu Yama says... > >Hi, >I want to know how I can enable data and instruction cache on MPC8260. The core >standard separate 16 KB each DCACHE and ICACHE. > >I am using WindRiver's wrSbc8260 BSP33 as the reference code. In this bsp, in >config.h file, I have defined these 3 macros. >#define INCLUDE_CACHE_SUPPORT >#define USER_I_CACHE_ENABLE >#define USER_D_CACHE_ENABLE >As a result, cacheLibInit(), cacheEnable() are being called in usrConfig.c. > >However, in HID0 register [ICE = Instruction Cache Enable, bit 16; DCE = Data >Cache Enable, bit 17] are not set and Caching is not enabled. You also need to define INCLUDE_MMU_BASIC, and make sure your sysPhysMemDesc[] and sysBatDesc[] arrays are properly setup for your particular hardware. You may also want to start small and just enable the MMU, to make sure you have that configured properly. Then enable the I-cache, then the d-cache. >If I set these bits 16,17 manually from visionICE emulator and then let the code >run, it >crashes with "unknown reason error". If you don't understand why this happens, you may need the "caching 101" class. See also cache flushing and invalidating. In this case, for sure just try enabling the MMU with INCLUDE_MMU_BASIC, and the go for the cache. hth, Dan Gold gold@ensemble.com --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Co-processer unuseable exception Date: Sat, 12 Jan 2002 00:53:46 GMT From: "Jeff Creem" Organization: ATT Broadband Message-ID: References: <3C3CC423.DC65B2E6@ssl.umd.edu> Are you checks looking for NaN and denormalized numbers? The error you are getting does not sound like this is the problem (I'd expect something like unimplemented FPA operation) but you were not super specific about exactly the error you are getting. "Stephen Roderick" wrote in message news:3C3CC423.DC65B2E6@ssl.umd.edu... > Hi > > We're intermittently getting one of these on a MIPS R4000 based board. The > function it occurs in is very simple, and is simply doing some unit conversion > from floats to integers. We've got overflow checking/protection built in there, > though this shouldn't be the issue. We've also looked at the MIPS manuals for > some clues but they're not particularly helpful (whatever happened to RTFM > working for you? :-) ) > > Is anyone able to shed some light on a possible reason? > > TIA > Stephen > --------------------------- Newsgroups: comp.os.vxworks Subject: source code from W. Richard Stevens book Date: Sat, 12 Jan 2002 01:49:35 GMT From: "David" Message-ID: Please help! I try to study source code from W. Richard Stevens book: Unix networking programming Networking APIs:Sockets and XTI in Linux red hat 7.0 platform. I download the file unpv12e.tar.gz, my operation is as following: gunzip -c unpv12e.tar.gz |tar -xvf- cd unpv12e ./configure cd lib make then appear in_pktinfo redifination error I am a biginner about Linux and Unix, I have no idea about it, what should I do next? Could I run the code in red hat 7 OS, please give me some guide, thank in advance! David email:davidyang_us@yahoo.com --------------------------- Newsgroups: comp.os.vxworks Subject: Re: source code from W. Richard Stevens book Date: Fri, 11 Jan 2002 21:02:50 -0500 From: joe durusau Organization: BELLSOUTH.net Message-ID: <3C3F994A.2EA0FFCB@bellsouth.net> References: Suggestion: post the question and at least one of the error messages in a linux ng, and you will get a fairly good answer. You will only get guesses here. Speaking only for myself, Joe Durusau David wrote: > Please help! > I try to study source code from W. Richard Stevens book: > Unix networking programming Networking APIs:Sockets > and XTI in Linux red hat 7.0 platform. > > I download the file unpv12e.tar.gz, my operation is as following: > > gunzip -c unpv12e.tar.gz |tar -xvf- > > cd unpv12e > > ./configure > > cd lib > > make > > then appear in_pktinfo redifination error > > I am a biginner about Linux and Unix, I have no idea about it, what should I > do next? > Could I run the code in red hat 7 OS, please give me some guide, thank in > advance! > > David > email:davidyang_us@yahoo.com --------------------------- Newsgroups: comp.os.vxworks Subject: ipAttach problem ?? ( two end driver) Date: Sat, 12 Jan 2002 15:16:29 +0900 From: "it2xpert" Organization: ONSE Telecom Message-ID: Hi, the senior vxworkers !! I'm using T2 Vxworks5.4, MPC850SR target board.... I want to use the two end driver ... configNet.h - ------------------------------ END_TBL_ENTRY endDevTbl [] = { { 0, CPM_LOAD_FUNC_IF0, CPM_LOAD_STRING_IF0, 1, NULL, FALSE}, { 1, CPM_LOAD_FUNC_IF1, CPM_LOAD_STRING_IF1, 1, NULL, FALSE}, { 0, END_TBL_END, NULL, 0, NULL, FALSE}, }; I successed the following - ---------------------------- muxDevLoad 0 muxDevStart 0 muxDevLoad 1 muxDevStart 1 Attached TCP/IP interface to cpm unit 0 Attaching interface lo0...done In targetserver, i lunch muxShow - ------------------------------ - -> muxShow Device: cpm Unit: 0 Description: MPC860 Power-QUICC Enhanced Network Driver Protocol: IP 4.4 ARP Type: 2054 Recv 0x3ebdc Shutdown 0x3eea0 Protocol: IP 4.4 TCP/IP Type: 2048 Recv 0x3ebdc Shutdown 0x3edd8 Device: cpm Unit: 1 Description: MPC860 Power-QUICC Enhanced Network Driver value = 0 = 0x0 - -> Then i want to lunch ipAttach(1,"cpm") the following - ---------------------------------- void netInit(void) { int junk; char sysName[128]; char ipAddr[32]; char gateway[32]; unsigned long netMask; LOCAL int i=1; /* device number */ /* initialize settings */ strcpy(sysName,"cpm1"); strcpy(ipAddr,"192.168.0.3"); strcpy(gateway,"192.168.0.1"); netMask = 0xffffff00; /* Setup Ethernet port */ sethostname(sysName, strlen(sysName)+1); usrNetEndDevStart("cpm", i, sysName, ipAddr, netMask); /* sysCpmEnetEnable(i); sysCpmEnetIntEnable(i); */ /* Setup info about others */ junk = routeAdd("0",gateway); } /* end netInit */ when i lunch usrNetEndDevStart() The error occurred on console. - ------------------------------------------------------- 0xf61978 (t1): Protocol is out of space. Increase IP_MAX_UNITS. Targetserver message - ---------------------------------------------------- - -> netInit Failed to attach to device cpm1 value = 0 = 0x0 what's wrong ??? Thanks you for your interest... regards... jwyu --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Tornado project launcher not working Date: 11 Jan 2002 22:16:40 -0800 From: john_94501@yahoo.com (John) Organization: http://groups.google.com/ Message-ID: <488e459a.0201112216.748b6ed4@posting.google.com> References: <71cd02f888c82c8631d4d31fbdb94d0e.8627@mygate.mailgate.org> Hello, Try using 'tornado' instead of 'launch' - that should get you the T2 project facility. From there you should be able to start target servers, windview etc. HTH, John... "Arne Henrichsen" wrote in message news:<71cd02f888c82c8631d4d31fbdb94d0e.8627@mygate.mailgate.org>... > I have installed Tornadoon a sun machine. When I type 'launch', the Tornado > launcher comes up. But when pressing the 'Project' or 'WindView' button, the > following happens: > > Tornado 2.0 > Copyright (C) 1998, Wind River Systems, Inc. > All Rights Reserved > > But no project window gets loaded... > > I am using tornado2.0.2 on sun-solaris. > > Does anybody have similar problems? > > Thanks > Arne Henrichsen --------------------------- End of New-News digest ********************** From vxwexplo-errs@csg.lbl.gov Sun Jan 13 04:03:07 2002 From: Vxworks Exploder Date: Sun Jan 13 04:03:09 PST 2002 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Sun Jan 13 04:03:04 PST 2002 Subject: Re: Libraries for MPC8260 Subject: Redirect telnet files to a device ------------------------------------------------------- Newsgroups: comp.os.vxworks,comp.sys.powerpc.tech Subject: Re: Libraries for MPC8260 Date: 12 Jan 2002 07:15:27 -0800 From: raj_rashid@hotmail.com (Rashid) Organization: http://groups.google.com/ Message-ID: <2a6118da.0201120715.7ba8cd45@posting.google.com> References: <2a6118da.0112302226.83b6acd@posting.google.com> <3fcd02dd.0201111124.23b9a838@posting.google.com> Hi Jim, Thanks for the reply. I have PPCEC603gnu library in my Tornado installation. I will try using those libraries. Thanks a lot, regards, Rashid jd3work@yahoo.com (jnd3) wrote in message news:<3fcd02dd.0201111124.23b9a838@posting.google.com>... > Hi Rashid, > > You should use the PPCEC603gnu library for the MPC8260 (the 8260 is > built on the 603 core). > > You'll have to use libraries that are specific to the 8260 (603), as > the 860 libraries are not compatible (voice of experience: I spent > about 8 months porting an embedded application from an 860 to an > 8260). > > Hope this helps... > Jim --------------------------- Newsgroups: comp.os.vxworks Subject: Redirect telnet files to a device Date: Sat, 12 Jan 2002 17:02:58 GMT From: "magicmoments4" Organization: AT&T Broadband Message-ID: <6%Z%7.18875$JF.138634@rwcrnsc52.ops.asp.att.net> I am trying to understand how to redirect console files from a telnet window to a device and then how to FTP the file to my work statation. Can anyone help me? Thanks in advance, William --------------------------- End of New-News digest ********************** From vxwexplo-errs@csg.lbl.gov Mon Jan 14 00:17:46 2002 From: "Saravana Kumar K" Date: Mon Jan 14 00:17:48 PST 2002 Subject: How to do DMA in VxWorks? Hi, I am new to VxWorks. I am using pentium processor, pc486 BSP and VxWorks 5.4. I want to do DMA transfer from my port to PC memory. I searched and got only cacheLib.h library supporting memory allocation for DMA. But how to set DMA channel, DMA address etc. for DMA transfer? Thanks in Advance Saravanan From vxwexplo-errs@csg.lbl.gov Mon Jan 14 04:03:10 2002 From: Vxworks Exploder Date: Mon Jan 14 04:03:12 PST 2002 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Mon Jan 14 04:03:06 PST 2002 Subject: building vxworks_rom for powerpc Subject: Re: ipAttach problem ?? ( two end driver) Subject: Re: Datacell Snapper on Synergy VGM5/VSS4 Subject: Re: ipAttach problem ?? ( two end driver) Subject: Questions of WindNet IPSEC Subject: a test(void) Subject: Re: ipAttach problem ?? ( two end driver) Subject: /var/tmp/....errors while compling assembly routines Subject: HDLC driver for MPC8260 ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: building vxworks_rom for powerpc Date: Sun, 13 Jan 2002 13:40:55 GMT From: "Kelly Hornsby" Organization: Excite@Home - The Leader in Broadband http://home.com/faster Message-ID: Hello All I was looking over the steps to build vxworks_rom, and I could not find out how to "split" the vxworks rom image into the rominit portion that lives high in flash (at 0xfff00000) and the rest of vxworks that lives in another section of flash, such as 0xff000000 for a 16 meg flash system. I have RCS0 set to 0xfff00000, and RCS1 set to 0xff000000, and I can place code anywhere I want, but I don't know if I can split the images into 2 pieces like that. I'm looking for tips and suggestions. thx --------------------------- Newsgroups: comp.os.vxworks Subject: Re: ipAttach problem ?? ( two end driver) Date: 13 Jan 2002 21:32:28 GMT From: Johan Borkhuis Organization: Agere Systems Message-ID: References: "it2xpert" wrote: > Hi, the senior vxworkers !! > > I'm using T2 Vxworks5.4, MPC850SR target board.... > > I want to use the two end driver ... .... > The error occurred on console. > ------------------------------------------------------- > 0xf61978 (t1): Protocol is out of space. Increase IP_MAX_UNITS. > what's wrong ??? Just what it says in the error: you have to increase IP_MAX_UNITS. The default value is 1, so if you want to use more devices you should increase this value. You can find this in your project under network components -> basic network init. -> network buffer init. Groeten, Johan - -- o o o o o o o . . . _____________________________ o _____ || Johan Borkhuis | .][__n_n_|DD[ ====_____ | borkhuis@agere.com | >(________|__|_[_________]_|__________________________| _/oo OOOOO oo` ooo ooo 'o!o!o o!o!o` === VxWorks FAQ: http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html === --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Datacell Snapper on Synergy VGM5/VSS4 Date: Mon, 14 Jan 2002 10:28:43 +1030 From: Tim Shaw Organization: Defence Science and Technology Organisation Message-ID: <3C421F33.8E4D180@dsto.defence.gov.au> References: Paolo, We haven't used the Snapper24, but one thing when moving a driver from Motorola boards to Synergy boards is that the Synergy boards have used pciAutoconfig for a long time (much longer then the MVME260x BSP). So with the Motorola driver you find some region of PCI space, manually reserve it and configure the driver to write the allocated region to the PCI Base Address Registers. But with the VGM5, your driver instead reads the region that has been allocated to the BAR (using pciConfigInLong), and assigns this value to some internal variable to use later when accessing the device. Apart from a couple of other little details, this has been the only change we've had to make when moving a driver from the MVME family to the VGM5. Tim Paolo Bellutta wrote: > Help! I'm trying to find out the correct configuration to use for a Datacell > Snapper24 PMC module (frame grabber) to be used on a Synergy VGM5 or a VSS4. > I currently have the configuration for a Motorola MVME2400 and a Cetia (Thales) > VMPC6C-Dual. Reading the BSP and hardware manual didn't help in figuring out > a setup that is working. > > My hope is that somebody on this forum is using a similar configuration: > Snapper24 and a Synergy board. If you do, and are willing to share some > info, please drop me a line. I would be extremely grateful. > > Thanks. > > Paolo Bellutta --------------------------- Newsgroups: comp.os.vxworks Subject: Re: ipAttach problem ?? ( two end driver) Date: 13 Jan 2002 16:54:28 -0800 From: youerming@163.com (Torna) Organization: http://groups.google.com/ Message-ID: References: you should set IP_MAX_UNITS to 2 surely in your application you can pass it by modifing the file prjparam.h in the project facility ! "it2xpert" wrote in message news:... > Hi, the senior vxworkers !! > > I'm using T2 Vxworks5.4, MPC850SR target board.... > > I want to use the two end driver ... > > configNet.h > ------------------------------ > END_TBL_ENTRY endDevTbl [] = > { > { 0, CPM_LOAD_FUNC_IF0, CPM_LOAD_STRING_IF0, 1, NULL, FALSE}, > { 1, CPM_LOAD_FUNC_IF1, CPM_LOAD_STRING_IF1, 1, NULL, FALSE}, > { 0, END_TBL_END, NULL, 0, NULL, FALSE}, > }; > > I successed the following > ---------------------------- > muxDevLoad 0 > muxDevStart 0 > muxDevLoad 1 > muxDevStart 1 > Attached TCP/IP interface to cpm unit 0 > Attaching interface lo0...done > > In targetserver, i lunch muxShow > ------------------------------ > -> muxShow > Device: cpm Unit: 0 > Description: MPC860 Power-QUICC Enhanced Network Driver > Protocol: IP 4.4 ARP Type: 2054 Recv 0x3ebdc Shutdown 0x3eea0 > Protocol: IP 4.4 TCP/IP Type: 2048 Recv 0x3ebdc Shutdown 0x3edd8 > Device: cpm Unit: 1 > Description: MPC860 Power-QUICC Enhanced Network Driver > value = 0 = 0x0 > -> > > Then i want to lunch ipAttach(1,"cpm") the following > ---------------------------------- > void netInit(void) > { > int junk; > char sysName[128]; > char ipAddr[32]; > char gateway[32]; > unsigned long netMask; > > LOCAL int i=1; /* device number */ > > /* initialize settings */ > > strcpy(sysName,"cpm1"); > strcpy(ipAddr,"192.168.0.3"); > strcpy(gateway,"192.168.0.1"); > netMask = 0xffffff00; > > /* Setup Ethernet port */ > sethostname(sysName, strlen(sysName)+1); > usrNetEndDevStart("cpm", i, sysName, ipAddr, netMask); > /* > sysCpmEnetEnable(i); > sysCpmEnetIntEnable(i); > */ > /* Setup info about others */ > junk = routeAdd("0",gateway); > > > } /* end netInit */ > > when i lunch usrNetEndDevStart() > The error occurred on console. > ------------------------------------------------------- > 0xf61978 (t1): Protocol is out of space. Increase IP_MAX_UNITS. > > Targetserver message > ---------------------------------------------------- > -> netInit > Failed to attach to device cpm1 > value = 0 = 0x0 > > > what's wrong ??? > > Thanks you for your interest... > > regards... > jwyu --------------------------- Newsgroups: comp.os.vxworks Subject: Questions of WindNet IPSEC Date: 13 Jan 2002 20:55:24 -0800 From: frankzhou@peoplemail.com.cn (frankzhou) Organization: http://groups.google.com/ Message-ID: <58706286.0201132055.27e82e19@posting.google.com> hi,all here are some question of WindNet IPSEC: 1.Is there any benchmark of WindNet IPSEC? 2.Now WindNet IPSEC can support the Motorola MPC180 security processor. Are there some interface functions of MPC180 or other hardware security device in WinNet IPSEC? 3.In WindNet IPSEC and IKE, the mode of the key manage is pre-shared, then we must set the fixed IP addrs, is that OK? Does WindNet Ipsec support dynamic IP addrs? for example,when we use the remote access, IP addrs is dynamic,then how to do it? Best Regards, Frank --------------------------- Newsgroups: comp.os.vxworks Subject: a test(void) Date: Mon, 14 Jan 2002 15:07:30 +0800 From: "r0uter" Organization: Changzhou Telecom Message-ID: --------------------------- Newsgroups: comp.os.vxworks Subject: Re: ipAttach problem ?? ( two end driver) Date: 13 Jan 2002 23:13:52 -0800 From: sardine_sz@163.net (sardine) Organization: http://groups.google.com/ Message-ID: References: ok, pls remember: increase IP_MAX_UNITS in your vxworks kernel project facility configuration macro. use "find" mechanism in the tornado project facility. you didn't changed IP_MAX_UNITS in your source code. --------------------------- Newsgroups: comp.os.vxworks Subject: /var/tmp/....errors while compling assembly routines Date: 13 Jan 2002 23:46:43 -0800 From: alimohd@logiceastern.com (Ali) Organization: http://groups.google.com/ Message-ID: Hi everybody I am trying to compile and build the assembly routines for MPC8260 .but every time i compile some tmp errors are generated that have really perturbed me . Can any one tell me why these erros come and how can i remove them . thanx in advance ali --------------------------- Newsgroups: comp.os.vxworks Subject: HDLC driver for MPC8260 Date: Mon, 14 Jan 2002 09:25:51 +0100 From: pjemec Organization: IskraTEL Message-ID: Hi, could anybody recomend to me which HDLC driver for MPC8260 should I use in VxWorks. Thanks in advance for help! Marjan Jemec http://www.iskratel.si --------------------------- End of New-News digest ********************** From vxwexplo-errs@csg.lbl.gov Tue Jan 15 02:16:50 2002 From: "Tyler, Paul A (DBY)" Date: Tue Jan 15 02:16:52 PST 2002 Subject: Re: Dumb question about the target-resident shell Try setting sysCplusEnable = 1 in your start script. This allows the VxWorks shell to mangle the entry name, and it can then find it in the symbol table. Paul Tyler Measurement Systems Specialist Rig Test Systems Team, Facility Engineering Dept. Rolls-Royce plc. e-mail Paul.A.Tyler@Rolls-Royce.com Tel. +44 (1332) 247808 Fax. +44 (1332) 247928 ----- Original Message ----- From: Trevor Ward Newsgroups: comp.os.vxworks Sent: Wednesday, January 02, 2002 4:08 PM Subject: Dumb question about the target-resident shell > Since there aren't any stupid questions (only stupid answers), I thought I'd > provide the exception that proves the rule... > > I've built a bootable image using Tornado 2, extending the default kernel with > some extra .o object modules of my own. The entry point into my (C++) > application is a function called 'fdl' that takes an int parameter. I want to > execute 'fdl(1)' from a startup script whenever the target boots. I have a > target resident symbol table, so I can perform 'lkup "fdl"' on the target shell > and it gives me an entry like: > > fdl(int) 0x0011e3b0 text > > If I type 'fdl(1)' I get the error 'undefined symbol: fdl', but if I type > '0x0011e3b0(1)' then it runs my application as desired. The problem is that the > address of fdl, i.e. 0x0011e3b0, moves around as I change to code and I don't > want to keep having to change the startup script every time I recompile. I'm > sure there must be a really easy way of resolving this problem, but I can't > find it... > > > -- > Posted via Mailgate.ORG Server - http://www.Mailgate.ORG The data contained in, or attached to, this e-mail, may contain confidential information. If you have received it in error you should notify the sender immediately by reply e-mail, delete the message from your system and contact +44(0)1332 248119 (the Rolls-Royce IT Security Director) if you need assistance. Please do not copy it for any purpose, or disclose its contents to any other person. An e-mail response to this address may be subject to interception or monitoring for operational reasons or for lawful business practices. (c) 2001 Rolls-Royce plc From vxwexplo-errs@csg.lbl.gov Tue Jan 15 03:36:52 2002 From: mmathers@landis-lund.co.uk Date: Tue Jan 15 03:36:54 PST 2002 Subject: Universe II PCI/VME bridge Hello all, Not a question directly about vxWorks, but something that someone in the group may have experience of. We are using a Pentium based VME card that uses the Universe II PCI to VME bridge. When accessing the VME memory space I have found that I get a big endian/ little endian problem. I've looked in the Universe II manual and have come across this statement in appendix E: "The Universe always performs Address Invariant translation between the PCI and VMEbus ports. Address Invariant mapping preserves the byte ordering of a data structure in a littleendian memory map and a big-endian memory map." The tables in the appendix indicate that the Universe device should perform the big/little endian translation (if I understand the table correctly). So why do I need to twist the bytes in my data? Does anybody else using the Universe II device have this problem. TIA Mick Mathers ======================================================= Mick Mathers Cranfield Precision, 90 Central Avenue, Wharley End, Cranfield, Bedfordshire MK43 0JR Tel: 01234 754075 Fax: 01234 754223 From vxwexplo-errs@csg.lbl.gov Tue Jan 15 04:03:19 2002 From: Vxworks Exploder Date: Tue Jan 15 04:03:22 PST 2002 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Tue Jan 15 04:03:15 PST 2002 Subject: Re: /var/tmp/....errors while compling assembly routines Subject: Re: compile assembly codes?? "unresolved expression that must be .." Subject: Re: LC82C169 Driver... Subject: Re: Redirect telnet files to a device Subject: Cannot get dhcpc parameters!!! Subject: Re: Co-processer unuseable exception Subject: FTP commands Subject: Re: Cannot get dhcpc parameters!!! Subject: why coulud not my ARM7TDMI and vxworks system support floating point? Subject: Re: why coulud not my ARM7TDMI and vxworks system support floating point? Subject: Re: FTP commands Subject: SingleStep Subject: Re: building vxworks_rom for powerpc Subject: Re: Does Tornado require write access to download links/libraries? Subject: Re: LC82C169 Driver... Subject: Tornado 3 and Windows 2000 Subject: DHCP over PPP Subject: ARM coff2elf conversion Subject: Re: why coulud not my ARM7TDMI and vxworks system support floating Subject: change usrconfig.c , but not works Subject: Re: LC82C169 Driver... Subject: Re: why coulud not my ARM7TDMI and vxworks system support floating Subject: Very important, I need the Login/password of my Hi-Focus ADSL modem (ECI) Subject: what program do i need to use to exploit the udp port 17185 of a system working under vxworks ? Subject: Re: what program do i need to use to exploit the udp port 17185 of a system working under vxworks ? Subject: Re: pci error in IDT Rc32332 Subject: Building GCC 2.95.x for vxWorks (PPC), Solaris host Subject: taskdelay and interrupts Subject: Hi All Subject: WRS Announces Transfer of FreeBSD Sponsorship Subject: Re: why coulud not my ARM7TDMI and vxworks system support floating point? Subject: Re: change usrconfig.c , but not works Subject: Re: taskdelay and interrupts Subject: Re: change usrconfig.c , but not works Subject: debuging fads860 interrupt problems "uninitialized interrupt" Subject: Re: taskdelay and interrupts Subject: lock timeout using shared memoy Subject: Re: BootRom and MMU Subject: Re: what program do i need to use to exploit the udp port 17185 of a system working under vxworks ? Subject: Waiting on multiple events Subject: Re: change usrconfig.c , but not works Subject: Re: why coulud not my ARM7TDMI and vxworks system support floating ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: /var/tmp/....errors while compling assembly routines Date: Mon, 14 Jan 2002 12:19:50 -0000 From: "Dave Korn" Organization: Lumber Cartel (tinlc) Members #2234-2237 (owing to browser refresh) Message-ID: References: Ali wrote in message ... >Hi everybody > I am trying to compile and build the assembly routines for MPC8260 >.but every time i compile some tmp errors are generated that have >really perturbed me . Can any one tell me why these erros come and how >can i remove them . >thanx in advance >ali You really need to tell us what the errors actually are and what system you're using, otherwise everyone can only guess. So my first two guesses are: do you have write perms to the tmp dir ? Is the drive nearly full ? DaveK - -- moderator of alt.talk.rec.soc.biz.news.comp.humanities.meow.misc.moderated.meow Burn your ID card! http://www.optional-identity.org.uk/ Help support the campaign, copy this into your .sig! Proud Member of the Exclusive "I have been plonked by Davee because he thinks I'm interesting" List Member # --------------------------- Newsgroups: comp.os.vxworks Subject: Re: compile assembly codes?? "unresolved expression that must be .." Date: Mon, 14 Jan 2002 12:32:59 -0000 From: "Dave Korn" Organization: Lumber Cartel (tinlc) Members #2234-2237 (owing to browser refresh) Message-ID: References: <9t1vih$75m$1@news.asu.edu> bcoxsprint12@earthlink.net wrote in message ... >In article <9t1vih$75m$1@news.asu.edu>, Okehee Goh writes: >>I'd like to compile assembly codes for target machine, MBX860. >>This assembly codes will not be associated with VxWorks OS. >>As a compiler, I tried to use ccppc(gcc) compiler that comes with TornadoII >>targeting MBX860. >>But it complained with "Error: unresolved expression that must be resolved". >>I don't know how to solve this problem. >> >>The following is the command line to compile assembly code "ae3.s". >>The next one includes error message that ccppc generated. >>The third part is "ae3.s" >>I will appreciate your help. Thank you so much. >> >>1. >> >>ccppc -g -mcpu=860 -ansi -nostdinc -fvolatile -fno-builtin -fno-for-scope - -P >> -x assembler-with-cpp -I. -IC:/Tornado/target/h -DCPU=PPC860 -c ae3.s >> >>2. >> >>C:\DOCUME~1\ogoh\LOCALS~1\Temp\cca02324.s: Assembler messages: >>C:\DOCUME~1\ogoh\LOCALS~1\Temp\cca02324.s:9: Error: unresolved expression >>that must be resolved >>C:\DOCUME~1\ogoh\LOCALS~1\Temp\cca02324.s:10: Error: unresolved expression >>that must be resolved >> >>.... >> >>3. "ae3.s" >> >> .text >> >> .align 2 >> .globl main >>main: >> .long 0 >> .long 0 >> .long 0 >> lis r7,0x0180 >> addi r7,r7,0x1000 >> addi r8,r0,0 >> stw r8,0(r7) >> stw r8,0x1000(r7) >> stw r8,0x2000(r7) [SNIP] >The line numbers don't match, but the local label "..L4" seems >to have too many leading periods. The reference to that label >it the branch instruction at the end only has one period. Given that the line numbers don't match, I suspect the real problem is that gas expects to find the registers referred to by numbers in the source file, rather than "rN". You can't refer to registers by the "rN" notation without first using: #include to pull in #defines for the register names. Alternatively, you have to write your .s file to look like: >> lis 7,0x0180 >> addi 7,7,0x1000 >> addi 8,0,0 >> stw 8,0(7) >> stw 8,0x1000(7) >> stw 8,0x2000(7) DaveK - -- moderator of alt.talk.rec.soc.biz.news.comp.humanities.meow.misc.moderated.meow Burn your ID card! http://www.optional-identity.org.uk/ Help support the campaign, copy this into your .sig! Proud Member of the Exclusive "I have been plonked by Davee because he thinks I'm interesting" List Member # --------------------------- Newsgroups: comp.os.vxworks Subject: Re: LC82C169 Driver... Date: Mon, 14 Jan 2002 13:47:54 +0100 From: "Werner Schiendl" Message-ID: <3c42d37d@brateggebdc5.br-automation.co.at> References: <21595172.0201072023.67f7fc78@posting.google.com> <3C3AF6D0.9F54E13F@varndellengineering.com> <3C3EE8EA.A08361FE@varndellengineering.com> "George Varndell" wrote in message news:3C3EE8EA.A08361FE@varndellengineering.com... - -- snip -- > > > Maybe he's gung ho and looking for a starting point. Who knows? > > Does his first post really seem gung ho to you? It reads like the > desperate cry of a person who took one step, stumbled, and then > gave up. He needed to have his ass kicked. Don't agree. Given the original post, you have absolutely no clue about the real expertise of the original poster. You only know that he tries to write a network driver for VxWorks and probably has not done that particular thing before. But he may have well a lot of experience with VxWorks in general and/or networking stuff and/or ethernet hardware etc. Asking for a good starting point does not require your kicking anybody's ass. Provide useful help or keep it. > I will be one of the first to jump in and help a person whose post > demonstrates that he/she has made a genuine attempt to solve the > problem and has become stuck on something specific. With aggressivly trying to sell your services? regards Werner --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Redirect telnet files to a device Date: Mon, 14 Jan 2002 14:00:05 +0100 From: "Werner Schiendl" Message-ID: <3c42d657$1@brateggebdc5.br-automation.co.at> References: <6%Z%7.18875$JF.138634@rwcrnsc52.ops.asp.att.net> Hi, what do you actually want to do? It is not clear to me, from your post... Do you want to capture the output of a telnet session with your target? How to do this, depends on which telnet client you are using... To FTP files to and from your target system, you need to include FTP support into it. (And of course some sort of file system, where you can store the files). hth Werner "magicmoments4" wrote in message news:6%Z%7.18875$JF.138634@rwcrnsc52.ops.asp.att.net... > I am trying to understand how to redirect console files from a telnet window > to a device and then how to FTP the file to my work statation. > > Can anyone help me? > > Thanks in advance, > > William > > --------------------------- Newsgroups: comp.os.vxworks Subject: Cannot get dhcpc parameters!!! Date: Mon, 14 Jan 2002 13:09:11 GMT From: "Jay Lee" Organization: Dacom Message-ID: Hi, everyone? I've got a problem with DHCP client. Even though I set the dhcp client options that I wanna get, I couldn't get some of the parameters like dns server ip and router ip. A strange thing is, when I print the parameters by using 'dhcpcParamsShow()', I can see what I wanna know. For reference, I attached my codes which are based on Tornado 2.0.2. Anything is gonna be helpful for me... Thanks to all.. ============[ source code : dhcp_client.c ]================= if (dhcpcOptionSet(pCookie, _DHCP_NAME_SERVER_TAG) == ERROR) gen_printf("Set _DHCP_NAME_SERVER_TAG failed with error %x\n", errno); if (dhcpcOptionSet(pCookie, _DHCP_DNS_SERVER_TAG) == ERROR) gen_printf("Set DHCP_DNS_SERVER_TAG failed with error %x\n", errno); if (dhcpcOptionSet(pCookie, _DHCP_DNS_DOMAIN_TAG) == ERROR) gen_printf("Set _DHCP_DNS_DOMAIN_TAG failed with error %x\n", errno); if (dhcpcOptionSet(pCookie, _DHCP_ROUTER_TAG) == ERROR) gen_printf("Set _DHCP_ROUTER_TAG failed with error %x\n", errno); result = dhcpcBind(pCookie, TRUE); memset((char *)&srvAddr, 0x00, sizeof(srvAddr)); if (dhcpcServerGet(pCookie, &srvAddr) == ERROR) { gen_printf("Get server address failed with error %x\n", errno); } else { gen_printf("DHCP Server address : %s\n", inet_ntoa(srvAddr)); } /*************************************************************************** ***/ /*************************************************************************** ***/ /************************ Here, I can see the parameters *************************/ /*************************************************************************** ***/ /*************************************************************************** ***/ dhcpcParamsShow(pCookie); memset((char *)¶mList, 0x00, sizeof(paramList)); paramList.subnet_mask = (struct in_addr *)malloc(sizeof(struct in_addr)); memset((char *)paramList.subnet_mask, 0x00, sizeof(paramList.subnet_mask)); paramList.sname = (char *)malloc(256); memset((char *)paramList.sname, 0x00, sizeof(paramList.sname)); paramList.dns_server = (struct in_addrs *)malloc(sizeof(struct in_addrs)); memset((char *)paramList.dns_server, 0x00, sizeof(paramList.dns_server)); paramList.dns_domain = (char *)malloc(256); if (dhcpcParamsGet(pCookie, ¶mList) == ERROR) gen_printf("Can not get parameter list with error %x \n", errno); else { gen_printf("DHCP server host name : %s\n", paramList.sname); gen_printf("DHCP dns domain name : %s\n", paramList.dns_domain); gen_printf("DHCP server's IP address : %s\n", inet_ntoa(paramList.server_id)); gen_printf("Pre-dhcp assigned address : %s\n", inet_ntoa(paramList.ciaddr)); gen_printf("Client's assigned address : %s\n", inet_ntoa(paramList.yiaddr)); gen_printf("Client's subnet mask : %s\n", inet_ntoa(*(paramList.subnet_mask))); /*************************************************************************** ***/ /*************************************************************************** ***/ /******************** I cannot get the following two parameters ********************/ /*************************************************************************** ***/ /*************************************************************************** ***/ gen_printf("DNS Server IP address : %s\n", inet_ntoa(*(paramList.router->addr))); gen_printf("Router IP address : %s\n", inet_ntoa(*(paramList.dns_server->addr))); =========================================================== Jay Lee --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Co-processer unuseable exception Date: Mon, 14 Jan 2002 09:05:36 -0500 From: Stephen Roderick Organization: University of Maryland College Park Message-ID: <3C42E5A7.AAAB167A@ssl.umd.edu> References: <3C3CC423.DC65B2E6@ssl.umd.edu> Jeff Creem wrote: > Are you checks looking for NaN and denormalized numbers? No they're not, but we're storing the numbers prior to the calculation and they don't result in either of these. Interestingly, if I change the intermediate result to a double instead of a float, the exception doesn't happen, even though the result is 0.0 > The error you are getting does not sound like this is the problem > (I'd expect something like unimplemented FPA operation) but you > were not super specific about exactly the error you are getting. I agree, I don't think it's a bounds issue. The error is "Co-processor unusable Exception" at the shell, along with some register values. The actual code looks like (simplified) int16_t convert(float in, /* other params */ ) float out; /* use 'other params' to determine 'conversionFactor' */ out = (int16_t)(in / conversionFactor); return out; This generated the exception, so I changed it to... int16_t convert(float in, /* other params */ ) float out; float result; /* use 'other params' to determine 'conversionFactor' */ result = in / conversionFactor; if (result < (float)INT16_MIN) out = INT16_MIN else if (result > (float)INT16_MAX) out = INT16_MAX else out = (int16_t)result endif return out; And the exception occurs on the 'out = (int16_t)result' line! We stored the in, out, conversionFactor, and result values, and they are nothing unusual, no NANs, etc. As soon as I changed result from a float to a double, the problem went away. A fine workaround, but unacceptable in general as I have no idea where else to do this in the program! Any ideas or suggestions appreciated! Stephen --------------------------- Newsgroups: comp.os.vxworks Subject: FTP commands Date: Mon, 14 Jan 2002 15:09:59 +0100 From: Nathalie BOCQUIER Organization: Service BEP Message-ID: <3C42E6B6.9BB4E6B9@sepro-robotique.com> Hello everybody ! Does anybody know the list of the available FTP commands for the Vxworks 5.4 FTP server ? Thanks very much Nathalie --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Cannot get dhcpc parameters!!! Date: Mon, 14 Jan 2002 14:10:39 +0000 (UTC) From: David Laight Organization: BT Openworld Message-ID: <3C42E6E0.40770FC6@l8s.co.uk> References: Jay Lee wrote: > > Hi, everyone? > > I've got a problem with DHCP client. > Even though I set the dhcp client options that I wanna get, I couldn't get > some of the parameters like dns server ip and router ip. > A strange thing is, when I print the parameters by using > 'dhcpcParamsShow()', I can see what I wanna know. > For reference, I attached my codes which are based on Tornado 2.0.2. Try something like: opt_len = sizeof rp->domainName; if (!dhcpcOptionGet( lease, _DHCP_DNS_SERVER_TAG, &opt_len, rp->domainName ) && opt_len >= 4 ) { /* Save IP address of DNS resolvers... */ if (opt_len > (int)sizeof dns_servers) opt_len = sizeof dns_servers; bcopy( rp->domainName, dns_servers, opt_len ); ns_count = opt_len / 4; /* Get current resolv options - we set the whole lot but might as well do a read-write in case anything new gets added. Also it lets is detect whether the library has been initialised. */ resolvParamsGet( rp ); /* Get any DNS domain name fron DHCP response, default to that from the current flash configuration */ opt_len = sizeof rp->domainName; if (dhcpcOptionGet( lease, _DHCP_DNS_DOMAIN_TAG, &opt_len, rp->domainName )) { char *svr; strcpy( rp->domainName, sysBootParams.other ); svr = index( rp->domainName, ':' ); if (svr) *svr = 0; } else rp->domainName[ opt_len ] = 0; /* Update DNS name server list, we use the existing value to detect whether the library has been initialised. */ init_resolv = rp->nameServersAddr[0][0]; bzero( &rp->nameServersAddr, sizeof rp->nameServersAddr ); do { ns_count--; inet_ntoa_b( dns_servers[ ns_count ], rp->nameServersAddr[ ns_count ] ); } while (ns_count); /* Finally give resolv code its info */ rp->queryOrder = QUERY_LOCAL_FIRST; if (!init_resolv || init_resolv == '0') resolvInit( rp->nameServersAddr[0], rp->domainName, 0 ); resolvParamsSet( rp ); Worked for me..... I discovered that requesting the default route causes it to be used. David --------------------------- Newsgroups: comp.os.vxworks Subject: why coulud not my ARM7TDMI and vxworks system support floating point? Date: 14 Jan 2002 06:17:05 -0800 From: hurunmin@sina.com (davis) Organization: http://groups.google.com/ Message-ID: <15decc28.0201140617.399b988@posting.google.com> Hello, everyone! Please help me about floating point support on ARM7TDMI. Now my develop enviroment is: CPU: ARM7TDMI(S3C4510B), RAM: SDRAM 16MBytes, OS: Tornado 2.0 + vxWorks 5.4 Host OS: windows98. But this system couldn't support floating point! The fllowing is a sample about floating point operation: #include "vxworks.h" #include "stdio.h" #include "math.h" float ft() { float value; float temp; value = 2.0; temp = value * 23.3344; printf("value: %2.1f\n", value); printf("temp: %2.1f\n", temp); return temp; } those fllowing is the result after downloading the object: - -> ft Exception number 16: Task: 0xd6e3c4 (t5) Data abort Program Counter: 0x9880c Status Register: 0x80000013 91e8 _vxTaskEntry +28 : 3936c () 39410 _wdbFuncCallLibInit+128: _ft () 1a5fb4 _ft +1c : ___extendsfdf2 () 9960c ___extendsfdf2 +1c : 987f4 () value = 0 = 0x0 Also, my project have those configuration: configAll.h #define INCLUDE_FLOATING_POINT /* floating point I/O */ usrConfig.c: #ifdef INCLUDE_FLOATING_POINT floatInit (); /*initialize floating point I/O */ #endif /* INCLUDE_FLOATING_POINT */ Now, how can I use floating point operation in my target board?? Thanks for your help! hurunmin@sina.com --------------------------- Newsgroups: comp.os.vxworks Subject: Re: why coulud not my ARM7TDMI and vxworks system support floating point? Date: 14 Jan 2002 10:51:20 -0500 From: Bill Pringlemeir Organization: Bell Sympatico Message-ID: References: <15decc28.0201140617.399b988@posting.google.com> Sender: bpringlemeir@DeadDuck >>>>> "davis" == davis writes: davis> Hello, everyone! Please help me about floating point support davis> on ARM7TDMI. [snip] davis> Also, my project have those configuration: configAll.h #define davis> INCLUDE_FLOATING_POINT /* floating point I/O */ davis> Now, how can I use floating point operation in my target davis> board?? Thanks for your help! You did include floating point support in both the vxWorks image and your project. If the `ft()' function is downloaded to a board, then you have to update the kernel to support floating point. The kernel (a bootable vxWorks image) must then be recompiled, and downloaded to the device (possibly upgrading Eprom or Flash etc). I have used floating point in ARM mode; I am unsure if I have ever used it in thumb mode. hth, Bill Pringlemeir. - -- A small bomb is in the air. Many people will ask you for advice about staying where you are. Don't forget to watch Dr. Who. You'll someday be unhappy. vxWorks FAQ, "http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html" --------------------------- Newsgroups: comp.os.vxworks Subject: Re: FTP commands Date: Mon, 14 Jan 2002 17:06:04 +0100 From: "Werner Schiendl" Message-ID: <3c4301ee$1@brateggebdc5.br-automation.co.at> References: <3C42E6B6.9BB4E6B9@sepro-robotique.com> Hi, try the help command (of the server, not the client!) From Windows 2000 command line FTP client it can be get with: quote help (you need to connect to your server for this to work before, of course) other FTP clients may have slightly different syntax to send (quote) a command to the server. hth Werner "Nathalie BOCQUIER" wrote in message news:3C42E6B6.9BB4E6B9@sepro-robotique.com... > Hello everybody ! > > Does anybody know the list of the available FTP commands for the Vxworks > 5.4 FTP server ? > Thanks very much > > Nathalie > --------------------------- Newsgroups: comp.os.vxworks Subject: SingleStep Date: 14 Jan 2002 11:07:43 -0600 From: "Alex" Message-ID: <3c42b08e$1@news.barak.net.il> Hello, Has somebody experience in work with SingleStep 7.6.2. together with VxWorks 5.4.? Thank you. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: building vxworks_rom for powerpc Date: Mon, 14 Jan 2002 12:10:00 -0500 From: "William Dennen" Organization: Motorola Message-ID: References: Kelly First, FLASH doesn't reside at FFF00000. That address is the reset vector for PowerPC processors. Flash Bank A (soldered) resides at FF000000 (or sometimes at F4000000 when using the MVME5100); Bank B (socketed) resides at FF800000. The jumper sets which bank is mapped onto the address space beginning at FFF00000. It's a bad idea, even though it does work, to flash to FFF00000 and rely on the jumper to redirect to the correct address space. This undocumented "feature" was broken a few releases ago in the PPCBugs with catastrophic results. It's best to flash where you mean to ... Regarding your real question, it depends on the memory organization within flash. Most of the board's are split odd/even bytes. Review the Programmer's Manual in the SMC section (for Hawk or Harrier) or Falcon to determine. Regards Bill Dennen CRC/GTAC Motorola Computer Group "Kelly Hornsby" wrote in message news:H7g08.48704$LQ1.14615046@news2.nash1.tn.home.com... > Hello All > > I was looking over the steps to build vxworks_rom, and I could not find out > how to "split" the vxworks rom image into the rominit portion that lives > high in flash (at 0xfff00000) and the rest of vxworks that lives in another > section of flash, such as 0xff000000 for a 16 meg flash system. > > I have RCS0 set to 0xfff00000, and RCS1 set to 0xff000000, and I can place > code anywhere I want, but I don't know if I can split the images into 2 > pieces like that. > > I'm looking for tips and suggestions. > > thx > > > > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Does Tornado require write access to download links/libraries? Date: 14 Jan 2002 10:37:02 -0800 From: matthew.becker@usa.xerox.com (Matthew Becker) Organization: http://groups.google.com/ Message-ID: <94e55b6.0201141037.7cafa04b@posting.google.com> References: <94e55b6.0201071416.23f691ff@posting.google.com> I hadn't received any response yet, so I'm reposting this question. Other debug tools typically do not require write access in order to download code and trace through it (e.g. visionClick, SingleStep, etc.). My experience so far indicates that the Tornado tools DO require write access. Could someone please explain if this is the case with Tornado, and, if possible, please provide reasons why. Thanks in advance. - -Matt - - Matthew Becker Xerox Corporation matthew.becker@usa.xerox.com matthew.becker@usa.xerox.com (Matthew Becker) wrote in message news:<94e55b6.0201071416.23f691ff@posting.google.com>... > Hello, > > I've noticed that if I don't have write access to a library/link, that > I am unable to download it to the target. > > I select "Project->Download" from the Tornado main menu. I then > select the appropriate file from the "Download objects" GUI. I then > receive the following message, > > > F:\Runtime.Phoenix8260\MainApp\obj.iphoenixPlus\appiphoenixPlus > > This file exists with Read Only attributes. > > Please use a different file name. > > I have read-execute access to this file, but not write access. Does > Tornado require write access in order to download links/libraries (I > am using Tornado 2.0.2)? If so, then could someone please explain why > this is the case. > > Thanks in advance. > > -Matt > > - > Matthew Becker > Xerox Corporation > matthew.becker@usa.xerox.com --------------------------- Newsgroups: comp.os.vxworks Subject: Re: LC82C169 Driver... Date: Mon, 14 Jan 2002 18:53:50 GMT From: Randy Ryan Organization: Our-Kids Message-ID: <3C432938.49002A35@pacbell.net> References: <21595172.0201072023.67f7fc78@posting.google.com> <3C3AF6D0.9F54E13F@varndellengineering.com> <3C3EE8EA.A08361FE@varndellengineering.com> <3c42d37d@brateggebdc5.br-automation.co.at> Werner Schiendl wrote: > > Don't agree. Given the original post, you have absolutely no clue about the > real expertise of the original poster. You only know that he tries to write > a network driver for VxWorks and probably has not done that particular thing > before. It is the same type of question I would have posted and have the past. Also the more difficult one to get answers for. Between proprietary issues and people wanting to make a buck consulting, it can get frustrating getting starting points and snippets. Randy, NovaSol, http://www.nova-sol.com Seeking Sr. Enginneers that want to live and work in Hawaii --------------------------- Newsgroups: comp.os.vxworks Subject: Tornado 3 and Windows 2000 Date: Mon, 14 Jan 2002 11:19:16 -0800 From: "mc" Message-ID: Hi all, I installed Tornado 3 on Windows 2000 SP1 w/IE 6 (Pentium 3) but when I try to launch it, nothing happens: no error message, no crash, nothing. I tried to reinstall it more than once, but it still doesn't work. Any idea? Thanks in advance. Regards, MC --------------------------- Newsgroups: comp.os.vxworks Subject: DHCP over PPP Date: 14 Jan 2002 11:46:01 -0800 From: jzhang@drs.ca (Jason) Organization: http://groups.google.com/ Message-ID: Hello All, I am going to implement the DHCP over PPP. By default, the vxWorks only supports DHCP over Ethernet. It would be greatly appreciated if anyone can give me some ideas or sample codes about DHCP over PPP. Thanks in advanges --------------------------- Newsgroups: comp.os.vxworks Subject: ARM coff2elf conversion Date: Mon, 14 Jan 2002 15:21:49 -0500 From: "Harjinder Chayra" Organization: Storm Internet Services Message-ID: Sender: terayon@gw-terayon.storm.ca I'd like to be able to build a vxWorks/bootrom image (for the ARM9) and then load it into the ARM debugger to look at the resultant code etc. BUT ELF is not a supported format for the ARM from Windriver. Are their any COFF to ELF converters out there? any other suggestions? Thanks Harjinder --------------------------- Newsgroups: comp.os.vxworks Subject: Re: why coulud not my ARM7TDMI and vxworks system support floating point? Date: Mon, 14 Jan 2002 20:25:56 +0000 (UTC) From: David Laight Organization: BT Openworld Message-ID: <3C433EDC.423EF732@l8s.co.uk> References: <15decc28.0201140617.399b988@posting.google.com> davis wrote: > > Hello, everyone! Please help me about floating point support on ARM7TDMI. > Now my develop enviroment is: > CPU: ARM7TDMI(S3C4510B), > RAM: SDRAM 16MBytes, > OS: Tornado 2.0 + vxWorks 5.4 > Host OS: windows98. > Exception number 16: Task: 0xd6e3c4 (t5) > > Data abort > Program Counter: 0x9880c > Status Register: 0x80000013 > 91e8 _vxTaskEntry +28 : 3936c () > 39410 _wdbFuncCallLibInit+128: _ft () > 1a5fb4 _ft +1c : ___extendsfdf2 () > 9960c ___extendsfdf2 +1c : 987f4 () Lets see: vxWorks found extendsfdf2 (float to double) at 9960c (about 600k) so this is likely to be within the kernel. Your module is loaded near 1a5fb4 (1.7M) which will usually be just above the top of the resident kernel. So the host tools think the fp routines are in the vxworks image you are running. I would disassemble a bit of code from 0x9960c and 0x987f4, you might find they are not the function prologue you might expect - indicating a mismatch between the host symbol table and the loaded image. David --------------------------- Newsgroups: comp.os.vxworks Subject: change usrconfig.c , but not works Date: Mon, 14 Jan 2002 20:35:12 GMT From: "Frank" Message-ID: <4iH08.15119$iM.14260@news1.bloor.is> hi, I met a "simple" problem, but made me mad. When I changed " target/config/all/usrconfig.c", but seems it make no sense when I re-build all my project. I doubt the "usrconfig.c" never be re-compile when I re-build my project. I use Tornado 2 and VxWorks 5.4 for 860. Pls give me some advice. Thank you Frank - -- --------------------------- Newsgroups: comp.os.vxworks Subject: Re: LC82C169 Driver... Date: 14 Jan 2002 15:42:16 -0500 From: Bill Pringlemeir Organization: Bell Sympatico Message-ID: References: <21595172.0201072023.67f7fc78@posting.google.com> <3C3AF6D0.9F54E13F@varndellengineering.com> <3C3EE8EA.A08361FE@varndellengineering.com> <3c42d37d@brateggebdc5.br-automation.co.at> <3C432938.49002A35@pacbell.net> Sender: bpringlemeir@DeadDuck >>> Werner Schiendl wrote: >> Don't agree. Given the original post, you have absolutely no clue >> about the real expertise of the original poster. You only know >> that he tries to write a network driver for VxWorks and probably >> has not done that particular thing before. >> >>>>> "Randy" == Randy Ryan writes: Randy> It is the same type of question I would have posted and have Randy> the past. Also the more difficult one to get answers for. Randy> Between proprietary issues and people wanting to make a buck Randy> consulting, it can get frustrating getting starting points and Randy> snippets. That is a bunch of bull. For one thing the title of this message is stupid. If the OP had looked around they would find a plethora of Linux driver (Ie Tulip) that take care of many oddities specifically related to the chipset in question. The second part of the equation is to look at {any*} ethernet driver for vxWorks; especially one for your target CPU. I agree consultants can be gregarious in their pursuit of cash. However, they have to be sales people as well as programmers. It is a pretty tough job and some of them do give advice for free. Many write magazine articles that you might read. It is unlikely that anyone here has code the specific CPU and ethernet chipset that the OP wants and even less likely that they will give away the source or a skeleton to one. There are lots of example drivers for vxWorks and the "Network Protocol Toolkit User's Guide" should tell you exactly what you need to know. The consultant was a bit of an ignoramous, but to say that no one on c.o.v takes any time to help people is crap. Generally if you can't ask a very specific question in regards to vxWorks, you are being *_LAZY_* and you shouldn't get a response. regards, Bill Pringlemeir. - -- Many people will ask you for advice about washing clothes. Minutes from now cleaning the fridge may seem like a good idea. You will discover that you have two left feet. Loving your child now will turn out to be unhealthy after a few hours. You'll never be shrunk. vxWorks FAQ, "http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html" --------------------------- Newsgroups: comp.os.vxworks Subject: Re: why coulud not my ARM7TDMI and vxworks system support floating point? Date: Mon, 14 Jan 2002 12:55:14 -0800 From: Lori Fraleigh Organization: Real-Time Innovations Message-ID: <3C4345B2.A0EAA44@rti.com> References: <15decc28.0201140617.399b988@posting.google.com> You've defined INCLUDE_FLOATING_POINT but that just includes the ability to do floating point i/o (for the printf to be able to print). Normally one would define INCLUDE_SW_FP to get the software floating point libraries in the kernel. However, the implementation of this define is architecture and BSP dependent. Your BSP may or may not support this. Did you have any undefined symbols when downloading your module? I have heard of certain ARM BSPs being unable to do basic floating point math in the past from other people. However, I do not know how they resolved it. Regards, Lori davis wrote: > > Hello, everyone! Please help me about floating point support on ARM7TDMI. > Now my develop enviroment is: > CPU: ARM7TDMI(S3C4510B), > RAM: SDRAM 16MBytes, > OS: Tornado 2.0 + vxWorks 5.4 > Host OS: windows98. > But this system couldn't support floating point! > The fllowing is a sample about floating point operation: > > #include "vxworks.h" > #include "stdio.h" > #include "math.h" > > float ft() > { > float value; > float temp; > > value = 2.0; > > temp = value * 23.3344; > printf("value: %2.1f\n", value); > printf("temp: %2.1f\n", temp); > return temp; > } > > those fllowing is the result after downloading the object: > > -> ft > > Exception number 16: Task: 0xd6e3c4 (t5) > > Data abort > Program Counter: 0x9880c > Status Register: 0x80000013 > 91e8 _vxTaskEntry +28 : 3936c () > 39410 _wdbFuncCallLibInit+128: _ft () > 1a5fb4 _ft +1c : ___extendsfdf2 () > 9960c ___extendsfdf2 +1c : 987f4 () > value = 0 = 0x0 > > Also, my project have those configuration: > configAll.h > #define INCLUDE_FLOATING_POINT /* floating point I/O */ > > usrConfig.c: > #ifdef INCLUDE_FLOATING_POINT > floatInit (); /*initialize floating point I/O */ > #endif /* INCLUDE_FLOATING_POINT */ > > Now, how can I use floating point operation in my target board?? > Thanks for your help! > > hurunmin@sina.com - -- Lori Fraleigh Director of Engineering Real-Time Innovations, Inc. http://www.rti.com Phone: 408-734-4200 x121 Fax: 408-734-5009 --------------------------- Newsgroups: comp.os.vxworks Subject: Very important, I need the Login/password of my Hi-Focus ADSL modem (ECI) Date: Mon, 14 Jan 2002 22:13:23 +0100 From: "Mirak" Organization: Guest of ProXad - France Message-ID: <3c4348e3$0$203$626a54ce@news.free.fr> Who knows how to get it ? I can't configurate the modem without this log/pass. Bye. --------------------------- Newsgroups: comp.os.vxworks Subject: what program do i need to use to exploit the udp port 17185 of a system working under vxworks ? Date: Mon, 14 Jan 2002 22:27:09 +0100 From: "Mirak" Organization: Guest of ProXad - France Message-ID: <3c434c15$0$194$626a54ce@news.free.fr> Hello. What program do i need to use to exploit the udp port 17185 of a system working under vxworks ? This a dsl modem , what program should i use to use this port ? I have eard that it is a RPC port of the vxworks, but do you have or know programs that can communicate with this port ? Bye. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: what program do i need to use to exploit the udp port 17185 of a system working under vxworks ? Date: 14 Jan 2002 17:16:52 -0500 From: Bill Pringlemeir Organization: Bell Sympatico Message-ID: References: <3c434c15$0$194$626a54ce@news.free.fr> Sender: bpringlemeir@DeadDuck >>>>> "Mirak" == Mirak writes: Mirak> Hello. What program do i need to use to exploit the udp port Mirak> 17185 of a system working under vxworks ? This a dsl modem , Mirak> what program should i use to use this port ? I have eard that Mirak> it is a RPC port of the vxworks, but do you have or know Mirak> programs that can communicate with this port ? Bye. They are freaking morones if they left WDB port active. The WDB port is a Gopher type protocol. You can google `WBD' to find out about this port. They should have conditionally enabled the WDB port. You would need a image with symbols in order to connect the vxWorks tools to this port. However, you can probably hack something up to examine (and modify) memory. As another suggestion, versions of GDB have been released with WBD backends. This back end interfaces to the WDB port. You basically have total access to the machine and all of its resources if you have access to this port... I hope this isn't an Acatel Speedtouch? [Please tell me if it is the speed touch, I would like to look at it if it is]. Maybe ECI/Innovia? I guess vxWorks is popular due to the PPPoE implementations. hth, Bill PRinglemeir. - -- BLOWN UP BY THREE BROKEN IRC SERVERS. SINKING. U-590. vxWorks FAQ, "http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html" --------------------------- Newsgroups: comp.os.vxworks Subject: Re: pci error in IDT Rc32332 Date: 14 Jan 2002 17:18:02 -0800 From: tonyzhao_2000@yahoo.com (tony zhao) Organization: http://groups.google.com/ Message-ID: <636ad94e.0201141718.24d9605@posting.google.com> References: <4ceedd43.0112282323.368465f8@posting.google.com> <4ceedd43.0201050420.7ee19f02@posting.google.com> <4ceedd43.0201100504.2690427a@posting.google.com> you can keep -O2 option on if you declare the register variable as _volatile_. sczhang@esteemnetworks.com (ShiChao Zhang) wrote in message news:<4ceedd43.0201100504.2690427a@posting.google.com>... > Johan Borkhuis wrote in message news:... > > sczhang@esteemnetworks.com (ShiChao Zhang) wrote: > > > > > I have tried it, but has no effect. And if I disable the exception, > > > all the value I got is the address of the PCI configuration Address > > > Register:0x18002cfc. > > > > Did you try the command "pciDeviceShow" and/or "pciHeaderShow"? > > > > Groeten, > > Johan > > Oh, I solved this problem. The reason is using the -O2 compiler > option, which reorder my c code(convert a write,read operation to > read,write operation). > Now the pciDeviceShow is OK! > > Thanks! --------------------------- Newsgroups: comp.os.vxworks Subject: Building GCC 2.95.x for vxWorks (PPC), Solaris host Date: Tue, 15 Jan 2002 01:22:54 GMT From: sj@shell1.aracnet.com (Scott Johnson) Organization: aracnet.com -- Portland's loudest electrons (info@aracnet.com) Message-ID: Hello, Has anyone managed to build GCC 2.95.x for vxWorks (powerpc 860 architecture) successfully on Solaris? I tried Dave Korn's website; but he has only binaries for Windoze, and the directions are windows-centric. (I have sent Dave an email; but am also interested in a wider response. Dave, if you HAVE responded already; I apologize--I might have accidentally tossed it into the bit bucket as SPAM; which I seem to get a lot of.... Everything seems to work fine up until it tries to build libgcc1-test.o. The linker then hangs (except for one instance where it segfaults and dumps core). Our system (I don't have root, so my ability to affect changes to the environemnt is limited): Target is a MPC860 design running vxWorks. Host platform consists of UltraSPARCs running Solaris 2.5.1 Host compiler is GCC 2.8.1 (can this compile a cross GCC 2.95, or do I need to have GCC 2.95 for Solaris running before I can compile a version for vxWorksB) Host binutils is 2.11.2 (the only one which works well with Solaris, so I am told...) Target binutils is 2.10, as gotten from Dave's site. In addition, I had to download lots of other packages not mentioned in the docs (Autoconf, flex) to even get it to compile that far. Finally, when fixinc runs; I get all sorts of error messages; usually caused by fixinc trying to find MIPS-related header files, which are not installed at my site. Can one pass preprocessor directives (ie - -DCPU=PPC860) to fixinc to make this junk go away--or does it even matter? Finally---if anyone HAS successfully gotten this to work on Solaris; does anyone have binaries they would like to share??? Tarballs would be preferred. Thanks, e_s - -- - ------------------------------------------------------------------------ engineer_scotty (no, not that one) -- consumer of fine ales and lagers some days you're the fire hydrant, some days you're the dog | go blazers no small furry creatures were harmed in the creation of this .signature --------------------------- Newsgroups: comp.os.vxworks Subject: taskdelay and interrupts Date: 14 Jan 2002 17:46:56 -0800 From: karahanduman@hotmail.com (Kara) Organization: http://groups.google.com/ Message-ID: Greetings, I am using 5.4 and tornado 2.1... I have a very simple application that is ================================================================= #include #include #include #include #include "idt79s334a.h" void intHandler(int arg); void readLink(int offset, long *data); void writeLink(int offset, long data); void Enter(void) { /* vars */ unsigned long *pReg; unsigned long *pBase=(unsigned long*)0xA4800000; unsigned long reg; printf("Initializing.. \n"); InitHW(); /* Do Stuff that inits HW */ taskDelay(10); intConnect(INUM_TO_IVEC(SCC_INT_NUM),(VOIDFUNCPTR)intHandler, 0); intEnable(SCC_INT_LVL); while(1) { taskDelay(100000); } } void intHandler(int arg) { unsigned long value; unsigned long *pBase=(unsigned long*)0xA4800000; unsigned long linkReg; printf("Interrupt occurred\n"); /* Cleanup */ /* Get the cause of the interrupt */ readLink(0, &value); /* Reset the Interrupt */ writeLink(0, 0); HandleMyISr(value); return; } void readLink(int offset, long *data) { unsigned long regValue; intKey = intLock(); /* Read Rgister */ (void)intUnlock(intKey); *data = regValue; } void writeLink(int offset, long data) { intKey = intLock(); /* Write Register */ (void)intUnlock(intKey); } ================================================= If I comment out the taskDelay in the main loop and run it, I get the printf messages as caused by interrupts without any problem. When I run it with task delay(i.e. taskDelay(100000);), I don't get the printf messages. if I run the taskDelay with some small value (i.e. 100), I get the messages with a little delay. If I take out the while(1); loop and let it return to the shell, I don't get interrupts at all. This is unbelievable.. Any help will be greatly appreciated. Thanks, Kara.. --------------------------- Newsgroups: comp.os.vxworks Subject: Hi All Date: 14 Jan 2002 20:55:27 -0800 From: aravinddk@tataelxsi.co.in (aravind) Organization: http://groups.google.com/ Message-ID: Hi, We are trying to capture the ppc keyboard key presses. We tried the Event function with which we could not get control into the Event function at all. So we used a UGL function as below, :: Func { while(1) { status = uglInputQStandardQEventGet(10000, &Event); if(ESC == inputEvent.event.keyboard.scanCode) { MyEvent.type = E_KEY; MyEvent.rawCode = U_KEY_ESC; Event(MyEvent); // calls the Event function below break; } } } ZafEventType ClassABC::Event(const ZafEventStruct &event) { ZafEventType ccode = LogicalEvent(event); //if the event is a keyBoard Event if(event.InputType() == E_KEY) { switch (event.rawCode) { // user defined action } } } But, when I use this function, sometimes a key has to be pressed twice in order to be trapped. This we think is happening since there is another GUI which is already running and which has called the Control () of the Zinc Event function. So the first key press is routed to the Control () loop and the second key press is routed to the uglInputQStandardQEventGet(). Is there a soultion to this problem ? OR How to make the existing Control () loop (in another application) to recognise the key presses in our application. The two are linked during compilation. Thanks in advance Regards Arvind d.K --------------------------- Newsgroups: comp.os.vxworks Subject: WRS Announces Transfer of FreeBSD Sponsorship Date: 14 Jan 2002 23:47:42 -0500 From: bpringlemeir@yahoo.com Organization: Bell Sympatico Message-ID: Sender: bpringle@DeadDuck (Applies to: WIND YHOO SNE) Wind River Announces Transfer of FreeBSD Sponsorship to FreeBSD Mall, Inc.; Sponsorship of Open Source Effort Returns to Roots ALAMEDA, Calif., Jan 14, 2002 (BUSINESS WIRE) -- Wind River Systems, Inc. (Nasdaq:WIND), a leading provider of software and services for connected devices, today announced a definitive agreement to transfer its FreeBSD operations to FreeBSD Mall, Inc. FreeBSD is an advanced open source UNIX operating system, derived from BSD UNIX and developed at the University of California. Well known for its performance and reliability, FreeBSD technology is widely used as a server operating system by many large Internet sites including Yahoo!, Hotmail, Sony Japan, Apache, Pair Networks and Whistle Communications. Like Linux, the source code for FreeBSD is freely available. Wind River assumed stewardship for the FreeBSD open source project in May 2001 when it acquired assets of Berkeley Software Design, Inc. (BSDi), the former sponsor of FreeBSD. In addition to the FreeBSD sponsorship, Wind River acquired the proprietary BSD UNIX-based OS (BSD/OSO). Wind River's interest in the BSDi assets continues to focus on BSD/OS. Divesting the FreeBSD business further sharpens that focus and provides continuity and increased support for FreeBSD. FreeBSD Mall Inc. is led by its founder, Bob Bruce. Bruce's involvement with FreeBSD dates back to 1993 when his former company, Walnut Creek CDROM, was the first and primary distributor of FreeBSD. "The FreeBSD community will be well served by this transaction," said Larry Macfarlane, senior director of Wind River's Application Platforms product division. "When we decided to divest the FreeBSD business, we looked for a successor organization that could meet high standards of customer service and maintain a mutually beneficial relationship with the FreeBSD community. After carefully evaluating many interested organizations, we decided that the best way to ensure the continuity and vitality of FreeBSD was to return it to its roots." Bob Bruce enthusiastically welcomes FreeBSD back home, and commented, "As we go forward, we will be able to build on strong relationships and friendships. We have all worked together before. FreeBSD has a promising future, and I am committed to helping it reach its full potential." Jordan Hubbard, co-founder of the open source FreeBSD Project, also endorses this change. "I'm happy to see that the FreeBSD Mall will be continuing without interruption," said Hubbard. "Through this transfer, FreeBSD will be back under the direction of the same people who started and know the FreeBSD CD product line perhaps better than anyone." Hubbard is an engineering manager at Apple, where he helps develop the highly acclaimed Mac OS X, and works on the open source Darwin Project. Both Mac OS X and Darwin are based on FreeBSD. FreeBSD Mall plans to aggressively promote and market FreeBSD. In addition to providing the standard FreeBSD distribution on CDROM, they will continue to offer a subscription service, snapshots of the current development branch, and published hardcopy editions of the FreeBSD Handbook. They will also offer several levels of professional support and services for FreeBSD. FreeBSD Mall has placed FreeBSD products in many mainstream retail stores, including Best Buy, CompUSA, Borders Books, Barnes & Noble and Amazon.com. In the near future, they plan to expand their retail presence, especially in Europe and Asia. Current support obligations and software subscriptions at Wind River will be transferred to FreeBSD Mall in this transaction. All current Wind River employees working with FreeBSD will be employed by FreeBSD Mall. Wind River will continue to develop and support BSD/OS, a professionally engineered and supported BSD operating system widely used in embedded systems. A major upgrade of BSD/OS that will focus on meeting the needs of networked embedded devices is scheduled for release later this year. Wind River will also continue to support the BSD open source community, and plans to cooperate with the FreeBSD Mall and other BSD organizations to help promote wide adoption of BSD technology. The transaction is expected to close at the end of January. Financial details are not being disclosed at this time. About Wind River Wind River is a worldwide leader in integrated embedded software solutions for creating reliable and innovative connected devices. Wind River provides development tools, real-time operating systems, and advanced connectivity software for use in products in carrier and enterprise networking, consumer electronics, automotive, industrial measurement and control, and aerospace/defense markets. Wind River is How Smart Things Think(TM). Founded in 1983, Wind River is headquartered in Alameda, California, with operations worldwide. About FreeBSD Mall FreeBSD Mall Inc. is a publisher and distributor of FreeBSD software, and a provider of FreeBSD services and support. Founded in 1991, FreeBSD Mall is located in Concord, California. Note to Editors: Wind River Systems and the Wind River Systems logo are registered trademarks; BSD/OS and How Smart Things Think are trademarks of Wind River Systems, Inc. All other names mentioned are trademarks, registered trademarks or service marks of their respective companies. CONTACT: Wind River Jim Barillo, 510/749-2284 jim.barillo@windriver.com www.windriver.com URL: http://www.businesswire.com Today's News On The Net - Business Wire's full file on the Internet with Hyperlinks to your home page. Copyright (C) 2002 Business Wire. All rights reserved. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: why coulud not my ARM7TDMI and vxworks system support floating point? Date: 14 Jan 2002 21:23:46 -0800 From: hurunmin@sina.com (davis) Organization: http://groups.google.com/ Message-ID: <15decc28.0201142123.22f5f5a8@posting.google.com> References: <15decc28.0201140617.399b988@posting.google.com> <3C4345B2.A0EAA44@rti.com> Thanks for all posted message! > Normally one would define INCLUDE_SW_FP to get the software > floating point libraries in the kernel. However, the > implementation of this define is architecture and BSP > dependent. Your BSP may or may not support this. Now, I try to include the INCLUDE_SW_FP macro in config.h. But when I complied the vxWorks image there are those errors and the complier stopped. vxWorks.tmp(.text+0x7e80):fake: undefined reference to `mathSoftInit' MAKE.EXE: *** [vxWorks] Error 0x1 mathSoftInit() is called by in usrConfig.c file. The fllowing is the source code: #ifdef INCLUDE_SW_FP mathSoftInit (); /* use software emulation for fp math */ #endif /* INCLUDE_SW_FP */ Does this means that my vxworks or GCC complier don't support floating point? is this true and how can I do in order to support floating point? --------------------------- Newsgroups: comp.os.vxworks Subject: Re: change usrconfig.c , but not works Date: Tue, 15 Jan 2002 08:43:06 +0100 From: "Michael Lawnick" Organization: Buergernetz Dillingen Message-ID: References: <4iH08.15119$iM.14260@news1.bloor.is> Reply-To: "Michael Lawnick" Sender: mlawnick@pd9524416.dip.t-dialin.net Hi Frank, usrConfig.c is 'out of duty' when compiling from workspace under T2. All work of it is done within the workspace and its derived file prjConfig.c. - -- Mit freundlichen Grüßen, Michael Lawnick ============================================== SOFTEC GmbH Tel +49-731-96600-0 Promenade 17 Fax +49-731-96600-23 D-89073 Ulm Michael Lawnick Germany lawnick@softec.de ============================================== "Frank" schrieb im Newsbeitrag news:4iH08.15119$iM.14260@news1.bloor.is... > hi, > I met a "simple" problem, but made me mad. > When I changed " target/config/all/usrconfig.c", but seems it make no sense > when I re-build all my project. > I doubt the "usrconfig.c" never be re-compile when I re-build my project. > I use Tornado 2 and VxWorks 5.4 for 860. > Pls give me some advice. > Thank you > > Frank > -- > > > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: taskdelay and interrupts Date: 15 Jan 2002 09:15:53 +0100 From: Reinhold Wirth Organization: WEB.DE Message-ID: <3c43e539$1@netnews.web.de> References: Hi Kara, you use printf() in a interrupt service routine. USE IN INTERRUPT SERVICE ROUTINES Because logMsg( ) does not directly cause output to I/O devices, but instead simply writes to a message queue, it can be called from an interrupt service routine as well as from tasks. Normal I/O, such as printf( ) output to a serial port, cannot be done from an interrupt service routine. karahanduman@hotmail.com (Kara) wrote: >Greetings, >I am using 5.4 and tornado 2.1... > >I have a very simple application that is > >================================================================= > >#include >#include >#include >#include >#include "idt79s334a.h" > > >void intHandler(int arg); > >void readLink(int offset, long *data); >void writeLink(int offset, long data); > > >void Enter(void) >{ > /* vars */ > unsigned long *pReg; > unsigned long *pBase=(unsigned long*)0xA4800000; > unsigned long reg; > > printf("Initializing.. \n"); > InitHW(); /* Do Stuff that inits HW */ > taskDelay(10); > > intConnect(INUM_TO_IVEC(SCC_INT_NUM),(VOIDFUNCPTR)intHandler, 0); > intEnable(SCC_INT_LVL); > > > while(1) > { > taskDelay(100000); > } > >} > >void intHandler(int arg) >{ > unsigned long value; > unsigned long *pBase=(unsigned long*)0xA4800000; > unsigned long linkReg; > > printf("Interrupt occurred\n"); > /* Cleanup */ > /* Get the cause of the interrupt */ > readLink(0, &value); > > /* Reset the Interrupt */ > writeLink(0, 0); > HandleMyISr(value); > return; >} >void readLink(int offset, long *data) >{ >unsigned long regValue; > >intKey = intLock(); >/* Read Rgister */ >(void)intUnlock(intKey); > >*data = regValue; >} > >void writeLink(int offset, long data) >{ > intKey = intLock(); > /* Write Register */ > (void)intUnlock(intKey); >} > > > >================================================= >If I comment out the taskDelay in the main loop and run it, I get the >printf messages as caused by interrupts without any problem. When I >run it with task delay(i.e. taskDelay(100000);), I don't get the >printf messages. if I run the taskDelay with some small value (i.e. >100), I get the messages with a little delay. >If I take out the while(1); loop and let it return to the shell, I >don't get interrupts at all. > > >This is unbelievable.. Any help will be greatly appreciated. >Thanks, >Kara.. - -- __________________________________________________________ News suchen, lesen, schreiben mit http://newsgroups.web.de --------------------------- Newsgroups: comp.os.vxworks Subject: Re: change usrconfig.c , but not works Date: 15 Jan 2002 08:23:10 GMT From: Johan Borkhuis Organization: Agere Systems Message-ID: References: <4iH08.15119$iM.14260@news1.bloor.is> "Frank" wrote: > hi, > I met a "simple" problem, but made me mad. > When I changed " target/config/all/usrconfig.c", but seems it make no > sense when I re-build all my project. > I doubt the "usrconfig.c" never be re-compile when I re-build my > project. I use Tornado 2 and VxWorks 5.4 for 860. > Pls give me some advice. > Thank you You are completely right: the file usrConfig.c is not compliled when using the project facility in Tornado 2. This mechanism is used in Tornado 1. Tornado 2 uses so-called configlettes. When you look in the direcory target/config/comps/src you will find al lot of source files for different components. These files are compiled using the file prjConfig.c. This file is generated when you create or change the properties of your project. Groeten, Johan - -- o o o o o o o . . . _____________________________ o _____ || Johan Borkhuis | .][__n_n_|DD[ ====_____ | borkhuis@agere.com | >(________|__|_[_________]_|__________________________| _/oo OOOOO oo` ooo ooo 'o!o!o o!o!o` === VxWorks FAQ: http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html === --------------------------- Newsgroups: comp.os.vxworks Subject: debuging fads860 interrupt problems "uninitialized interrupt" Date: 15 Jan 2002 00:38:43 -0800 From: yong_bvcom@sina.com (bigtornado) Organization: http://groups.google.com/ Message-ID: <5624c1bf.0201150038.74cbea67@posting.google.com> hi,everybody: I use fads860 board,and now I want add an device,then using MPC860 external interrupt IRQ1, I connect my ISR routine use "intConnect(INUM_TO_IVEC(2),myIsr,0)",after this i enable this interrupt use "intEnable(2)",but when interrupt occurs,system give me information "uninitialized interrupt",I debug my interrupt drive routine "ppc860ctrl.c",all seems right,why about this? Thanks for anyone's suggestions very much! Jeffy 1.15 --------------------------- Newsgroups: comp.os.vxworks Subject: Re: taskdelay and interrupts Date: 15 Jan 2002 08:42:21 GMT From: Johan Borkhuis Organization: Agere Systems Message-ID: References: karahanduman@hotmail.com (Kara) wrote: > If I comment out the taskDelay in the main loop and run it, I get the > printf messages as caused by interrupts without any problem. When I > run it with task delay(i.e. taskDelay(100000);), I don't get the > printf messages. if I run the taskDelay with some small value (i.e. > 100), I get the messages with a little delay. > If I take out the while(1); loop and let it return to the shell, I > don't get interrupts at all. NEVER use printf statements in an interrupt routine! The fact that you were able to get output at all surprised me. There are more routines that you cannot call from within an interrupt routine. Take a look at chapter 2.5 in the Programmer's Guide (Interrupt Service Code), and espcially 2.5.3 (Special Limitations of ISRs) If you want to create output from within an ISR use logMsg instead of printf. Groeten, Johan - -- o o o o o o o . . . _____________________________ o _____ || Johan Borkhuis | .][__n_n_|DD[ ====_____ | borkhuis@agere.com | >(________|__|_[_________]_|__________________________| _/oo OOOOO oo` ooo ooo 'o!o!o o!o!o` === VxWorks FAQ: http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html === --------------------------- Newsgroups: comp.os.vxworks Subject: lock timeout using shared memoy Date: 15 Jan 2002 00:49:32 -0800 From: ruchama@septier.com (Ruchama Goldshmidt) Organization: http://groups.google.com/ Message-ID: <5df340e0.0201150049.7399ef82@posting.google.com> hi vxWorkers!! I am using 6 PPC8260 boards which sent to an Mcp750 board messages over a UDP socket , many times I get a lot of missed packets (especlially when enable the cache in the ppc8260 boards) I have an idea to write another retry alghorithm instead of the vxWorks' algorithm (in the "smLib.c" file because the original algorithm has a busy wait that is no appropriate to running with a cache enable. does anyone has an idea of another alghorithm?? any idea should help!! TIA Ruchama --------------------------- Newsgroups: comp.os.vxworks Subject: Re: BootRom and MMU Date: Tue, 15 Jan 2002 09:35:21 +0100 From: Sebastien Fauris Organization: Nortel Networks Message-ID: <3C43E9C9.8630FBE3@europem01.nt.com> References: <0y%_7.14882$8e.753501@news> Reply-To: sfauris@europem01.nt.com Hi vxWorkers, what I find strange is that in a bootrom the PPC processor runs in real mode and so the MMU is disabled. The MMU will be enabled only in the kernel downloaded afterwards. Sebastien. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: what program do i need to use to exploit the udp port 17185 of a system working under vxworks ? Date: Tue, 15 Jan 2002 09:59:08 +0100 From: "Mirak" Organization: Guest of ProXad - France Message-ID: <3c43ee3d$0$201$626a54ce@news.free.fr> References: <3c434c15$0$194$626a54ce@news.free.fr> Yes it a ECI/innovia B-Focus. Other ports (8080 for me) on this modem are open only when the modem is booted desynchronised. I am interested in modifiny the password and login of the modem, myself, to be sure tat nobody else will do it before me, like for the alcatel speed touch. If you have any information, on this or the procedure to fo that i am interested. Bye. "Bill Pringlemeir" wrote in message news:uelkszigb.fsf@yahoo.com... > >>>>> "Mirak" == Mirak writes: > > Mirak> Hello. What program do i need to use to exploit the udp port > Mirak> 17185 of a system working under vxworks ? This a dsl modem , > Mirak> what program should i use to use this port ? I have eard that > Mirak> it is a RPC port of the vxworks, but do you have or know > Mirak> programs that can communicate with this port ? Bye. > > They are freaking morones if they left WDB port active. The WDB port > is a Gopher type protocol. You can google `WBD' to find out about > this port. They should have conditionally enabled the WDB port. You > would need a image with symbols in order to connect the vxWorks tools > to this port. However, you can probably hack something up to examine > (and modify) memory. As another suggestion, versions of GDB have been > released with WBD backends. This back end interfaces to the WDB port. > You basically have total access to the machine and all of its > resources if you have access to this port... I hope this isn't an > Acatel Speedtouch? [Please tell me if it is the speed touch, I would > like to look at it if it is]. Maybe ECI/Innovia? I guess vxWorks is > popular due to the PPPoE implementations. > > hth, > Bill PRinglemeir. > > -- > BLOWN UP BY THREE BROKEN IRC SERVERS. SINKING. U-590. > vxWorks FAQ, "http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html" --------------------------- Newsgroups: comp.os.vxworks Subject: Waiting on multiple events Date: 15 Jan 2002 01:58:13 -0800 From: niall.leonard@ncr.com (Niall Leonard) Organization: http://groups.google.com/ Message-ID: <14e76c83.0201150158.696e793f@posting.google.com> I would like to cause a task to pend on a group of semaphores. I would like the task to be made ready when the first semaphore is given. There is a similar function available in the Win32 API called waitForMultipleObjects() (or something like that..). It is given an array of "objects" that a task can pend on, and returns when the first of these objects is available. Anyone got any ideas on how to do this in VxWorks ? Thanks, Niall Leonard Consultant - Real Time Software, Platform Technology Development, NCR Financial Solutions Group Ltd, Discovery Centre, 3 Fulton Road, Dundee, DD2 4SW Tel: +44 (0)1382 592866 (Direct) e-mail: niall.leonard@ncr.com --------------------------- Newsgroups: comp.os.vxworks Subject: Re: change usrconfig.c , but not works Date: Tue, 15 Jan 2002 12:46:16 +0200 From: "Ofer Goren" Organization: Verio Message-ID: References: <4iH08.15119$iM.14260@news1.bloor.is> Hi. The way I understand it, usrConfig.c is taken into consideration only when you first create you project. After that, you might need to change prjConfig.c for changes to take effect. However, I'm not sure it will work either. You have two options: 1. the bad solution: try to change to project file itself 2. if this is something of macro or parameter changing, try to find it using the GUI, and change it from there. Not like as Tornado 1.0.1, the vxWorks expects that the changes to the OS will be done only after usrAppInit(). HTH. - -- Ofer Goren Iamba Technologies "Frank" wrote in message news:4iH08.15119$iM.14260@news1.bloor.is... > hi, > I met a "simple" problem, but made me mad. > When I changed " target/config/all/usrconfig.c", but seems it make no sense > when I re-build all my project. > I doubt the "usrconfig.c" never be re-compile when I re-build my project. > I use Tornado 2 and VxWorks 5.4 for 860. > Pls give me some advice. > Thank you > > Frank > -- > > > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: why coulud not my ARM7TDMI and vxworks system support floating point? Date: Tue, 15 Jan 2002 11:37:30 +0000 (UTC) From: David Laight Organization: BT Openworld Message-ID: <3C4414B7.6C0C2556@l8s.co.uk> References: <15decc28.0201140617.399b988@posting.google.com> <3C4345B2.A0EAA44@rti.com> <15decc28.0201142123.22f5f5a8@posting.google.com> d > Now, I try to include the INCLUDE_SW_FP macro in config.h. But when > I complied the vxWorks image there are those errors and the complier > stopped. > > vxWorks.tmp(.text+0x7e80):fake: undefined reference to `mathSoftInit' > MAKE.EXE: *** [vxWorks] Error 0x1 > > mathSoftInit() is called by in usrConfig.c file. The fllowing is > the source code: > > #ifdef INCLUDE_SW_FP > mathSoftInit (); /* use software emulation for fp math */ > #endif /* INCLUDE_SW_FP */ > > Does this means that my vxworks or GCC complier don't support > floating point? > is this true and how can I do in order to support floating point? You don't need mathSoftInt(), the arm cpus dont have an FPU (ok one did...) gcc always generates subroutine calls - indeed your program is making one of them (the correct one). As I said before, check that the host tools are using the correct symbol table. FYI the fp routines WRS supply for arm systems suck! They are too slow... David --------------------------- End of New-News digest ********************** From vxwexplo-errs@csg.lbl.gov Wed Jan 16 04:03:25 2002 From: Vxworks Exploder Date: Wed Jan 16 04:03:28 PST 2002 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Wed Jan 16 04:03:20 PST 2002 Subject: Re: Waiting on multiple events Subject: help on telnet problem Subject: Re: what program do i need to use to exploit the udp port 17185 of a system working under vxworks ? Subject: How can I update the file system date Subject: Testc Only Subject: Looking for COFF to ELF utility for Windows 2000 Subject: Re: taskdelay and interrupts Subject: Tornado 2.1 For MIPS seems to obsolete existing BSP. Subject: Re: why coulud not my ARM7TDMI and vxworks system support floating point? Subject: Re: Hi All Subject: Re: taskdelay and interrupts Subject: Re: LC82C169 Driver... Subject: Re: Very important, I need the Login/password of my Hi-Focus ADSL modem (ECI) Subject: Re: help on telnet problem Subject: Re: How can I update the file system date Subject: Re: taskdelay and interrupts Subject: Re: change usrconfig.c , but not works Subject: Re: debuging fads860 interrupt problems "uninitialized interrupt" Subject: Re: ARM coff2elf conversion Subject: Re: Waiting on multiple events Subject: Re: Building GCC 2.95.x for vxWorks (PPC), Solaris host Subject: Re: change usrconfig.c , but not works Subject: Re: BootRom and MMU Subject: Serial Port Configuration Subject: Re: change usrconfig.c , but not works Subject: Re: taskdelay and interrupts Subject: Tornado 2.1 for PPC860 Subject: Re: Tornado 2.1 for PPC860 Subject: PMC module on MVME5100 (mv5100) Subject: Unable to connect to 2nd target through 1st target using Target Server Subject: Tornado 2.1 for PPC860 Subject: Re: BootRom and MMU Subject: Re: Tornado 2.1 For MIPS seems to obsolete existing BSP. Subject: Re: debuging fads860 interrupt problems "uninitialized interrupt" Subject: Re: Waiting on multiple events Subject: Re: ipAttach problem ?? ( two end driver) Subject: How to connect a routine to IRQ1?(MPC860T on FADS BOARD) Subject: Re: Serial Port Configuration Subject: /var/tmp/....errors while compling assembly routines Subject: Re: Building GCC 2.95.x for vxWorks (PPC), Solaris host Subject: CRC and Alignment Errors. Subject: Re: How to connect a routine to IRQ1?(MPC860T on FADS BOARD) Subject: serial i/o Subject: Re: ARM coff2elf conversion Subject: Re: change usrconfig.c , but not works Subject: Re: why coulud not my ARM7TDMI and vxworks system support floating Subject: Re: BootRom and MMU Subject: BSP PCpentium & boot on floppy & Tornado2 Subject: a WPJ forPCpentium BSP with a floppy boot Subject: Re: serial i/o Subject: CRC and Alignment Errors. Subject: Need Help in locating a retailor for purchasing PMC plugs Subject: Re: serial i/o ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: Waiting on multiple events Date: 15 Jan 2002 12:25:37 GMT From: Johan Borkhuis Organization: Agere Systems Message-ID: References: <14e76c83.0201150158.696e793f@posting.google.com> niall.leonard@ncr.com (Niall Leonard) wrote: > I would like to cause a task to pend on a group of semaphores. I would > like the task to be made ready when the first semaphore is given. > There is a similar function available in the Win32 API called > waitForMultipleObjects() (or something like that..). It is given an > array of "objects" that a task can pend on, and returns when the first > of these objects is available. There are a number of ways to do this. Instead of waiting for multiple events with one call you can create a loop that waits for all events. You can do this inline, or create a function that does this for you. Or you could even create a task that waits for the signals and releases a semaphore when all signals are received. Groeten, Johan - -- o o o o o o o . . . _____________________________ o _____ || Johan Borkhuis | .][__n_n_|DD[ ====_____ | borkhuis@agere.com | >(________|__|_[_________]_|__________________________| _/oo OOOOO oo` ooo ooo 'o!o!o o!o!o` === VxWorks FAQ: http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html === --------------------------- Newsgroups: comp.os.vxworks Subject: help on telnet problem Date: 15 Jan 2002 04:34:45 -0800 From: kevinatnj@yahoo.com.cn (kewin) Organization: http://groups.google.com/ Message-ID: <221caa76.0201150434.e1e7cf2@posting.google.com> hi all I have added the macro ,INCLUDE_TELNET,in the config.h. And make a vxWorks in the command line enviroment.After booting my target board,i can telnet it. If i make the vxWorks file in project enviroment, i can telnet the target board .Anybody can tell me that there is any special configuration comand line enviroment when making vxworks image file. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: what program do i need to use to exploit the udp port 17185 of a system working under vxworks ? Date: Tue, 15 Jan 2002 15:15:09 +0100 From: "Mirak" Organization: Guest of ProXad - France Message-ID: <3c4438da$0$3191$626a54ce@news.free.fr> References: <3c434c15$0$194$626a54ce@news.free.fr> I have GDB , it's included in cygwin for windows. but there is only a TCP connection but not UDP does it matter ? What am i supposed to do with that ? bye. "Bill Pringlemeir" wrote in message news:uelkszigb.fsf@yahoo.com... > >>>>> "Mirak" == Mirak writes: > > Mirak> Hello. What program do i need to use to exploit the udp port > Mirak> 17185 of a system working under vxworks ? This a dsl modem , > Mirak> what program should i use to use this port ? I have eard that > Mirak> it is a RPC port of the vxworks, but do you have or know > Mirak> programs that can communicate with this port ? Bye. > > They are freaking morones if they left WDB port active. The WDB port > is a Gopher type protocol. You can google `WBD' to find out about > this port. They should have conditionally enabled the WDB port. You > would need a image with symbols in order to connect the vxWorks tools > to this port. However, you can probably hack something up to examine > (and modify) memory. As another suggestion, versions of GDB have been > released with WBD backends. This back end interfaces to the WDB port. > You basically have total access to the machine and all of its > resources if you have access to this port... I hope this isn't an > Acatel Speedtouch? [Please tell me if it is the speed touch, I would > like to look at it if it is]. Maybe ECI/Innovia? I guess vxWorks is > popular due to the PPPoE implementations. > > hth, > Bill PRinglemeir. > > -- > BLOWN UP BY THREE BROKEN IRC SERVERS. SINKING. U-590. > vxWorks FAQ, "http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html" --------------------------- Newsgroups: comp.os.vxworks Subject: How can I update the file system date Date: 15 Jan 2002 06:32:50 -0800 From: eiad.makled@att.net (Edward) Organization: http://groups.google.com/ Message-ID: <454b12eb.0201150632.3702a40a@posting.google.com> Could you please display me the code that allows me to update the file system date as it still reading an old date. Thanks Edward --------------------------- Newsgroups: comp.os.vxworks Subject: Testc Only Date: Tue, 15 Jan 2002 10:07:26 -0500 From: "Harjinder Chayra" Organization: Storm Internet Services Message-ID: Sender: terayon@gw-terayon.storm.ca My last posting never made it to the group so I am just checking if this one does. --------------------------- Newsgroups: comp.os.vxworks Subject: Looking for COFF to ELF utility for Windows 2000 Date: Tue, 15 Jan 2002 10:09:51 -0500 From: "Harjinder Chayra" Organization: Storm Internet Services Message-ID: Sender: terayon@gw-terayon.storm.ca I would appreciate it if someone could let me know where to get the utility. Thanks Harjinder --------------------------- Newsgroups: comp.os.vxworks Subject: Re: taskdelay and interrupts Date: 15 Jan 2002 07:25:25 -0800 From: mynamvs@nsc.network.com (Venkat Mynampati) Organization: http://groups.google.com/ Message-ID: <65b2abfc.0201150725.4f5f9903@posting.google.com> References: Johan Borkhuis wrote in message news:... > karahanduman@hotmail.com (Kara) wrote: > > > If I comment out the taskDelay in the main loop and run it, I get the > > printf messages as caused by interrupts without any problem. When I > > run it with task delay(i.e. taskDelay(100000);), I don't get the > > printf messages. if I run the taskDelay with some small value (i.e. > > 100), I get the messages with a little delay. > > If I take out the while(1); loop and let it return to the shell, I > > don't get interrupts at all. > > NEVER use printf statements in an interrupt routine! The fact that you > were able to get output at all surprised me. > > There are more routines that you cannot call from within an interrupt > routine. Take a look at chapter 2.5 in the Programmer's Guide (Interrupt > Service Code), and espcially 2.5.3 (Special Limitations of ISRs) > > If you want to create output from within an ISR use logMsg instead of > printf. > > Groeten, > Johan Hi, I am relatively new to vxWorks anyway i will stick my neck out... Fisrt thing according to doc's is: "printf() and any call that blocks on a resource cannot be used in an interrupt." However, what would happen if i call printf() in an ISR, as is in this case, and assuming there are no other tasks or interrupt vieing for CPU, then, i think printf() would work, 'cause, the ISR blocks on some resource only when that resource is unavailable and what if that resource is available? Or how can we explain the fact that printf() worked when taskDelay() is less? Venkat --------------------------- Newsgroups: comp.os.vxworks Subject: Tornado 2.1 For MIPS seems to obsolete existing BSP. Date: Tue, 15 Jan 2002 10:43:41 -0500 From: Doug Dotson Organization: Verio Message-ID: <3C444E2D.80ADE8B2@digidata.com> Greetings! Just received the upgrade to Tornado 2.1 For MIPS. In trying to rebuild our BSP I have noticed that the changing of macros seems to have made the BSP obsolete. Currently the macro CPU=R4000 in the Makefile selects the architecture, but in 2.1 there is no R4000 CPU. It has been changed to either MIPS32 or MIPS64. In our case, MIPS64 is appropriate for our R7000 processor. Changing the arcitecture to MIPS64 causes the MIPS64gnu toolchain to be properly selected, but then there is no R4000 macro defines to allow proper building of the BSP. Is there any easy workaround to this? Thanks, Doug --------------------------- Newsgroups: comp.os.vxworks Subject: Re: why coulud not my ARM7TDMI and vxworks system support floating point? Date: 15 Jan 2002 10:42:42 -0500 From: Bill Pringlemeir Organization: Bell Sympatico Message-ID: References: <15decc28.0201140617.399b988@posting.google.com> <3C4345B2.A0EAA44@rti.com> <15decc28.0201142123.22f5f5a8@posting.google.com> <3C4414B7.6C0C2556@l8s.co.uk> Sender: bpringlemeir@DeadDuck >>>>> "David" == David Laight writes: [snip] David> FYI the fp routines WRS supply for arm systems suck! They are David> too slow... They have been updated in T2.1.1. I guess you meant "the fp routines for ARM and thumb"? They seem to be particularly bad for the ARM. regards, Bill Pringlemeir. - -- Good languages are hefty. WHY? vxWorks FAQ, "http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html" --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Hi All Date: 15 Jan 2002 11:00:40 -0500 From: Bill Pringlemeir Organization: Bell Sympatico Message-ID: References: Sender: bpringlemeir@DeadDuck >>>>> "aravind" == aravind writes: aravind> Hi, We are trying to capture the ppc keyboard key aravind> presses. We tried the Event function with which we could not [snip] aravind> running and which has called the Control () of the Zinc aravind> Event function. So the first key press is routed to the aravind> Control () loop and the second key press is routed to the aravind> uglInputQStandardQEventGet(). aravind> Is there a soultion to this problem ? OR How to make the aravind> existing Control () loop (in another application) to aravind> recognise the key presses in our application. The two are aravind> linked during compilation. Much closer to giving enough information to solve this problem. However, I still don't understand why you wish to do the UGL key board processing. If you have Zinc (for vxWorks/WindML) then in the directory ../target/src/zinc/vxworks, there will be a file v_keybrd.cpp. This file is generally responsible for placing events on the Zinc device queue [see EventPoll() in v_osevnt.cpp and Get() in v_eventmgr.cpp]. The standard Zinc 6.0/WinML 2.0 [what are you using?] use the WindML windows facility, which use the callback mechanism to put events on the Zinc queue. You should be able to pend multiple task on the queue and each will get the key. Alternatively, I have hooked the keyboard driver to use a totally different mechanism. You just open "/dev/keyboard" and read from it (with select maybe) and post this to whatever queue you like. At any rate, WindML support multiple queues; Zinc does not. I still don't get what you are trying to do. You need to show a system diagram of some sort. There maybe a better design. hth, Bill Pringlemeir. - -- If I gave a damn, you'd be the first person I'd give it to. vxWorks FAQ, "http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html" --------------------------- Newsgroups: comp.os.vxworks Subject: Re: taskdelay and interrupts Date: 15 Jan 2002 15:58:31 GMT From: Johan Borkhuis Organization: Agere Systems Message-ID: References: <65b2abfc.0201150725.4f5f9903@posting.google.com> mynamvs@nsc.network.com (Venkat Mynampati) wrote: > Fisrt thing according to doc's is: "printf() and any call that blocks > on a resource cannot be used in an interrupt." > However, what would happen if i call printf() in an ISR, as is in this > case, and assuming there are no other tasks or interrupt vieing for > CPU, then, i think printf() would work, 'cause, the ISR blocks on some > resource only when that resource is unavailable and what if that > resource is available? printf always blocks on a resource: you are writing data to a serial port and this port has a certain buffer, probably a couple of chars. When you run out of buffer printf blocks. But there is another issue: your ISR should be as short as possible. If you are doing a lot of work inside an ISR this will block all other processes until your ISR is finished. And a printf will probably not be as important as your operational SW. Groeten, Johan - -- o o o o o o o . . . _____________________________ o _____ || Johan Borkhuis | .][__n_n_|DD[ ====_____ | borkhuis@agere.com | >(________|__|_[_________]_|__________________________| _/oo OOOOO oo` ooo ooo 'o!o!o o!o!o` === VxWorks FAQ: http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html === --------------------------- Newsgroups: comp.os.vxworks Subject: Re: LC82C169 Driver... Date: Tue, 15 Jan 2002 16:56:02 +0100 From: "Werner Schiendl" Message-ID: <3c445113$1@brateggebdc5.br-automation.co.at> References: <21595172.0201072023.67f7fc78@posting.google.com> <3C3AF6D0.9F54E13F@varndellengineering.com> <3C3EE8EA.A08361FE@varndellengineering.com> <3c42d37d@brateggebdc5.br-automation.co.at> <3C432938.49002A35@pacbell.net> "Bill Pringlemeir" wrote in message news:un0zgzmtz.fsf@yahoo.com... > ... > > The consultant was a bit of an ignoramous, but to say that no one on > c.o.v takes any time to help people is crap. Generally if you can't > ask a very specific question in regards to vxWorks, you are being > *_LAZY_* and you shouldn't get a response. > I agree with that, but I think they shouldn't get a response that keeps them from posting to or even reading the group any further, too. The original reply obviously didn't take this minimum of kindness into account. regards Werner --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Very important, I need the Login/password of my Hi-Focus ADSL modem (ECI) Date: Tue, 15 Jan 2002 17:05:38 +0100 From: "Werner Schiendl" Message-ID: <3c445354@brateggebdc5.br-automation.co.at> References: <3c4348e3$0$203$626a54ce@news.free.fr> Hi, Default username and passwort for VxWorks is target and password - at least for VxWorks 5.4. (Default means, it is shipped with this settings, not that this is kind of a backdoor in a properly configured system. Therefore, you'll probably have no luck with this information unless they did a real poor job...) regards Werner "Mirak" wrote in message news:3c4348e3$0$203$626a54ce@news.free.fr... > Who knows how to get it ? > I can't configurate the modem without this log/pass. > Bye. > > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: help on telnet problem Date: Tue, 15 Jan 2002 17:11:23 +0100 From: "Werner Schiendl" Message-ID: <3c4454ac$1@brateggebdc5.br-automation.co.at> References: <221caa76.0201150434.e1e7cf2@posting.google.com> Hi, I guess there is a typo somewhere in your post? What does not work? When working with the project facility you should (normally) not change configuration files manually. The settings will be overwritten (partly, sometimes) with the settings in the project facility. So include the telnet server with the project facility, you can search for INCLUDE_TELNET there to quickly find the right place. If it was the command line build that fails, you need to provide details. hth Werner "kewin" wrote in message news:221caa76.0201150434.e1e7cf2@posting.google.com... > hi all > > I have added the macro ,INCLUDE_TELNET,in the config.h. And make a > vxWorks in the command line enviroment.After booting my target board,i > can telnet it. > If i make the vxWorks file in project enviroment, i can telnet the > target board > .Anybody can tell me that there is any special configuration comand > line enviroment when making vxworks image file. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: How can I update the file system date Date: Tue, 15 Jan 2002 17:15:32 +0100 From: "Werner Schiendl" Message-ID: <3c4455a6@brateggebdc5.br-automation.co.at> References: <454b12eb.0201150632.3702a40a@posting.google.com> Which file system? dosFs? look at dosFsTimeSet(), dosFsDateSet() and dosFsDateTimeInstall() IIRC, dosFs2 has a built-in handling for getting the correct date/time from the system clock. You need to provide your own code to read your RTC (at startup), unless your BSP cares for it already. hth Werner "Edward" wrote in message news:454b12eb.0201150632.3702a40a@posting.google.com... > Could you please display me the code that allows me to update the file > system date as it still reading an old date. > > Thanks > Edward --------------------------- Newsgroups: comp.os.vxworks Subject: Re: taskdelay and interrupts Date: Tue, 15 Jan 2002 17:28:00 +0100 From: "Werner Schiendl" Message-ID: <3c445891$1@brateggebdc5.br-automation.co.at> References: <65b2abfc.0201150725.4f5f9903@posting.google.com> "Venkat Mynampati" wrote in message news:65b2abfc.0201150725.4f5f9903@posting.google.com... ... > > Hi, > I am relatively new to vxWorks anyway i will stick my neck out... > Fisrt thing according to doc's is: "printf() and any call that blocks > on a resource cannot be used in an interrupt." > However, what would happen if i call printf() in an ISR, as is in this > case, and assuming there are no other tasks or interrupt vieing for CPU, > then, i think printf() would work, 'cause, the ISR blocks on some resource > only when that resource is unavailable and what if that resource is available? No, you cannot make sure that printf will work, even if the ISR were the one and only place where you had I/O in the application. This is because printf calls IO functions and it depends on the underlying driver and IO media what happens. Interrupts are often locked in an ISR (depending on architecture, etc.) and what would happend if that IO needed to wait for an interrupt on a serial line or network board? It will get dead-locked. Further, there are a lot of VxWorks functions that will refuse to work completely and return an appropriate error code when called anyway. Semaphore and message queue functions have this behaviour, IIRC. If the driver uses one of this functions, it will fail again. The conclusion can only be to use the appropriate means for that task: logMsg() > > Or how can we explain the fact that printf() worked when taskDelay() is less? > E. g. on x86 PC based architecture, the default output device for printf is the VGA screen which is 'always ready' and therefore printf does usually work fine, as long as no normal task uses it at the same time. This causes problems with newbies any now and then, since it silently locks the whole system at random intervals when interupt and task context usage of printf collide. hth Werner --------------------------- Newsgroups: comp.os.vxworks Subject: Re: change usrconfig.c , but not works Date: Tue, 15 Jan 2002 10:57:37 -0600 From: Dinker Charak Organization: Fermi National Accelerator Laboratory Message-ID: <3C445F81.CA0ACE5A@fnal.gov> References: <4iH08.15119$iM.14260@news1.bloor.is> > You have two options: > 1. the bad solution: try to change to project file itself This won't work. You can make changes. But the moment you doa build using the GUI/Workspace, this file is re-written and those changes will be lost. The _only_ way is to use various macro options in VxWorks Tab. If you want to do something very custom, look at file in target/comps 00VxWorks.cdf, 00Bsp.cdf and 00Network.cdf (hope I remember the names right) and see how they are laid out and how that created the vxworks tab. Add you custom folder into it and hence define macros, init functions etc for you custom action. HTH, Dinker --------------------------- Newsgroups: comp.os.vxworks Subject: Re: debuging fads860 interrupt problems "uninitialized interrupt" Date: 15 Jan 2002 09:02:11 -0800 From: john_94501@yahoo.com (John) Organization: http://groups.google.com/ Message-ID: <488e459a.0201150902.4634461b@posting.google.com> References: <5624c1bf.0201150038.74cbea67@posting.google.com> Hello, Just a hunch, and maybe it was a typo in your message, but why if your device is on IRQ1 were you trying to connect the handler to interrupt level 2? The "uninitialized" message is telling you that there is no handler attached to the interrupt you have generated, so this would also be consistent. HTH, John... yong_bvcom@sina.com (bigtornado) wrote in message news:<5624c1bf.0201150038.74cbea67@posting.google.com>... > hi,everybody: > I use fads860 board,and now I want add an device,then using MPC860 > external interrupt IRQ1, I connect my ISR routine use > "intConnect(INUM_TO_IVEC(2),myIsr,0)",after this i enable this > interrupt use "intEnable(2)",but when interrupt occurs,system give me > information "uninitialized interrupt",I debug my interrupt drive > routine "ppc860ctrl.c",all seems right,why about this? > Thanks for anyone's suggestions very much! > Jeffy 1.15 --------------------------- Newsgroups: comp.os.vxworks Subject: Re: ARM coff2elf conversion Date: 15 Jan 2002 09:05:55 -0800 From: john_94501@yahoo.com (John) Organization: http://groups.google.com/ Message-ID: <488e459a.0201150905.4f9e5a99@posting.google.com> References: Hello, Your chances are not good, though COFF to ELF is possible. The problem you will face is that the debug information is probably not going to transfer, so the best you'd get from your debugger is assembler level. If all you want is to look at the code, try 'objdumparm -S' out. If you want to debug it for real, then you are stuck with CrossWind, or a third party debugger that supports VxWorks on ARM. HTH, John... "Harjinder Chayra" wrote in message news:... > I'd like to be able to build a vxWorks/bootrom image (for the ARM9) and then > load it into the ARM debugger to look at the resultant code etc. BUT ELF is > not a supported format for the ARM from Windriver. > > Are their any COFF to ELF converters out there? any other suggestions? > > Thanks > Harjinder --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Waiting on multiple events Date: Tue, 15 Jan 2002 17:37:45 -0000 From: "Dave Korn" Organization: Lumber Cartel (tinlc) Members #2234-2237 (owing to browser refresh) Message-ID: References: <14e76c83.0201150158.696e793f@posting.google.com> Niall Leonard wrote in message <14e76c83.0201150158.696e793f@posting.google.com>... >I would like to cause a task to pend on a group of semaphores. I would >like the task to be made ready when the first semaphore is given. >There is a similar function available in the Win32 API called >waitForMultipleObjects() (or something like that..). It is given an >array of "objects" that a task can pend on, and returns when the first >of these objects is available. > >Anyone got any ideas on how to do this in VxWorks ? Alas, VxWorks doesn't have the same concept as windows of a generic waitable dispatcher object: this means that you need one routine to wait for a semaphore, another to wait for a message queue, yet a third to wait for a file descriptor. VxWorks also doesn't have a generalized ability to wait on multiple objects either. I can think of a work-around, because there is one function that can be used to wait for multiple objects: select (). So rather than a bunch of semaphores, you could have a bunch of pipes; rather than giving a semaphore when you want to wake the task, you could send a one byte message down the pipe; then the task could call select on all the pipe fd's, and would return when the first one was written. This trick might be sufficient for your purposes, although it would be tricky to recreate the atomicity and can't-be-given-twice-without-being-taken-inbetween properties of real semaphores. DaveK - -- moderator of alt.talk.rec.soc.biz.news.comp.humanities.meow.misc.moderated.meow Burn your ID card! http://www.optional-identity.org.uk/ Help support the campaign, copy this into your .sig! Proud Member of the Exclusive "I have been plonked by Davee because he thinks I'm interesting" List Member # --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Building GCC 2.95.x for vxWorks (PPC), Solaris host Date: Tue, 15 Jan 2002 17:57:46 -0000 From: "Dave Korn" Organization: Lumber Cartel (tinlc) Members #2234-2237 (owing to browser refresh) Message-ID: References: Scott Johnson wrote in message ... > >Hello, > >Has anyone managed to build GCC 2.95.x for vxWorks (powerpc 860 >architecture) successfully on Solaris? I tried Dave Korn's website; but >he has only binaries for Windoze, and the directions are windows-centric. >(I have sent Dave an email; but am also interested in a wider response. >Dave, if you HAVE responded already; I apologize--I might have >accidentally tossed it into the bit bucket as SPAM; which I seem to get a >lot of.... Hi Scott, I remember seeing it arrive; I'm afraid I just haven't had time to reply it yet, sorry! >Everything seems to work fine up until it tries to build libgcc1-test.o. >The linker then hangs (except for one instance where it segfaults and >dumps core). I reckon that's almost certainly solaris-specific, and probably a bug in ld. >Our system (I don't have root, so my ability to affect changes to the >environemnt is limited): AFAIR, that's no problem, as long as you use a --prefix setting pointing to somewhere you do have full rights. >Host compiler is GCC 2.8.1 (can this compile a cross GCC 2.95, or do I >need to have GCC 2.95 for Solaris running before I can compile a version >for vxWorksB) >Host binutils is 2.11.2 (the only one which works well with Solaris, so I >am told...) Hmm. I dunno about that. I found that binutils series 2.11 wouldn't work with Gcc 2.95.x ppc-vxworks. If it's the only version that works well with Solaris, but it doesn't work well with pre-3.0-gcc, you have a problem. 2.8.x is very old too: it might be just a bit too old for such a recent binutils. I was working under cygwin when I tried it, so this might not apply to your Solaris build, but I think that the binutils 2.11.x series has changes made to remain in-sync with gcc-3.0.x, that have broken backward compatibility with the 2.95.x and earlier versions of gcc. Since you're using a pretty old version of gcc (2.8.x), I'd consider trying an earlier binutils, but I think the best solution for Solaris might well be to install a more recent (specifically, 2.95.3) version of Gcc, configured to use the native toolchain - the sun linker and assembler, etc - by using the --without-gnu-as --without-gnu-ld options at configure time. That is, *if* you have the native sun dev tools. >In addition, I had to download lots of other packages not mentioned in the >docs (Autoconf, flex) to even get it to compile that far. Agh, mea culpa; the package was meant to build straight out of the box, and if you download the whole source archive I believe it does, but I screwed up when I made the diffs and left an out-of-date inclhack.def in the fixincs subdir, so if you build by downloading the patches and applying them to the GNU source tarball, you find it wants the auto-* tools during the build process. Check the FAQ/Docs page (http://www.lysergic.u-net.com/faqdocs.html), the final paragraph under 'UWOT? "You need to install autogen!"' >Finally, when fixinc runs; I get all sorts of error messages; usually >caused by fixinc trying to find MIPS-related header files, which are not >installed at my site. Can one pass preprocessor directives (ie >-DCPU=PPC860) to fixinc to make this junk go away--or does it even matter? You'll need to elaborate on this. Where are these errors coming from? If they're the errors when it tries to build the binary fixincs, fails, and falls back to the shell script version, then don't worry about it. If they aren't, then I don't think the compiler is involved. And I'm curious to know how you go about compiling native programs for your MIPS if you don't have any MIPS-related header files. >Finally---if anyone HAS successfully gotten this to work on Solaris; does >anyone have binaries they would like to share??? Tarballs would be >preferred. I'm fairly sure it has been made to work on Solaris; when I first started building cross gcc compilers, I tried it on an old Solaris, and the more-or-less standard procedure worked. I think I've communicated either here or in email with people who've built it on Solaris in the past, but I can't remember who off the top of my head. DaveK - -- moderator of alt.talk.rec.soc.biz.news.comp.humanities.meow.misc.moderated.meow Burn your ID card! http://www.optional-identity.org.uk/ Help support the campaign, copy this into your .sig! Proud Member of the Exclusive "I have been plonked by Davee because he thinks I'm interesting" List Member # --------------------------- Newsgroups: comp.os.vxworks Subject: Re: change usrconfig.c , but not works Date: 15 Jan 2002 11:16:49 -0800 From: vloscomp@yahoo.com (Vinh) Organization: http://groups.google.com/ Message-ID: References: <4iH08.15119$iM.14260@news1.bloor.is> Hi, I do not know if you can used Cygwin enviroment to build your project. But if you do this is a way to get around this problem. In the Cygwin enviroment, used the rules.x86-win32, rules.bsp, defs.bsp, defs.x86-win32gnu, make.PPC604gnu to generate rules and dependencies. From here on i modified the rules to compile usrConfig.c, and point them to project generated files and add them into the build process. This will take a bit of time, since you need to generate the file in your project GUI then copy them into your build enviroment. Vinh Lam "Ofer Goren" wrote in message news:... > Hi. > The way I understand it, usrConfig.c is taken into consideration only when > you first create you project. After that, you might need to change > prjConfig.c for changes to take effect. However, I'm not sure it will work > either. > > You have two options: > 1. the bad solution: try to change to project file itself > 2. if this is something of macro or parameter changing, try to find it using > the GUI, and change it from there. > > Not like as Tornado 1.0.1, the vxWorks expects that the changes to the OS > will be done only after usrAppInit(). > > HTH. > > -- > Ofer Goren > Iamba Technologies > "Frank" wrote in message > news:4iH08.15119$iM.14260@news1.bloor.is... > > hi, > > I met a "simple" problem, but made me mad. > > When I changed " target/config/all/usrconfig.c", but seems it make no > sense > > when I re-build all my project. > > I doubt the "usrconfig.c" never be re-compile when I re-build my project. > > I use Tornado 2 and VxWorks 5.4 for 860. > > Pls give me some advice. > > Thank you > > > > Frank > > -- > > > > > > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: BootRom and MMU Date: 15 Jan 2002 11:20:20 -0800 From: vloscomp@yahoo.com (Vinh) Organization: http://groups.google.com/ Message-ID: References: <0y%_7.14882$8e.753501@news> <3C43E9C9.8630FBE3@europem01.nt.com> Hi Sebastien, This is correct, and i'm having problem with VxWorks kernel. Thus, if any one have any ideas how we can step into the code during this phase is greatly appreciated. Vinh Lam Sebastien Fauris wrote in message news:<3C43E9C9.8630FBE3@europem01.nt.com>... > Hi vxWorkers, > > what I find strange is that in a bootrom > the PPC processor runs in real mode and so > the MMU is disabled. The MMU will be enabled > only in the kernel downloaded afterwards. > > Sebastien. --------------------------- Newsgroups: comp.os.vxworks Subject: Serial Port Configuration Date: Tue, 15 Jan 2002 20:02:18 GMT From: "Chad" Organization: Excite@Home - The Leader in Broadband http://home.com/faster Message-ID: Is there a standard interface to check the number of parity bits, data bits, stop bits etc that a serial driver is using? I can't seem to find any IOCTL functions for checking/setting these values. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: change usrconfig.c , but not works Date: 15 Jan 2002 12:20:31 -0800 From: akarpov@my-deja.com (Alex) Organization: http://groups.google.com/ Message-ID: <4f09bb68.0201151220.4b12e193@posting.google.com> References: <4iH08.15119$iM.14260@news1.bloor.is> Frank, I'm not sure if this is relevant, but if you got used to the manual configuring VxWorks and command line builds you would use torVars.bat located in your Tornado2 tree. Alex "Frank" wrote in message news:<4iH08.15119$iM.14260@news1.bloor.is>... > hi, > I met a "simple" problem, but made me mad. > When I changed " target/config/all/usrconfig.c", but seems it make no sense > when I re-build all my project. > I doubt the "usrconfig.c" never be re-compile when I re-build my project. > I use Tornado 2 and VxWorks 5.4 for 860. > Pls give me some advice. > Thank you > > Frank > -- --------------------------- Newsgroups: comp.os.vxworks Subject: Re: taskdelay and interrupts Date: Tue, 15 Jan 2002 15:48:35 -0500 From: Doug Dotson Organization: Verio Message-ID: <3C4495A2.9564F1C4@digidata.com> References: <65b2abfc.0201150725.4f5f9903@posting.google.com> As far as I know, output to the serial port is interrupt driven in most BSPs. So by defintion, serial output is impossible from within an ISR using the normal mechanisms. You can use logMsg() if timing is not a problem and you don't exit the ISR. We have a similar problem in that we want to output panic information from within an ISR when an unrecoverable error happens. doug Venkat Mynampati wrote: > Johan Borkhuis wrote in message news:... > > karahanduman@hotmail.com (Kara) wrote: > > > > > If I comment out the taskDelay in the main loop and run it, I get the > > > printf messages as caused by interrupts without any problem. When I > > > run it with task delay(i.e. taskDelay(100000);), I don't get the > > > printf messages. if I run the taskDelay with some small value (i.e. > > > 100), I get the messages with a little delay. > > > If I take out the while(1); loop and let it return to the shell, I > > > don't get interrupts at all. > > > > NEVER use printf statements in an interrupt routine! The fact that you > > were able to get output at all surprised me. > > > > There are more routines that you cannot call from within an interrupt > > routine. Take a look at chapter 2.5 in the Programmer's Guide (Interrupt > > Service Code), and espcially 2.5.3 (Special Limitations of ISRs) > > > > If you want to create output from within an ISR use logMsg instead of > > printf. > > > > Groeten, > > Johan > > Hi, > I am relatively new to vxWorks anyway i will stick my neck out... > Fisrt thing according to doc's is: "printf() and any call that blocks > on a resource cannot be used in an interrupt." > However, what would happen if i call printf() in an ISR, as is in this > case, and assuming there are no other tasks or interrupt vieing for CPU, > then, i think printf() would work, 'cause, the ISR blocks on some resource > only when that resource is unavailable and what if that resource is available? > > Or how can we explain the fact that printf() worked when taskDelay() is less? > > Venkat --------------------------- Newsgroups: comp.os.vxworks Subject: Tornado 2.1 for PPC860 Date: Tue, 15 Jan 2002 21:51:04 GMT From: sj@shell1.aracnet.com (Scott Johnson) Organization: aracnet.com -- Portland's loudest electrons (info@aracnet.com) Message-ID: Wind River seems to have hidden information on its website about a new version of Tornado, namely Tornado 2.1; which seems to be based on the 5.x kernel (as opposed to AE) but have a more up-to-date toolchain. However, certain architectures (notably the embedded PowerPCs) do not yet have 2.1 support (interesting, since they have ported 2.1 to the G4); those of us that have low-end PPC targets seem to be stuck with Tornado 2.0.x and its antiquated toolset (unless we build our own; which I've been trying to do. See separate thread). Does anyone from WRS (or anyone else) know and care to comment on if and when 2.1 will make an appearance for additional architectures; specifcally the Motorola MPC860 (and 823 for that matter)? I certainly would hope that the PowerPC embedded targets aren't relegated to second-class status. e_s - -- - ------------------------------------------------------------------------ engineer_scotty (no, not that one) -- consumer of fine ales and lagers some days you're the fire hydrant, some days you're the dog | go blazers no small furry creatures were harmed in the creation of this .signature --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Tornado 2.1 for PPC860 Date: 15 Jan 2002 17:13:09 -0500 From: Bill Pringlemeir Organization: Bell Sympatico Message-ID: References: Sender: bpringlemeir@DeadDuck [snip] Scott> However, certain architectures (notably the embedded PowerPCs) Scott> do not yet have 2.1 support (interesting, since they have Scott> ported 2.1 to the G4); those of us that have low-end PPC Scott> targets seem to be stuck with Tornado 2.0.x and its antiquated Scott> toolset (unless we build our own; which I've been trying to Scott> do. See separate thread). I feel your pain... The same thing applies to the ARM toolkit. They have dropped support for ARM v3 and only support ARM v4. I don't blame them, the ARM v3 is brain dead and only the stupid hardware people would have included it in a design any newer, than three years ago. However, it still leaves you out in the cold if you are supporting `Legacy' products. Hopefully this is just a stepping stone to the other processors. I am more amazed that they didn't upgrade for the MPC860. This thing that I helped make in a previous life use it, "http://www.currentdirections.com/hardware/teklogix/9150.html" "http://www.psionpress.com/gpress/releases/pr550933.asp" etc... The '860 is a pretty good network processor even by today's standards. I would think that WRS will accommodate you at some point. I have been writing some `address mode three' emulation code for the last two weeks so that I can get the ARM ARCH3 working with T2.1.1... I would like to take a big stick to one of their product managers! regards, Bill Pringlemeir. - -- Your efforts at starting a revolution turn out to be costly. Look for your spouse with one eye. You will discover that Michael Jackson is abusing your life insurance. Pet piglet with the help of two other people put together. You'll eventually be unhappy. vxWorks FAQ, "http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html" --------------------------- Newsgroups: comp.os.vxworks Subject: PMC module on MVME5100 (mv5100) Date: 15 Jan 2002 14:17:48 -0800 From: strick@hiwaay.net (BS) Organization: http://groups.google.com/ Message-ID: The problem I have is when I install a PMC module into my MVME5100 I am unable to boot the system. The PMC module I have supports both 33MHZ and 66 MHZ operation and work just fine on a MVME2700. Without the PMC card the system boots just fine. I have also been able to boot the system with a PMC module that does not support 66MHZ operations, so it doesn't seem that anything is wrong with the 5100 itself. Has anyone seen similar problems like this that might be able offer a suggestion for resolving this issue. --------------------------- Newsgroups: comp.os.vxworks Subject: Unable to connect to 2nd target through 1st target using Target Server Date: 15 Jan 2002 15:12:10 -0800 From: chriss@redwavenet.com (Chris Satish) Organization: http://groups.google.com/ Message-ID: <6d6a076a.0201151512.6b71baed@posting.google.com> Hi all, I have a setup where there are two PPC 8260 targets running VxWorks 5.4 and a Win2K PC running Tornado 2.0. Target_1 has two ethernet interfaces and Target_2 has only one interface. The connection between them is as follows PC (192.168.1.163) ---- (192.168.1.38) Target_1 (192.168.21.2) ---- (192.168.21.66) Target_2 Static routes are setup on both the PC and Target_2 to use Target_1 as the gateway and pinging both ways cofirms that the setup is correct. The problem is that when on my PC, I configure a Target Server for the Target_2 at 192.168.21.66, Target_1 at 192.168.1.38 always "hijacks" this connection. This seems to be a bug in Tornado/VxWorks and I have even gone to the extent of trying to setup a Target Server for a non-existant IP of 192.168.21.123 and Target_1 STILL responds to this. Is there something that I have configured wrongly or is this a genuine bug in VxWorks? Has anybody else come across a similar scenario and if so, what is the workaround that you employed to get this to work? Thanks in advance, Chris Satish --------------------------- Newsgroups: comp.os.vxworks Subject: Tornado 2.1 for PPC860 Date: Tue, 15 Jan 2002 23:18:17 GMT From: Andray Kaganovsky Organization: Primus Canada Message-ID: References: sj@shell1.aracnet.com (Scott Johnson) wrote in news:cv118.8944$WM6.1609540 @e420r-atl1.usenetserver.com: > > Wind River seems to have hidden information on its website about a new > version of Tornado, namely Tornado 2.1; which seems to be based on the 5.x > kernel (as opposed to AE) but have a more up-to-date toolchain. > > > However, certain architectures (notably the embedded PowerPCs) do not yet > have 2.1 support (interesting, since they have ported 2.1 to the G4); > those of us that have low-end PPC targets seem to be stuck with Tornado > 2.0.x and its antiquated toolset (unless we build our own; which I've been > trying to do. See separate thread). > > > Does anyone from WRS (or anyone else) know and care to comment on if and > when 2.1 will make an appearance for additional architectures; > specifcally the Motorola MPC860 (and 823 for that matter)? I certainly > would hope that the PowerPC embedded targets aren't relegated to second- class > status. > > e_s [AndrayK]. I think WindRiver plans to skip general T-2.1/PowerPC release, and go straight to T-2.2 instead. According to the partners e-mail bulletin that I have received from WindRiver, they will start shipping "release candidate" version of T-2.2 to partners in May 2002, and release T-2.2 for general public around July 2002. My understanding is that T-2.2 will be free upgrade to all registered T-2.0x users. Andray --------------------------- Newsgroups: comp.os.vxworks Subject: Re: BootRom and MMU Date: 15 Jan 2002 16:41:32 -0800 From: brian_doe@hotmail.com (brian) Organization: http://groups.google.com/ Message-ID: References: <0y%_7.14882$8e.753501@news> <3C43E9C9.8630FBE3@europem01.nt.com> Hi All, Don't have an answer here but I have a question. I havent used the PPC but always thought (or at least it's true in all the processors I've worked on so far) that vxWorks didn't use the MMU as it always ran in flat memory model mode. Except for AE I assume. thanx brian vloscomp@yahoo.com (Vinh) wrote in message news:... > Hi Sebastien, > This is correct, and i'm having problem with VxWorks kernel. Thus, > if any one have any ideas how we can step into the code during this > phase is greatly appreciated. > > Vinh Lam > > Sebastien Fauris wrote in message news:<3C43E9C9.8630FBE3@europem01.nt.com>... > > Hi vxWorkers, > > > > what I find strange is that in a bootrom > > the PPC processor runs in real mode and so > > the MMU is disabled. The MMU will be enabled > > only in the kernel downloaded afterwards. > > > > Sebastien. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Tornado 2.1 For MIPS seems to obsolete existing BSP. Date: Tue, 15 Jan 2002 16:40:19 -0800 From: Lori Fraleigh Organization: Real-Time Innovations Message-ID: <3C44CBF3.F0091538@rti.com> References: <3C444E2D.80ADE8B2@digidata.com> Doug - I think if you go through the release notes and architecture supplement for Tornado 2.1 for MIPS it says to use the - -mips4 compiler option in conjunction with the CPU type of MIPS64. The old CPU types are no longer applicable in this release. Regards, Lori Doug Dotson wrote: > > Greetings! > > Just received the upgrade to Tornado 2.1 For MIPS. In trying > to rebuild our BSP I have noticed that the changing of macros > seems to have made the BSP obsolete. Currently the macro > CPU=R4000 in the Makefile selects the architecture, but in 2.1 there is > no R4000 CPU. It has been changed to either MIPS32 or > MIPS64. In our case, MIPS64 is appropriate for our R7000 > processor. Changing the arcitecture to MIPS64 causes the > MIPS64gnu toolchain to be properly selected, but then > there is no R4000 macro defines to allow proper building of the > BSP. Is there any easy workaround to this? > > Thanks, > Doug - -- Lori Fraleigh Director of Engineering Real-Time Innovations, Inc. http://www.rti.com Phone: 408-734-4200 x121 Fax: 408-734-5009 --------------------------- Newsgroups: comp.os.vxworks Subject: Re: debuging fads860 interrupt problems "uninitialized interrupt" Date: 15 Jan 2002 17:28:22 -0800 From: yong_bvcom@sina.com (bigtornado) Organization: http://groups.google.com/ Message-ID: <5624c1bf.0201151728.23cb516e@posting.google.com> References: <5624c1bf.0201150038.74cbea67@posting.google.com> <488e459a.0201150902.4634461b@posting.google.com> hi,thanks I use INUM_TO_IVEC(2) for IRQ1 because mpc860's interrupt priority order is NUM INTERRUPT SOURCE 0 IRQ0 1 LEVEL0 2 IRQ1 3 LEVEL1 . . . . Is it correct?I think the IRQ1 should be 2,because only after intEnable(2),the information appears! john_94501@yahoo.com (John) wrote in message news:<488e459a.0201150902.4634461b@posting.google.com>... > Hello, > > Just a hunch, and maybe it was a typo in your message, but why if your > device is on IRQ1 were you trying to connect the handler to interrupt > level 2? The "uninitialized" message is telling you that there is no > handler attached to the interrupt you have generated, so this would > also be consistent. > > HTH, > > John... > > yong_bvcom@sina.com (bigtornado) wrote in message news:<5624c1bf.0201150038.74cbea67@posting.google.com>... > > hi,everybody: > > I use fads860 board,and now I want add an device,then using MPC860 > > external interrupt IRQ1, I connect my ISR routine use > > "intConnect(INUM_TO_IVEC(2),myIsr,0)",after this i enable this > > interrupt use "intEnable(2)",but when interrupt occurs,system give me > > information "uninitialized interrupt",I debug my interrupt drive > > routine "ppc860ctrl.c",all seems right,why about this? > > Thanks for anyone's suggestions very much! > > Jeffy 1.15 --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Waiting on multiple events Date: 15 Jan 2002 17:52:57 -0800 From: john_94501@yahoo.com (John) Organization: http://groups.google.com/ Message-ID: <488e459a.0201151752.6556a99d@posting.google.com> References: <14e76c83.0201150158.696e793f@posting.google.com> Hello all, > I can think of a work-around, because there is one function that can be > used to wait for multiple objects: select (). So rather than a bunch of > semaphores, you could have a bunch of pipes; rather than giving a semaphore > when you want to wake the task, you could send a one byte message down the > pipe; then the task could call select on all the pipe fd's, and would return > when the first one was written. This trick might be sufficient for your > purposes, although it would be tricky to recreate the atomicity and > can't-be-given-twice-without-being-taken-inbetween properties of real > semaphores. One option here would be to create a pseudo-device for semaphores, much like the pipe driver is in fact simply pseudo-device on top of a message queue. That should be relatively simple to do. You could ask your FAE about it, perhaps they'll let you have the pipe code to base it on ;-) HTH, John... --------------------------- Newsgroups: comp.os.vxworks Subject: Re: ipAttach problem ?? ( two end driver) Date: 15 Jan 2002 19:08:24 -0800 From: yong_bvcom@sina.com (bigtornado) Organization: http://groups.google.com/ Message-ID: <5624c1bf.0201151908.2a20e18d@posting.google.com> References: two reasons: 1.you must increase you IP_MAX_UNITS to 2 in your config files; 2.you can't use the same ip address for both your cpm units,and must be in difference subnet; I hope this may help you! sardine_sz@163.net (sardine) wrote in message news:... > ok, > pls remember: > increase IP_MAX_UNITS in your vxworks kernel project facility > configuration macro. > > use "find" mechanism in the tornado project facility. > > you didn't changed IP_MAX_UNITS in your source code. --------------------------- Newsgroups: comp.os.vxworks Subject: How to connect a routine to IRQ1?(MPC860T on FADS BOARD) Date: 15 Jan 2002 19:18:53 -0800 From: yong_bvcom@sina.com (bigtornado) Organization: http://groups.google.com/ Message-ID: <5624c1bf.0201151918.37b4e6f6@posting.google.com> hi,all: I want to connect a ISR to IRQ1 use code as follow: intConnect(INUM_TO_VEC(2),myIsr,0); intEnable(2); but failed! someone can tell me whether it is correct? I use windriver FADS MPC860T target board. thanks! --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Serial Port Configuration Date: Wed, 16 Jan 2002 05:01:34 GMT From: Jerry Hudgins Message-ID: <3C45092C.D9EDAABA@e-farm.com> References: Reply-To: Jerry Hudgins Chad wrote: > > Is there a standard interface to check the number of parity bits, data bits, > stop bits etc that a serial driver is using? I can't seem to find any IOCTL > functions for checking/setting these values. For SIO drivers, see SIO_HW_OPTS_GET/SIO_HW_OPTS_SET in sioLib.h. --------------------------- Newsgroups: comp.os.vxworks Subject: /var/tmp/....errors while compling assembly routines Date: 15 Jan 2002 21:16:20 -0800 From: alimohd@logiceastern.com (Ali) Organization: http://groups.google.com/ Message-ID: hi DaveK Thanx for the answer but i would like to tell u that i have write perms to /var as well as tmp directory. Also the dir is not full. I am using sun sparc system and runs tornado 2.0 on it . I am using the PPCEC403 tool chain for compilation but when i compile evry time a new error is generated in the /var/tmp/**** directory .the **** here means that every time this file changes here. i don't understand why thus happens. waiting fo rur reply and if any body can help me in resolving it. thanx in advance ali --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Building GCC 2.95.x for vxWorks (PPC), Solaris host Date: Wed, 16 Jan 2002 05:30:09 GMT From: sj@shell1.aracnet.com (Scott Johnson) Organization: aracnet.com -- Portland's loudest electrons (info@aracnet.com) Message-ID: References: In article you write: >Scott Johnson wrote in message ... >> >>Hello, >> >>Has anyone managed to build GCC 2.95.x for vxWorks (powerpc 860 >>architecture) successfully on Solaris? I tried Dave Korn's website; but >>he has only binaries for Windoze, and the directions are windows-centric. >>(I have sent Dave an email; but am also interested in a wider response. >>Dave, if you HAVE responded already; I apologize--I might have >>accidentally tossed it into the bit bucket as SPAM; which I seem to get a >>lot of.... > > Hi Scott, > > I remember seeing it arrive; I'm afraid I just haven't had time to reply >it yet, sorry! Not a problem. >>Everything seems to work fine up until it tries to build libgcc1-test.o. >>The linker then hangs (except for one instance where it segfaults and >>dumps core). > > I reckon that's almost certainly solaris-specific, and probably a bug in >ld. Linking libgcc1-test? That would be the target linker involved, NOT the host linker??? In other words, the linker in binutils. >>Our system (I don't have root, so my ability to affect changes to the >>environemnt is limited): > > AFAIR, that's no problem, as long as you use a --prefix setting pointing >to somewhere you do have full rights. That part does seem to work. One other clue for your FAQ (by the way... what does UWOT stand for? Inquiring minds want to know....) is that if you do this, make sure to add {prefix-dir}/lib to LD_LIBRARY_PATH; otherwise the make may pull in stuff from who-knows-where, or fail altogether. >>Host compiler is GCC 2.8.1 (can this compile a cross GCC 2.95, or do I >>need to have GCC 2.95 for Solaris running before I can compile a version >>for vxWorksB) Update. I've since tried building GCC 2.95.3 for Solaris. The host compiler seems to work; I still get the segfault when compiling the cross compiler. >>Host binutils is 2.11.2 (the only one which works well with Solaris, so I >>am told...) > > Hmm. I dunno about that. I found that binutils series 2.11 wouldn't work >with Gcc 2.95.x ppc-vxworks. I'm using binutils 2.10 as my target binutils; 2.11 is only the native binutils for building things. If it's the only version that works well with >Solaris, but it doesn't work well with pre-3.0-gcc, you have a problem. >2.8.x is very old too: it might be just a bit too old for such a recent >binutils. > > I was working under cygwin when I tried it, so this might not apply to >your Solaris build, but I think that the binutils 2.11.x series has changes >made to remain in-sync with gcc-3.0.x, that have broken backward >compatibility with the 2.95.x and earlier versions of gcc. Since you're >using a pretty old version of gcc (2.8.x), I'd consider trying an earlier >binutils, but I think the best solution for Solaris might well be to install >a more recent (specifically, 2.95.3) version of Gcc, configured to use the >native toolchain - the sun linker and assembler, etc - by using >the --without-gnu-as --without-gnu-ld options at configure time. That is, >*if* you have the native sun dev tools. As mentioned above, I have done so (though using the Gnu binutils rather than the native ones). I will try this though. I do seem to have a native linker and assembler hanging around (/usr/ccs/ld and /usr/ccs/as; are those the right ones)? Don't have the native Sun compiler though... since GCC native seems to work fine, I doubt I need it. Anyway, I'll give rebuild gcc (native) with --without-gnu-ls|ad and compile everything up again. >>In addition, I had to download lots of other packages not mentioned in the >>docs (Autoconf, flex) to even get it to compile that far. > > Agh, mea culpa; the package was meant to build straight out of the box, >and if you download the whole source archive I believe it does, but I >screwed up when I made the diffs and left an out-of-date inclhack.def in the >fixincs subdir, so if you build by downloading the patches and applying them >to the GNU source tarball, you find it wants the auto-* tools during the >build process. Check the FAQ/Docs page >(http://www.lysergic.u-net.com/faqdocs.html), the final paragraph under >'UWOT? "You need to install autogen!"' A point of clarification: Is it better to use the pre-patched sources (the so-called gcc 2.95.4) or download the pristine gcc 2.95.3 code and apply your patch? I did the latter; you seem to suggest the former is better. The build instructions at the top of your FAQ/Docs page seem to call for the latter; if the former is better you might modify your page to suggest that. Since I am not an expert at GCC internals, I was trying to follow your procedure to the letter...) >>Finally, when fixinc runs; I get all sorts of error messages; usually >>caused by fixinc trying to find MIPS-related header files, which are not >>installed at my site. Can one pass preprocessor directives (ie >>-DCPU=PPC860) to fixinc to make this junk go away--or does it even matter? > > You'll need to elaborate on this. Where are these errors coming from? If >they're the errors when it tries to build the binary fixincs, fails, and >falls back to the shell script version, then don't worry about it. It doesn't stop the build procedure, at any rate. If they >aren't, then I don't think the compiler is involved. And I'm curious to >know how you go about compiling native programs for your MIPS if you don't >have any MIPS-related header files. Because my target is PowerPC. :) It appears that fixinc is trying to patch the header files, and is going about including sub-headers as needed. For some reason or other, rather than including the stuff in arch/ppc it is going after arch/mips. As we do not have the MIPS target installed; arch/mips is empty. Why it is doing this, I've no idea--unless it thought for some reason that MIPS was defined (or is otherwise brainless about how it handles conditional includes....) A stupid thought (as the tools running on the host shouldn't care about the Tornado headers): If fixinc isn't fixing certain braindead Wind River header files (arch/ppc/toolPpc.h, with its broken stdarg/vararg implementation, comes to mind...) might that somehow break the compiler/linker? It shouldn't AFAIK--that might cause the compiler to produce bad code, but it shouldn't cause the compiler to dump core. On the other hand, stranger things have happened. Just thinking out loud. >>Finally---if anyone HAS successfully gotten this to work on Solaris; does >>anyone have binaries they would like to share??? Tarballs would be >>preferred. > I'm fairly sure it has been made to work on Solaris; when I first started >building cross gcc compilers, I tried it on an old Solaris, and the >more-or-less standard procedure worked. I think I've communicated either >here or in email with people who've built it on Solaris in the past, but I >can't remember who off the top of my head. I'll keep looking around... Thanks for your help! e_s > DaveK >-- >moderator of >alt.talk.rec.soc.biz.news.comp.humanities.meow.misc.moderated.meow >Burn your ID card! http://www.optional-identity.org.uk/ >Help support the campaign, copy this into your .sig! >Proud Member of the Exclusive "I have been plonked by Davee because he >thinks I'm interesting" List >Member # - ------------------------------------------------------------------------ engineer_scotty (no, not that one) -- consumer of fine ales and lagers some days you're the fire hydrant, some days you're the dog | go blazers no small furry creatures were harmed in the creation of this .signature - ------------------------------------------------------------------------ - -- - ------------------------------------------------------------------------ engineer_scotty (no, not that one) -- consumer of fine ales and lagers some days you're the fire hydrant, some days you're the dog | go blazers no small furry creatures were harmed in the creation of this .signature --------------------------- Newsgroups: comp.sys.powerpc.tech,comp.os.vxworks,comp.sys.m68k,comp.sys.powerpc.advocacy Subject: CRC and Alignment Errors. Date: 15 Jan 2002 21:41:14 -0800 From: ashokchandran@chequemail.com (akosoto) Organization: http://groups.google.com/ Message-ID: Hi all, While doing performance test for our new adapter which runs on PSOS 2.5 and MPC8260 chip we face CRC and Alignment Errors. Conditions: - --------- 1. Performance test with 4k data packet size. This results in ethernet packets of size 1500 bytes. With packet sizes less than 1400, these errors are very few. Configuration: - ------------- 1. LAN driver configures FCC with Full Duplex (FPSMR[FDE], FPSMR[LPB] set. 2. MRBLR is 1536, MAX DMA counts are 1520 each. 3. TX and RX clock driven by external clocks, i.e., from PHY. Problem: - ------- When the packet size is 1500 bytes, sum of CRC and alignment errors is as high as 25 %. This results in a very poor LAN quality and results in some performance tests to fail. On an average, the percent of errors to total number of frames received is 20 % Have anyone faced such a problem with MPC8260 with PSOS running on it? Please convey your ideas in resolving this problem. Thanks in advance Ashok Chandran. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: How to connect a routine to IRQ1?(MPC860T on FADS BOARD) Date: Wed, 16 Jan 2002 08:57:51 +0000 From: Graham Baxter Organization: Graham Baxter (Software) Limited Message-ID: <3C45408F.128653AF@NOSPAM.bcs.org.uk> References: <5624c1bf.0201151918.37b4e6f6@posting.google.com> It looks OK but check that intConnect() and intEnable() both return OK Regards, Graham Baxter (VxWorks and pSOS BSP'S) Freelance Software Engineer gbaxter@NOSPAM.bcs.org.uk --------------------------- Newsgroups: comp.os.vxworks Subject: serial i/o Date: Wed, 16 Jan 2002 09:01:52 -0000 From: "Martin Dowie" Message-ID: <3c45413f@pull.gecm.com> Can I use the ioLib routines open/close/read/write to transmit/receive via the RS232 channel on our DY4 SVME-179 card? Or the RS422 & RS485 we have as well? If so, what 'file name' should we supply to the open routine? "/dev/???"? TIA --------------------------- Newsgroups: comp.os.vxworks Subject: Re: ARM coff2elf conversion Date: 16 Jan 2002 09:26:57 GMT From: Diamand@btinternet.com (Luke Diamand) Message-ID: References: Harjinder Chayra wrote: >I'd like to be able to build a vxWorks/bootrom image (for the ARM9) and then >load it into the ARM debugger to look at the resultant code etc. BUT ELF is >not a supported format for the ARM from Windriver. > >Are their any COFF to ELF converters out there? any other suggestions? You can build objdump from the GNU sources to binutils with support for ARM/COFF and ARM/ELF in the same image. It might then be possible to copy from one format to the other. HTH! Luke Diamand --------------------------- Newsgroups: comp.os.vxworks Subject: Re: change usrconfig.c , but not works Date: Wed, 16 Jan 2002 11:47:34 +0200 From: "Ofer Goren" Organization: Verio Message-ID: References: <4iH08.15119$iM.14260@news1.bloor.is> <3C445F81.CA0ACE5A@fnal.gov> well, providing you close your project before doing those changes, it WILL work. Next time you'll open your project, the changes will take effect. - -- Ofer Goren Iamba Technologies "Dinker Charak" wrote in message news:3C445F81.CA0ACE5A@fnal.gov... > > You have two options: > > 1. the bad solution: try to change to project file itself > > This won't work. You can make changes. But the moment you doa build using the > GUI/Workspace, this file is re-written and those changes will be lost. The > _only_ way is to use various macro options in VxWorks Tab. If you want to do > something very custom, look at file in target/comps 00VxWorks.cdf, 00Bsp.cdf > and 00Network.cdf (hope I remember the names right) and see how they are laid > out and how that created the vxworks tab. Add you custom folder into it and > hence define macros, init functions etc for you custom action. > > HTH, > Dinker > > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: why coulud not my ARM7TDMI and vxworks system support floating point? Date: Wed, 16 Jan 2002 10:31:50 +0000 (UTC) From: David Laight Organization: BT Openworld Message-ID: <3C4556E9.BFFB2FF7@l8s.co.uk> References: <15decc28.0201140617.399b988@posting.google.com> <3C4345B2.A0EAA44@rti.com> <15decc28.0201142123.22f5f5a8@posting.google.com> <3C4414B7.6C0C2556@l8s.co.uk> Bill Pringlemeir wrote: > > >>>>> "David" == David Laight writes: > > [snip] > > David> FYI the fp routines WRS supply for arm systems suck! They are > David> too slow... > > They have been updated in T2.1.1. I guess you meant "the fp routines > for ARM and thumb"? They seem to be particularly bad for the ARM. dunno about T2.1.1 - but I speeded up our app by a factor of 100 or so by spending a week writing some FP routines in ARM asm. Maybe WRS have finally took the hint and got the ASM routines from ARM. The ARM port used to use the C routines - which are particularly badly written! David --------------------------- Newsgroups: comp.os.vxworks Subject: Re: BootRom and MMU Date: Wed, 16 Jan 2002 10:35:25 +0000 (UTC) From: David Laight Organization: BT Openworld Message-ID: <3C4557C0.5F74BC87@l8s.co.uk> References: <0y%_7.14882$8e.753501@news> <3C43E9C9.8630FBE3@europem01.nt.com> > Don't have an answer here but I have a question. I havent used the PPC > but always thought (or at least it's true in all the processors I've > worked on so far) that vxWorks didn't use the MMU as it always ran in > flat memory model mode. Except for AE I assume. It depends... They don't do any dynamic mapping nor set any restrictive permissions, but on some systems you have to enable the mmu in to : 1) enable the data and instruction caches 2) get dram contiguous and at address 0 3) get page faults for invalid addresses David --------------------------- Newsgroups: comp.os.vxworks Subject: BSP PCpentium & boot on floppy & Tornado2 Date: 16 Jan 2002 08:53:51 GMT From: Charly Message-ID: <2002116-95351-971517@foorum.com> Hello, I'd like to use PCpentium BSP with a floppy boot. I am not sure to understand. So , could you correct me from these afirmation: 1)To create an uncompressed floppy, the files necessary are "bootrom_uncmp" and "vxworks" 2)To create an compressed floppy, the files necessary are "bootrom" and "vxWorks_romCompress". Do I need to rename "vxWorks_romCompress" to "vxWorks.st". Thanks a lot. charly - -- Use our news server 'news.foorum.com' from anywhere. More details at: http://nnrpinfo.go.foorum.com/ --------------------------- Newsgroups: comp.os.vxworks Subject: a WPJ forPCpentium BSP with a floppy boot Date: 16 Jan 2002 09:19:54 GMT From: Charly Message-ID: <2002116-101954-931982@foorum.com> Hello everyboby, I'd like to create a bootable floppy for PCpentium BSP. I will appriciate a VxWorks project example able to use the shell via the serial link Thanks Charly - -- Use our news server 'news.foorum.com' from anywhere. More details at: http://nnrpinfo.go.foorum.com/ --------------------------- Newsgroups: comp.os.vxworks Subject: Re: serial i/o Date: Wed, 16 Jan 2002 11:19:46 +0000 From: Graham Baxter Organization: Graham Baxter (Software) Limited Message-ID: <3C4561D2.2A12DFB8@NOSPAM.bcs.org.uk> References: <3c45413f@pull.gecm.com> See usrConfig.c. On my 8260 PowerPC, it creates the NUM_TTY devices with names like: /tyCo/0 /tyC0/1 and opens the console device and directs STDIO to it. Regards, Graham Baxter (VxWorks and pSOS BSP'S) Freelance Software Engineer gbaxter@NOSPAM.bcs.org.uk --------------------------- Newsgroups: comp.os.vxworks Subject: CRC and Alignment Errors. Date: 16 Jan 2002 03:25:08 -0800 From: ashokchandran@chequemail.com (akosoto) Organization: http://groups.google.com/ Message-ID: Hi all, While doing performance test for our new adapter which runs on PSOS 2.5 and MPC8260 chip we face CRC and Alignment Errors. Conditions: - --------- 1. Performance test with 4k data packet size. This results in ethernet packets of size 1500 bytes. With packet sizes less than 1400, these errors are very few. Configuration: - ------------- 1. LAN driver configures FCC with Full Duplex (FPSMR[FDE], FPSMR[LPB] set. 2. MRBLR is 1536, MAX DMA counts are 1520 each. 3. TX and RX clock driven by external clocks, i.e., from PHY. Problem: - ------- When the packet size is 1500 bytes, sum of CRC and alignment errors is as high as 25 %. This results in a very poor LAN quality and results in some performance tests to fail. On an average, the percent of errors to total number of frames received is 20 % Have anyone faced such a problem with MPC8260 with PSOS running on it? Please convey your ideas in resolving this problem. Thanks in advance Ashok Chandran. --------------------------- Newsgroups: comp.os.vxworks Subject: Need Help in locating a retailor for purchasing PMC plugs Date: 16 Jan 2002 03:28:55 -0800 From: nongji.chen@vil.ite.mee.com (N Chen) Organization: http://groups.google.com/ Message-ID: I am at the initial stage of designing a prototype card complying to the PMC form factor. I need to purchase a few 64w plugs as a result. After some web searches and a number of phone calls, I am still no where near finding someone who sells them. I would therefore appreciate it much if any reader could provide some help in this matter. Thanks, Nongji Chen Mitsubishi Electric ITE Guildford, Surrey, UK nongji.chen@vil.ite.mee.com --------------------------- Newsgroups: comp.os.vxworks Subject: Re: serial i/o Date: Wed, 16 Jan 2002 11:47:33 -0000 From: "Martin Dowie" Message-ID: <3c456813$1@pull.gecm.com> References: <3c45413f@pull.gecm.com> <3C4561D2.2A12DFB8@NOSPAM.bcs.org.uk> "Graham Baxter" wrote in message news:3C4561D2.2A12DFB8@NOSPAM.bcs.org.uk... > See usrConfig.c. > > On my 8260 PowerPC, it creates the NUM_TTY devices with names like: > /tyCo/0 > /tyC0/1 > and opens the console device and directs STDIO to it. Thanks - that looks very interesting... --------------------------- End of New-News digest ********************** From vxwexplo-errs@csg.lbl.gov Wed Jan 16 11:24:44 2002 From: "Chang, Mark" Date: Wed Jan 16 11:24:46 PST 2002 Subject: Need Vxworks upgrade stories... Hello, We are currently using Vxworks 5.3.1 and T1. We have talked to WindRiver about upgrading to T2, Vxworks 5.4 - They recommend getting their consultants in and doing the work for us. I would like to know if any of you have done your own upgrade and how hard or easy, how much time it took, what road blocks you encounter etc... This is for MIPS, DevSys on NT. Thanks, Mark From vxwexplo-errs@csg.lbl.gov Thu Jan 17 02:36:50 2002 From: "amanullah khan" Date: Thu Jan 17 02:36:53 PST 2002 Subject: Tornado with MIPS RM7000A processor

Hi

I am working on MIPS RM7000A processor. I have a device which understands only 64bit. To see whether Tornado supports 64bit I wrote a small code. I have executed the code with #define LLINT64 unsigned long long in Tornado 2.0 with GNU compile version 2.7.2 and also in Tornado 2.1 with GNU version 2.96.  The assembly code seems to be same in both the version. Tornado 2.0 runs in 32 bit version whereas T2.1 runs in 64bit version.

I wanted to know whether Tornado supports 64bit instruction for MIPS RM7000A.

Regards

Aman

 


Chat with friends online, try MSN Messenger: Click Here
From vxwexplo-errs@csg.lbl.gov Thu Jan 17 02:38:51 2002 From: "suresh" Date: Thu Jan 17 02:38:53 PST 2002 Subject: tornado This is a multi-part message in MIME format. ------=_NextPartTM-000-bce01d5c-0b2c-11d6-a941-00b0d0d06be8 Content-Type: multipart/alternative; boundary="----=_NextPart_000_0013_01C19F70.DB07A770" ------=_NextPart_000_0013_01C19F70.DB07A770 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable Hi, I am basically a linux user and now there is a requirement for porting = of code in linux to vxworks. I just want to know which message queues(vxworksqueues/posixqueues) = should i use while porting the code. I also want to is there any chance to findout the limits of message = queues in vxworks like maximum size of one message, maximum number of = message queues that can be created and maximum number of messages that = can be queued etc.. thx in advance.. rgds.. suresh. ------=_NextPart_000_0013_01C19F70.DB07A770 Content-Type: text/html; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable
Hi,
I am basically a linux user and now = there is a=20 requirement for porting of code in linux to vxworks.
I just want to know which message = queues(vxworksqueues/posixqueues) should i use while porting the=20 code.
I also want to is there = any chance to findout=20 the limits of message queues in vxworks like maximum size of one = message,=20 maximum number of message queues that can be created and maximum number = of=20 messages that can be queued etc..
thx in advance..
rgds..
suresh.
------=_NextPart_000_0013_01C19F70.DB07A770-- ------=_NextPartTM-000-bce01d5c-0b2c-11d6-a941-00b0d0d06be8 Content-Type: text/plain; name="Wipro_Disclaimer.txt" Content-Transfer-Encoding: 7bit Content-Disposition: attachment; filename="Wipro_Disclaimer.txt" ----------------------------------------------------------------------------------------------------------------------- Information transmitted by this E-MAIL is proprietary to Wipro and/or its Customers and is intended for use only by the individual or entity to which it is addressed, and may contain information that is privileged, confidential or exempt from disclosure under applicable law. If you are not the intended recipient or it appears that this mail has been forwarded to you without proper authority, you are notified that any use or dissemination of this information in any manner is strictly prohibited. In such cases, please notify us immediately at mailto:mailadmin@wipro.com and delete this mail from your records. ------------------------------------------------------------------------------------------------------------------------ ------=_NextPartTM-000-bce01d5c-0b2c-11d6-a941-00b0d0d06be8-- From vxwexplo-errs@csg.lbl.gov Thu Jan 17 04:03:26 2002 From: Vxworks Exploder Date: Thu Jan 17 04:03:29 PST 2002 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Thu Jan 17 04:03:21 PST 2002 Subject: Re: serial i/o Subject: Re: Tornado 3 and Windows 2000 Subject: Re: Tornado 2.1 for PPC860 Subject: Re: /var/tmp/....errors while compling assembly routines Subject: Re: Does Tornado require write access to download links/libraries? Subject: troubles with "new" and "delete" Subject: Re: CRC and Alignment Errors. Subject: Re: a WPJ forPCpentium BSP with a floppy boot Subject: Re: a WPJ forPCpentium BSP with a floppy boot Subject: How do I for annotation // Subject: Re: serial i/o Subject: driver for uart synchronous Subject: prova Subject: Re: SingleStep Subject: how use Time slot Assign ? Subject: telnet login on 2400s Subject: Logging passwords with windweb Subject: Re: Tornado 3 and Windows 2000 Subject: Re: Tornado 3 and Windows 2000 Subject: SNMP: nman application for management Subject: Re: DHCP over PPP Subject: Re: Serial Port Configuration Subject: WindNet PPP problem Subject: Support for 8 bits + even parity in ttyDrv? Subject: Re: PMC module on MVME5100 (mv5100) Subject: Re: How do I for annotation // Subject: Re: taskdelay and interrupts Subject: Re: taskdelay and interrupts Subject: Re: taskdelay and interrupts Subject: Re: Support for 8 bits + even parity in ttyDrv? Subject: Re: How do I for annotation // Subject: Re: How do I for annotation // Subject: Re: [HELP]UDP TX/RX packet mismatch problem... Subject: torVars and make Subject: why "make "do not go? Subject: Re: flash and MMU Subject: Who can give me the source code of pppoe on vxworks? Subject: Re: Hi All Subject: Re: Is that bug of MUX or mine ? Subject: Re: torVars and make Subject: Uninitialized Interrupt! Vector number 47 Subject: mpc860 interrupt debug problem "uninitialized interrupt"? Subject: Problem of HTTP Server for vxworks on MPC860 Subject: Re: How do I for annotation // Subject: Re: why "make "do not go? Subject: Re: _protoSwIndex and inetsw[ ] used in our OSPF implementation Subject: Re: [HELP]UDP TX/RX packet mismatch problem... Subject: Re: telnet login on 2400s Subject: Re: Waiting on multiple events Subject: Re: Who can give me the source code of pppoe on vxworks? Subject: Re: torVars and make Subject: Re: driver for uart synchronous Subject: Problem with font in Tornado 2 Subject: Problem with font in Tornado 2 - Part 2 Subject: Using a PCI ethernet adapter on PCPentium BSP Subject: Re: driver for uart synchronous Subject: Re: Visual Studio Integration ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: serial i/o Date: Wed, 16 Jan 2002 12:25:27 -0000 From: "Martin Dowie" Message-ID: <3c4570f5$1@pull.gecm.com> References: <3c45413f@pull.gecm.com> <3C4561D2.2A12DFB8@NOSPAM.bcs.org.uk> "Graham Baxter" wrote in message news:3C4561D2.2A12DFB8@NOSPAM.bcs.org.uk... > See usrConfig.c. > > On my 8260 PowerPC, it creates the NUM_TTY devices with names like: > /tyCo/0 > /tyC0/1 > and opens the console device and directs STDIO to it. ok, I've done a iosDevShow and I we seems to have 4 devices hanging off our 'drv 1' /tyCo/0 /tyCo/1 /tyCo/2 /tyCo/3 what on earth are the extra devices? --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Tornado 3 and Windows 2000 Date: Wed, 16 Jan 2002 12:13:42 -0000 From: "Dave Korn" Organization: Lumber Cartel (tinlc) Members #2234-2237 (owing to browser refresh) Message-ID: References: mc wrote in message ... >Hi all, > >I installed Tornado 3 on Windows 2000 SP1 w/IE 6 (Pentium 3) but when I try >to launch it, nothing happens: no error message, no crash, nothing. I tried >to reinstall it more than once, but it still doesn't work. Any idea? Thanks >in advance. >Regards, > >MC Perhaps a problem with permissions, were you logged in as Administrator when you installed it? Are you logged in as Administrator when you try to run it? If you were an ordinary user when you installed it, are you still logged in as that same user? DaveK - -- moderator of alt.talk.rec.soc.biz.news.comp.humanities.meow.misc.moderated.meow Burn your ID card! http://www.optional-identity.org.uk/ Help support the campaign, copy this into your .sig! Proud Member of the Exclusive "I have been plonked by Davee because he thinks I'm interesting" List Member # --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Tornado 2.1 for PPC860 Date: Wed, 16 Jan 2002 11:30:07 -0000 From: "Dave Korn" Organization: Lumber Cartel (tinlc) Members #2234-2237 (owing to browser refresh) Message-ID: References: Bill Pringlemeir wrote in message ... > >[snip] > Scott> However, certain architectures (notably the embedded PowerPCs) > Scott> do not yet have 2.1 support (interesting, since they have > Scott> ported 2.1 to the G4); those of us that have low-end PPC > Scott> targets seem to be stuck with Tornado 2.0.x and its antiquated > Scott> toolset (unless we build our own; which I've been trying to > Scott> do. See separate thread). > > I feel your pain... The same >thing applies to the ARM toolkit. They have dropped support for ARM >v3 and only support ARM v4. I don't blame them, the ARM v3 is brain >dead and only the stupid hardware people would have included it in a >design any newer, than three years ago. "Oh, the programmers can always work round *that* in software......" DaveK - -- moderator of alt.talk.rec.soc.biz.news.comp.humanities.meow.misc.moderated.meow Burn your ID card! http://www.optional-identity.org.uk/ Help support the campaign, copy this into your .sig! Proud Member of the Exclusive "I have been plonked by Davee because he thinks I'm interesting" List Member # --------------------------- Newsgroups: comp.os.vxworks Subject: Re: /var/tmp/....errors while compling assembly routines Date: Wed, 16 Jan 2002 12:55:56 -0000 From: "Dave Korn" Organization: Lumber Cartel (tinlc) Members #2234-2237 (owing to browser refresh) Message-ID: References: Ali wrote in message ... >hi DaveK > Thanx for the answer but i would like to tell u that i have write >perms to /var as well as tmp directory. Also the dir is not full. > I am using sun sparc system and runs tornado 2.0 on it . I am using >the PPCEC403 tool chain for compilation but when i compile evry time a >new error is generated in the /var/tmp/**** directory .the **** here >means that every time this file changes here. i don't understand why >thus happens. >waiting fo rur reply and if any body can help me in resolving it. > thanx in advance >ali Those aren't errors, those are temporary files. DaveK - -- moderator of alt.talk.rec.soc.biz.news.comp.humanities.meow.misc.moderated.meow Burn your ID card! http://www.optional-identity.org.uk/ Help support the campaign, copy this into your .sig! Proud Member of the Exclusive "I have been plonked by Davee because he thinks I'm interesting" List Member # --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Does Tornado require write access to download links/libraries? Date: Wed, 16 Jan 2002 11:43:16 -0000 From: "Dave Korn" Organization: Lumber Cartel (tinlc) Members #2234-2237 (owing to browser refresh) Message-ID: References: <94e55b6.0201071416.23f691ff@posting.google.com> <94e55b6.0201141037.7cafa04b@posting.google.com> Matthew Becker wrote in message <94e55b6.0201141037.7cafa04b@posting.google.com>... >I hadn't received any response yet, so I'm reposting this question. > >Other debug tools typically do not require write access in order to >download code and trace through it (e.g. visionClick, SingleStep, >etc.). My experience so far indicates that the Tornado tools DO >require write access. Could someone please explain if this is the >case with Tornado, and, if possible, please provide reasons why. > >Thanks in advance. I don't use the gui downloader myself, so I can't confirm your experience, but the error message you posted seems entirely unambiguous to me. As to *why* it needs rw access, the answer can only be that it doesn't need rw access but someone wrote O_RDWR probably just from habit, on the grounds that 'well it's no harm to open it for writing if you don't actually go and do any writes to it' or similar. DaveK - -- moderator of alt.talk.rec.soc.biz.news.comp.humanities.meow.misc.moderated.meow Burn your ID card! http://www.optional-identity.org.uk/ Help support the campaign, copy this into your .sig! Proud Member of the Exclusive "I have been plonked by Davee because he thinks I'm interesting" List Member # > >-Matt > >- >Matthew Becker >Xerox Corporation >matthew.becker@usa.xerox.com > > >matthew.becker@usa.xerox.com (Matthew Becker) wrote in message news:<94e55b6.0201071416.23f691ff@posting.google.com>... >> Hello, >> >> I've noticed that if I don't have write access to a library/link, that >> I am unable to download it to the target. >> >> I select "Project->Download" from the Tornado main menu. I then >> select the appropriate file from the "Download objects" GUI. I then >> receive the following message, >> >> > F:\Runtime.Phoenix8260\MainApp\obj.iphoenixPlus\appiphoenixPlus >> > This file exists with Read Only attributes. >> > Please use a different file name. >> >> I have read-execute access to this file, but not write access. Does >> Tornado require write access in order to download links/libraries (I >> am using Tornado 2.0.2)? If so, then could someone please explain why >> this is the case. >> >> Thanks in advance. >> >> -Matt >> >> - >> Matthew Becker >> Xerox Corporation >> matthew.becker@usa.xerox.com --------------------------- Newsgroups: comp.os.vxworks Subject: troubles with "new" and "delete" Date: 16 Jan 2002 05:06:01 -0800 From: qf_liu@163.com (strong) Organization: http://groups.google.com/ Message-ID: I have to reload the operators new and delete which confilct with the "C++ runtime support". I tried to exclude it only to find more troubles although the two operators did work. Could anybody tell me a safe method to reload them? Thanks! --------------------------- Newsgroups: comp.sys.powerpc.tech,comp.os.vxworks,comp.sys.m68k,comp.sys.powerpc.advocacy Subject: Re: CRC and Alignment Errors. Date: Wed, 16 Jan 2002 14:19:22 +0100 From: "Michael Lawnick" Organization: Buergernetz Dillingen Message-ID: References: Reply-To: "Michael Lawnick" Sender: mlawnick@pd9e65904.dip.t-dialin.net Hi Ashok, don't know whether it helps, but on our current project with a selfmade board we had a similar problem. The cause was the clock for the PHY, that had a chitter. When packets were sent with little gap's, the receiver shut down and retarted (=resynced) again. When sending with no gap (big high level packets or many small), receiver didn't shutdown and got out of sync after some time. After replacing with a high quality quartz, the problems where gone. HTH - -- Mit freundlichen Grüßen, Michael Lawnick ============================================== SOFTEC GmbH Tel +49-731-96600-0 Promenade 17 Fax +49-731-96600-23 D-89073 Ulm Michael Lawnick Germany lawnick@softec.de ============================================== "akosoto" schrieb im Newsbeitrag news:d64bec.0201152141.6d24d388@posting.google.com... > Hi all, > While doing performance test for our new adapter which runs on PSOS > 2.5 and MPC8260 chip we face CRC and Alignment Errors. > > Conditions: > --------- > > 1. Performance test with 4k data packet size. This results in > ethernet > packets of size 1500 bytes. With packet sizes less than 1400, > these > errors are very few. > > Configuration: > ------------- > 1. LAN driver configures FCC with Full Duplex (FPSMR[FDE], FPSMR[LPB] > set. > 2. MRBLR is 1536, MAX DMA counts are 1520 each. > 3. TX and RX clock driven by external clocks, i.e., from PHY. > > Problem: > ------- > > When the packet size is 1500 bytes, sum of CRC and alignment errors is > as > high as 25 %. This results in a very poor LAN quality and results in > some performance tests to fail. On an average, the percent of errors > to > total number of frames received is 20 % > > Have anyone faced such a problem with MPC8260 with PSOS running on it? > Please > convey your ideas in resolving this problem. > Thanks in advance > Ashok Chandran. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: a WPJ forPCpentium BSP with a floppy boot Date: Wed, 16 Jan 2002 14:49:47 +0100 From: "Werner Schiendl" Message-ID: <3c4584fe@brateggebdc5.br-automation.co.at> References: <2002116-101954-931982@foorum.com> Hi, you need to modify config.h in your BSP directory for your needs. Particularly check the DEFAULT_BOOTLINE macro. Then you create bootrom_uncmp with make bootrom_uncmp Then you create a boot disk with mkboot a: bootrom_uncmp For all ops, you need to work in your BSP directory. After that, you can copy your VxWorks image with the proper name. IIRC, the default bootrom for x86 targets does not support compressed images? I'd start with a plain old bootable image project and compile vxWorks. .st is just a shorthand for stand-alone (working without the host) and has nothing to do with compressed or not. hth Werner "Charly" wrote in message news:2002116-101954-931982@foorum.com... > > Hello everyboby, > I'd like to create a bootable floppy for PCpentium BSP. > I will appriciate a VxWorks project example able to use the shell via the serial > link > Thanks > Charly > > > > > -- > Use our news server 'news.foorum.com' from anywhere. > More details at: http://nnrpinfo.go.foorum.com/ --------------------------- Newsgroups: comp.os.vxworks Subject: Re: a WPJ forPCpentium BSP with a floppy boot Date: Wed, 16 Jan 2002 14:49:30 +0000 (UTC) From: don@mars.dgrc.crc.ca (Donald McLachlan) Organization: The unconfigured xvnews people Message-ID: References: <2002116-101954-931982@foorum.com> Reply-To: don@mars.dgrc.crc.ca Charly, There are instructions in the manuals but they are not in obvious places. What is your development host? If Solaris: From TSR 115781 on the wrs web site (which includes vxld.bin) : aoutToBinDec < bootrom_uncmp > bootrom.sys fdformat -U -d -B vxld.bin volcheck cp bootrom.sys /floppy/floppy0 cp default/vxWorks /floppy/floppy0/vxWorks.st When the boot program starts counting down, hit enter to interrupt it, and make sure the "file" boot param is "/fd0/vxWorks.st". Then boot and you are away to the races. Gee .. bet I know what you'll want to do next ... modify the boot code to support your network card ... but one step at a time. Don - -- Donald McLachlan E-mail Donald.McLachlan@crc.ca Communications Research Centre / RNS Tel (613) 998-2845 3701 Carling Ave., Fax (613) 998-9648 Ottawa, Ontario K2H 8S2 Canada --------------------------- Newsgroups: comp.os.vxworks Subject: How do I for annotation // Date: Wed, 16 Jan 2002 20:19:40 +0900 From: "it2xpert" Organization: ONSE Telecom Message-ID: Hi... I use T2 Vxworks 5.4 on windows 2000 I want use // for annotation replace by /* */ How do ? Have a nice day !!! --------------------------- Newsgroups: comp.os.vxworks Subject: Re: serial i/o Date: Wed, 16 Jan 2002 15:45:57 +0000 From: Graham Baxter Organization: Graham Baxter (Software) Limited Message-ID: <3C45A035.33F9D1D5@NOSPAM.bcs.org.uk> References: <3c45413f@pull.gecm.com> <3C4561D2.2A12DFB8@NOSPAM.bcs.org.uk> <3c4570f5$1@pull.gecm.com> Look in sysSerial.c in your BSP. You will see which driver functions are being called. Regards, Graham Baxter (VxWorks and pSOS BSP'S) Freelance Software Engineer gbaxter@NOSPAM.bcs.org.uk --------------------------- Newsgroups: comp.os.vxworks Subject: driver for uart synchronous Date: Wed, 16 Jan 2002 16:03:01 +0000 (UTC) From: "Daniele" Organization: Mailgate.ORG Server - http://www.Mailgate.ORG Message-ID: <8a709af968c8df1ed5bf5be38ef4af71.60945@mygate.mailgate.org> Hello First my english is terrible...but I try I must use an uart in synchronous mode Before I used a driver for uart asynchronous mode with mbx860, all ok ! . Now I must use the uart in synchronous mode. I've set the register psmr[SYN]=1; gsmr_l[tenc,renc]=NRZ; gsmr_l[rdcr,tdcr]=1x i used an external clock Configured port A pins to enable the CLK3 pin. set PAPAR[5] and clear PADIR[5] Connect CLK3 to SCC2 using th SI register. set SICR[R2CS] and SICR[T2CS]. Connected the external clock to bsp and go the result is... :(((((( i do not see data in Tx pin are there driver for this modality ? - -- Posted via Mailgate.ORG Server - http://www.Mailgate.ORG --------------------------- Newsgroups: comp.os.vxworks Subject: prova Date: Wed, 16 Jan 2002 16:09:15 +0000 (UTC) From: "Daniele Rossi" Organization: Mailgate.ORG Server - http://www.Mailgate.ORG Message-ID: test... - -- Posted via Mailgate.ORG Server - http://www.Mailgate.ORG --------------------------- Newsgroups: comp.os.vxworks Subject: Re: SingleStep Date: Wed, 16 Jan 2002 16:24:06 GMT From: "Koen Van Hulle" Organization: Xeikon Message-ID: References: <3c42b08e$1@news.barak.net.il> Yep, I've been doing some basic debugging on my project using SDS 7.6.6, Tornado 2.0 on PPC 405 and am quite satisfied with the debugger (a hell of a lot better than the standard GNU debugger). I haven't been able to use it is system mode yet since the current version of the PPC405 BSP doesn't support it. Regards, Koen "Alex" wrote in message news:3c42b08e$1@news.barak.net.il... > Hello, > Has somebody experience in work with SingleStep 7.6.2. together with VxWorks > 5.4.? > Thank you. > > --------------------------- Newsgroups: comp.os.vxworks Subject: how use Time slot Assign ? Date: Wed, 16 Jan 2002 16:24:40 +0000 (UTC) From: "Daniele Rossi" Organization: Mailgate.ORG Server - http://www.Mailgate.ORG Message-ID: <84f4f52c9d9c545e4bbeebcd77e05bb1.60945@mygate.mailgate.org> I must use Time Slot Assign with 860 but i don't have any driver anybody used TDM with bsp860 can help me with some example ??? thanks Daniele - -- Posted via Mailgate.ORG Server - http://www.Mailgate.ORG --------------------------- Newsgroups: comp.os.vxworks Subject: telnet login on 2400s Date: 16 Jan 2002 08:32:28 -0800 From: vouxx@yahoo.com (vouxx) Organization: http://groups.google.com/ Message-ID: <86d4086f.0201160832.5b11256@posting.google.com> hi, has anyone else noticed it too? if i do crtl+c on a telnet login prompt, i get logged in. the vxworks simply restarts the shell. what could be wrong here? Connected to xxx (xxx). Escape character is '^]'. VxWorks login: 1a36fc vxTaskEntry +60 : shell () 18495c shell +18c: 184988 () 184a1c shell +24c: loginPrompt () 182338 loginPrompt +120: 181f38 () 181fdc loginUserShow +128: taskDelay () tShell restarted. - -> i tried this on 2300s and they work fine. - - vouxx --------------------------- Newsgroups: comp.os.vxworks Subject: Logging passwords with windweb Date: Wed, 16 Jan 2002 11:39:19 -0500 From: david lindauer Message-ID: <3C45ACB7.6050208@notifier-is.net> hi, we want to have a history of what users have viewed our WindWeb served HTML pages and when, based on the entered username/password. Is there any way to do this? Also we want to modify the content of a page based on whether the user has administrative access (as opposed to directing them to type in a different page we would like to add a link to an existing page and have it removed for users that don't have administrative access). Is this possible as well? Thanks, David --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Tornado 3 and Windows 2000 Date: Wed, 16 Jan 2002 09:26:42 -0800 From: "mc" Message-ID: <7Ii18.4289$oP4.36024@news.uswest.net> References: I had full admin privileges when I installed it as well as when I run it. "Dave Korn" wrote in message news:SIe18.544$qS6.1287000@newsr2.u-net.net... > mc wrote in message ... > >Hi all, > > > >I installed Tornado 3 on Windows 2000 SP1 w/IE 6 (Pentium 3) but when I try > >to launch it, nothing happens: no error message, no crash, nothing. I tried > >to reinstall it more than once, but it still doesn't work. Any idea? Thanks > >in advance. > >Regards, > > > >MC > > Perhaps a problem with permissions, were you logged in as Administrator > when you installed it? Are you logged in as Administrator when you try to > run it? If you were an ordinary user when you installed it, are you still > logged in as that same user? > > DaveK > -- > moderator of > alt.talk.rec.soc.biz.news.comp.humanities.meow.misc.moderated.meow > Burn your ID card! http://www.optional-identity.org.uk/ > Help support the campaign, copy this into your .sig! > Proud Member of the Exclusive "I have been plonked by Davee because he > thinks I'm interesting" List > Member # > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Tornado 3 and Windows 2000 Date: Wed, 16 Jan 2002 09:35:57 -0800 From: Lori Fraleigh Organization: Real-Time Innovations Message-ID: <3C45B9FD.6EE85D8B@rti.com> References: Have you ever run it successfully? If so, delete all the files in c:\.wind\tornado3.1 and try again. If you've never run it successfully, look in the files in that directory for errors. (and if you haven't upgraded from Tornado 3.0 to 3.1, I'd highly recommend it) There is an updated version of Java required for Pentium 4 computers. Maybe it is also required for your pentium 3? See WindSurf for details. Regards, Lori mc wrote: > > Hi all, > > I installed Tornado 3 on Windows 2000 SP1 w/IE 6 (Pentium 3) but when I try > to launch it, nothing happens: no error message, no crash, nothing. I tried > to reinstall it more than once, but it still doesn't work. Any idea? Thanks > in advance. > Regards, > > MC - -- Lori Fraleigh Director of Engineering Real-Time Innovations, Inc. http://www.rti.com Phone: 408-734-4200 x121 Fax: 408-734-5009 --------------------------- Newsgroups: comp.os.vxworks Subject: SNMP: nman application for management Date: Wed, 16 Jan 2002 19:39:14 +0200 From: "Ofer Goren" Organization: Verio Message-ID: Hi. Does anybody know where can I get the 'nman' aplication to manage an SNMP client? the 'nman' application is a command-line network management utility. Thanks, - -- Ofer Goren Iamba Technologies --------------------------- Newsgroups: comp.os.vxworks Subject: Re: DHCP over PPP Date: Wed, 16 Jan 2002 10:10:21 -0800 From: "Yusuf Motiwala" Organization: Texas Instruments Message-ID: References: You mean client right. Why do you need to implement, you can use WRS supplied DHCP client. However, by default it uses boot device to get DHCP lease which you need to change. Yusuf "Jason" wrote in message news:e555a185.0201141146.5099300b@posting.google.com... > I am going to implement the DHCP over PPP. By default, the vxWorks > only supports DHCP over Ethernet. It would be greatly appreciated if > anyone can give me some ideas or sample codes about DHCP over PPP. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Serial Port Configuration Date: 16 Jan 2002 10:38:26 -0800 From: vloscomp@yahoo.com (Vinh) Organization: http://groups.google.com/ Message-ID: References: Hi, I'm not quite sure of your board. However, in my limited experience, every board used different UART chip. Therefore, you will need to look into the chip manual and the corresponding BSP. Vinh Lam "Chad" wrote in message news:... > Is there a standard interface to check the number of parity bits, data bits, > stop bits etc that a serial driver is using? I can't seem to find any IOCTL > functions for checking/setting these values. --------------------------- Newsgroups: comp.os.vxworks Subject: WindNet PPP problem Date: 16 Jan 2002 11:27:26 -0800 From: agraham@openglobe.net (Aaron Graham) Organization: http://groups.google.com/ Message-ID: <95ebbe31.0201161127.64980d93@posting.google.com> Has anyone had a problem getting WindNet PPP to establish a connection? I can get it to dial out, and then I call pppConnectionOpen(), and nothing happens. I tried to set up the interface manually, but to no avail. I set up the stack and connection to call all of the callbacks that it possibly could (several dozen). However, the only callback that ever gets called is the stackAddComplete (meaning nothing, except that nothing has gone wrong yet). When I finally give up and close the connection, the pppClosed callback also gets called. No LCP, AUTH, NCP, or pppOpen callbacks are ever touched, which seems to me that PPP is just sitting on the connection, waiting for something ??? to happen. Since we are a customer of WRS, paying for WindNet PPP, we have access to source. However, all important function calls in the source code are via function pointers, and the unreasonable complexity of the code makes it all but impossible to follow. I would have thought that we could just set it up-- and not have to become experts in PPP, reading RFC's, manuals, books, scrounging through library/driver source code, and tinkering with and tweaking each of the hundreds of configuration parameters, just to get a connection. I called WRS support to see if they had sample code bringing up an interface. Obviously they must, I thought, since they probably TESTED their code before they sold it! So far, nothing... Anyone out there done this before and have some sample code? Aaron --------------------------- Newsgroups: comp.os.vxworks Subject: Support for 8 bits + even parity in ttyDrv? Date: Wed, 16 Jan 2002 19:28:46 GMT From: Jon Organization: EarthLink Inc. -- http://www.EarthLink.net Message-ID: <3C45D4D8.5FE5F442@bitworkssystemsDELETETHIS.com> I have a serial device that uses 8 bits plus even parity and at first glance didn't see away to set this in the tyLib ioctl options. Is there a way to get the ttyDrv to transmit and receive 1 start bit, 8 data bits, 1 even parity bit and 1 stop bit? Thats 11 bits total. Platform is Pentium running VxWorks AE 3.1 Jon Newbill Bitworks Systems Inc. jon @ bitworkssystemsNOSPAM.com --------------------------- Newsgroups: comp.os.vxworks Subject: Re: PMC module on MVME5100 (mv5100) Date: Wed, 16 Jan 2002 17:08:26 -0500 From: "William Dennen" Organization: Motorola Message-ID: References: Problem was at least side stepped by adding the PMC to the PCI AutoConfiguration exclusion list. >@< "BS" wrote in message news:d64227b2.0201151417.32ca5a1c@posting.google.com... > The problem I have is when I install a PMC module into my MVME5100 I > am unable to boot the system. The PMC module I have supports both > 33MHZ and 66 MHZ operation and work just fine on a MVME2700. > > Without the PMC card the system boots just fine. I have also been able > to boot the system with a PMC module that does not support 66MHZ > operations, so it doesn't seem that anything is wrong with the 5100 > itself. > > Has anyone seen similar problems like this that might be able offer a > suggestion for resolving this issue. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: How do I for annotation // Date: Wed, 16 Jan 2002 17:36:22 -0500 From: joe durusau Organization: Bellsouth.Net Message-ID: <3C460066.437F68F@bellsouth.net> References: Using C++, you can do that. It's illegal in C with the compiler version that comes with Tornado. Speaking only for myself, Joe Durusau it2xpert wrote: > Hi... > I use T2 Vxworks 5.4 on windows 2000 > I want use // for annotation replace by /* */ > How do ? > > Have a nice day !!! --------------------------- Newsgroups: comp.os.vxworks Subject: Re: taskdelay and interrupts Date: 16 Jan 2002 14:39:28 -0800 From: karahanduman@hotmail.com (Kara) Organization: http://groups.google.com/ Message-ID: References: <65b2abfc.0201150725.4f5f9903@posting.google.com> <3C4495A2.9564F1C4@digidata.com> Thanks to everybody involved, I have found the problem.. It was not printfs, I guess although they are not advisable they are ok. I was working on a Mips and for some reason VxWorks changes the value of SR (interrupt mask) register when a context switch occurs. Therefore SR is valid only in the task context. That's why my printf's were slow because the interrupt is enabled only when my task is active. The way around is the a call called taskSRSet(arg) which sets the default SR value for all tasks, it has to be made from BSP.. I appreciate your input and kind help. Have a good day.. Kara- Doug Dotson wrote in message news:<3C4495A2.9564F1C4@digidata.com>... > As far as I know, output to the serial port is interrupt driven in > most BSPs. So by defintion, serial output is impossible from > within an ISR using the normal mechanisms. You can use > logMsg() if timing is not a problem and you don't exit the > ISR. We have a similar problem in that we want to output > panic information from within an ISR when an unrecoverable > error happens. > > doug > > > Venkat Mynampati wrote: > > > Johan Borkhuis wrote in message news:... > > > karahanduman@hotmail.com (Kara) wrote: > > > > > > > If I comment out the taskDelay in the main loop and run it, I get the > > > > printf messages as caused by interrupts without any problem. When I > > > > run it with task delay(i.e. taskDelay(100000);), I don't get the > > > > printf messages. if I run the taskDelay with some small value (i.e. > > > > 100), I get the messages with a little delay. > > > > If I take out the while(1); loop and let it return to the shell, I > > > > don't get interrupts at all. > > > > > > NEVER use printf statements in an interrupt routine! The fact that you > > > were able to get output at all surprised me. > > > > > > There are more routines that you cannot call from within an interrupt > > > routine. Take a look at chapter 2.5 in the Programmer's Guide (Interrupt > > > Service Code), and espcially 2.5.3 (Special Limitations of ISRs) > > > > > > If you want to create output from within an ISR use logMsg instead of > > > printf. > > > > > > Groeten, > > > Johan > > > > Hi, > > I am relatively new to vxWorks anyway i will stick my neck out... > > Fisrt thing according to doc's is: "printf() and any call that blocks > > on a resource cannot be used in an interrupt." > > However, what would happen if i call printf() in an ISR, as is in this > > case, and assuming there are no other tasks or interrupt vieing for CPU, > > then, i think printf() would work, 'cause, the ISR blocks on some resource > > only when that resource is unavailable and what if that resource is available? > > > > Or how can we explain the fact that printf() worked when taskDelay() is less? > > > > Venkat --------------------------- Newsgroups: comp.os.vxworks Subject: Re: taskdelay and interrupts Date: 16 Jan 2002 17:56:36 -0500 From: Bill Pringlemeir Organization: Bell Sympatico Message-ID: References: <65b2abfc.0201150725.4f5f9903@posting.google.com> <3C4495A2.9564F1C4@digidata.com> Sender: bpringlemeir@DeadDuck >>>>> "Kara" == Kara writes: Kara> Thanks to everybody involved, I have found the problem.. It Kara> was not printfs, I guess although they are not advisable they Kara> are ok. I was working on a Mips and for some reason VxWorks Kara> changes the value of SR (interrupt mask) register when a Kara> context switch occurs. Therefore SR is valid only in the task Kara> context. That's why my printf's were slow because the interrupt Kara> is enabled only when my task is active. The way around is the a Kara> call called taskSRSet(arg) which sets the default SR value for Kara> all tasks, it has to be made from BSP.. I appreciate your input Kara> and kind help. Is this running in a Nuclear power plant control system or a heart monitor? Perhaps a child's toy? You did not listen to the people here... some of them are quite knowledgeable about vxWorks. The one who gave you the answer writes the FAQ for this newsgroup. !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! !!!!!!!!!!!YOU SHOULD NOT USE PRINTFS IN ISRS!!!!!!!!!!!!!!!!!!!!!!!!!! !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! That said, I will be kinder and point you to the vxWorks header file called . In it there is a macro called `sioPollOutput'. This is a function that can be used to print out data in an ISR, if you are desperate for such a thing. I don't know a good way to get ahold of the `SIO_CHAN' structure needed for these macros. However, you can change your BSP to expose the structure and use it to call these functions. This is safe to do in an interrupt. The reason you must mess around with the SR interrupt mask is because you should not be using printfs in an ISR. Or to say it another way, you should not be using printfs in an ISR. More over, you should not be using printfs in an ISR. However, you should not be using printfs in an ISR. And Finally, you should not be using printfs in an ISR. hth, Bill Pringlemeir. - -- SDI Peking North Korea spy colonel ammunition class struggle kibo PLO Rule Psix FBI South Africa Albanian Soviet $400 million in gold bullion vxWorks FAQ, "http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html" --------------------------- Newsgroups: comp.os.vxworks Subject: Re: taskdelay and interrupts Date: Wed, 16 Jan 2002 23:13:51 GMT From: "Jeff Creem" Organization: ATT Broadband Message-ID: References: <65b2abfc.0201150725.4f5f9903@posting.google.com> <3C4495A2.9564F1C4@digidata.com> And in case the prior poster was not clear..Don't use printfs in an ISR. "Bill Pringlemeir" wrote in message news:uy9ixx5uj.fsf@yahoo.com... > >>>>> "Kara" == Kara writes: > > Kara> Thanks to everybody involved, I have found the problem.. It > Kara> was not printfs, I guess although they are not advisable they > Kara> are ok. I was working on a Mips and for some reason VxWorks > Kara> changes the value of SR (interrupt mask) register when a > Kara> context switch occurs. Therefore SR is valid only in the task > Kara> context. That's why my printf's were slow because the interrupt > Kara> is enabled only when my task is active. The way around is the a > Kara> call called taskSRSet(arg) which sets the default SR value for > Kara> all tasks, it has to be made from BSP.. I appreciate your input > Kara> and kind help. > > Is this running in a Nuclear power plant control system or a heart > monitor? Perhaps a child's toy? You did not listen to the people > here... some of them are quite knowledgeable about vxWorks. The one > who gave you the answer writes the FAQ for this newsgroup. > > !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! > > !!!!!!!!!!!YOU SHOULD NOT USE PRINTFS IN ISRS!!!!!!!!!!!!!!!!!!!!!!!!!! > > !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! > > > > That said, I will be kinder and point you to the vxWorks header file > called . In it there is a macro called `sioPollOutput'. > This is a function that can be used to print out data in an ISR, if > you are desperate for such a thing. I don't know a good way to get > ahold of the `SIO_CHAN' structure needed for these macros. However, > you can change your BSP to expose the structure and use it to call > these functions. This is safe to do in an interrupt. > > The reason you must mess around with the SR interrupt mask is because > you should not be using printfs in an ISR. Or to say it another way, > you should not be using printfs in an ISR. More over, you should not > be using printfs in an ISR. However, you should not be using printfs > in an ISR. And Finally, you should not be using printfs in an ISR. > > hth, > Bill Pringlemeir. > > -- > SDI Peking North Korea spy colonel ammunition class struggle kibo PLO > Rule Psix FBI South Africa Albanian Soviet $400 million in gold > bullion > > vxWorks FAQ, "http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html" --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Support for 8 bits + even parity in ttyDrv? Date: 16 Jan 2002 16:06:57 -0800 From: john_94501@yahoo.com (John) Organization: http://groups.google.com/ Message-ID: <488e459a.0201161606.1b6c2db@posting.google.com> References: <3C45D4D8.5FE5F442@bitworkssystemsDELETETHIS.com> Hello, Take a look in the driver for your serial chip. For the standard pcPentiumX BSPs this is i8250Sio.c (under target/src/drv/sio). You will see in there that there is a function called i8250OptsSet() which is called as a result of an ioctl() on the device with the operation code SIO_HW_OPTS_SET. The argument of the ioctl() is passed through to i8250OptsSet(). The options you are interested in are PARENB and PARODD. Since you want even parity, you should only set the PARENB (parity enable), and not PARODD (select odd). These values are defined in sioLib.h, so you'll want to include that into whatever is making the ioctl() call. HTH, John... Jon wrote in message news:<3C45D4D8.5FE5F442@bitworkssystemsDELETETHIS.com>... > I have a serial device that uses 8 bits plus even parity and at first > glance didn't see away to set this in the tyLib ioctl options. Is there > a way to get the ttyDrv to transmit and receive 1 start bit, 8 data > bits, 1 even parity bit and 1 stop bit? Thats 11 bits total. > > Platform is Pentium running VxWorks AE 3.1 > > Jon Newbill > Bitworks Systems Inc. > jon @ bitworkssystemsNOSPAM.com --------------------------- Newsgroups: comp.os.vxworks Subject: Re: How do I for annotation // Date: Wed, 16 Jan 2002 16:29:20 -0800 From: "Michael R. Kesti" Organization: MK Associates Message-ID: <3C461AE0.9EEA48CE@gv.net> References: <3C460066.437F68F@bellsouth.net> Reply-To: mkesti@gv.net joe durusau wrote: >Using C++, you can do that. And there's nothing to stop a C programmer from using the C++ compiler, and, in addition to better type checking and such, enjoy using the C++ comment delimiter. All that one needs to do is change one's source files' name extension from ".c" to ".cpp". > It's illegal in C with the compiler version >that comes with Tornado. If memory serves (And it doesn't always!), this is configurable in the GNU C compiler using a command line switch. - -- ======================================================================== Michael Kesti | "And like, one and one don't make | two, one and one make one." mkesti@gv.net | - The Who, Bargain --------------------------- Newsgroups: comp.os.vxworks Subject: Re: How do I for annotation // Date: 16 Jan 2002 20:14:34 -0500 From: Bill Pringlemeir Organization: Bell Sympatico Message-ID: References: Sender: bpringlemeir@DeadDuck >>>>> "it2xpert" == it2xpert writes: it2xpert> Hi... I use T2 Vxworks 5.4 on windows 2000 I want use // it2xpert> for annotation replace by /* */ How do ? In a makefile, specify... ADDED_CFLAGS += -Wp,-lang-c In a project/GUI environment, click on the `build' tab open a tree and right click to select properties. From the properties click on the "C/C++ compiler" tab. Then enter the string "-Wp,-lang-c" somewhere in the edit box. [or you can use a makefile...] This command tells the C pre-processor to process C++ comments. Comments aren't really part of the translation phase of either C or C++. They should always be done by the pre-processor. Please check the FAQ entry at the bottom of the message. If the solution is not there, I would be very surprised. It probably should be... but you should look at it anyways it has good information. hth, Bill Pringlemeir. - -- Michael Jackson is mentally retarded, huh? Let me tell you something, you beast, just because you read it in the Daily Mirror that doesn't mean it's true. vxWorks FAQ, "http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html" --------------------------- Newsgroups: comp.os.vxworks Subject: Re: [HELP]UDP TX/RX packet mismatch problem... Date: Thu, 17 Jan 2002 11:00:39 +0900 From: "chihwankim" Organization: ELIMnet Internet News Message-ID: References: <3C3D7D66.2D999429@bellsouth.net> <3C3DB452.4DFBFCBF@l8s.co.uk> <3C3F1A42.C2014EE2@NOSPAM.bcs.org.uk> Oh, thanks for your comments. Would you let me know how to solve that probelm? Did you fix BSP or hardware? thanks. chkim "Graham Baxter" wrote in message news:3C3F1A42.C2014EE2@NOSPAM.bcs.org.uk... > I had a similar problem when I configured the TX and RX clocks the wrong > way around. > > They were close enough to each other to work most of the time! > > Packet loss was 3-4%. > > -- > > > Regards, > > > Graham Baxter (VxWorks and pSOS BSP'S) > Freelance Software Engineer > gbaxter@NOSPAM.bcs.org.uk --------------------------- Newsgroups: comp.os.vxworks Subject: torVars and make Date: 16 Jan 2002 18:23:20 -0800 From: welcome21cn@hotmail.com (yuhang) Organization: http://groups.google.com/ Message-ID: I have one question. when i use the command "c:\tornado\host\x86-win32\bin\torVars.bat",it appears " c:\tornado\host\x86-win32\bin\torvars c:\tornado\host\x86-win32\bin\rem Command Line Build Environments c:\tornado\host\x86-win32\bin\set WIND_HOST_TYPE=x86-win32 c:\tornado\host\x86-win32\bin\set WIND_BASE_=C:\Tornado out of the environment space c:\tornado\host\x86-win32\bin\set PATH=\host\\bin;D:\PROGRA~1\UEDIT;C :\WINDOWS;c:\WINDOWS\COMMAND out of the environment space and I wonder if any way to change it ,and in "c:\tornado\target\src\snmpv1\agent",use the command"make ", in' c:\tornado\target\src\snmpv1\agent make CPU=ARM7TDMI TOOL=gnu clean'", it does not go , "bad command or file name",why? Thankyou in advance! --------------------------- Newsgroups: comp.os.vxworks Subject: why "make "do not go? Date: 16 Jan 2002 18:32:25 -0800 From: welcome21cn@hotmail.com (yuhang) Organization: http://groups.google.com/ Message-ID: I have one question. when i use the command "c:\tornado\host\x86-win32\bin\torVars.bat",it appears " c:\tornado\host\x86-win32\bin\torvars c:\tornado\host\x86-win32\bin\rem Command Line Build Environments c:\tornado\host\x86-win32\bin\set WIND_HOST_TYPE=x86-win32 c:\tornado\host\x86-win32\bin\set WIND_BASE_=C:\Tornado out of the environment space c:\tornado\host\x86-win32\bin\set PATH=\host\\bin;D:\PROGRA~1\UEDIT;C :\WINDOWS;c:\WINDOWS\COMMAND out of the environment space and I wonder if any way to change it ,and in "c:\tornado\target\src\snmpv1\agent",use the command"make ", in' c:\tornado\target\src\snmpv1\agent make CPU=ARM7TDMI TOOL=gnu clean'", it does not go , "bad command or file name",why? Thankyou in advance! --------------------------- Newsgroups: comp.os.vxworks Subject: Re: flash and MMU Date: Thu, 17 Jan 2002 11:53:28 +0900 From: "chihwankim" Organization: ELIMnet Internet News Message-ID: References: Well, I think you should also update sysBatDesc[] in sysLib.c just above the sysPhysMemDesc []. Maybe cache for your flash is enabled there. So you can fix like following. As you can see MMU_LBAT_CACHE_INHIBIT is set for flash ROM. Try it. UINT32 sysBatDesc [2 * (_MMU_NUM_IBAT + _MMU_NUM_DBAT)] = { /* I BAT 0 */ ((ROM_BASE_ADRS & _MMU_UBAT_BEPI_MASK) | _MMU_UBAT_BL_2M | _MMU_UBAT_VS | _MMU_UBAT_VP), ((ROM_BASE_ADRS & _MMU_LBAT_BRPN_MASK) | _MMU_LBAT_PP_RW | _MMU_LBAT_CACHE_INHIBIT), /* I BAT 1 */ 0, 0, /* I BAT 2 */ 0, 0, /* I BAT 3 */ 0, 0, /* D BAT 0 */ ((ROM_BASE_ADRS & _MMU_UBAT_BEPI_MASK) | _MMU_UBAT_BL_2M | _MMU_UBAT_VS | _MMU_UBAT_VP), ((ROM_BASE_ADRS & _MMU_LBAT_BRPN_MASK) | _MMU_LBAT_PP_RW | _MMU_LBAT_CACHE_INHIBIT), /* D BAT 1 */ 0, 0, /* D BAT 2 */ 0, 0, /* D BAT 3 */ 0, 0 }; "ruready" wrote in message news:ae663a93.0201101832.26760eb5@posting.google.com... > Why I cannot read or write flash, when I enable MMU in BSP? > When reading flash, I can only get the data of 0xff. > But if I disable the MMU, everything is OK. > That's what I write in sysPhysMemDesc[] > PHYS_MEM_DESC sysPhysMemDesc [] = > { > { > /* Vector Table and Interrupt Stack */ > > (void *) LOCAL_MEM_LOCAL_ADRS, > (void *) LOCAL_MEM_LOCAL_ADRS, > RAM_LOW_ADRS, > VM_STATE_MASK_VALID | VM_STATE_MASK_WRITABLE | VM_STATE_MASK_CACHEABLE, > VM_STATE_VALID | VM_STATE_WRITABLE | VM_STATE_CACHEABLE_NOT > }, > > { > /* RAM on 60x bus - Must be second entry for Auto Sizing */ > > (void *) RAM_LOW_ADRS, > (void *) RAM_LOW_ADRS, > LOCAL_MEM_SIZE - RAM_LOW_ADRS, > VM_STATE_MASK_VALID | VM_STATE_MASK_WRITABLE | VM_STATE_MASK_CACHEABLE | > VM_STATE_MASK_MEM_COHERENCY, > VM_STATE_VALID | VM_STATE_WRITABLE | VM_STATE_CACHEABLE | > VM_STATE_MEM_COHERENCY > }, > > { > /* Board Control and Status Registers */ > > (void *) BCSR_BASE_ADRS, > (void *) BCSR_BASE_ADRS, > BCSRS_SIZE, > VM_STATE_MASK_VALID | VM_STATE_MASK_WRITABLE | VM_STATE_MASK_CACHEABLE | > VM_STATE_MASK_GUARDED, > VM_STATE_VALID | VM_STATE_WRITABLE | VM_STATE_CACHEABLE_NOT | > VM_STATE_GUARDED > }, > > { > /* MPC8260 Internal Memory Map */ > > (void *) DEFAULT_IMM_ADRS, > (void *) DEFAULT_IMM_ADRS, > IMM_SIZE, > VM_STATE_MASK_VALID | VM_STATE_MASK_WRITABLE | VM_STATE_MASK_CACHEABLE | > VM_STATE_MASK_GUARDED, > VM_STATE_VALID | VM_STATE_WRITABLE | VM_STATE_CACHEABLE_NOT | > VM_STATE_GUARDED > }, > > { > /* Flash */ > > (void *) ROM_BASE_ADRS, > (void *) ROM_BASE_ADRS, > ROM_SIZE, > VM_STATE_MASK_VALID | VM_STATE_MASK_WRITABLE | VM_STATE_MASK_CACHEABLE | > VM_STATE_MASK_GUARDED, > VM_STATE_VALID | VM_STATE_WRITABLE | VM_STATE_CACHEABLE_NOT | > VM_STATE_GUARDED > } > }; > > Could someone do me a faver helping me to solve this problem? --------------------------- Newsgroups: comp.os.vxworks Subject: Who can give me the source code of pppoe on vxworks? Date: Thu, 17 Jan 2002 11:43:56 +0800 From: "5982" <5982851@ihw.com.cn> Organization: Bentium Ltd. (CN99) Message-ID: Who can give me the source code of pppoe on vxworks? --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Hi All Date: 16 Jan 2002 20:06:37 -0800 From: aravinddk@tataelxsi.co.in (aravind) Organization: http://groups.google.com/ Message-ID: References: Hi, When we try to use InputQueuePoll and Build the File We always get the undefined symbol for InputQueuePoll when we look into the symbol table and so cannot download the executable to the target board. I have included all the header files as in the v_keybrd.cpp. Is there anything else to be included? Regards Arvind D.K Bill Pringlemeir wrote in message news:... > >>>>> "aravind" == aravind writes: > > aravind> Hi, We are trying to capture the ppc keyboard key > aravind> presses. We tried the Event function with which we could not > > [snip] > > aravind> running and which has called the Control () of the Zinc > aravind> Event function. So the first key press is routed to the > aravind> Control () loop and the second key press is routed to the > aravind> uglInputQStandardQEventGet(). > > aravind> Is there a soultion to this problem ? OR How to make the > aravind> existing Control () loop (in another application) to > aravind> recognise the key presses in our application. The two are > aravind> linked during compilation. > > Much closer to giving enough information to solve this problem. > However, I still don't understand why you wish to do the UGL key board > processing. If you have Zinc (for vxWorks/WindML) then in the > directory ../target/src/zinc/vxworks, there will be a file > v_keybrd.cpp. This file is generally responsible for placing events > on the Zinc device queue [see EventPoll() in v_osevnt.cpp and Get() in > v_eventmgr.cpp]. > > The standard Zinc 6.0/WinML 2.0 [what are you using?] use the WindML > windows facility, which use the callback mechanism to put events on > the Zinc queue. You should be able to pend multiple task on the queue > and each will get the key. Alternatively, I have hooked the keyboard > driver to use a totally different mechanism. You just open > "/dev/keyboard" and read from it (with select maybe) and post this to > whatever queue you like. At any rate, WindML support multiple queues; > Zinc does not. I still don't get what you are trying to do. You need > to show a system diagram of some sort. There maybe a better design. > > hth, > Bill Pringlemeir. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Is that bug of MUX or mine ? Date: 16 Jan 2002 21:33:14 -0800 From: houcheng@ms17.hinet.net (houcheng) Organization: http://groups.google.com/ Message-ID: References: <4ceedd43.0112210136.62edbe0d@posting.google.com> Thanks all, I've found the reason. The nptLoad function is invoked by muxDevLoad two times, In the first time, the nptLoad function is calling to ask the driver's name, so the nptLoad function must fill driver's name into initString and return NULL. In the second time, the nptLoad is called and perform the real initialization. Note: the templateEnd.c provide by WR$ is not correct! Its templateLoad missing a piece of code to process the first time invocation. The SENS developer do not mention about it!!! END_OBJ * sarLoad(char * initString) { if(initString) { if(initString[0] == 0) { bcopy("sar", initString, 4); return(NULL); } } .... /* driver initialization */ } --------------------------- Newsgroups: comp.os.vxworks Subject: Re: torVars and make Date: Thu, 17 Jan 2002 07:03:11 +0100 From: "Martin Raabe" Organization: T-Online Message-ID: References: Hello yuhang, the "_" is too much in the line: c:\tornado\host\x86-win32\bin\set WIND_BASE_=C:\Tornado It should be: c:\tornado\host\x86-win32\bin\set WIND_BASE=C:\Tornado Are you using an alternative shell instead of the cmd.exe or command.com? If yes, check the correctness of the syntax in torvars.bat for your shell. Usually you do a call torvars.bat. Did you use this mechanism? Hope it helps! BTW: Did you manually edit the torvars.bat file? Ciao Martin Raabe "yuhang" schrieb im Newsbeitrag news:bffd1323.0201161823.3471df94@posting.google.com... > I have one question. > when i use the command "c:\tornado\host\x86-win32\bin\torVars.bat",it > appears > " > c:\tornado\host\x86-win32\bin\torvars > c:\tornado\host\x86-win32\bin\rem Command Line Build Environments > c:\tornado\host\x86-win32\bin\set WIND_HOST_TYPE=x86-win32 > c:\tornado\host\x86-win32\bin\set WIND_BASE_=C:\Tornado > out of the environment space > c:\tornado\host\x86-win32\bin\set PATH=\host\\bin;D:\PROGRA~1\UEDIT;C > :\WINDOWS;c:\WINDOWS\COMMAND > out of the environment space > and I wonder if any way to change it ,and in > "c:\tornado\target\src\snmpv1\agent",use the command"make ", > in' c:\tornado\target\src\snmpv1\agent make CPU=ARM7TDMI TOOL=gnu > clean'", > it does not go , > "bad command or file name",why? > > Thankyou in advance! --------------------------- Newsgroups: comp.os.vxworks Subject: Uninitialized Interrupt! Vector number 47 Date: 16 Jan 2002 22:06:07 -0800 From: nukala_s@yahoo.com (Srinivas) Organization: http://groups.google.com/ Message-ID: I am using VxWorks for PentiumPro with Intel PRO/100B. When I am doing continous read/write over the network I get the following message: Uninitialized Interrupt! Vector number 47 (0-255) .... .... If I am not mistaken this vector number corresponds to IRQ 15. My Intel PRO/100B NIC card is assigned IRQ 11. But I suspect that the occurance of this spurious interrupt has something to do with the NIC card. I would like to know if anyone had similar problems and already have a solution for it. I would greatly appreciate any pointers. Thanks, - --Srinivas. --------------------------- Newsgroups: comp.os.vxworks Subject: mpc860 interrupt debug problem "uninitialized interrupt"? Date: 16 Jan 2002 23:19:53 -0800 From: yong_bvcom@sina.com (bigtornado) Organization: http://groups.google.com/ Message-ID: <5624c1bf.0201162319.ebe6cd6@posting.google.com> hi,all: I want connect my routine to mpc860 extern interrupt IRQ1,as follows: intConnect(INUM_TO_IVEC(2),MYISR,0); intEnable(2); but when interrupt on IRQ1 occurs,system tell me "uninitizlized interrupt",so I debug interrupt driver ppc860intr.c,as follows: STATUS ppc860IntrInit ( VOIDFUNCPTR * cpmIntrVec /* Intr level of the CPM Intr ctrl */ ) { VOIDFUNCPTR defaultVec; /* INTR3 default vector */ UINT32 regBase; /* device register base address */ UINT32 cicrIntLevel; /* CICR interupt level */ UINT32 simaskIntLevel; int vector; /* Get the default vector connected to the External Interrupt (0x500) */ defaultVec = (VOIDFUNCPTR) excVecGet ((FUNCPTR *) _EXC_OFF_INTR); /* Connect the interrupt demultiplexer to External Interrupt (0x500) */ excIntConnect ((VOIDFUNCPTR *) _EXC_OFF_INTR, ppc860IntrDeMux); /* Install `system' intConnect routine */ if (_func_intConnectRtn == NULL) _func_intConnectRtn = ppc860IntConnect; if (_func_intEnableRtn == NULL) _func_intEnableRtn = ppc860IntEnable; if (_func_intDisableRtn == NULL) _func_intDisableRtn = ppc860IntDisable; /* Set all vectors to default handler */ for (vector = 0; vector < NUM_VEC_MAX; vector++) intConnect (INUM_TO_IVEC(vector), defaultVec, 0); /* Get the CPM interrupt level */ switch (IVEC_TO_INUM(cpmIntrVec)) { case IVEC_TO_INUM(IV_LEVEL0): cicrIntLevel = CICR_IRL_LVL0; simaskIntLevel = SIMASK_LVM0; break; case IVEC_TO_INUM(IV_LEVEL1): cicrIntLevel = CICR_IRL_LVL1; simaskIntLevel = SIMASK_LVM1; break; case IVEC_TO_INUM(IV_LEVEL2): cicrIntLevel = CICR_IRL_LVL2; simaskIntLevel = SIMASK_LVM2; break; case IVEC_TO_INUM(IV_LEVEL3): cicrIntLevel = CICR_IRL_LVL3; simaskIntLevel = SIMASK_LVM3; break; case IVEC_TO_INUM(IV_LEVEL4): cicrIntLevel = CICR_IRL_LVL4; simaskIntLevel = SIMASK_LVM4; break; case IVEC_TO_INUM(IV_LEVEL5): cicrIntLevel = CICR_IRL_LVL5; simaskIntLevel = SIMASK_LVM5; break; case IVEC_TO_INUM(IV_LEVEL6): cicrIntLevel = CICR_IRL_LVL6; simaskIntLevel = SIMASK_LVM6; break; case IVEC_TO_INUM(IV_LEVEL7): cicrIntLevel = CICR_IRL_LVL7; simaskIntLevel = SIMASK_LVM7; break; default: return (ERROR); } /* * Connect the CPM interrupt demultiplexer at the level specified by * `cpmIntLevel' */ #ifdef INCLUDE_WINDVIEW cpmVecNum = IVEC_TO_INUM(cpmIntrVec); #endif /* INCLUDE_WINDVIEW */ intConnect (cpmIntrVec, ppc860CpmIntrDeMux, simaskIntLevel); /* Get the register base address */ regBase = vxImmrGet(); /* * Set the request interrupt level from the CPM and enable CPM interrupt * generation. */ * CICR(regBase) = (CICR_SCCP_DEF | cicrIntLevel | CICR_IEN | CICR_HP_MSK); /* Enable required CPM interrupt */ * SIMASK(regBase) = simaskIntLevel; return (OK); } is it correct? Anyone's suggestions are appreciated! --------------------------- Newsgroups: comp.os.vxworks Subject: Problem of HTTP Server for vxworks on MPC860 Date: 17 Jan 2002 00:10:04 -0800 From: e-tonyc@163.net (Tonyc) Organization: http://groups.google.com/ Message-ID: Hi all, I have written a simple Http Server for Vxworks on MPC 860, but I met a problem, it "sleeps" for a period between 30 sec to 1 min after 12 requests, and it begins to response again after that. Though I can ping the server successfully, I guess it must be somthing wrong with TCP of my http server, is there any one who can help me? Thanks in advance! Tonyc --------------------------- Newsgroups: comp.os.vxworks Subject: Re: How do I for annotation // Date: 17 Jan 2002 08:09:27 GMT From: Johan Borkhuis Organization: Agere Systems Message-ID: References: "it2xpert" wrote: > Hi... > I use T2 Vxworks 5.4 on windows 2000 > I want use // for annotation replace by /* */ > How do ? Remove the -ansi switch from the compiler options. Groeten, Johan - -- o o o o o o o . . . _____________________________ o _____ || Johan Borkhuis | .][__n_n_|DD[ ====_____ | borkhuis@agere.com | >(________|__|_[_________]_|__________________________| _/oo OOOOO oo` ooo ooo 'o!o!o o!o!o` === VxWorks FAQ: http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html === --------------------------- Newsgroups: comp.os.vxworks Subject: Re: why "make "do not go? Date: 17 Jan 2002 08:41:33 GMT From: Denis Perelyubskiy Message-ID: References: * yuhang [01-Wed-02 18:32 -0800]: > > I have one question. > when i use the command "c:\tornado\host\x86-win32\bin\torVars.bat",it > appears > " > c:\tornado\host\x86-win32\bin\torvars > c:\tornado\host\x86-win32\bin\rem Command Line Build Environments > c:\tornado\host\x86-win32\bin\set WIND_HOST_TYPE=x86-win32 > c:\tornado\host\x86-win32\bin\set WIND_BASE_=C:\Tornado > out of the environment space [...] so, why not use an excellent tool: google, and type in "out of the environment space"? .... try it, the results may just surprise you :) i just tried, and solutions depend on your operating system, but all seem to talk about passing a flag to command.com in any case, go search in google, and you shall find your answer. denis - -- if you want to reply by mail, please tweak my address in a reasonable way. sorry for the inconvenience. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: _protoSwIndex and inetsw[ ] used in our OSPF implementation Date: 17 Jan 2002 00:53:45 -0800 From: lihuapeng@21cn.com (DragonSpring) Organization: http://groups.google.com/ Message-ID: <1d411918.0201170053.b929e4e@posting.google.com> References: <1d411918.0201091833.3205a89f@posting.google.com> Why can not get OSPF packets via inetsw[IP_PROTO_NUM_MAX]? Is there any other method to get the OSPF packet from all END-drivered network interfaces besides _protoSwIndex and inetsw[]? Please don't be silent! Any information is valueable, and you will be appreciated. Thanks! lihuapeng@21cn.com (DragonSpring) wrote in message news:<1d411918.0201091833.3205a89f@posting.google.com>... > Hi, > > A question about _protoSwIndex and > inetsw[IP_PROTO_NUM_MAX].IP_PROTO_NUM_MAX is defined as 8 in VxWorks. > > Environment: target: PPC860 rev. D4 > Host: WinNT 4.0, Tornado 2.0/VxWorks 5.4 for PPC, Tornado > 2.0 Cumulative Patch 2 (SPR T2CP2), NPT 5.4(TDK-13274), NPT patch(SPR > 30894) > > We are developing our OSPF protocol (rfc2328). We want to get all OSPF > packet from IP layer. So we use the code as following: > > int ospf_init() > { > FAST struct protosw *p_ospf_protocol_switch; > > if (_protoSwIndex >= ((sizeof(inetsw))/(sizeof(inetsw[0])))) > return (-1) ; > > p_ospf_protocol_switch = &inetsw [_protoSwIndex]; > > if (p_ospf_protocol_switch == NULL) > return (-1); > > if (p_ospf_protocol_switch->pr_domain != NULL) > return (1); /* already initialized */ > > p_ospf_protocol_switch->pr_type = SOCK_RAW; > p_ospf_protocol_switch->pr_domain = &inetdomain; > p_ospf_protocol_switch->pr_protocol = IPPROTO_OSPF; > p_ospf_protocol_switch->pr_flags = PR_ATOMIC|PR_ADDR; > p_ospf_protocol_switch->pr_input = ospf_receive; > p_ospf_protocol_switch->pr_output = ospf_send; > p_ospf_protocol_switch->pr_ctlinput = raw_ctlinput; > p_ospf_protocol_switch->pr_ctloutput = rip_ctloutput; > p_ospf_protocol_switch->pr_usrreq = rip_usrreq; > p_ospf_protocol_switch->pr_init = 0 ; > p_ospf_protocol_switch->pr_fasttimo = 0; > p_ospf_protocol_switch->pr_slowtimo = 0; > p_ospf_protocol_switch->pr_drain = 0; > p_ospf_protocol_switch->pr_sysctl = 0; > > _protoSwIndex++; > > ... > } > > void ospf_receive(struct mbuf *m) > { > /* determine which interface this OSPF packet is received from */ > ... > } > > void ospf_send(struct mbuf *m) > { > ... > } > > We invoke ospf_init() in usrAppInit() (which is in usrAppInit.c) to > start the OSPF process. > > Now my problem is that I can never get any OSPF packet in > ospf_recieve(). It seems that the IP protocol stack never call > ospf_receive(). OSPF takes inetsw[7]. inetsw[IP_PROTO_NUM_MAX] seems > take no effect. Why? Does the IP protocol discard the OSPF packet? Is > there any problem in the VxWorks configurtion? BTW, we have observed > the motFecEnd.c (END driver) has already passed the OSPF packet to MUX > layer. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: [HELP]UDP TX/RX packet mismatch problem... Date: Thu, 17 Jan 2002 08:56:57 +0000 From: Graham Baxter Organization: Graham Baxter (Software) Limited Message-ID: <3C4691D9.177D7991@NOSPAM.bcs.org.uk> References: <3C3D7D66.2D999429@bellsouth.net> <3C3DB452.4DFBFCBF@l8s.co.uk> <3C3F1A42.C2014EE2@NOSPAM.bcs.org.uk> I just had to modify the CMXFCR register to set the correct clocks for my hardware. The BSP used CLK10 for TX and CLK9 for RX. My hardware had the clocks the other way around. Regards, Graham Baxter (VxWorks and pSOS BSP'S) Freelance Software Engineer gbaxter@NOSPAM.bcs.org.uk --------------------------- Newsgroups: comp.os.vxworks Subject: Re: telnet login on 2400s Date: 17 Jan 2002 09:28:36 GMT From: Johan Borkhuis Organization: Agere Systems Message-ID: References: <86d4086f.0201160832.5b11256@posting.google.com> vouxx@yahoo.com (vouxx) wrote: > has anyone else noticed it too? if i do crtl+c on a telnet login > prompt, i get logged in. the vxworks simply restarts the shell. what > could be wrong here? I did not notice it before, but the same happens with the MIPS32 telnet (arch = MIPS32sfgnu). This is a nice security hole! Did you already report it to WindRiver? Groeten, Johan - -- o o o o o o o . . . _____________________________ o _____ || Johan Borkhuis | .][__n_n_|DD[ ====_____ | borkhuis@agere.com | >(________|__|_[_________]_|__________________________| _/oo OOOOO oo` ooo ooo 'o!o!o o!o!o` === VxWorks FAQ: http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html === --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Waiting on multiple events Date: 17 Jan 2002 01:50:21 -0800 From: niall.leonard@ncr.com (Niall Leonard) Organization: http://groups.google.com/ Message-ID: <14e76c83.0201170150.1a4f6ae5@posting.google.com> References: <14e76c83.0201150158.696e793f@posting.google.com> <488e459a.0201151752.6556a99d@posting.google.com> In this particular case, I used a message queue instead of semaphores. The other tasks write a byte to the queue indicating which one is 'ready'. The task waiting then simply has to look at the content of the message to find out which event occured. It might not be the best way, but in this case it worked. Thanks for the suggestions. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Who can give me the source code of pppoe on vxworks? Date: 17 Jan 2002 10:17:39 GMT From: Johan Borkhuis Organization: Agere Systems Message-ID: References: "5982" <5982851@ihw.com.cn> wrote: > Who can give me the source code of pppoe on vxworks? WindRiver can, if you pay them. Groeten, Johan - -- o o o o o o o . . . _____________________________ o _____ || Johan Borkhuis | .][__n_n_|DD[ ====_____ | borkhuis@agere.com | >(________|__|_[_________]_|__________________________| _/oo OOOOO oo` ooo ooo 'o!o!o o!o!o` === VxWorks FAQ: http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html === --------------------------- Newsgroups: comp.os.vxworks Subject: Re: torVars and make Date: 17 Jan 2002 10:16:51 GMT From: Johan Borkhuis Organization: Agere Systems Message-ID: References: welcome21cn@hotmail.com (yuhang) wrote: > I have one question. > when i use the command "c:\tornado\host\x86-win32\bin\torVars.bat",it > appears > " > c:\tornado\host\x86-win32\bin\torvars > c:\tornado\host\x86-win32\bin\rem Command Line Build Environments > c:\tornado\host\x86-win32\bin\set WIND_HOST_TYPE=x86-win32 > c:\tornado\host\x86-win32\bin\set WIND_BASE_=C:\Tornado > out of the environment space > c:\tornado\host\x86-win32\bin\set PATH=\host\\bin;D:\PROGRA~1\UEDIT;C >:\WINDOWS;c:\WINDOWS\COMMAND > out of the environment space > and I wonder if any way to change it ,and in > "c:\tornado\target\src\snmpv1\agent",use the command"make ", > in' c:\tornado\target\src\snmpv1\agent make CPU=ARM7TDMI TOOL=gnu > clean'", > it does not go , > "bad command or file name",why? The answer is in the output of torVars.bat: you are out of environment space. You have to increase it for the extra variables to be set and to adjust the path. If you fix that you can run the make command. Groeten, Johan - -- o o o o o o o . . . _____________________________ o _____ || Johan Borkhuis | .][__n_n_|DD[ ====_____ | borkhuis@agere.com | >(________|__|_[_________]_|__________________________| _/oo OOOOO oo` ooo ooo 'o!o!o o!o!o` === VxWorks FAQ: http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html === --------------------------- Newsgroups: comp.os.vxworks Subject: Re: driver for uart synchronous Date: Thu, 17 Jan 2002 10:26:54 +0000 From: Graham Baxter Organization: Graham Baxter (Software) Limited Message-ID: <3C46A6EE.FCDCF173@NOSPAM.bcs.org.uk> References: <8a709af968c8df1ed5bf5be38ef4af71.60945@mygate.mailgate.org> Daniele, Did you set ENR and ENT in GSMR_L? Regards, Graham Baxter (VxWorks and pSOS BSP'S) Freelance Software Engineer gbaxter@NOSPAM.bcs.org.uk --------------------------- Newsgroups: comp.os.vxworks Subject: Problem with font in Tornado 2 Date: Thu, 17 Jan 2002 14:07:29 +0300 From: "Shrike" Organization: MTU-Intel ISP Message-ID: How can I change font in Tornado 2 editor? Or I will be blind soon. --------------------------- Newsgroups: comp.os.vxworks Subject: Problem with font in Tornado 2 - Part 2 Date: Thu, 17 Jan 2002 14:12:55 +0300 From: "Shrike" Organization: MTU-Intel ISP Message-ID: I know about Font/Size tab in Options, but there is a blank space in size selector. And nothing can be written there. --------------------------- Newsgroups: comp.os.vxworks Subject: Using a PCI ethernet adapter on PCPentium BSP Date: 17 Jan 2002 09:13:55 GMT From: Charly Message-ID: <2002117-101355-104040@foorum.com> Good morning everybody, Is it possible to use a PCI ethernet adapter with the PCPentium BSP? I would like to use "EDIMAX EN-5200PLT". Did someone use it? Thanks a lot charly - -- Use our news server 'news.foorum.com' from anywhere. More details at: http://nnrpinfo.go.foorum.com/ --------------------------- Newsgroups: comp.os.vxworks Subject: Re: driver for uart synchronous Date: Thu, 17 Jan 2002 11:43:47 +0000 (UTC) From: "Daniele Rossi" Organization: Mailgate.ORG Server - http://www.Mailgate.ORG Message-ID: <6924389e9f7583fc7715253abbda3cdd.60945@mygate.mailgate.org> References: <8a709af968c8df1ed5bf5be38ef4af71.60945@mygate.mailgate.org> <3C46A6EE.FCDCF173@NOSPAM.bcs.org.uk> yes. ENR and ENT in GSMR_L are ok. Regards, Daniele Marconi Mobile "Graham Baxter" wrote in message news:3C46A6EE.FCDCF173@NOSPAM.bcs.org.uk... > Daniele, > > Did you set ENR and ENT in GSMR_L? > > > Regards, > > > Graham Baxter (VxWorks and pSOS BSP'S) > Freelance Software Engineer > gbaxter@NOSPAM.bcs.org.uk - -- Posted via Mailgate.ORG Server - http://www.Mailgate.ORG --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Visual Studio Integration Date: Thu, 17 Jan 2002 11:52:09 -0000 From: "Dave Korn" Organization: Lumber Cartel (tinlc) Members #2234-2237 (owing to browser refresh) Message-ID: References: Chad wrote in message ... >Hi, > >I got a sample makefile from the TornadoII FAQ I'm working with to try and >allow me to browse and compile my ppc project source from visual studio. >Does anybody know where I can get more info on doing this? Johan's FAQ has a section on that. http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html DaveK - -- moderator of alt.talk.rec.soc.biz.news.comp.humanities.meow.misc.moderated.meow Burn your ID card! http://www.optional-identity.org.uk/ Help support the campaign, copy this into your .sig! Proud Member of the Exclusive "I have been plonked by Davee because he thinks I'm interesting" List Member # --------------------------- End of New-News digest ********************** From vxwexplo-errs@csg.lbl.gov Thu Jan 17 17:55:40 2002 From: "Dean Thompson" Date: Thu Jan 17 17:55:42 PST 2002 Subject: Tornado and Windows XP Anyone have any luck getting Tornado 2 running on a Windows XP host? ===================================================== T h e P T R G r o u p, I n c. ===================================================== Embedded, Real-Time Solutions and Training Dean Thompson mailto:dean@ThePTRGroup.com Director of Engineering http://www.ThePTRGroup.com V: (703) 858-4723 F: (703) 729-5499 ===================================================== From vxwexplo-errs@csg.lbl.gov Fri Jan 18 04:03:24 2002 From: Vxworks Exploder Date: Fri Jan 18 04:03:27 PST 2002 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Fri Jan 18 04:03:18 PST 2002 Subject: POSIX timer handler not executed reliably Subject: VxWorks, Walnut 405GP tftp download error Subject: Re: How do I for annotation // Subject: Re: taskdelay and interrupts Subject: Re: taskdelay and interrupts Subject: Re: Problem with font in Tornado 2 - Part 2 Subject: linking problem Subject: trapping alphanumeric keys Subject: Re: Using a PCI ethernet adapter on PCPentium BSP Subject: mem_part structure and it's meaning. Subject: Re: trapping alphanumeric keys Subject: x86 mmu and physical memory Subject: Re: How do I for annotation // Subject: Re: linking problem Subject: Interrupt Latency Problem!!! Subject: vxsim from Tornado 2 on Windows XP Subject: Re: POSIX timer handler not executed reliably Subject: Re: VxWorks, Walnut 405GP tftp download error Subject: Re: linking problem Subject: pppInit() (IPCP address negotiation with PPP) Subject: Re: Interrupt Latency Problem!!! Subject: Re: x86 mmu and physical memory Subject: Adding a toolchain to T2 Subject: Re: linking problem Subject: Re: Symbol table errors compiling an image when using STL Subject: Re: Support for 8 bits + even parity in ttyDrv? Subject: Re: Hi All Subject: Getting UART errors from i8250Sio driver Subject: Machine Check Exception Subject: dhcp client options in boot Subject: Re: Machine Check Exception Subject: help urgent.... how to enable MULTICAST flag.... Subject: Re: Who can give me the source code of pppoe on vxworks? Subject: Re: POSIX timer handler not executed reliably Subject: GPIB Driver Subject: help urgent.... how to enable MULTICAST flag.... Subject: why extern interrupt cause vxworks reboot?mpc860 Subject: Re: Symbol table errors compiling an image when using STL Subject: Re: Symbol table errors compiling an image when using STL Subject: Re: why extern interrupt cause vxworks reboot?mpc860 Subject: Re: mpc850 upmx program question Subject: workQPanic: Kernel work queue overflow PROBLEM Subject: problem while receiving from my END driver Subject: Re: Machine Check Exception Subject: Re: why extern interrupt cause vxworks reboot?mpc860 Subject: Couldn't open C:\TORNADO\kuku Subject: Re: Machine Check Exception Subject: Re: _protoSwIndex and inetsw[ ] used in our OSPF implementation Subject: Re: VxWorks, Walnut 405GP tftp download error Subject: Re: /var/tmp/....errors while compling assembly routines Subject: Re: pppInit() (IPCP address negotiation with PPP) Subject: Re: workQPanic: Kernel work queue overflow PROBLEM ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: POSIX timer handler not executed reliably Date: 17 Jan 2002 04:51:26 -0800 From: a16157@email.mot.com (Li Xin) Organization: http://groups.google.com/ Message-ID: <5c83e0.0201170451.79fd40ec@posting.google.com> For the portability reason, I am using the POSIX timer in my driver code. When the timer expires, the timer handler routine gets executed sometimes but does not at other times. Is it possible that vxWorks doesn't execute the POSIX timer handler even the timer expires? Thanks in advance. --------------------------- Newsgroups: comp.os.vxworks,comp.sys.powerpc.tech Subject: VxWorks, Walnut 405GP tftp download error Date: 17 Jan 2002 05:06:10 -0800 From: arun4vxworks@indiatimes.com (Arun Prasad) Organization: http://groups.google.com/ Message-ID: Hi, We are using Vxworks on IBM's Walnut PPC 405GP board. The bootrom on it is the same one which came with this board. We are trying to boot the system(walnut) using TFTP and download the vxWorks image. The following error message is displayed: >Not a valid boot image file We got this sequence a couple of times too: Loading file "/tftpboot/mydir1/vxWorks" ... Sending tftp boot request ... Unknown destination, block 1 Unknown destination, block 1 Unknown destination, block 1 Unknown destination, block 1 Unknown destination, block 1 Unknown destination, block 1 Unknown destination, block 1 Unknown destination, block 1 Unknown destination, block 1 TFTP boot failed. I am looking for info on how to build a tftp downloadable vxworks image for the Walnut 405GP system. Is it looking for some thing specific at some specific location ... Does the file type i.e., Binary or Hex matter? Do we need use elfHex or linker with any specific options, or any changes to the config? Looking forward for an early reply. Thanks in advance, - -- --------------------------- Newsgroups: comp.os.vxworks Subject: Re: How do I for annotation // Date: 17 Jan 2002 05:26:27 -0800 From: sczhang@esteemnetworks.com (ShiChao Zhang) Organization: http://groups.google.com/ Message-ID: <4ceedd43.0201170526.6967b81a@posting.google.com> References: <3C460066.437F68F@bellsouth.net> joe durusau wrote in message news:<3C460066.437F68F@bellsouth.net>... > Using C++, you can do that. It's illegal in C with the compiler version > > that comes with Tornado. > > Speaking only for myself, > > Joe Durusau > > > it2xpert wrote: > > > Hi... > > I use T2 Vxworks 5.4 on windows 2000 > > I want use // for annotation replace by /* */ > > How do ? > > > > Have a nice day !!! Hi, You can modify the macro CC_COMPILER, remove the -ansi option, then you can use the '//' for annotation. The macro is defined in makefiles: defs.bsp and defs.default. If you use tornado project environment, you can edit the compile options. HTH! Shichao Zhang Esteem Inc. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: taskdelay and interrupts Date: Thu, 17 Jan 2002 12:20:44 -0000 From: "Dave Korn" Organization: Lumber Cartel (tinlc) Members #2234-2237 (owing to browser refresh) Message-ID: References: <65b2abfc.0201150725.4f5f9903@posting.google.com> <3C4495A2.9564F1C4@digidata.com> [rearranged for added crunchiness and cascadeability] Jeff Creem wrote in message ... >"Bill Pringlemeir" wrote in message >news:uy9ixx5uj.fsf@yahoo.com... >> >>>>> "Kara" == Kara writes: >> >> Kara> Thanks to everybody involved, I have found the problem.. It >> Kara> was not printfs, I guess although they are not advisable they >> Kara> are ok. I was working on a Mips and for some reason VxWorks >> Kara> changes the value of SR (interrupt mask) register when a >> Kara> context switch occurs. Therefore SR is valid only in the task >> Kara> context. That's why my printf's were slow because the interrupt >> Kara> is enabled only when my task is active. The way around is the a >> Kara> call called taskSRSet(arg) which sets the default SR value for >> Kara> all tasks, it has to be made from BSP.. I appreciate your input >> Kara> and kind help. >> >> Is this running in a Nuclear power plant control system or a heart >> monitor? Perhaps a child's toy? You did not listen to the people >> here... some of them are quite knowledgeable about vxWorks. The one >> who gave you the answer writes the FAQ for this newsgroup. >> >> !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! >> >> !!!!!!!!!!!YOU SHOULD NOT USE PRINTFS IN ISRS!!!!!!!!!!!!!!!!!!!!!!!!!! >> >> !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! >> >> >> >> That said, I will be kinder and point you to the vxWorks header file >> called . In it there is a macro called `sioPollOutput'. >> This is a function that can be used to print out data in an ISR, if >> you are desperate for such a thing. I don't know a good way to get >> ahold of the `SIO_CHAN' structure needed for these macros. However, >> you can change your BSP to expose the structure and use it to call >> these functions. This is safe to do in an interrupt. >> >> The reason you must mess around with the SR interrupt mask is because >> you should not be using printfs in an ISR. Or to say it another way, >> you should not be using printfs in an ISR. More over, you should not >> be using printfs in an ISR. However, you should not be using printfs >> in an ISR. And Finally, you should not be using printfs in an ISR. >> >> hth, >> Bill Pringlemeir. >> >> -- >> SDI Peking North Korea spy colonel ammunition class struggle kibo PLO >> Rule Psix FBI South Africa Albanian Soviet $400 million in gold >> bullion >> >> vxWorks FAQ, "http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html" > > >And in case the prior poster was not clear..Don't use printfs in an ISR. At this moment in time, I'd just like to make it clear: Don't use printfs in an ISR. DaveK - -- moderator of alt.talk.rec.soc.biz.news.comp.humanities.meow.misc.moderated.meow Burn your ID card! http://www.optional-identity.org.uk/ Help support the campaign, copy this into your .sig! Proud Member of the Exclusive "I have been plonked by Davee because he thinks I'm interesting" List Member # --------------------------- Newsgroups: comp.os.vxworks Subject: Re: taskdelay and interrupts Date: Thu, 17 Jan 2002 14:44:55 +0100 From: patrick@hotmail.com Organization: Siemens Inc. Message-ID: References: <65b2abfc.0201150725.4f5f9903@posting.google.com> <3C4495A2.9564F1C4@digidata.com> Reply-To: patrick@hotmail.com Does anyone know wether it is OK to use printfs in an ISR? :-) regards, Patrick --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Problem with font in Tornado 2 - Part 2 Date: Thu, 17 Jan 2002 14:47:43 +0100 From: "Adam Ciszewski" Organization: T.O.P.S.E.C.R.E.T Message-ID: References: "Shrike" wrote: > I know about Font/Size tab in Options, but there is a blank space in size > selector. And nothing can be written there. Please, download Service Pack 3 for Tornado 2 from ftp.windriver.com /pub/updates/tornado/SPR/sprT2CP3.tar Regards AC --------------------------- Newsgroups: comp.os.vxworks Subject: linking problem Date: Thu, 17 Jan 2002 08:52:41 -0500 From: david lindauer Message-ID: <3C46D729.70802@notifier-is.net> In one C++ file I have: typedef char * sz ; sz const monthlist[12] = { "jan","feb",... } ; and in another: typedef char *sz ; extern sz const monthlist[12] ; For some reason LD insists that the reference in the second file can not be resolved. I have used nm on the object files and both list '_monthlist', unadorned. One is type 't' and one is type 'U'. So it should work. Both files are definitely in the project. I have tried changing the name of the variable with no success. Anyone know what I am doing wrong? David --------------------------- Newsgroups: comp.os.vxworks Subject: trapping alphanumeric keys Date: 17 Jan 2002 06:14:57 -0800 From: aravinddk@tataelxsi.co.in (aravind) Organization: http://groups.google.com/ Message-ID: Hi, I am facing problem in differentiating between alphanumeric characters. Alphanumeric keypad is one with alphabets and numbers on it, like ABC on one key, DEF on another and GHI on another and so on (similar to a telephone keypad). Suppose a user wants to enter B, then he has to press the ABC button twice, in which case I should be displaying only B and not AB. But, the problem is how to decide that only B has to be displayed. I am using InputQueuePoll() to trap the key presses. hth, ADK --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Using a PCI ethernet adapter on PCPentium BSP Date: Thu, 17 Jan 2002 16:32:35 +0100 From: "Werner Schiendl" Message-ID: <3c46ee95$1@brateggebdc5.br-automation.co.at> References: <2002117-101355-104040@foorum.com> Hi, "Charly" wrote in message news:2002117-101355-104040@foorum.com... > > Good morning everybody, > > Is it possible to use a PCI ethernet adapter with the PCPentium BSP? Yes, definitely possible. > > I would like to use "EDIMAX EN-5200PLT". Did someone use it? > Thanks a lot > charly > Never heard about that particular thing. Is it something 'compatible'? I strongly recommend using one of VxWorks' supported ethernet boards for your first tests. E. g. 3com Etherlink 3c905 (elPci) or Intel Etherexpress Pro 100B (fei or eeE). But in theory you can use any board you like as long as you have or write your own driver. hth Werner --------------------------- Newsgroups: comp.os.vxworks Subject: mem_part structure and it's meaning. Date: 17 Jan 2002 08:23:46 -0800 From: mynamvs@nsc.network.com (Venkat Mynampati) Organization: http://groups.google.com/ Message-ID: <65b2abfc.0201170823.495f029a@posting.google.com> Hi, I am trying to pry open the mem_part structure,and i thought that by following one of it's member DL_LIST freeList,(memPartLibp.h) i can get to see the num of freeblocks (FREE_BLOCKS) and the associated BLOK_HDR's. But, i am unable to do so, even though the 'freeList' field says that it's value is valid. What wrong am i doing? How to get to see the freeList? Also, i am confused about the values of memSysPartId->totalWords and other cum/cur[Blocks/Words]Allocated fields. How to interpret these values? and why does totalWords shows me very less (totalWords=[52428] ), even though i have about 8 MB of ram? Anyhelp is greatly appreciated. Regards Venkat --------------------------- Newsgroups: comp.os.vxworks Subject: Re: trapping alphanumeric keys Date: Thu, 17 Jan 2002 17:23:49 +0100 From: "Werner Schiendl" Message-ID: <3c46fa96$1@brateggebdc5.br-automation.co.at> References: Hi, can you tweak in a backspace keypress when you find out you need B instead of A? Then you'd have proper visual feedback for the user, too. hth Werner "aravind" wrote in message news:a2df8e26.0201170614.2290ecaf@posting.google.com... > Hi, > I am facing problem in differentiating between alphanumeric > characters. > > Alphanumeric keypad is one with alphabets and numbers on it, like > ABC on one key, DEF on another and GHI on another and so on (similar > to a telephone keypad). > > Suppose a user wants to enter B, then he has to press the ABC button > twice, in which case I should be displaying only B and not AB. > > But, the problem is how to decide that only B has to be displayed. > > I am using InputQueuePoll() to trap the key presses. > > hth, > ADK --------------------------- Newsgroups: comp.os.vxworks Subject: x86 mmu and physical memory Date: 17 Jan 2002 08:44:58 -0800 From: buckeyebobm@yahoo.com (Bob Martin) Organization: http://groups.google.com/ Message-ID: <432d6bc1.0201170844.1dfb8de3@posting.google.com> Please excuse my lack on knowledge on this subject! I'm trying to pick an OS for a project. I've never used a processor with mmu in an embedded system before. Anyway, I'll be using an x86 with 2 gB of ram. I need a DMA buffer 1.6Gb in size. I'd like to avoid using scatter gather since that's close to 400k pages - I also don't have the overhead time (let me know if this is a stupid assumption!). These will be 64 bit xfers at 66Mhz using about half of the pci bandwidth (bursts of transfers). What I'm trying to figure out is if VXWorks will let me grab a huge, contiguous, physical buffer for this purpose. Any advise would be apreciated also. Bob --------------------------- Newsgroups: comp.os.vxworks Subject: Re: How do I for annotation // Date: 17 Jan 2002 12:22:08 -0500 From: Bill Pringlemeir Organization: Bell Sympatico Message-ID: References: Sender: bpringlemeir@DeadDuck >>> "it2xpert" wrote: >> Hi... I use T2 Vxworks 5.4 on windows 2000 I want use // for >> annotation replace by /* */ How do ? >>>>> "JB" == Johan Borkhuis writes: JB> Remove the -ansi switch from the compiler options. This is a solution. However, you may wish to keep your source base ANSI compliant; my point is somewhat pedantic :-). I do use builtin alloca() and other functions occasionally; but I prefer the code to compile anywhere. Passing the "-Wp,-lang-c" argument should only enable CPP comments. Here is the comment from the pre-processor docs, `-lang-c', `-lang-c89', `-lang-c++' `-lang-objc', `-lang-objc++' Specify the source language. `-lang-c' is the default; it allows recognition of C++ comments (comments that begin with `//' and end at end of line), since this is a common feature and it will most likely be in the next C standard. `-lang-c89' disables recognition of C++ comments. `-lang-c++' handles C++ comment syntax and includes extra default include directories for C++. `-lang-objc' enables the Objective C `#import' directive. `-lang-objc++' enables both C++ and Objective C extensions. These options are generated by the compiler driver gcc, but not passed from the `gcc' command line unless you use the driver's `-Wp' option . hth, Bill Pringlemeir. - -- Actually, what I'd like is a little toy spaceship!! vxWorks FAQ, "http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html" --------------------------- Newsgroups: comp.os.vxworks Subject: Re: linking problem Date: Thu, 17 Jan 2002 18:42:58 GMT From: "H. E. Taylor" Organization: MTS Internet Message-ID: <3C473642.D04@despam.autobahn.mb.ca> References: <3C46D729.70802@notifier-is.net> In article <3C46D729.70802@notifier-is.net>, david lindauer wrote: > > In one C++ file I have: > > typedef char * sz ; > sz const monthlist[12] = { "jan","feb",... } ; > > and in another: > > typedef char *sz ; > extern sz const monthlist[12] ; > > For some reason LD insists that the reference in the second file can not > be resolved. I have used nm on the object files and both list > '_monthlist', unadorned. One is type 't' and one is type 'U'. So it > should work. Both files are definitely in the project. > > I have tried changing the name of the variable with no success. Anyone > know what I am doing wrong? > No. But this makes me really curious. My first thought was that one was extern "C" and one not, but your nm evidence would seem to disprove that. What happens if you remove the 'const'? Let us know how you resolve this. - -het - -- "If a packet hits a pocket on a socket on a port, And the bus is interrupted as a very last resort, And the address of the memory makes your floppy disk abort, Then the socket packet pocket has an error to report!" - -Gene Ziegler (with a nod to Dr. Suess) Name your Poison: http://www.autobahn.mb.ca/~het/catastrophes.html H.E. Taylor http://www.autobahn.mb.ca/~het/ --------------------------- Newsgroups: comp.os.vxworks Subject: Interrupt Latency Problem!!! Date: 17 Jan 2002 19:27:39 GMT From: Chris Smith Organization: Concentric Internet Services Message-ID: We are using Tornado2/VxWorks5.4 on a MVME230x target running at 350MHz. The application is industrial equipment where we move around a dozen or so mechanical parts as fast as we can over a one foot stroke, but to very high precision (sub-micron). This means we require fast servo sample rates (say 10kHz). We have set up a fixed-rate, external IRQ for the servo interrupt which we set using the Raven MPIC at the highest priority of 15. The default ethernet IRQ is set at 14 which is one lower than highest. All other IRQ priorities are lower than these two. This same bug happens whether we use our external PCI IRQ or, instead, use the VxWorks sysAuxClk routines (note: the sysAuxClk routines use the Raven Timer0 counter at IRQ priority 15 also). We measure interrupt latency the usual way -- our external device generates the external PCI IRQ, then the first thing the ISR does is toggle a bit on our external device. We use a logic analyzer to capture this time difference. Alternatively, I made a simple test case when I reported this bug to WindRiver that shows the exact same bad behavior using only the target board and software (code is below). The software- only and logic analyzer timings matched very well to each other which gives me confidence in both techniques. With no network activity, the interrupt latency is an acceptable 3-5us. If I make the PCI bus busy or the serial port busy, it stays at the same 3-5us. However, if I make the network busy, then the latency can glitch up to 47us for the given simple test case. WindRiver reproduced this problem recently and they said it glitched as high as 78us for a very similar target (though I don't know their CPU speed). They also reproduced it on a MVME5100 target with latencies around 14us (much better but still too high in my opinion). I have been hounding WindRiver for a month and a half about this. Up until jan 7th their response was that I should pay for special consulting services from them. Finally WindRiver opened up an SPR (# 72433), I think just to shut me up. My uneducated guess as to what's happening is that during the ethernet ISR, the network driver must do some kind of intLock which prevents my ISR from starting on time, even though my IRQ has the higher IRQ priority. I just bought WindView and there is something strange going on when my ISR interrupts the network ISR (strange meaning the network ISR takes a lot longer to finish than if it was not interrupted). I have yet to synchronize the latency glitch with something I can see in WindView, as I'm still getting used to this new tool. I guess the docs say I can genereate a user event, so I'll try that next...... So, finally, my questions are: 1. Have other hard-real-time users on PPC targets seen this behavior? Am I the only one who thinks this is serious? 2. Can anyone suggest some possible work-arounds? Here is the test code: /**************************************************************** Note that mySysTimeBaseLGet() is a direct copy of the assembly routine sysTimeBaseLGet() from target\config\mv2304\sysALib.s. It simply reads the lower PPC Time Base Register using the "mftb" assembly call. For our board each tick is 60ns (divide 4 of 66.667MHz bus). *****************************************************************/ volatile unsigned long isrCount=0; volatile unsigned long cnt=0; volatile unsigned long cntPrev=0; volatile unsigned long delta=0; volatile unsigned long deltaMax=0; volatile unsigned long deltaMin=LONG_MAX; void myIsr(int level) { cntPrev = cnt; cnt = mySysTimeBaseLGet(); delta = cnt - cntPrev; if(delta < deltaMin) deltaMin = delta; if(delta > deltaMax) deltaMax = delta; if( isrCount < 10 ) { deltaMin = LONG_MAX; deltaMax = 0; } isrCount++; } long myInit(void) { sysAuxClkRateSet(1000); sysAuxClkConnect(myIsr, 0); sysAuxClkEnable(); } /**********************************************/ --------------------------- Newsgroups: comp.os.vxworks Subject: vxsim from Tornado 2 on Windows XP Date: Thu, 17 Jan 2002 20:06:55 GMT From: "Kim Letkeman" Message-ID: Anyone come up with a way to install the ULIP driver under Windows XP? I've managed to get a number of older drivers to work on XP, but this is a network adapter emulator and it is currently stumping me. Ideas please? Thankx, Kim --------------------------- Newsgroups: comp.os.vxworks Subject: Re: POSIX timer handler not executed reliably Date: 17 Jan 2002 13:11:43 -0800 From: john_94501@yahoo.com (John) Organization: http://groups.google.com/ Message-ID: <488e459a.0201171311.146f995c@posting.google.com> References: <5c83e0.0201170451.79fd40ec@posting.google.com> Hello, Firstly, you told us nothing about the version of VxWorks or the architecture you are using. Not smart if you want a good answer. Given the limited info you gave though, the POSIX timers work by sending a signal to the task when the timer expires. There is almost no way that the timer would not send the signal since that is fired by a watchdog routine. There are however some possibilities I can think of: 1) You are generating the next signal before the current one is handled. Often, signals are blocked during their handler's execution time. You don't say how frequent the timer event is, nor what kind of handling you are doing. 2) Since the signal is handled at task level, even though it is delivered when the timer expires, the handler may not execute at that time. Instead, it will execute the next time the task is scheduled. If your task is a low priority, perhaps this is the cause of the "missed" signals. Without more information, that's the best I can come up with. Please try to be more specific in future when asking for help. HTH, John... a16157@email.mot.com (Li Xin) wrote in message news:<5c83e0.0201170451.79fd40ec@posting.google.com>... > For the portability reason, I am using the POSIX timer in my driver > code. When the timer expires, the timer handler routine gets executed > sometimes but does not at other times. Is it possible that vxWorks > doesn't execute the POSIX timer handler even the timer expires? > > Thanks in advance. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: VxWorks, Walnut 405GP tftp download error Date: 17 Jan 2002 13:16:15 -0800 From: john_94501@yahoo.com (John) Organization: http://groups.google.com/ Message-ID: <488e459a.0201171316.26b1891d@posting.google.com> References: Hello, You need to replace the boot ROM on the target system. Perhaps you were sent an actual VxWorks boot ROM with Tornado, if not, you will need to program an AMD29F040 flash device with the image that was on the CD. I just used the one that was on the board, and replaced the IBM monitor, but you may want to keep that somewhere safe and use a new part. To program the boot ROM (actually flash), you will need either an external programmer, or one of the JTAG debug tools that IBM has for the board and a copy of RISCwatch. See the BSP's documentation for more information. HTH, John... arun4vxworks@indiatimes.com (Arun Prasad) wrote in message news:... > Hi, > > We are using Vxworks on IBM's Walnut PPC 405GP board. > The bootrom on it is the same one which came with this board. > We are trying to boot the system(walnut) using TFTP and download the > vxWorks image. > > The following error message is displayed: > >Not a valid boot image file > > > We got this sequence a couple of times too: > Loading file "/tftpboot/mydir1/vxWorks" ... > Sending tftp boot request ... > Unknown destination, block 1 > Unknown destination, block 1 > Unknown destination, block 1 > Unknown destination, block 1 > Unknown destination, block 1 > Unknown destination, block 1 > Unknown destination, block 1 > Unknown destination, block 1 > Unknown destination, block 1 > TFTP boot failed. > > I am looking for info on how to build a tftp downloadable vxworks > image for the Walnut 405GP system. > > Is it looking for some thing specific at some specific location ... > Does the file type i.e., Binary or Hex matter? > Do we need use elfHex or linker with any specific options, > or any changes to the config? > > Looking forward for an early reply. > > Thanks in advance, > > -- --------------------------- Newsgroups: comp.os.vxworks Subject: Re: linking problem Date: 17 Jan 2002 13:19:17 -0800 From: john_94501@yahoo.com (John) Organization: http://groups.google.com/ Message-ID: <488e459a.0201171319.6ab418e@posting.google.com> References: <3C46D729.70802@notifier-is.net> Hello, david lindauer wrote in message news:<3C46D729.70802@notifier-is.net>... > In one C++ file I have: > > typedef char * sz ; > sz const monthlist[12] = { "jan","feb",... } ; > > and in another: > > typedef char *sz ; > extern sz const monthlist[12] ; > > For some reason LD insists that the reference in the second file can not > be resolved. I have used nm on the object files and both list > '_monthlist', unadorned. One is type 't' and one is type 'U'. So it > should work. Both files are definitely in the project. No, you cannot link against type 't' for two reasons: 1) Since it is lower case, that implies that it is a local symbol (making it impossible to link against) 2) Since it is a 't' it refers to a function, and not to a data type, and your array is clearly data ;-) HTH, John... > I have tried changing the name of the variable with no success. Anyone > know what I am doing wrong? > > David --------------------------- Newsgroups: comp.os.vxworks Subject: pppInit() (IPCP address negotiation with PPP) Date: Thu, 17 Jan 2002 22:28:44 +0100 From: Gerhard Juen Organization: 'FH Gelsenkirchen, Abt. Bocholt' Message-ID: <3C47420C.C591228F@fh-gelsenkirchen.de> Hi all, I want to connect a target ( a standard 80486 PC ) over the serial interface and a modem to an internet provider. In the IPCP address negotiation phase, our provider expects us to send a "0.0.0.0" as recommendated local and remote IP adsress. We use the command PPPInit( .. ). Any IP-Adress in the corresponding parameter fields which is not "0.0.0.0" is passed correctly as it is to our provider. If we take "0.0.0.0" or a NULL pointer (as proposed in the vxworks documentation for sending "0.0.0.0") our program ignores our parameter setting and passes the IP-address of the Ethernet network card which is also installed in the PC. Does anyone know how to pass a "0.0.0.0" to the provider during address negotiation? Thanks in advance, Norbert Rosskamp --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Interrupt Latency Problem!!! Date: 17 Jan 2002 13:32:10 -0800 From: john_94501@yahoo.com (John) Organization: http://groups.google.com/ Message-ID: <488e459a.0201171332.553345cd@posting.google.com> References: Hello, There are a few things to be careful of in PPC-land with interrupts. In particular, the fact that while interrupts may be nested, this does require some careful setup of the priorities to get the correct behaviour. It sounds like you have arranged those quite carefully, but it might be worth double checking the priority list, especially since, IIRC, the PICs are cascaded on these boards, and the second one appears in the middle of the first, not at an end. You might also want to check when in your ISR you acknowledge the interrupt on the PIC. If you do this too early, then lower priority interrupts will be allowed to interrupt you again (defeating the priority mechanism). It would be worth trying to find out whether the problem is your interrupt being delayed by a lock, or your ISR being interrupted (perhaps before it actually starts executing your code). Since you have an LA, it might be a good idea to watch the e'net chip's IRQ line and yours and see which order the interrupts arrived in when the problem is seen. Finally, you didn't tell us which network driver you are using. Since the network seems to be tied into this problem so tightly perhaps you should tell us that too (maybe somebody will have seen a similar problem with that driver on some other architecture, or could have a look in the source for locks). HTH, John... Chris Smith wrote in message news:... > We are using Tornado2/VxWorks5.4 on a MVME230x target running at 350MHz. > The application is industrial equipment where we move around a dozen or > so mechanical parts as fast as we can over a one foot stroke, but to > very high precision (sub-micron). This means we require fast servo > sample rates (say 10kHz). > > We have set up a fixed-rate, external IRQ for the servo interrupt which > we set using the Raven MPIC at the highest priority of 15. The default > ethernet IRQ is set at 14 which is one lower than highest. All other > IRQ priorities are lower than these two. This same bug happens whether > we use our external PCI IRQ or, instead, use the VxWorks sysAuxClk > routines (note: the sysAuxClk routines use the Raven Timer0 counter at > IRQ priority 15 also). > > We measure interrupt latency the usual way -- our external device > generates the external PCI IRQ, then the first thing the ISR does is > toggle a bit on our external device. We use a logic analyzer to capture > this time difference. Alternatively, I made a simple test case when I > reported this bug to WindRiver that shows the exact same bad behavior > using only the target board and software (code is below). The software- > only and logic analyzer timings matched very well to each other which > gives me confidence in both techniques. > > With no network activity, the interrupt latency is an acceptable 3-5us. > If I make the PCI bus busy or the serial port busy, it stays at the same > 3-5us. However, if I make the network busy, then the latency can glitch > up to 47us for the given simple test case. WindRiver reproduced this > problem recently and they said it glitched as high as 78us for a very > similar target (though I don't know their CPU speed). They also > reproduced it on a MVME5100 target with latencies around 14us (much > better but still too high in my opinion). > > I have been hounding WindRiver for a month and a half about this. Up > until jan 7th their response was that I should pay for special > consulting services from them. Finally WindRiver opened up an SPR (# > 72433), I think just to shut me up. > > My uneducated guess as to what's happening is that during the ethernet > ISR, the network driver must do some kind of intLock which prevents my > ISR from starting on time, even though my IRQ has the higher IRQ > priority. I just bought WindView and there is something strange going > on when my ISR interrupts the network ISR (strange meaning the network > ISR takes a lot longer to finish than if it was not interrupted). I > have yet to synchronize the latency glitch with something I can see in > WindView, as I'm still getting used to this new tool. I guess the docs > say I can genereate a user event, so I'll try that next...... > > So, finally, my questions are: > > 1. Have other hard-real-time users on PPC targets seen this behavior? > Am I the only one who thinks this is serious? > > 2. Can anyone suggest some possible work-arounds? > > > Here is the test code: > > /**************************************************************** > Note that mySysTimeBaseLGet() is a direct copy of the assembly > routine sysTimeBaseLGet() from target\config\mv2304\sysALib.s. > It simply reads the lower PPC Time Base Register using the "mftb" > assembly call. For our board each tick is 60ns (divide 4 of > 66.667MHz bus). > *****************************************************************/ > > volatile unsigned long isrCount=0; > volatile unsigned long cnt=0; > volatile unsigned long cntPrev=0; > volatile unsigned long delta=0; > volatile unsigned long deltaMax=0; > volatile unsigned long deltaMin=LONG_MAX; > > void myIsr(int level) > { > cntPrev = cnt; > cnt = mySysTimeBaseLGet(); > > delta = cnt - cntPrev; > > if(delta < deltaMin) > deltaMin = delta; > > if(delta > deltaMax) > deltaMax = delta; > > if( isrCount < 10 ) > { > deltaMin = LONG_MAX; > deltaMax = 0; > } > > isrCount++; > } > > long myInit(void) > { > sysAuxClkRateSet(1000); > sysAuxClkConnect(myIsr, 0); > sysAuxClkEnable(); > } > > /**********************************************/ --------------------------- Newsgroups: comp.os.vxworks Subject: Re: x86 mmu and physical memory Date: 17 Jan 2002 13:36:29 -0800 From: john_94501@yahoo.com (John) Organization: http://groups.google.com/ Message-ID: <488e459a.0201171336.6fb94ed2@posting.google.com> References: <432d6bc1.0201170844.1dfb8de3@posting.google.com> Hello, Yes, both VxWorks 5.x and VxWorks AE will let you allocate physically contiguous blocks of memory. In VxWorks 5.x, this is all you get since the translations are always physical = virtual; the MMU is used only to control the cacheability of the pages (which, btw, is something you'll need to think about too). In VxWorks AE, you have a number of ways to do this, but given the size of the block you are talking about, the simplest solution is probably to reserve the space in the system's memory configuration table (the region table). FYI, all heap allocations in AE 1.x are guaranteed to be physically contiguous to make drivers easier to migrate from 5.x too. The docs do say that this might change in the future, though there will always be an interface to get physically contiguous buffers for exactly the type of thing you are trying to do. HTH, John... buckeyebobm@yahoo.com (Bob Martin) wrote in message news:<432d6bc1.0201170844.1dfb8de3@posting.google.com>... > Please excuse my lack on knowledge on this subject! I'm trying to pick > an OS for a project. I've never used a processor with mmu in an > embedded system before. Anyway, I'll be using an x86 with 2 gB of ram. > I need a DMA buffer 1.6Gb in size. I'd like to avoid using scatter > gather since that's close to 400k pages - I also don't have the > overhead time (let me know if this is a stupid assumption!). These > will be 64 bit xfers at 66Mhz using about half of the pci bandwidth > (bursts of transfers). What I'm trying to figure out is if VXWorks > will let me grab a huge, contiguous, physical buffer for this purpose. > Any advise would be apreciated also. > > Bob --------------------------- Newsgroups: comp.os.vxworks Subject: Adding a toolchain to T2 Date: 17 Jan 2002 13:51:29 -0800 From: pkockritz@mac.com (Pete Kockritz) Organization: http://groups.google.com/ Message-ID: <9423c55b.0201171351.75fbb1f@posting.google.com> I'm attempting to add a toolchain to Tornado for the Green Hills C/C++ compiler. I've followed the instructions in the T2 manuals (Tornado API Programmer's Guide, section 7, Integrating Third-Party Toolchains) but I don't see the new toolchain listed when I create a new project. Has anyone done this? I tried the example in the manual for the Diab compiler, but that didn't seem to work either. Regards, Pete Kockritz pkockritz@mac.com --------------------------- Newsgroups: comp.os.vxworks Subject: Re: linking problem Date: Thu, 17 Jan 2002 17:00:51 -0500 From: david lindauer Message-ID: <3C474993.4020903@notifier-is.net> References: <3C46D729.70802@notifier-is.net> <488e459a.0201171319.6ab418e@posting.google.com> - --------------070305090507020000070307 Content-Type: text/plain; charset=us-ascii; format=flowed Content-Transfer-Encoding: 7bit John wrote: >Hello, > >david lindauer wrote in message news:<3C46D729.70802@notifier-is.net>... > >>In one C++ file I have: >> >>typedef char * sz ; >>sz const monthlist[12] = { "jan","feb",... } ; >> >>and in another: >> >>typedef char *sz ; >>extern sz const monthlist[12] ; >> >>For some reason LD insists that the reference in the second file can not >>be resolved. I have used nm on the object files and both list >>'_monthlist', unadorned. One is type 't' and one is type 'U'. So it >>should work. Both files are definitely in the project. >> > >No, you cannot link against type 't' for two reasons: > >1) Since it is lower case, that implies that it is a local symbol >(making it impossible to link against) > >2) Since it is a 't' it refers to a function, and not to a data type, >and your array is clearly data ;-) > ok so then why did the compiler make it a local symbol? It is defined without 'static' keyword! And why did it make it a function? And if it did make it a function why didn't it mangle the name? Thanks, David > >HTH, > >John... > >>I have tried changing the name of the variable with no success. Anyone >>know what I am doing wrong? >> >>David >> - --------------070305090507020000070307 Content-Type: text/html; charset=us-ascii Content-Transfer-Encoding: 7bit

John wrote:
Hello,

david lindauer <dlindauer@notifier-is.net> wrote in message news:<3C46D729.70802@notifier-is.net>...
In one C++ file I have:

typedef char * sz ;
sz const monthlist[12] = { "jan","feb",... } ;

and in another:

typedef char *sz ;
extern sz const monthlist[12] ;

For some reason LD insists that the reference in the second file can not 
be resolved.  I have used nm on the object files and both list 
'_monthlist', unadorned.  One is type 't' and one is type 'U'.  So it 
should work.  Both files are definitely in the project.

No, you cannot link against type 't' for two reasons:

1) Since it is lower case, that implies that it is a local symbol
(making it impossible to link against)

2) Since it is a 't' it refers to a function, and not to a data type,
and your array is clearly data ;-)
ok so then why did the compiler make it a local symbol?   It is defined without 'static' keyword!
And why did it make it a function?  And if it did make it a function why didn't it mangle the name?

Thanks,

David

HTH,

John...

I have tried changing the name of the variable with no success.  Anyone 
know what I am doing wrong?

David

- --------------070305090507020000070307-- --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Symbol table errors compiling an image when using STL Date: 17 Jan 2002 14:12:29 -0800 From: sasacoh@europe.com (Sasa) Organization: http://groups.google.com/ Message-ID: <769ac895.0201171412.346dc387@posting.google.com> References: <3c29bbec$1@news.barak.net.il> <3c3833dd$1@news.barak.net.il> <488e459a.0201071346.16dad6f2@posting.google.com> john_94501@yahoo.com (John) wrote in message news:<488e459a.0201071346.16dad6f2@posting.google.com>... > Hello, > > Your problem is probably the use of namespaces - I don't believe that > they are supported by any version of Tornado/VxWorks at the moment. > > HTH, > > John... > > "Arnon W" wrote in message news:<3c3833dd$1@news.barak.net.il>... > > Hi, > > > > Didn't work. > > I added only these 3 lines to one of my cpp's: > > > > #include (or #include ) > > using namespace std; > > list test; > > > > The funny thing is that the module containing the STL part compiles with no > > problems, > > and the code runs well as a downloadable module. > > The minute I try to integrate it into the image - boom! > > > > > > Arnon > > > > "Dave Korn" wrote in message > > news:W4XY7.206$qS6.464044@newsr2.u-net.net... > > > Arnon W wrote in message <3c29bbec$1@news.barak.net.il>... > > > >Hi, > > > > > > > >I'm new to this environment so it might be a minor issue I'm having > problems > > > >with. > > > >(On the other hand - I couldn't find earlyer posts in this specific > > > matter:) > > > > > > >When I try to integrate it to the image, I get vicious responses at this > > > >point: > > > > > > > > >ccppc -c -fdollars-in-identifiers -O3 -mstrict-align -nostdinc -DRW_MULTI_T > > > H > > > >READ -D_REENTR > > > > > >ANT -fvolatile -fno-builtin -fno-for-scope -DCPU=PPC604 -DPRJ_BUILD -DHAL_L > > > I > > > >B > > > >...paths... symTbl.c -o symTbl.o > > > >symTbl.c:2013: syntax error before `.120' > > > >symTbl.c:2014: syntax error before `.139' > > > >symTbl.c:2015: syntax error before `.148' > > > >symTbl.c:8448: > `__sn_t24__default_alloc_template2b1i0$__node_allocator_lock' > > > >undeclared he > > > >re (not in a function) > > > >symTbl.c:8448: syntax error before `.120' > > > >symTbl.c:8449: initializer element for `standTbl[1997].value' is not > > > >constant > > > >symTbl.c:8449: warning: initialization from incompatible pointer type > > > >symTbl.c:8449: > `__sn_t24__default_alloc_template2b1i0$__node_allocator_lock' > > > >undeclared he > > > >re (not in a function) > > > >symTbl.c:8449: syntax error before `.139' > > > >symTbl.c:8450: initializer element for `standTbl[1998].name' is not > constant > > > >symTbl.c:8450: syntax error before string constant > > > > >When I took out STL related stuff (I was using ) It compiled well. > > > >Is there a cure? > > > > > > I'm just passing on something I read somewhere else here, STL is really > > > not my field, but ISTR someone saying that to use STL you should only > > > #include - i.e. you use that header and it includes all the other > > > stl headers like and for you. You might like to try this > > > approach and see if it fixes your problem. > > > > > > DaveK > > > -- > > > Burn your ID card! http://www.optional-identity.org.uk/ > > > Help support the campaign, copy this into your .sig! > > > Proud Member of the Exclusive "I have been plonked by Davee because he > > > thinks I'm interesting" List > > > Member # > > > > > > Hello! We're having exactly the same problems (on ARM9 board) with STL compiled together with symbol table (INCLUDE_SYM_TBL_INIT) and we don't use namespaces. If INCLUDE_SYM_TBL_INIT macro is excluded it works well but without possibility to execute commands from shell. But we need this! Any suggestions? Sasa --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Support for 8 bits + even parity in ttyDrv? Date: Thu, 17 Jan 2002 22:55:45 GMT From: Jon Organization: EarthLink Inc. -- http://www.EarthLink.net Message-ID: <3C4756DC.79DBB9D7@bitworkssystemsDELETETHIS.com> References: <3C45D4D8.5FE5F442@bitworkssystemsDELETETHIS.com> <488e459a.0201161606.1b6c2db@posting.google.com> Thanks I did find it. A person really needs this group to do this since none of this is in the printed manuals or the online documentation. Searching for SIO_HW_OPTS_SET returns "No topics found". Thanks again! John wrote: > Hello, > > Take a look in the driver for your serial chip. For the standard > pcPentiumX BSPs this is i8250Sio.c (under target/src/drv/sio). You > will see in there that there is a function called i8250OptsSet() which > is called as a result of an ioctl() on the device with the operation > code SIO_HW_OPTS_SET. The argument of the ioctl() is passed through to > i8250OptsSet(). > > The options you are interested in are PARENB and PARODD. Since you > want even parity, you should only set the PARENB (parity enable), and > not PARODD (select odd). These values are defined in sioLib.h, so > you'll want to include that into whatever is making the ioctl() call. > > HTH, > > John... > > Jon wrote in message news:<3C45D4D8.5FE5F442@bitworkssystemsDELETETHIS.com>... > > I have a serial device that uses 8 bits plus even parity and at first > > glance didn't see away to set this in the tyLib ioctl options. Is there > > a way to get the ttyDrv to transmit and receive 1 start bit, 8 data > > bits, 1 even parity bit and 1 stop bit? Thats 11 bits total. > > > > Platform is Pentium running VxWorks AE 3.1 > > > > Jon Newbill > > Bitworks Systems Inc. > > jon @ bitworkssystemsNOSPAM.com --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Hi All Date: 17 Jan 2002 18:01:12 -0500 From: Bill Pringlemeir Organization: Bell Sympatico Message-ID: References: Sender: bpringlemeir@DeadDuck >>>>> "aravind" == aravind writes: aravind> Hi, When we try to use InputQueuePoll and Build the File We aravind> always get the undefined symbol for InputQueuePoll when we aravind> look into the symbol table and so cannot download the aravind> executable to the target board. aravind> I have included all the header files as in the aravind> v_keybrd.cpp. Is there anything else to be included? Please restrict the conversation to the newsgroup. Even if it seems irrelevant, your questions will help other gauge popularity, defects, challenges to using UGL and Zinc on vxWorks. Also, I don't wish to take time to answer your questions just for you. I actually read this newsgroup more often than the email address in my header. I marked your article for later reply... From private email, aravind> Thanks for giving the inputs on using InputQueuePoll , aravind> Double key press avoided, In Single Press we are now getting aravind> Now, But really I want to understand why the aravind> uglInputQStandardQEventGet(10000, &inputEvent) isn't able to aravind> achieve the same, is there reason behind, How InputQueuePoll aravind> is able to acheive it, please reply so i do only use but aravind> understand the reason, I have grepped the Zinc source and I can not find InputQueuePoll. What are you talking about? Please try to make this easy for me. From another email, aravind> I want to know how do we trap the Alpha Numeric Keys, is aravind> there any way to get those keys trapped using Zinc or UGL and another email... [snip] aravind> Alphanumeric keypad is one with alphabets and numbers on it, aravind> like ABC on one key, DEF on another and GHI on another and aravind> so on (similar to a telephone keypad). aravind> Suppose a user wants to enter B, then he has to press the aravind> ABC button twice, in which case I should be displaying only aravind> B and not AB. [snip] You have described this problem well and I think that this is very common for developers of an embedded GUI to encounter. For the most part WRS/Zinc seem [*1] to assume an `embedded PC' with 101 keyboard, mouse and VGA. I had the same problem and concluded that the Zinc layer was the best place to do this. I have some experience with hand helds, PDA and have used different cell phones. The best way [*1] is to enable the `key value' cycling when you are in the ZafFormatString class/gui element. The format string specifies whether the keyboard entry is going to be numeric, alpha, alphanumeric, etc. I have modified the file "src/zinc/generic/i_fmtstr.cpp"; specifically the Event() functions (especially E_KEY case (i hope you guessed that)). case E_KEY: { static const char special[30] = { '$', '%', '.', ',', '?', '!', '+', '-', '*', '=', '^', '#', [snip, pressing zero several times allows punctuation.] static const char decode[10][8] = { {'0', ' ', '~'}, {'1', 'q', 'z', 'Q', 'Z'}, {'2', 'a', 'b', 'c', 'A', 'B', 'C'}, [snip, describes cycling sequence] // Search for legal characters. ZafIChar tChar = event.key.value; bool stepForward = true; /* move cursor forward */ // If special key, overwrite previous value (space). if('~' == lastKeyValue) { for(int i = 0; i < sizeof(special); i++) [snip] // Check to see if we should not repeat a value. if(tChar != lastKeyValue) { if(D_ON == zKeyTimer->Event(ZafEventStruct(D_STATE))) { zKeyTimer->Event(ZafEventStruct(D_OFF)); MoveCursor(L_RIGHT); [snip... ok, this is somewhat ugly. I did this under time pressure as always. Run through the `decode' array and look for two possible characters from IsLegalChar() virtual (or callback) function. If two are found and then we cycle through the values. Don't move the cursor forward and set a timer to move the cursor when that happens... ] // Cycle through keys. if(tChar >= '0' && tChar <= '9') { const char *p = &decode[tChar - '0'][0]; int i = curKeyIndex; do { if(IsLegalChar(p[i], cursor)) { tChar = p[i]; // multiple key values for this key? for(int j = i + 1; i != j; j = (j+1) %8) { if(IsLegalChar(p[j], cursor)) { stepForward = false; curKeyIndex = j; lastCursor = cursor; break; } } break; } i = (i+1) % 8; } while(i != curKeyIndex); } if(tChar == '~') { [snip, if zero is pressed several times bring up the ZafStringEditor class. I have totatlly re-written this as well...] ZafStringEditor *stringEditor = new ZafStringEditor; stringEditor->SetRequestor(this); } else if (IsLegalChar(tChar, cursor)) { ZafEventStruct tEvent = event; tEvent.key.value = tChar; ZafUpdateType updateType = stringData->Update(); stringData->SetUpdate(ZAF_UPDATE_NONE); stringData->Replace(cursor, 1, &tChar, 1); stringData->SetUpdate(updateType); [only move forward if this is no possible cycling...] if(stepForward) { MoveCursor(L_RIGHT); // Reset cycle code. lastKeyValue = 0; curKeyIndex = 0; } else { // Restart timer... zKeyTimer->Event(ZafEventStruct(D_INITIALIZE)); zKeyTimer->Event(ZafEventStruct(D_ON)); } [...] I have given you a lot Aravind. Please try to work something out for yourself. That is part of being a software developer! ;-) You must modify this for your application. I suggest that you get together with someone who knows your market and customize the UI to best fit your customers needs. This could be a key (*bad pun*) part of the product. Depending on your market, you may need some characters like "accente` gui`" [excuse my French] or other odd characters. One strength of Zinc [*1] is that you can over-ride the behaviour of a class to fit your UI requirements; especially over plan vanilla WindML. A weakness is the resource overhead [*1]. hth, Bill Pringlemeir. [*1] That's my opinion! - -- The Kibo Bible is the symbol of atoms. vxWorks FAQ, "http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html" --------------------------- Newsgroups: comp.os.vxworks Subject: Getting UART errors from i8250Sio driver Date: Thu, 17 Jan 2002 23:16:59 GMT From: Jon Organization: EarthLink Inc. -- http://www.EarthLink.net Message-ID: <3C475BD6.F5E5A512@bitworkssystemsDELETETHIS.com> Now that I've been able to get my serial port set to the appropriate protocol (8+even+1 stop), now I also need to get UART error codes perferable on a character by character basis. Looking at the driver source i8250Sio.c it appears the line status register where things like parity errors, framing errors and overrun errors are reported is read by the driver but then ignored. Am I reading this correctly? Am I going to have to modify the driver to get access to UART error codes? Thanks in advance. Jon Newbill Bitworks Systems Inc. jon@bitworkssystemsSPAMOUT.com --------------------------- Newsgroups: comp.os.vxworks Subject: Machine Check Exception Date: Fri, 18 Jan 2002 00:14:17 +0000 (UTC) From: "martin brook" Organization: BT Openworld Message-ID: Hi, I'm having problems with my application getting random machine check exceptions (0x200). The exception address reported does not correspond to any code which may cause a machine check. I understand that the exception reporting can be delayed so giving no clue as to where the exception is occuring. Has anyone got a method of debugging these exceptions ? or a way of turning the machine check off ? My app is doing a lot of serial I/O on ports 0/1 which are set to 115200baud. TIA martin - --- Outgoing mail is certified Virus Free. Checked by AVG anti-virus system (http://www.grisoft.com). Version: 6.0.313 / Virus Database: 174 - Release Date: 1/3/02 --------------------------- Newsgroups: comp.os.vxworks Subject: dhcp client options in boot Date: Thu, 17 Jan 2002 19:24:01 -0500 From: "Apu" Message-ID: Hi Does anybody know how to add DHCP client options (like router) in the boot, and successfully retrieve the params that a DHCP server may provide? dhcpcOptionAdd and dhcpOptionSet do not work in the boot. Any help is appreciated. Thanks Apu --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Machine Check Exception Date: Fri, 18 Jan 2002 00:28:52 GMT From: sj@shell1.aracnet.com (Scott Johnson) Organization: aracnet.com -- Portland's loudest electrons (info@aracnet.com) Message-ID: <8%J18.387$h31.69966@e420r-atl1.usenetserver.com> References: In article , martin brook wrote: >Hi, > >I'm having problems with my application getting random machine check >exceptions (0x200). The exception address reported does not correspond to >any code which may cause a machine check. I understand that the exception >reporting can be delayed so giving no clue as to where the exception is >occuring. Quite likely you have a stray pointer somewhere in your code. Have you done a stack trace of the offending task to see where the exception is occurring? >Has anyone got a method of debugging these exceptions ? or a way of turning >the machine check off ? In hardware, you cannot "turn off" a machine check. (Well, you can--but the alternative to enabling machine check exceptions is to halt the processor when a machine chek occurs). You can try to disable the exception handler in vxWorks; but that is a BAD idea. The machine check is telling you that your code is buggy, and that you had better go fix it. (Else your hardware is buggy.) Attempting to ignore it will only lead to worse pain. >My app is doing a lot of serial I/O on ports 0/1 which are set to >115200baud. I don't know if this would cause the problem. What are you using for your serial port hardware? (BTW, since you say "machine check" rather than "bus error"; I am assuming you have a PowerPC of some sort. If not, some of the above comments might not apply.) e_S >TIA martin > > > >--- >Outgoing mail is certified Virus Free. >Checked by AVG anti-virus system (http://www.grisoft.com). >Version: 6.0.313 / Virus Database: 174 - Release Date: 1/3/02 > > - -- - ------------------------------------------------------------------------ engineer_scotty (no, not that one) -- consumer of fine ales and lagers some days you're the fire hydrant, some days you're the dog | go blazers no small furry creatures were harmed in the creation of this .signature --------------------------- Newsgroups: comp.os.vxworks Subject: help urgent.... how to enable MULTICAST flag.... Date: 17 Jan 2002 17:04:21 -0800 From: hrkim@hiper.co.kr (Hongrae Kim) Organization: http://groups.google.com/ Message-ID: <505213fc.0201171704.17c9cc06@posting.google.com> how to enable MULTICAST flags ? when I run ifShow(), it does not shows MULTICAST flag. Here's a little info on my VxWorks box. PPC860 CPU Motorola, and vxWorks 5.3.1 cpm (unit number 0): Flags: (0x63) UP BROADCAST ARP RUNNING Type: ETHERNET_CSMACD Internet address: 61.97.37.228 Broadcast address: 61.97.37.255 Netmask 0xff000000 Subnetmask 0xffffff00 Ethernet address is 00:03:a1:19:97:3e Metric is 0 Maximum Transfer Unit size is 1500 17 packets received; 5 packets sent 15 multicast packets received 1 multicast packets sent 0 input errors; 0 output errors 0 collisions; 0 dropped --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Who can give me the source code of pppoe on vxworks? Date: Fri, 18 Jan 2002 09:20:35 +0800 From: "5982" <5982851@ihw.com.cn> Organization: Bentium Ltd. (CN99) Message-ID: References: good idea. :) --------------------------- Newsgroups: comp.os.vxworks Subject: Re: POSIX timer handler not executed reliably Date: 17 Jan 2002 19:11:41 -0800 From: a16157@email.mot.com (Li Xin) Organization: http://groups.google.com/ Message-ID: <5c83e0.0201171911.7c56c069@posting.google.com> References: <5c83e0.0201170451.79fd40ec@posting.google.com> <488e459a.0201171311.146f995c@posting.google.com> Thanks John very much. I think I have found the cause of the my POSIX timer problem. I am using vxWorks 5.4 and my code is like: void timerhandler() { post the semaphore; } int inited = 0; sem_t *SemId; timer_t wdgId; void mytest() { if (!inited) { inited = 1; create posix timer; create a semaphore; } arm the posix timer; wait the semaphore; print a message; disarm the timer; } When I execute "mytest" from wind shell, the timer handler always executes. When I execute "sp mytest" from wind shell, the timer handler executes only the first time I execute "sp mytest". This is because the task that created the posix timer has exited when I execute the second "sp mytest". john_94501@yahoo.com (John) wrote in message news:<488e459a.0201171311.146f995c@posting.google.com>... > Hello, > > Firstly, you told us nothing about the version of VxWorks or the > architecture you are using. Not smart if you want a good answer. > > Given the limited info you gave though, the POSIX timers work by > sending a signal to the task when the timer expires. There is almost > no way that the timer would not send the signal since that is fired by > a watchdog routine. There are however some possibilities I can think > of: > > 1) You are generating the next signal before the current one is > handled. Often, signals are blocked during their handler's execution > time. You don't say how frequent the timer event is, nor what kind of > handling you are doing. > > 2) Since the signal is handled at task level, even though it is > delivered when the timer expires, the handler may not execute at that > time. Instead, it will execute the next time the task is scheduled. If > your task is a low priority, perhaps this is the cause of the "missed" > signals. > > Without more information, that's the best I can come up with. Please > try to be more specific in future when asking for help. > > HTH, > > John... > > a16157@email.mot.com (Li Xin) wrote in message news:<5c83e0.0201170451.79fd40ec@posting.google.com>... > > For the portability reason, I am using the POSIX timer in my driver > > code. When the timer expires, the timer handler routine gets executed > > sometimes but does not at other times. Is it possible that vxWorks > > doesn't execute the POSIX timer handler even the timer expires? > > > > Thanks in advance. --------------------------- Newsgroups: comp.os.vxworks Subject: GPIB Driver Date: Fri, 18 Jan 2002 03:42:29 GMT From: Francisco J Pataro Organization: Prodigy Internet http://www.prodigy.com Message-ID: <3C4799A7.3070103@swbell.net> Gents, I'm looking for a GPIB Driver for VxWorks 4.5 on an Intel platform. The GPIB chip that is planned to be use is a NTN488. Thanks, Francisco --------------------------- Newsgroups: comp.os.vxworks Subject: help urgent.... how to enable MULTICAST flag.... Date: 17 Jan 2002 20:44:16 -0800 From: hrkim@hiper.co.kr (Hongrae Kim) Organization: http://groups.google.com/ Message-ID: <505213fc.0201172044.5349276@posting.google.com> how to enable MULTICAST flag. when I run ifShow(), it does not shows MULTICAST falg cpm0 interface. Here's a info on my VxWorks box: PPC860 CPU, Motorola cpm (unit number 0): Flags: (0x63) UP BROADCAST ARP RUNNING Type: ETHERNET_CSMACD Internet address: 61.97.37.228 Broadcast address: 61.97.37.255 Netmask 0xff000000 Subnetmask 0xffffff00 Ethernet address is 00:03:a1:19:97:3e Metric is 0 Maximum Transfer Unit size is 1500 81 packets received; 5 packets sent 78 multicast packets received 0 multicast packets sent 0 input errors; 0 output errors 0 collisions; 0 dropped --------------------------- Newsgroups: comp.os.vxworks Subject: why extern interrupt cause vxworks reboot?mpc860 Date: 17 Jan 2002 20:56:50 -0800 From: yong_bvcom@sina.com (bigtornado) Organization: http://groups.google.com/ Message-ID: <5624c1bf.0201172056.1b0dc0ef@posting.google.com> hi,all: I use windriver FADS mpc860 board,I connect a ISR to extern interrupt IRQ1,but when program running in full speed and interrupt occurs,the system reboot,I debug the interrupt driver ppc860intr.c step by step, then the system don't reboot and work ok,why? Your answer is appreciated!Thanks! jeffy --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Symbol table errors compiling an image when using STL Date: 17 Jan 2002 20:59:34 -0800 From: john_94501@yahoo.com (John) Organization: http://groups.google.com/ Message-ID: <488e459a.0201172059.2268dd7d@posting.google.com> References: <3c29bbec$1@news.barak.net.il> <3c3833dd$1@news.barak.net.il> <488e459a.0201071346.16dad6f2@posting.google.com> <769ac895.0201171412.346dc387@posting.google.com> sasacoh@europe.com (Sasa) wrote in message news:<769ac895.0201171412.346dc387@posting.google.com>... [history deleted to save space] > Hello! > We're having exactly the same problems (on ARM9 board) with STL > compiled together with symbol table (INCLUDE_SYM_TBL_INIT) and we > don't use namespaces. If INCLUDE_SYM_TBL_INIT macro is excluded it > works well but without possibility to execute commands from shell. But > we need this! Any suggestions? The problem looks to be related to the generation of '.' characters in the mangled names. I believe that this can be turned off for GNU, but I cannot remember how. As a workaround, you can run sed over the symTbl.c file to remove all the lines that contain identifiers with a '.' in them. This can be added to the make rules (under target/h/make) so that you don't need to do it manually each time. This will just make them unavailable to the target symbol table, which will be OK provided you don't need to download anything that binds to them (or call them directly from the shell - which is unlikely). If you do need them, then you need to find a way of persuading the compiler not to generate '.' in symbols. HTH, John... --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Symbol table errors compiling an image when using STL Date: 17 Jan 2002 20:59:34 -0800 From: john_94501@yahoo.com (John) Organization: http://groups.google.com/ Message-ID: <488e459a.0201172059.1eeb7d1d@posting.google.com> References: <3c29bbec$1@news.barak.net.il> <3c3833dd$1@news.barak.net.il> <488e459a.0201071346.16dad6f2@posting.google.com> <769ac895.0201171412.346dc387@posting.google.com> sasacoh@europe.com (Sasa) wrote in message news:<769ac895.0201171412.346dc387@posting.google.com>... [history deleted to save space] > Hello! > We're having exactly the same problems (on ARM9 board) with STL > compiled together with symbol table (INCLUDE_SYM_TBL_INIT) and we > don't use namespaces. If INCLUDE_SYM_TBL_INIT macro is excluded it > works well but without possibility to execute commands from shell. But > we need this! Any suggestions? The problem looks to be related to the generation of '.' characters in the mangled names. I believe that this can be turned off for GNU, but I cannot remember how. As a workaround, you can run sed over the symTbl.c file to remove all the lines that contain identifiers with a '.' in them. This can be added to the make rules (under target/h/make) so that you don't need to do it manually each time. This will just make them unavailable to the target symbol table, which will be OK provided you don't need to download anything that binds to them (or call them directly from the shell - which is unlikely). If you do need them, then you need to find a way of persuading the compiler not to generate '.' in symbols. HTH, John... --------------------------- Newsgroups: comp.os.vxworks Subject: Re: why extern interrupt cause vxworks reboot?mpc860 Date: Fri, 18 Jan 2002 05:38:41 GMT From: sj@shell1.aracnet.com (Scott Johnson) Organization: aracnet.com -- Portland's loudest electrons (info@aracnet.com) Message-ID: References: <5624c1bf.0201172056.1b0dc0ef@posting.google.com> In article <5624c1bf.0201172056.1b0dc0ef@posting.google.com>, bigtornado wrote: >hi,all: > I use windriver FADS mpc860 board,I connect a ISR to extern >interrupt IRQ1,but when program running in full speed and interrupt >occurs,the system reboot,I debug the interrupt driver ppc860intr.c >step by step, then the system don't reboot and work ok,why? > Your answer is appreciated!Thanks! > jeffy Make sure your interrupt handler ACKNOWLEDGES the interrupt and clears it. Otherwise, you may get a nifty cascade of nested interrupts; each with a stack frame lower in memory than the previous. Eventually, this interrupt cascade can wipe out your kernel and vector table; causing a crash or reboot (depending on various things like the phase of the moon, whether checkstops cause reboots, etc.) e_s - -- - ------------------------------------------------------------------------ engineer_scotty (no, not that one) -- consumer of fine ales and lagers some days you're the fire hydrant, some days you're the dog | go blazers no small furry creatures were harmed in the creation of this .signature --------------------------- Newsgroups: comp.os.vxworks Subject: Re: mpc850 upmx program question Date: Fri, 18 Jan 2002 06:01:20 GMT From: "Gary M" Organization: AT&T Broadband Message-ID: References: I highly recommend the following application note from Motorola (they recently updated it). It describes most of the issues you need to understand to program the UPM to interface to SDRAM: http://e-www.motorola.com/brdata/PDFDB/docs/AN2066.pdf There is also a tool provided by Motorola for generating the UPM program using interactive timing diagrams. It isn't perfect, but it was very helpful. Buggy as far as exporting the output on WinNT and Win2K systems. Go to http://mot-sps.com/ , click on Tools Library link, select "32-bit Embedded Processors", "PowerPC ISA", "MPC82XX and MPC8XX Integrated Communications Processors", *do not* select a product, select "Software" and then search for MPC860COD09, the MPC860 Programming Tool. You'll need to register to download the tool. There's a lot of useful stuff on the Motorola site....you just have to look for it.... "juliedan" wrote in message news:d6763629.0201100250.7cf3243d@posting.google.com... > Could anybody tell me how to use SDRAM timing diagram to make upmTableSdram? > I hope you can help me in details(especially the process). > thank you in advance. > Liedan Ju > mail: jlddotcn@hotmail.com --------------------------- Newsgroups: comp.os.vxworks Subject: workQPanic: Kernel work queue overflow PROBLEM Date: Fri, 18 Jan 2002 06:11:28 GMT From: "Nejhdeh Ghevondian" Organization: Integrated Spectronics Message-ID: Reply-To: "Nejhdeh Ghevondian" Dear anyone! I lately increased my options in the VxWorks image file by including things like POSIX timers etc. Things that require further ISR routines to be attached. However, in my application i further require routines to become ISRs, like SCSI devices. Now I get the error during bootup of workQPanic: Kernel work queue overflow error. The Windriver help file suggests that I ether: 1. intLock() then intUnlock the place were the work Q becomes error. I tried this but still get the error. 2. Create a variable that can find out Work Q is full. HOW DO I DO THIS????? 3. Increase the priority of netTask. I COULDNT find the file to do this. Reagrds Nejhdeh Ghevondian --------------------------- Newsgroups: comp.os.vxworks Subject: problem while receiving from my END driver Date: 17 Jan 2002 22:37:47 -0800 From: dhruvesh_patel@yahoo.com (Dhruvesh) Organization: http://groups.google.com/ Message-ID: <84d8b1e9.0201172237.3370005@posting.google.com> The problem I am facing with my END driver is: I am sending a file of 12K from client to server. When I am calling recv() from my server application to read the data from socket buffers, it gets PAGE FAULT during second or third call. Same application works with the ETHERNET driver. At the driver level I am getting the complete file in correctly. I guess there is some problem with my 'receive handler' in driver. My receivehandler is as below. #################################################### /* Get a receive frame buffer */ pData = netClusterGet(pCS->end.pNetPool, pCS->pClPoolId); /* Grab a cluster block to join to the cluster we received. */ pClBlk = netClBlkGet (pCS->end.pNetPool, M_DONTWAIT); /* * OK we've got a spare, let's get an M_BLK_ID and join it to the * one in the ring. */ pMblk = mBlkGet(pCS->end.pNetPool, M_DONTWAIT, MT_DATA); /* If no buffer available free whatever was allocated*/ if ( pData == NULL || pClBlk == NULL || pMblk == NULL) { logMsg("BUFFER UNAVAILABLE \n",0,0,0,0,0,0); if (pData) { logMsg("cluster UNAVAILABLE \n",0,0,0,0,0,0); netClFree(pCS->end.pNetPool, pData); } if (pClBlk) { logMsg("cluster BLK UNAVAILABLE \n",0,0,0,0,0,0); netClBlkFree(pCS->end.pNetPool, pClBlk); } if ( pMblk ) { logMsg("MBLK UNAVAILABLE \n",0,0,0,0,0,0); netMblkFree(pCS->end.pNetPool, pMblk); } /* Increment the input error count */ END_ERR_ADD(&pCS->end, MIB2_IN_ERRS, +1); return; } /* Copy the received frame from the chip to the buffer - data is already available in rxFramePtr */ memcpy(pData, rxFramePtr->data, rxFramePtr->data_len, cpRetval); /* Join the cluster to the MBlock */ *clblkretval = netClBlkJoin(pClBlk, pData, DataLen, NULL, 0, 0, 0); *mblkval = netMblkClJoin(pMblk, pClBlk); /* set flags, pkthdr and flags of the Mblks */ pMblk->mBlkHdr.mLen = DataLen; pMblk->mBlkHdr.mFlags |= M_PKTHDR; pMblk->mBlkPktHdr.len = DataLen; END_RCV_RTN_CALL(&pCS->end, pMblk); logMsg("#####DATA SENT TO UPPER LAYERS#### \n",0,0,0,0,0,0); #################################################### Any suggestion/help from Vxworks gurus is welcome, Dhruvesh --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Machine Check Exception Date: Fri, 18 Jan 2002 06:48:40 GMT From: "Gary M" Organization: AT&T Broadband Message-ID: References: I agree with Scott: attempting to disable the machine check exception will only postpone death. Have you eliminated hardware issues, such as bus transaction aborts, RAM refresh problems, or bad configuration of SIU registers? "martin brook" wrote in message news:a27pcp$pv8$1@knossos.btinternet.com... > Hi, > > I'm having problems with my application getting random machine check > exceptions (0x200). The exception address reported does not correspond to > any code which may cause a machine check. I understand that the exception > reporting can be delayed so giving no clue as to where the exception is > occuring. > > Has anyone got a method of debugging these exceptions ? or a way of turning > the machine check off ? > > My app is doing a lot of serial I/O on ports 0/1 which are set to > 115200baud. > > TIA martin > > > > --- > Outgoing mail is certified Virus Free. > Checked by AVG anti-virus system (http://www.grisoft.com). > Version: 6.0.313 / Virus Database: 174 - Release Date: 1/3/02 > > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: why extern interrupt cause vxworks reboot?mpc860 Date: Fri, 18 Jan 2002 06:52:55 GMT From: "Gary M" Organization: AT&T Broadband Message-ID: References: <5624c1bf.0201172056.1b0dc0ef@posting.google.com> And what does your interrupt service routine actually do? That would be the likely cause of the crash. (Oh my, I hope it's not a "printf"...) "bigtornado" wrote in message news:5624c1bf.0201172056.1b0dc0ef@posting.google.com... > hi,all: > I use windriver FADS mpc860 board,I connect a ISR to extern > interrupt IRQ1,but when program running in full speed and interrupt > occurs,the system reboot,I debug the interrupt driver ppc860intr.c > step by step, then the system don't reboot and work ok,why? > Your answer is appreciated!Thanks! > jeffy --------------------------- Newsgroups: comp.os.vxworks Subject: Couldn't open C:\TORNADO\kuku Date: 18 Jan 2002 07:42:25 GMT From: Christophe Organization: Infosources Message-ID: <1011339745.881464@remus.infonie.fr> Hi, i am trying to use VxWorks in order to obtain some informations about a device. I have installed Tornado 2 Demo License, but when i try to connect to the device, even if the device is recognized from Tornado BSP Hi-FOCuS SU board - PowerPC 860 SAR CPU PPC860 OS VxWorks version 5.3.1 i receive the error message Error: Couldn't open C:\TORNADO\kuku: status = 0x2 Thanks PS: Hi-FOCuS is an ADSL model with a lot of possibilities (routing, NAT, DHCP, etc...), but ECI refuse to say how to extend the modem to use these extra features, and i can just use this modem as a modem :( even if he can act as a routeur --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Machine Check Exception Date: 18 Jan 2002 00:48:22 -0800 From: john_94501@yahoo.com (John) Organization: http://groups.google.com/ Message-ID: <488e459a.0201180048.23f0ed9d@posting.google.com> References: Hello, Which type of PPC are you using, and what version of VxWorks? Also, what else is running on your target at the time, or is it just the serial output? In particular, are you using things like watchdogs or POSIX timers? Rgds, John... "martin brook" wrote in message news:... > Hi, > > I'm having problems with my application getting random machine check > exceptions (0x200). The exception address reported does not correspond to > any code which may cause a machine check. I understand that the exception > reporting can be delayed so giving no clue as to where the exception is > occuring. > > Has anyone got a method of debugging these exceptions ? or a way of turning > the machine check off ? > > My app is doing a lot of serial I/O on ports 0/1 which are set to > 115200baud. > > TIA martin > > > > --- > Outgoing mail is certified Virus Free. > Checked by AVG anti-virus system (http://www.grisoft.com). > Version: 6.0.313 / Virus Database: 174 - Release Date: 1/3/02 --------------------------- Newsgroups: comp.os.vxworks Subject: Re: _protoSwIndex and inetsw[ ] used in our OSPF implementation Date: 18 Jan 2002 01:40:21 -0800 From: sitalakshmi.gautham@wipro.com (Sitalakshmi Gautham) Organization: http://groups.google.com/ Message-ID: <2fe901df.0201180140.58208894@posting.google.com> References: <1d411918.0201091833.3205a89f@posting.google.com> In addition to this, you will need to do the following: ip_protox[IPPROTO_OSPF] = p_ospf_protocol_switch - inetsw; (basically the index 7) since ipintr() invokes the receive routine by (*inetsw[ip_protox[ip->ip_id]].pr_input)(); - - Sita lihuapeng@21cn.com (DragonSpring) wrote in message news:<1d411918.0201091833.3205a89f@posting.google.com>... > Hi, > > A question about _protoSwIndex and > inetsw[IP_PROTO_NUM_MAX].IP_PROTO_NUM_MAX is defined as 8 in VxWorks. > > Environment: target: PPC860 rev. D4 > Host: WinNT 4.0, Tornado 2.0/VxWorks 5.4 for PPC, Tornado > 2.0 Cumulative Patch 2 (SPR T2CP2), NPT 5.4(TDK-13274), NPT patch(SPR > 30894) > > We are developing our OSPF protocol (rfc2328). We want to get all OSPF > packet from IP layer. So we use the code as following: > > int ospf_init() > { > FAST struct protosw *p_ospf_protocol_switch; > > if (_protoSwIndex >= ((sizeof(inetsw))/(sizeof(inetsw[0])))) > return (-1) ; > > p_ospf_protocol_switch = &inetsw [_protoSwIndex]; > > if (p_ospf_protocol_switch == NULL) > return (-1); > > if (p_ospf_protocol_switch->pr_domain != NULL) > return (1); /* already initialized */ > > p_ospf_protocol_switch->pr_type = SOCK_RAW; > p_ospf_protocol_switch->pr_domain = &inetdomain; > p_ospf_protocol_switch->pr_protocol = IPPROTO_OSPF; > p_ospf_protocol_switch->pr_flags = PR_ATOMIC|PR_ADDR; > p_ospf_protocol_switch->pr_input = ospf_receive; > p_ospf_protocol_switch->pr_output = ospf_send; > p_ospf_protocol_switch->pr_ctlinput = raw_ctlinput; > p_ospf_protocol_switch->pr_ctloutput = rip_ctloutput; > p_ospf_protocol_switch->pr_usrreq = rip_usrreq; > p_ospf_protocol_switch->pr_init = 0 ; > p_ospf_protocol_switch->pr_fasttimo = 0; > p_ospf_protocol_switch->pr_slowtimo = 0; > p_ospf_protocol_switch->pr_drain = 0; > p_ospf_protocol_switch->pr_sysctl = 0; > > _protoSwIndex++; > > ... > } > > void ospf_receive(struct mbuf *m) > { > /* determine which interface this OSPF packet is received from */ > ... > } > > void ospf_send(struct mbuf *m) > { > ... > } > > We invoke ospf_init() in usrAppInit() (which is in usrAppInit.c) to > start the OSPF process. > > Now my problem is that I can never get any OSPF packet in > ospf_recieve(). It seems that the IP protocol stack never call > ospf_receive(). OSPF takes inetsw[7]. inetsw[IP_PROTO_NUM_MAX] seems > take no effect. Why? Does the IP protocol discard the OSPF packet? Is > there any problem in the VxWorks configurtion? BTW, we have observed > the motFecEnd.c (END driver) has already passed the OSPF packet to MUX > layer. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: VxWorks, Walnut 405GP tftp download error Date: 18 Jan 2002 02:43:01 -0800 From: arun4vxworks@indiatimes.com (Arun Prasad) Organization: http://groups.google.com/ Message-ID: References: <488e459a.0201171316.26b1891d@posting.google.com> Hi, Thank you so much for your kind/appropriate information and that's what we were looking for also. Thanks again. Regards Arun john_94501@yahoo.com (John) wrote in message news:<488e459a.0201171316.26b1891d@posting.google.com>... > Hello, > > You need to replace the boot ROM on the target system. Perhaps you > were sent an actual VxWorks boot ROM with Tornado, if not, you will > need to program an AMD29F040 flash device with the image that was on > the CD. I just used the one that was on the board, and replaced the > IBM monitor, but you may want to keep that somewhere safe and use a > new part. > > To program the boot ROM (actually flash), you will need either an > external programmer, or one of the JTAG debug tools that IBM has for > the board and a copy of RISCwatch. See the BSP's documentation for > more information. > > HTH, > > John... > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: /var/tmp/....errors while compling assembly routines Date: 18 Jan 2002 03:01:08 -0800 From: alimohd@logiceastern.com (Ali) Organization: http://groups.google.com/ Message-ID: References: "Dave Korn" wrote in message news:... > Ali wrote in message ... > >hi DaveK > > Thanx for the answer but i would like to tell u that i have write > >perms to /var as well as tmp directory. Also the dir is not full. > > I am using sun sparc system and runs tornado 2.0 on it . I am using > >the PPCEC403 tool chain for compilation but when i compile evry time a > >new error is generated in the /var/tmp/**** directory .the **** here > >means that every time this file changes here. i don't understand why > >thus happens. > >waiting fo rur reply and if any body can help me in resolving it. > > thanx in advance > >ali > > Those aren't errors, those are temporary files. > > DaveK Hi DaveK Thanx for the you are right these are not errors but why then the the file does not build ,,as there are no errors except this. thanx ali --------------------------- Newsgroups: comp.os.vxworks Subject: Re: pppInit() (IPCP address negotiation with PPP) Date: Fri, 18 Jan 2002 11:15:08 +0000 (UTC) From: David Laight Organization: BT Openworld Message-ID: <3C48040E.D2871B13@nohope.l8s.co.uk> References: <3C47420C.C591228F@fh-gelsenkirchen.de> Gerhard Juen wrote: > > Hi all, > > I want to connect a target ( a standard 80486 PC ) over the serial > interface and a modem to an internet provider. In the > IPCP address negotiation phase, our provider expects us > to send a "0.0.0.0" as recommendated local and remote IP adsress. > > We use the command PPPInit( .. ). Any IP-Adress in the > corresponding parameter fields which is not "0.0.0.0" is passed > correctly as it is to our provider. If we take "0.0.0.0" or a NULL pointer > (as proposed in the vxworks documentation for sending "0.0.0.0") > our program ignores our parameter setting and passes the > IP-address of the Ethernet network card which is also installed in the PC. > > Does anyone know how to pass a "0.0.0.0" to the provider during > address negotiation? Don't think you can without patching the WRS code. The whole thing is stuck deeper into the mire than you can possibly image. If you pass 0.0.0.0 (or null) the hostname is grabbed and a name to address lookup done. The function used will try to use DNS (if configured) even when passed a numeric address. If a route to your DNS server exists (and an NFS mount might leave one lurking) it will have to timeout the DNS request. David --------------------------- Newsgroups: comp.os.vxworks Subject: Re: workQPanic: Kernel work queue overflow PROBLEM Date: Fri, 18 Jan 2002 06:37:11 -0500 From: joe durusau Organization: BELLSOUTH.net Message-ID: <3C4808E7.B77F4E6C@bellsouth.net> References: The usual suspects are: 1. Doing to damn much in an ISR. Ideally, an ISR acknowledges the int, gathers hardware info as required, and does a semGive to allow the main wor to proceed. 2. An ISR that exits earlier than the real hardware ack. In some systems, (mostly powerPCs,) sending an ack does not mean that it got onto the external bus. Look for 'write-posting' or maybe 'caching' with respect to devices on a VME bus or some such. You need to turn caching off for external (otuside the CPU card) devices. You need to ensure that wrwites are completed before exiting an ISR. 3. A device that requires a hardware command be sent to turn off an int, and the ISR is not sending the required command. 4. An attempt to do a whole bunch of things at once, like 20 semGives in an ISR. Good luck, and Speaking only for myself, Joe Durusau Nejhdeh Ghevondian wrote: > Dear anyone! > > I lately increased my options in the VxWorks image file by including things > like POSIX timers etc. Things that require further ISR routines to be > attached. > > However, in my application i further require routines to become ISRs, like > SCSI devices. > Now I get the error during bootup of workQPanic: Kernel work queue overflow > error. > The Windriver help file suggests that I ether: > 1. intLock() then intUnlock the place were the work Q becomes error. > I tried this but still get the error. > 2. Create a variable that can find out Work Q is full. > HOW DO I DO THIS????? > 3. Increase the priority of netTask. > I COULDNT find the file to do this. > > Reagrds > > Nejhdeh Ghevondian --------------------------- End of New-News digest ********************** From vxwexplo-errs@csg.lbl.gov Fri Jan 18 05:52:41 2002 From: Rainer Stelzer Date: Fri Jan 18 05:52:43 PST 2002 Subject: Highlander CORBA Hi vxworkers, this question is addressed to all "Highlander communication's Visibroker for C++ CORBA for Tornado" users. We have trouble migrating from version 3.2.2.1 to 4.1c. Under heavy load conditions the system behavior is'nt stable anymore, which wasn't the case when running version 3.2.2.1 Our CORBA application is unchanged except we now use interoperable Naming Service and the POA instead of the BOA. Has somebody already moved to version 4.1c ? Did you made similar experiences ? Any hint is welcome. rgds R.Stelzer vxworks Tornado ===================================================== Gebrüder Stoye GmbH department: traffic control systems name: Rainer Stelzer adr: Longericher Strasse 177 city: Cologne, Germany zip: 50739 mobil: 0173/7266090 phone: ++49/221/1707019 fax: ++49/221/1707010 mailto: r.stelzer@stoye.de www: http://www.stoye.de ===================================================== From vxwexplo-errs@csg.lbl.gov Fri Jan 18 07:05:23 2002 From: "Jeffrey Rada" Date: Fri Jan 18 07:05:25 PST 2002 Subject: VxWorks, Walnut 405GP tftp download error This is a multi-part message in MIME format. ------_=_NextPart_001_01C1A030.CE9B1D6D Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable Hi, This may not be what you are looking for but we replaced the default boot rom that came with the walnut with a version of ppcboot (open source Power PC boot loader) This works great with tftp and downloading binary images. We take the vxWorks.st file and convert it to a binary file which we then tftp to the target walnut board. =20 This the command we use to create the binary image file from the vxWorks.st file =20 >powerpc-eabi-objcopy -O binary vxWorks.st walnut.bin =20 good luck =20 Original message ********************************************************************** We are using Vxworks on IBM's Walnut PPC 405GP board. The bootrom on it is the same one which came with this board. We are trying to boot the system(walnut) using TFTP and download the=20 vxWorks image. The following error message is displayed: >Not a valid boot image file =20 We got this sequence a couple of times too: Loading file "/tftpboot/mydir1/vxWorks" ... Sending tftp boot request ... Unknown destination, block 1 Unknown destination, block 1 Unknown destination, block 1 Unknown destination, block 1 Unknown destination, block 1 Unknown destination, block 1 Unknown destination, block 1 Unknown destination, block 1 Unknown destination, block 1 TFTP boot failed. I am looking for info on how to build a tftp downloadable vxworks image for the Walnut 405GP system. Is it looking for some thing specific at some specific location ... Does the file type i.e., Binary or Hex matter? Do we need use elfHex or linker with any specific options, or any changes to the config? Looking forward for an early reply. Thanks in advance, ******************************************************************** Jeffrey A. Rada=20 SR. Software Engineer=20 Wave7 Optics Inc.=20 1075 Windward Ridge Pkwy. Suite 170 Alpharetta, GA. 30005=20 678-339-1034=20 mailto:jeffrey.rada@wave7optics.com=20 http://www.wave7optics.com =20 =20 ------_=_NextPart_001_01C1A030.CE9B1D6D Content-Type: text/html; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable

Hi,

This = may not be what=20 you are looking for but we replaced the default boot rom that came with = the=20 walnut with a version of ppcboot (open source Power PC boot loader) This = works=20 great with tftp and downloading binary images. We take the vxWorks.st = file and=20 convert it to a binary file which we then tftp to the target walnut=20 board.
 
This the command = we use to=20 create the binary image file from the vxWorks.st = file
 
>powerpc-eabi-objcopy -O binary = vxWorks.st =20 walnut.bin
 
good=20 luck
 
Original=20 message

****************************************************************= ******

We are using Vxworks on IBM's Walnut = PPC 405GP=20 board.
The bootrom on it is the same one which = came with=20 this board.
We are trying to boot the = system(walnut) using TFTP=20 and download the
vxWorks image.
The following error message is=20 displayed:
>Not a valid boot image = file
 
We got this sequence a couple of times=20 too:
Loading file "/tftpboot/mydir1/vxWorks" = ...
Sending tftp boot request = ...
Unknown destination, block = 1
Unknown destination, block = 1
Unknown destination, block = 1
Unknown destination, block = 1
Unknown destination, block = 1
Unknown destination, block = 1
Unknown destination, block = 1
Unknown destination, block = 1
Unknown destination, block = 1
TFTP boot failed.
I am looking for info on how to build a = tftp=20 downloadable vxworks
image for the Walnut 405GP = system.
Is it looking for some thing specific = at some=20 specific location ...
Does the file type i.e., Binary or Hex=20 matter?
Do we need use elfHex or linker with = any specific=20 options,
or any changes to the = config?
Looking forward for an early = reply.
Thanks in advance,

**********************************************= **********************

Jeffrey A. Rada
SR. Software Engineer
Wave7 Optics=20 Inc.
1075 Windward Ridge Pkwy. = Suite=20 170
Alpharetta, GA. 30005
678-339-1034
mailto:jeffrey.rada@wave7opt= ics.com=20
http://www.wave7optics.com

 
------_=_NextPart_001_01C1A030.CE9B1D6D-- From vxwexplo-errs@csg.lbl.gov Fri Jan 18 20:19:41 2002 From: Dale Luck Date: Fri Jan 18 20:19:43 PST 2002 Subject: reserving a block udp ports This message is in MIME format. Since your mail reader does not understand this format, some or all of this message may not be legible. ------_=_NextPart_001_01C1A09F.B3C7E230 Content-Type: text/plain; charset="iso-8859-1" I working on a project where I need to listen on dozens of different udp port numbers. I'd prefer to not need to open a socket for each one, but would like to listen intercept the packets in the ethernet driver, look to see if they are destined for me, pull them out, and let the rest go on up through the stack. I'm worried about the possible inefficiency of vxworks udp stack for determining who gets the packets, if there are dozens of ports it needs to check for each packet coming in, its likely to not be very efficient. So is there away I can reserve a udp port address space for myself and vxworks won't try to assign the same udp ports to some other process that is creating its own sockets? Dale ------_=_NextPart_001_01C1A09F.B3C7E230 Content-Type: text/html; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable reserving a block udp ports

I working on a project where I need to = listen on dozens of different udp port numbers. I'd prefer to not need = to open a socket for each one, but would like to listen intercept the = packets in the ethernet driver, look to see if they are destined for = me, pull them out, and let the rest go on up through the = stack.

I'm worried about the possible = inefficiency of vxworks udp stack for determining who gets the packets, = if there are dozens of ports it needs to check for each packet coming = in, its likely to not be very efficient.

So is there away I can reserve a udp = port address space for myself and vxworks won't try to assign the same = udp ports to some other process that is creating its own = sockets?

Dale

------_=_NextPart_001_01C1A09F.B3C7E230-- From vxwexplo-errs@csg.lbl.gov Sat Jan 19 04:03:17 2002 From: Vxworks Exploder Date: Sat Jan 19 04:03:20 PST 2002 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Sat Jan 19 04:03:13 PST 2002 Subject: Re: Machine Check Exception Subject: vxWorks image execution problem (target reboots continously) Subject: Re: help urgent.... how to enable MULTICAST flag.... Subject: Re: workQPanic: Kernel work queue overflow PROBLEM Subject: Re: vxWorks image execution problem (target reboots continously) Subject: Re: Couldn't open C:\TORNADO\kuku Subject: Re: PMC module on MVME5100 (mv5100) Subject: Any experiences with integrated debugging within Sniff+ (4.0x) Subject: Re: why extern interrupt cause vxworks reboot?mpc860 Subject: Re: Any experiences with integrated debugging within Sniff+ (4.0x) Subject: Re: Interrupt Latency Problem!!! Subject: Re: help urgent.... how to enable MULTICAST flag.... Subject: Re: Using a PCI ethernet adapter on PCPentium BSP Subject: Re: Interrupt Latency Problem!!! Subject: test Subject: Re: VxWorks, Walnut 405GP tftp download error Subject: Re: VxWorks, Walnut 405GP tftp download error Subject: How to allocate memory for PCI driver? Subject: multicast over PPP Subject: Multicast over ppp Subject: PCM Module operability Subject: Re: linking problem Subject: Re: linking problem Subject: Re: PMC module on MVME5100 (mv5100) Subject: SSH for VxWorks Subject: porting dbm onto vxworks Subject: OT : Spambot Fodder Subject: Checksum do not match [read carefully] Subject: FILE DESCIPTOR LOCKING using fcntl Subject: Re: How to allocate memory for PCI driver? Subject: WindNet PPP、PPPoE Subject: Re: FILE DESCIPTOR LOCKING using fcntl Subject: Re: Checksum do not match [read carefully] ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: Machine Check Exception Date: 18 Jan 2002 04:03:07 -0800 From: kannan@comneti.com (kannan) Organization: http://groups.google.com/ Message-ID: <2ede7703.0201180403.48974006@posting.google.com> References: Hi, Whether your code is accessing a memory area which is not existing or you are storing a value in a memory area which is not existing. This may lead to machine check exception. you can simple hack the the machine check exception by excVecSet your own C routine. And inside your own handler you can increment PC(program counter). But this is worst thing to do ....Instead you can verify your code. You can go step by step first inside your handler you can check the taskName and taskId of the task which creates exception. Confirm that there is a bug in a single task or many tasks are creating those exceptions.So that u can isolate that code alone. Inside your own handler do a taskRestart for that particular task and check whether that particular task is again creating exceptions frequently. If so then verify your task with stack trace are see where exactly that exception occurs and handle accordingly. kannan "Gary M" wrote in message news:... > I agree with Scott: attempting to disable the machine check exception will > only postpone death. > > Have you eliminated hardware issues, such as bus transaction aborts, RAM > refresh problems, or bad configuration of SIU registers? > > "martin brook" wrote in message > news:a27pcp$pv8$1@knossos.btinternet.com... > > Hi, > > > > I'm having problems with my application getting random machine check > > exceptions (0x200). The exception address reported does not correspond > to > > any code which may cause a machine check. I understand that the exception > > reporting can be delayed so giving no clue as to where the exception is > > occuring. > > > > Has anyone got a method of debugging these exceptions ? or a way of > turning > > the machine check off ? > > > > My app is doing a lot of serial I/O on ports 0/1 which are set to > > 115200baud. > > > > TIA martin > > > > > > > > --- > > Outgoing mail is certified Virus Free. > > Checked by AVG anti-virus system (http://www.grisoft.com). > > Version: 6.0.313 / Virus Database: 174 - Release Date: 1/3/02 > > > > --------------------------- Newsgroups: comp.os.vxworks Subject: vxWorks image execution problem (target reboots continously) Date: 18 Jan 2002 04:06:15 -0800 From: godha_rohit@yahoo.com (Rohit) Organization: http://groups.google.com/ Message-ID: <33cac97f.0201180406.6aeebfaf@posting.google.com> HI, I am new to vxworks and facing a problem. Please help me with this if any one knows about it. I am using a tornada 2.o. I did a make to get a Vxworks image and a vxworks.sym file. Then I loaded that image on my target (ppc860 compatible board). I got the following: boot device : cpm unit number : 0 processor number : 0 host name : sun file name : /usr/mysw/vxWorks inet on ethernet (e) : host inet (h) : user (u) : ftp password (pw) : flags (f) : 0x3 target name (tn) : turret startup script (s) : /usr/mysw/myscript.cmd Attached TCP/IP interface to cpm0. Subnet Mask: 0xffffff00 Attaching network interface lo0... done. Loading... 1662844 Starting at 0x100000... after this my target get reboots continously. I don't able to decode why it is happening. Please help me with this and let me know if more information is needed to solve this. Thanks Rohit --------------------------- Newsgroups: comp.os.vxworks Subject: Re: help urgent.... how to enable MULTICAST flag.... Date: 18 Jan 2002 12:32:28 GMT From: Johan Borkhuis Organization: Agere Systems Message-ID: References: <505213fc.0201171704.17c9cc06@posting.google.com> hrkim@hiper.co.kr (Hongrae Kim) wrote: > how to enable MULTICAST flags ? > when I run ifShow(), it does not shows MULTICAST flag. That is something that your driver enables on startup. Probably this driver does not support multicast. Groeten, Johan - -- o o o o o o o . . . _____________________________ o _____ || Johan Borkhuis | .][__n_n_|DD[ ====_____ | borkhuis@agere.com | >(________|__|_[_________]_|__________________________| _/oo OOOOO oo` ooo ooo 'o!o!o o!o!o` === VxWorks FAQ: http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html === --------------------------- Newsgroups: comp.os.vxworks Subject: Re: workQPanic: Kernel work queue overflow PROBLEM Date: 18 Jan 2002 12:38:53 GMT From: Johan Borkhuis Organization: Agere Systems Message-ID: References: "Nejhdeh Ghevondian" wrote: > Dear anyone! > > I lately increased my options in the VxWorks image file by including > things like POSIX timers etc. Things that require further ISR routines > to be attached. > > However, in my application i further require routines to become ISRs, > like SCSI devices. > Now I get the error during bootup of workQPanic: Kernel work queue > overflow error. This error usually indicates that there are more interrupts coming in than the kernel can deal with, or that you do not clear the source of the interrupt. Do you clear the interrupt source in your ISR when you are ready with the ISR? If not the interrupt can remain active and start another ISR-call. One way to check where the source of the problem is, is to enable the devices by hand from the shell (call your init routines from the target shell). Groeten, Johan - -- o o o o o o o . . . _____________________________ o _____ || Johan Borkhuis | .][__n_n_|DD[ ====_____ | borkhuis@agere.com | >(________|__|_[_________]_|__________________________| _/oo OOOOO oo` ooo ooo 'o!o!o o!o!o` === VxWorks FAQ: http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html === --------------------------- Newsgroups: comp.os.vxworks Subject: Re: vxWorks image execution problem (target reboots continously) Date: 18 Jan 2002 12:41:48 GMT From: Johan Borkhuis Organization: Agere Systems Message-ID: References: <33cac97f.0201180406.6aeebfaf@posting.google.com> godha_rohit@yahoo.com (Rohit) wrote: > Attached TCP/IP interface to cpm0. > Subnet Mask: 0xffffff00 > Attaching network interface lo0... done. > Loading... 1662844 > Starting at 0x100000... > > after this my target get reboots continously. I don't able to decode > why it is happening. Please help me with this and let me know if more > information is needed to solve this. This seems like a problem in your application. Try to start with an image that does not have any functionality. This is to get a starting point. If you have an image that downloads start integrating all your components one by one. Groeten, Johan - -- o o o o o o o . . . _____________________________ o _____ || Johan Borkhuis | .][__n_n_|DD[ ====_____ | borkhuis@agere.com | >(________|__|_[_________]_|__________________________| _/oo OOOOO oo` ooo ooo 'o!o!o o!o!o` === VxWorks FAQ: http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html === --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Couldn't open C:\TORNADO\kuku Date: 18 Jan 2002 12:50:00 GMT From: Johan Borkhuis Organization: Agere Systems Message-ID: References: <1011339745.881464@remus.infonie.fr> Christophe wrote: > I have installed Tornado 2 Demo License, You mean Tornado Prototyper? If so, Tornado Prototyper is only a simulator that cannot be used to work with "real" VxWorks systems. > but when i try to connect to the device, > even if the device is recognized from Tornado What do you mean with "recognized from Tornado"? Can you connect to it using Tornado, or can you connect to it from Hyperterm? > BSP Hi-FOCuS SU board - PowerPC 860 SAR > CPU PPC860 > OS VxWorks version 5.3.1 This is a pre-Tornado 2.0 version. > PS: Hi-FOCuS is an ADSL model with a lot of possibilities > (routing, NAT, DHCP, etc...), but ECI refuse to say > how to extend the modem to use these extra features, > and i can just use this modem as a modem :( even if he can > act as a routeur They probably have good reasons to refuse this. The extra features are probably enabled in a more expensive model, or are not reliable enough. Groeten, Johan - -- o o o o o o o . . . _____________________________ o _____ || Johan Borkhuis | .][__n_n_|DD[ ====_____ | borkhuis@agere.com | >(________|__|_[_________]_|__________________________| _/oo OOOOO oo` ooo ooo 'o!o!o o!o!o` === VxWorks FAQ: http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html === --------------------------- Newsgroups: comp.os.vxworks Subject: Re: PMC module on MVME5100 (mv5100) Date: 18 Jan 2002 05:04:29 -0800 From: vincent_hue@yahoo.com (Vincent) Organization: http://groups.google.com/ Message-ID: <67bf868f.0201180504.56268c84@posting.google.com> References: Is INCLUDE_IPMC761 defined in your BSP's config.h file ? Vincent strick@hiwaay.net (BS) wrote in message news:... > The problem I have is when I install a PMC module into my MVME5100 I > am unable to boot the system. The PMC module I have supports both > 33MHZ and 66 MHZ operation and work just fine on a MVME2700. > > Without the PMC card the system boots just fine. I have also been able > to boot the system with a PMC module that does not support 66MHZ > operations, so it doesn't seem that anything is wrong with the 5100 > itself. > > Has anyone seen similar problems like this that might be able offer a > suggestion for resolving this issue. --------------------------- Newsgroups: comp.os.vxworks Subject: Any experiences with integrated debugging within Sniff+ (4.0x) Date: 18 Jan 2002 05:30:32 -0800 From: heiko_elger@arburg.com (Heiko Elger) Organization: http://groups.google.com/ Message-ID: <97e6274e.0201180530.7859b0c1@posting.google.com> Hello, we are using VxWorks 5.4 (Tornado 2.02) with Inten x86 targets with Windows NT Hosts. Cause we have lots of problems in using the crosswind debugger, we think about to use the integrated Sniff+ debugger. Our FAE told us that this will solve lots of our Problems. Cause Sniff+ is very expensive ..... Perhaps other users use another debuggers ... Best regards Heiko Elger --------------------------- Newsgroups: comp.os.vxworks Subject: Re: why extern interrupt cause vxworks reboot?mpc860 Date: Fri, 18 Jan 2002 14:11:21 GMT From: "Koen Van Hulle" Organization: Xeikon Message-ID: References: <5624c1bf.0201172056.1b0dc0ef@posting.google.com> Just an idea... The source of this problem may be of the same kind as the 'workQpanic' problem (incorrect interrupt handling or too high interrupt load). I've seen the workQpanic problem cause my PPC system to reboot before the appropriate error message could be displayed. You'll find tens of discussions on this in this newsgroup and on WindRiver's website. Regards, Koen "bigtornado" wrote in message news:5624c1bf.0201172056.1b0dc0ef@posting.google.com... > hi,all: > I use windriver FADS mpc860 board,I connect a ISR to extern > interrupt IRQ1,but when program running in full speed and interrupt > occurs,the system reboot,I debug the interrupt driver ppc860intr.c > step by step, then the system don't reboot and work ok,why? > Your answer is appreciated!Thanks! > jeffy --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Any experiences with integrated debugging within Sniff+ (4.0x) Date: Fri, 18 Jan 2002 14:17:48 GMT From: "Koen Van Hulle" Organization: Xeikon Message-ID: References: <97e6274e.0201180530.7859b0c1@posting.google.com> Hi Heiko, We also had a lot of problems (reliability, lack of features) with the standard T2/PPC GNU debugger. We arranged that WindRiver gave us the SingeStep debugger as one of the WindPower tools that come with Tornado Power Pro. Thus far, we're very happy with it (PPC 405 target) Regards, Koen "Heiko Elger" wrote in message news:97e6274e.0201180530.7859b0c1@posting.google.com... > Hello, > > we are using VxWorks 5.4 (Tornado 2.02) with Inten x86 targets with > Windows NT Hosts. > Cause we have lots of problems in using the crosswind debugger, we > think about to use the integrated Sniff+ debugger. > Our FAE told us that this will solve lots of our Problems. > > Cause Sniff+ is very expensive ..... > > Perhaps other users use another debuggers ... > > > Best regards > > Heiko Elger --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Interrupt Latency Problem!!! Date: Fri, 18 Jan 2002 14:30 +0000 (GMT Standard Time) From: rew@cp.cix.co.uk (Richard Wenham) Organization: Nextra UK Message-ID: References: Reply-To: rew@cp.cix.co.uk We have always had exactly the same problem. Our target is PCx86 or PCPentium, and we have tried most of the available network interfaces. I came to the same conclusion, that the interrupts are switched off somewhere in the network interface or stack. The worst latency we have seen is around 220us. No assistance could be obtained from WRS under the standard maintenance contract. We can survive the poor latency since our ISR is 1 to 2kHZ, with important data staticised on the clock edge in hardware, so we just put up with it. Richard In article , anonymous@nospam.org (Chris Smith) wrote: > > > We are using Tornado2/VxWorks5.4 on a MVME230x target running at > 350MHz. The application is industrial equipment where we move around a > dozen or so mechanical parts as fast as we can over a one foot stroke, > but to very high precision (sub-micron). This means we require fast > servo sample rates (say 10kHz). > > We have set up a fixed-rate, external IRQ for the servo interrupt which > we set using the Raven MPIC at the highest priority of 15. The default > ethernet IRQ is set at 14 which is one lower than highest. All other > IRQ priorities are lower than these two. This same bug happens whether > we use our external PCI IRQ or, instead, use the VxWorks sysAuxClk > routines (note: the sysAuxClk routines use the Raven Timer0 counter at > IRQ priority 15 also). > > We measure interrupt latency the usual way -- our external device > generates the external PCI IRQ, then the first thing the ISR does is > toggle a bit on our external device. We use a logic analyzer to > capture this time difference. Alternatively, I made a simple test case > when I reported this bug to WindRiver that shows the exact same bad > behavior using only the target board and software (code is below). The > software- > only and logic analyzer timings matched very well to each other which > gives me confidence in both techniques. > > With no network activity, the interrupt latency is an acceptable 3-5us. > If I make the PCI bus busy or the serial port busy, it stays at the > same 3-5us. However, if I make the network busy, then the latency can > glitch up to 47us for the given simple test case. WindRiver reproduced > this problem recently and they said it glitched as high as 78us for a > very similar target (though I don't know their CPU speed). They also > reproduced it on a MVME5100 target with latencies around 14us (much > better but still too high in my opinion). > > I have been hounding WindRiver for a month and a half about this. Up > until jan 7th their response was that I should pay for special > consulting services from them. Finally WindRiver opened up an SPR (# > 72433), I think just to shut me up. > > My uneducated guess as to what's happening is that during the ethernet > ISR, the network driver must do some kind of intLock which prevents my > ISR from starting on time, even though my IRQ has the higher IRQ > priority. I just bought WindView and there is something strange going > on when my ISR interrupts the network ISR (strange meaning the network > ISR takes a lot longer to finish than if it was not interrupted). I > have yet to synchronize the latency glitch with something I can see in > WindView, as I'm still getting used to this new tool. I guess the docs > say I can genereate a user event, so I'll try that next...... > > So, finally, my questions are: > > 1. Have other hard-real-time users on PPC targets seen this behavior? > Am I the only one who thinks this is serious? > > 2. Can anyone suggest some possible work-arounds? > > > Here is the test code: > > /**************************************************************** > Note that mySysTimeBaseLGet() is a direct copy of the assembly > routine sysTimeBaseLGet() from target\config\mv2304\sysALib.s. > It simply reads the lower PPC Time Base Register using the "mftb" > assembly call. For our board each tick is 60ns (divide 4 of > 66.667MHz bus). > *****************************************************************/ > > volatile unsigned long isrCount=0; > volatile unsigned long cnt=0; > volatile unsigned long cntPrev=0; > volatile unsigned long delta=0; > volatile unsigned long deltaMax=0; > volatile unsigned long deltaMin=LONG_MAX; > > void myIsr(int level) > { > cntPrev = cnt; > cnt = mySysTimeBaseLGet(); > > delta = cnt - cntPrev; > > if(delta < deltaMin) > deltaMin = delta; > > if(delta > deltaMax) > deltaMax = delta; > > if( isrCount < 10 ) > { > deltaMin = LONG_MAX; > deltaMax = 0; > } > > isrCount++; > } > > long myInit(void) > { > sysAuxClkRateSet(1000); > sysAuxClkConnect(myIsr, 0); > sysAuxClkEnable(); > } > > /**********************************************/ --------------------------- Newsgroups: comp.os.vxworks Subject: Re: help urgent.... how to enable MULTICAST flag.... Date: 18 Jan 2002 06:40:33 -0800 From: jzhang@drs.ca (Jason) Organization: http://groups.google.com/ Message-ID: References: <505213fc.0201172044.5349276@posting.google.com> Hello, try this function: ifFlagChange. Regards, hrkim@hiper.co.kr (Hongrae Kim) wrote in message news:<505213fc.0201172044.5349276@posting.google.com>... > how to enable MULTICAST flag. > when I run ifShow(), it does not shows MULTICAST falg cpm0 interface. > > Here's a info on my VxWorks box: PPC860 CPU, Motorola > > cpm (unit number 0): > Flags: (0x63) UP BROADCAST ARP RUNNING > Type: ETHERNET_CSMACD > Internet address: 61.97.37.228 > Broadcast address: 61.97.37.255 > Netmask 0xff000000 Subnetmask 0xffffff00 > Ethernet address is 00:03:a1:19:97:3e > Metric is 0 > Maximum Transfer Unit size is 1500 > 81 packets received; 5 packets sent > 78 multicast packets received > 0 multicast packets sent > 0 input errors; 0 output errors > 0 collisions; 0 dropped --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Using a PCI ethernet adapter on PCPentium BSP Date: 18 Jan 2002 12:38:44 GMT From: Charly Message-ID: <2002118-133844-311193@foorum.com> References: <2002117-101355-104040@foorum.com> <3c46ee95$1@brateggebdc5.br-automation.co.at> Thanks Verner, I leaved "EDIMAX" ( which seems NE2000 compatible) to "3C905C-TX" But no success Now I use a simple ISA adapter "3C509TP ": no success. Perhaps someone could correct me. Test1: I use a direct cross cable bitween VxWorks and a PC Windows + WFTPD. bootline for a boot on network: "elt(0,0)host:vxWorks h=192.0.1.1 b=192.0.1.2 e=192.0.1.2 tn=target u=vxworks pw=vxworks f=0x80 o=elt" => Get VxWorks by TFTP timout Test2:Boot on floppy "fd=0,0(0,0)host:/fd0/vxWorks h=192.0.1.1 b=192.0.1.2 u=target tn=target u=vxworks pw=vxworks o=elt" => start ok in the Shell, I type "ifShow" => Flags 0x8069 UP LOOPBACK MULTICAST ARP RUNNING Type SOFTWARE LOOPBACK Internet address: 127.0.0.1 netmask 0xff00000000 subnet 0xff00000000 0 packets receive 0 packet sent .. I appriciate your suggestion to start my network Thanks Charly - -- Use our news server 'news.foorum.com' from anywhere. More details at: http://nnrpinfo.go.foorum.com/ --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Interrupt Latency Problem!!! Date: Fri, 18 Jan 2002 15:09:22 +0000 (UTC) From: David Laight Organization: BT Openworld Message-ID: <3C483AAB.29D43BCD@nohope.l8s.co.uk> References: Richard Wenham wrote: > > I came to the same conclusion, that the interrupts are switched off > somewhere in the network interface or stack. The worst latency we have > seen is around 220us. I tracked down the place where interrupts were disabled on one system (not vxWorks) by using the following: Connect a logic analiser to the IRQ of the interrupt you are having trouble with, set it to trigger if the interrupt is pending for more than (say) 100us. Connect the 'trigger output' of the analiser to the NMI interrupt of your system. Make the NMI trap into some debug code - the interrupted PC might be enough, but a stack trace is better. Sit back and enjoy the diagnostics. (In our case the PC console screen was being scrolled by reading/writing video memory on a ISA video adapter (on the motherboard on a dual pentium pro motherboard)) David --------------------------- Newsgroups: comp.os.vxworks Subject: test Date: Fri, 18 Jan 2002 09:14:09 -0600 From: PC User Organization: ExecPC Internet - Milwaukee, WI Message-ID: <3c483ea4$0$43577$272ea4a1@news.execpc.com> --------------------------- Newsgroups: comp.os.vxworks Subject: Re: VxWorks, Walnut 405GP tftp download error Date: 18 Jan 2002 08:12:31 -0800 From: arun4vxworks@indiatimes.com (Arun Prasad) Organization: http://groups.google.com/ Message-ID: References: <488e459a.0201171316.26b1891d@posting.google.com> Hi, <> The CD which we have doesn't contain the image which you were mentioning. Can you please send across? Thanks again for your answer. Regards Arun john_94501@yahoo.com (John) wrote in message news:<488e459a.0201171316.26b1891d@posting.google.com>... > Hello, > > You need to replace the boot ROM on the target system. Perhaps you > were sent an actual VxWorks boot ROM with Tornado, if not, you will > need to program an AMD29F040 flash device with the image that was on > the CD. I just used the one that was on the board, and replaced the > IBM monitor, but you may want to keep that somewhere safe and use a > new part. > > To program the boot ROM (actually flash), you will need either an > external programmer, or one of the JTAG debug tools that IBM has for > the board and a copy of RISCwatch. See the BSP's documentation for > more information. > > HTH, > > John... > > arun4vxworks@indiatimes.com (Arun Prasad) wrote in message news:... > > Hi, > > > > We are using Vxworks on IBM's Walnut PPC 405GP board. > > The bootrom on it is the same one which came with this board. > > We are trying to boot the system(walnut) using TFTP and download the > > vxWorks image. > > > > The following error message is displayed: > > >Not a valid boot image file > > > > > > We got this sequence a couple of times too: > > Loading file "/tftpboot/mydir1/vxWorks" ... > > Sending tftp boot request ... > > Unknown destination, block 1 > > Unknown destination, block 1 > > Unknown destination, block 1 > > Unknown destination, block 1 > > Unknown destination, block 1 > > Unknown destination, block 1 > > Unknown destination, block 1 > > Unknown destination, block 1 > > Unknown destination, block 1 > > TFTP boot failed. > > > > I am looking for info on how to build a tftp downloadable vxworks > > image for the Walnut 405GP system. > > > > Is it looking for some thing specific at some specific location ... > > Does the file type i.e., Binary or Hex matter? > > Do we need use elfHex or linker with any specific options, > > or any changes to the config? > > > > Looking forward for an early reply. > > > > Thanks in advance, > > > > -- --------------------------- Newsgroups: comp.os.vxworks Subject: Re: VxWorks, Walnut 405GP tftp download error Date: Fri, 18 Jan 2002 10:30:39 -0600 From: "Tim" Organization: Nicolet Technologies Message-ID: References: <488e459a.0201171316.26b1891d@posting.google.com> Reply-To: "Tim" Sender: moeller@niti.com I had to do the same thing for our Walnut board, even though the WRS documentation suggests that they supply a ROM already programmed. Incidentally, the file I used was BOOTROM_UNCMP.HEX, which is supplied in the installed Tornado\target\config\walnut directory. Also note that they recommend patching that file with the correct start address; this can be done easily with a simple text editor, prior to programming the part. -- Tim "John" wrote in message news:488e459a.0201171316.26b1891d@posting.google.com... > Hello, > > You need to replace the boot ROM on the target system. Perhaps you > were sent an actual VxWorks boot ROM with Tornado, if not, you will > need to program an AMD29F040 flash device with the image that was on > the CD. I just used the one that was on the board, and replaced the > IBM monitor, but you may want to keep that somewhere safe and use a > new part. > > To program the boot ROM (actually flash), you will need either an > external programmer, or one of the JTAG debug tools that IBM has for > the board and a copy of RISCwatch. See the BSP's documentation for > more information. > > HTH, > > John... > --------------------------- Newsgroups: comp.os.vxworks Subject: How to allocate memory for PCI driver? Date: 18 Jan 2002 08:48:33 -0800 From: igtorque@eliop.es (=?ISO-8859-1?Q?Ignacio_Gonz=E1lez?=) Organization: http://groups.google.com/ Message-ID: <9ec53d11.0201180848.30bc26f0@posting.google.com> This should be simple, but I cannot get it working. I have a PC Pentium with the standard pcPentium BSP (with little modifications) and it works fine. The PC has 128 MB, so I set the following BSP Memory Configuration Parameters: LOCAL_MEM_AUTOSIZE = FALSE LOCAL_MEM_SIZE = 0x07000000 (112 MB) USER_RESERVED_MEMORY = 0x01000000 (16 MB, for special purposes) Now, I'm developing a driver for a PCI based frame-grabber (the only PCI board in the PC, by the way), so I set this in config.h: #define INCLUDE_PCI This let me use the PCI library routines. Everything continues to work, but I cannot access the grabber's memory-mapped registers. The frame-grabber registers are located in 0xF0804008 - 0xF0805008 (almost at the top of the 4 GB address space), so I tried to create a partition (memPartCreate) and get memory from it (memPartAlloc) but I can create only a partition of 38 bytes or less (if I try more, I get an exception), and, if I use this small partition, any memPartAlloc (even as low as 10 bytes) returns me 0 reserved bytes. So I changed the above parameters to these: LOCAL_MEM_SIZE = 0x07000000 (112 MB) USER_RESERVED_MEMORY = 0xF0000000 (3,75 GB, enough to reach the above addresses when added to LOCAL_MEM_SIZE) This time, the VxWorks image won't get control after loading it and jumping to its starting address (the console simply get frozen after the jump.) So perhaps it's time to ask the experts. By the way, I'm not using the PCI configuration parameters from the PCI configuration component (I don't know if if must use them.) Any kind of pointer <> NULL would be much appreciated. --------------------------- Newsgroups: comp.os.vxworks Subject: multicast over PPP Date: Fri, 18 Jan 2002 16:50:31 GMT From: "alain tamdem" Organization: Rauland Borg Corporation Message-ID: Reply-To: "alain tamdem" Hi all, I am experiencing problem now to connect my configurtaion rpogram running on my PC to a target under Vxworks using serial port over PPP. In fact my application on my PC is sending the multicast ( 224.1.0.1) over the ppp link but this multicast does not seem to be received on the target side ( the target has also a Network card) Can anybody let me know why the multicast is not received on the PPP0 interface on the target ( on my target I have a socket with is bind to be listen for a multicast message on the serial port and a hook routine as describer in vxworks network programmer guide)?? any hint ?? thank in advance Alain TAMDEM Intern student RAULAND - BORG CORPORATION --------------------------- Newsgroups: comp.os.vxworks Subject: Multicast over ppp Date: Fri, 18 Jan 2002 16:53:34 GMT From: "alain tamdem" Organization: Rauland Borg Corporation Message-ID: Reply-To: "alain tamdem" just to complete my question by saying that the multicast flag is enabled on my ppp interface ( I saw that by doing a ifShow) Alain --------------------------- Newsgroups: comp.os.vxworks Subject: PCM Module operability Date: Fri, 18 Jan 2002 18:53:41 GMT From: Randy Ryan Organization: Our-Kids Message-ID: <3C486F38.FAE28DFA@pacbell.net> In researching use of PCM modules on SBCs, just getting back into this enhancement to single board computers after a too long layover : I'm getting the impression that it is not exactly plug and play and there are variances that get a response that the SBC manufacturer would need the PMC card and drivers, or the PMC card manufacturer would need the SBC to prove out their drivers on first. So how plug and playable in general are PCM modules and associated drivers. Is this a case that it "generally works if you know how to tweak out the SBC BSP", or is there really variable differences in how the PCI PCM models are accessed/implmented per SBC manufacturer that indeed makes it impossible for PCM cards to be generically used, with the exception being Motorla cards which appears to be the standard baseline to which drivers are developed. Guess I'm trying to determine what all the caveats and gotcha's are here... Thanks, Randy NovaSol, http://www.nova-sol.com NovaSol is looking for Sr. Engineers that want to work and live in Hawaii --------------------------- Newsgroups: comp.os.vxworks Subject: Re: linking problem Date: Fri, 18 Jan 2002 20:53:12 GMT From: "Joe Hagen" Organization: TDS.NET Internet Services www.tds.net Message-ID: References: <3C46D729.70802@notifier-is.net> David, If this was "C" code, then the linkage would work. However, in C++, "const" will cause the variable to have internal linkage, so the behavior you're seeing is because the internal linkage "hides" the variable from other modules. Joe "david lindauer" wrote in message news:3C46D729.70802@notifier-is.net... > In one C++ file I have: > > typedef char * sz ; > sz const monthlist[12] = { "jan","feb",... } ; > > and in another: > > typedef char *sz ; > extern sz const monthlist[12] ; > > For some reason LD insists that the reference in the second file can not > be resolved. I have used nm on the object files and both list > '_monthlist', unadorned. One is type 't' and one is type 'U'. So it > should work. Both files are definitely in the project. > > I have tried changing the name of the variable with no success. Anyone > know what I am doing wrong? > > David > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: linking problem Date: Fri, 18 Jan 2002 21:08:45 GMT From: "Joe Hagen" Organization: TDS.NET Internet Services www.tds.net Message-ID: References: <3C46D729.70802@notifier-is.net> David, One way to "undo" the fact that const causes internal linkage on this variable is to declare it like this in your first module: extern sz const monthlist[12] = { "jan","feb",... } ; I believe the linker will be able to resolve it with this change. Joe "Joe Hagen" wrote in message news:YW%18.3850$TH3.1031133@kent.svc.tds.net... > David, > > If this was "C" code, then the linkage would work. > > However, in C++, "const" will cause the variable to > have internal linkage, so the behavior you're seeing > is because the internal linkage "hides" the variable > from other modules. > > Joe > > "david lindauer" wrote in message > news:3C46D729.70802@notifier-is.net... > > In one C++ file I have: > > > > typedef char * sz ; > > sz const monthlist[12] = { "jan","feb",... } ; > > > > and in another: > > > > typedef char *sz ; > > extern sz const monthlist[12] ; > > > > For some reason LD insists that the reference in the second file can not > > be resolved. I have used nm on the object files and both list > > '_monthlist', unadorned. One is type 't' and one is type 'U'. So it > > should work. Both files are definitely in the project. > > > > I have tried changing the name of the variable with no success. Anyone > > know what I am doing wrong? > > > > David > > > > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: PMC module on MVME5100 (mv5100) Date: 18 Jan 2002 13:46:03 -0800 From: strick@hiwaay.net (BS) Organization: http://groups.google.com/ Message-ID: References: <67bf868f.0201180504.56268c84@posting.google.com> Thanks for all of you help. With the help of Bill Dennen and Yves I was able to get this working. Inside sysLibPci.c there is an excluded list of PCI devices that prevent those devices from being auto configured. Once I added my id's into this list I was able to get the 5100 to boot correctly. I have an init call inside sysHwInit that does all of the PCI configuration for the device. Thanks again vincent_hue@yahoo.com (Vincent) wrote in message news:<67bf868f.0201180504.56268c84@posting.google.com>... > Is INCLUDE_IPMC761 defined in your BSP's config.h file ? > > Vincent > > strick@hiwaay.net (BS) wrote in message news:... > > The problem I have is when I install a PMC module into my MVME5100 I > > am unable to boot the system. The PMC module I have supports both > > 33MHZ and 66 MHZ operation and work just fine on a MVME2700. > > > > Without the PMC card the system boots just fine. I have also been able > > to boot the system with a PMC module that does not support 66MHZ > > operations, so it doesn't seem that anything is wrong with the 5100 > > itself. > > > > Has anyone seen similar problems like this that might be able offer a > > suggestion for resolving this issue. --------------------------- Newsgroups: comp.os.vxworks Subject: SSH for VxWorks Date: Fri, 18 Jan 2002 17:03:04 -0500 From: Marty Stich Message-ID: <3C489B98.7470BFE1@hotmail.com> Can anyone make a recommendations where to find or from who to purchase a port of SSH to VxWorks? Thanks. --------------------------- Newsgroups: comp.os.vxworks Subject: porting dbm onto vxworks Date: 18 Jan 2002 16:17:01 -0800 From: ganesh_kumarg@rediffmail.com (ganesh godavari) Organization: http://groups.google.com/ Message-ID: <4e22fcd6.0201181617.7100d150@posting.google.com> hai everyone, i am new to VxWorks and also this group. i am trying to port gdbm-1.8.0 on to VxWorks. i need sys/file.h for compiling. i am not able to find this file in my VxWorks header file list. Does VxWorks have this header file?. if so from where can i download it? if VxWorks doesn't have it? how can i overcome this problem?. this problem must have been encountered by a lot of you previously. so please help me out. Thanks in advance, gkgodava --------------------------- Newsgroups: comp.os.vxworks Subject: OT : Spambot Fodder Date: Sat, 19 Jan 2002 01:59:20 GMT From: alanstv@ntlworld.com Organization: Virgin Net Usenet Service Message-ID: Hi My Name is Jeff, I thought it might be amusing to give a few people on one of the newsgroups some grief, however looks like I messed with the wrong person becuase they seem to have posted my email addy on 250,000 newsgroups. alanstv@ntlworld.com alanstv@ntlworld.com alanstv@ntlworld.com alanstv@ntlworld.com alanstv@ntlworld.com alanstv@ntlworld.com alanstv@ntlworld.com alanstv@ntlworld.com alanstv@ntlworld.com alanstv@ntlworld.com alanstv@ntlworld.com alanstv@ntlworld.com alanstv@ntlworld.com alanstv@ntlworld.com alanstv@ntlworld.com alanstv@ntlworld.com alanstv@ntlworld.com alanstv@ntlworld.com alanstv@ntlworld.com alanstv@ntlworld.com alanstv@ntlworld.com alanstv@ntlworld.com alanstv@ntlworld.com alanstv@ntlworld.com alanstv@ntlworld.com alanstv@ntlworld.com alanstv@ntlworld.com alanstv@ntlworld.com alanstv@ntlworld.com alanstv@ntlworld.com alanstv@ntlworld.com alanstv@ntlworld.com alanstv@ntlworld.com alanstv@ntlworld.com alanstv@ntlworld.com alanstv@ntlworld.com alanstv@ntlworld.com alanstv@ntlworld.com alanstv@ntlworld.com alanstv@ntlworld.com alanstv@ntlworld.com alanstv@ntlworld.com alanstv@ntlworld.com alanstv@ntlworld.com alanstv@ntlworld.com alanstv@ntlworld.com alanstv@ntlworld.com alanstv@ntlworld.com alanstv@ntlworld.com alanstv@ntlworld.com alanstv@ntlworld.com alanstv@ntlworld.com alanstv@ntlworld.com alanstv@ntlworld.com alanstv@ntlworld.com alanstv@ntlworld.com alanstv@ntlworld.com alanstv@ntlworld.com alanstv@ntlworld.com alanstv@ntlworld.com alanstv@ntlworld.com alanstv@ntlworld.com alanstv@ntlworld.com alanstv@ntlworld.com alanstv@ntlworld.com alanstv@ntlworld.com alanstv@ntlworld.com alanstv@ntlworld.com alanstv@ntlworld.com alanstv@ntlworld.com alanstv@ntlworld.com alanstv@ntlworld.com alanstv@ntlworld.com alanstv@ntlworld.com alanstv@ntlworld.com alanstv@ntlworld.com alanstv@ntlworld.com alanstv@ntlworld.com alanstv@ntlworld.com alanstv@ntlworld.com alanstv@ntlworld.com alanstv@ntlworld.com alanstv@ntlworld.com alanstv@ntlworld.com alanstv@ntlworld.com alanstv@ntlworld.com alanstv@ntlworld.com alanstv@ntlworld.com alanstv@ntlworld.com alanstv@ntlworld.com alanstv@ntlworld.com alanstv@ntlworld.com alanstv@ntlworld.com alanstv@ntlworld.com alanstv@ntlworld.com alanstv@ntlworld.com alanstv@ntlworld.com alanstv@ntlworld.com alanstv@ntlworld.com alanstv@ntlworld.com alanstv@ntlworld.com alanstv@ntlworld.com alanstv@ntlworld.com alanstv@ntlworld.com alanstv@ntlworld.com alanstv@ntlworld.com alanstv@ntlworld.com alanstv@ntlworld.com alanstv@ntlworld.com alanstv@ntlworld.com alanstv@ntlworld.com alanstv@ntlworld.com alanstv@ntlworld.com alanstv@ntlworld.com alanstv@ntlworld.com alanstv@ntlworld.com alanstv@ntlworld.com alanstv@ntlworld.com alanstv@ntlworld.com alanstv@ntlworld.com alanstv@ntlworld.com alanstv@ntlworld.com alanstv@ntlworld.com alanstv@ntlworld.com alanstv@ntlworld.com alanstv@ntlworld.com --------------------------- Newsgroups: comp.os.vxworks Subject: Checksum do not match [read carefully] Date: 18 Jan 2002 21:18:36 -0500 From: Bill Pringlemeir Organization: Bell Sympatico Message-ID: Sender: bpringlemeir@DeadDuck I have this error, Connecting to target agent... succeeded. Attaching C++ interface... succeeded. Attaching coff OMF reader for ARM CPU family... succeeded. Warning: Target checksum: 0xb781 (computed from 0x1000 to 0x201400). Host checksum: 0x5e3a (computed from 0xc70118 to 0xe70518). Warning: Core file checksums do not match. The reason is that my processor is an ARM 710A and the vxWorks image has been compiled with Tornado 2.1.1. In order to run the image, I must translate the hex code of some instruction to another that will cause the undefined opcode exception to be raised. I then emulate the instruction there. So, the checksums do not match, they shouldn't match... how can I make them match? Is there some way to notify the target server that the BSP has patched the image or can I modify the checksum routine? regards, Bill Pringlemeir. - -- I think you once were untainted three-quarters of the time. Could be.. vxWorks FAQ, "http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html" --------------------------- Newsgroups: comp.os.vxworks Subject: FILE DESCIPTOR LOCKING using fcntl Date: 18 Jan 2002 18:47:51 -0800 From: ganesh_kumarg@rediffmail.com (ganesh godavari) Organization: http://groups.google.com/ Message-ID: <4e22fcd6.0201181847.792ff168@posting.google.com> hello, i am working on porting my gdbm-1.8.0 onto vxworks. the dbm relies on file locking while updating the file. i was looking into VxWorks fcntl.h header file for fcntl function. i could not find the function. is there any releated function of fcntl call in VxWorks? if so please inform me. has anyone written a wrap around fucntion?. if so can u tell me how u have written one. Thanks in advance, gkgodava --------------------------- Newsgroups: comp.os.vxworks Subject: Re: How to allocate memory for PCI driver? Date: Sat, 19 Jan 2002 03:44:47 GMT From: "Iosif Harutyunov" Organization: AT&T Broadband Message-ID: References: <9ec53d11.0201180848.30bc26f0@posting.google.com> Ignacio, You need to create MMU mapping by adding address space to sysPhysMemDesc[] You can do it by calling: sysMmuMapAdd((void *)(membase & PCI_MEMBASE_MASK), memLength, VM_STATE_MASK_FOR_ALL, VM_STATE_FOR_PCI); Make this call after pciIntLibInit() in sysLib.c. Note that this will create one-to-one mapping between physical and logical memory. After that you should be able access your device memory space. Hope this will help. Iosif,_ "Ignacio González" wrote in message news:9ec53d11.0201180848.30bc26f0@posting.google.com... > This should be simple, but I cannot get it working. > > I have a PC Pentium with the standard pcPentium BSP (with little > modifications) and it works fine. The PC has 128 MB, so I set the > following BSP Memory Configuration Parameters: > > LOCAL_MEM_AUTOSIZE = FALSE > LOCAL_MEM_SIZE = 0x07000000 (112 MB) > USER_RESERVED_MEMORY = 0x01000000 (16 MB, for special purposes) > > Now, I'm developing a driver for a PCI based frame-grabber (the only > PCI board in the PC, by the way), so I set this in config.h: > > #define INCLUDE_PCI > > This let me use the PCI library routines. Everything continues to > work, but I cannot access the grabber's memory-mapped registers. > > The frame-grabber registers are located in 0xF0804008 - 0xF0805008 > (almost at the top of the 4 GB address space), so I tried to create a > partition (memPartCreate) and get memory from it (memPartAlloc) but I > can create only a partition of 38 bytes or less (if I try more, I get > an exception), and, if I use this small partition, any memPartAlloc > (even as low as 10 bytes) returns me 0 reserved bytes. > > So I changed the above parameters to these: > > LOCAL_MEM_SIZE = 0x07000000 (112 MB) > USER_RESERVED_MEMORY = 0xF0000000 (3,75 GB, enough to reach the above > addresses when added to LOCAL_MEM_SIZE) > > This time, the VxWorks image won't get control after loading it and > jumping to its starting address (the console simply get frozen after > the jump.) > > So perhaps it's time to ask the experts. By the way, I'm not using the > PCI configuration parameters from the PCI configuration component (I > don't know if if must use them.) > > Any kind of pointer <> NULL would be much appreciated. --------------------------- Newsgroups: comp.os.vxworks Subject: WindNet PPP、PPPoE Date: 18 Jan 2002 23:28:46 -0800 From: sardine_sz@163.net (sardine) Organization: http://groups.google.com/ Message-ID: hi gurus, is there have anyone with experence about WindNet PPP or PPPoE ? in high appreciate ! --------------------------- Newsgroups: comp.os.vxworks Subject: Re: FILE DESCIPTOR LOCKING using fcntl Date: Sat, 19 Jan 2002 11:25:10 +0200 From: "Leonid Rosenboim" Organization: Verio Message-ID: References: <4e22fcd6.0201181847.792ff168@posting.google.com> There eint no file locking in VxWorks anywhere. You should write your own wrapper and lock the DBM access using a Mutex semaphore, which can be done at file access level or at the upper DBM API level. If small amount of processing is anticipated, it eint matter where you do the lock, top o rbottom. "ganesh godavari" wrote in message news:4e22fcd6.0201181847.792ff168@posting.google.com... > hello, > i am working on porting my gdbm-1.8.0 onto vxworks. the dbm relies > on file locking while updating the file. i was looking into VxWorks > fcntl.h header file for fcntl function. i could not find the function. > > is there any releated function of fcntl call in VxWorks? if so please > inform me. > > has anyone written a wrap around fucntion?. if so can u tell me how u > have written one. > > Thanks in advance, > gkgodava --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Checksum do not match [read carefully] Date: Sat, 19 Jan 2002 10:26:00 +0000 (UTC) From: David Laight Organization: BT Openworld Message-ID: <3C494A08.AAC74C88@nohope.l8s.co.uk> References: Bill Pringlemeir wrote: > > I have this error, > > Connecting to target agent... succeeded. > Attaching C++ interface... succeeded. > Attaching coff OMF reader for ARM CPU family... succeeded. > Warning: Target checksum: 0xb781 (computed from 0x1000 to 0x201400). > Host checksum: 0x5e3a (computed from 0xc70118 to 0xe70518). > Warning: Core file checksums do not match. > > The reason is that my processor is an ARM 710A and the vxWorks image > has been compiled with Tornado 2.1.1. In order to run the image, I > must translate the hex code of some instruction to another that will > cause the undefined opcode exception to be raised. I then emulate the > instruction there. > > So, the checksums do not match, they shouldn't match... how can I make > them match? Is there some way to notify the target server that the > BSP has patched the image or can I modify the checksum routine? I don't think it matters that the checksums don't match, it is only a warning. It is important that the symbol tables match! I never saw a checksum success! probably because I was patching the loaded code (to fix things which cannot be bugs or WRS wouldn't ship them - would they?) before the checksums were calculated. David --------------------------- End of New-News digest ********************** From vxwexplo-errs@csg.lbl.gov Sun Jan 20 04:03:10 2002 From: Vxworks Exploder Date: Sun Jan 20 04:03:12 PST 2002 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Sun Jan 20 04:03:07 PST 2002 Subject: why dscc4 dead on 2Mbits/s ? Subject: Tornado 2.0 Prototyper for Linux? Subject: Re: Tornado 2.0 Prototyper for Linux? Subject: Re: WindNet PPP、PPPoE Subject: Re: Checksum do not match [read carefully] Subject: Re: BootRom and MMU Subject: errno 226 Subject: Senior Embedded Systems Developer Subject: Booting from ATA Compact Flash Subject: VxWorks, Walnut 405GP tftp download error - Thank you!!! ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: why dscc4 dead on 2Mbits/s ? Date: 19 Jan 2002 05:13:27 -0800 From: ljn_98@163.net (Tim) Organization: http://groups.google.com/ Message-ID: i am also writting driver of dscc4 for router,but i met a insolved problem.so i wish you experience on that,because i dont know if this is a bug of dscc4. OS: VXWORKS , Tornado II CPU: MPC8240 problem: Four channel work on address 0 model with 2M bits/s.i spawn a receive task to receive frame.when receive task is working,if a higher task preempte the cpu sourse,that result in whole system dead.i use logical analyser observe PCI signals.i found that /FRAME,/DEVSEL,/IRDY are allway low level ,so i think maybe dscc4 dont release PCI bus resluting in cpu is in a PCI cycle.when i reset dscc4 by put /RST low level,the system is come back again . well,these is bug of dscc4? dscc4 can work on 2Mbits/s stably?this problem afflict me for one month. i wish you help sincerely,thank you! Tim 2002.1.18 --------------------------- Newsgroups: comp.os.vxworks Subject: Tornado 2.0 Prototyper for Linux? Date: Sat, 19 Jan 2002 15:44:36 GMT From: "David" Message-ID: Hi,All; Please tell me if is possible to installed Tornado 2.0 Prototyper in Linux. Wind River Tornado 2.0 Prototyper can be installed in Solaris OS, I have no idea about Solaris OS. Thanks David davidyang_us@yahoo.com --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Tornado 2.0 Prototyper for Linux? Date: 19 Jan 2002 09:40:58 -0800 From: john_94501@yahoo.com (John) Organization: http://groups.google.com/ Message-ID: <488e459a.0201190940.65a9756a@posting.google.com> References: Hello, No, Tornado 2.0 only supports Windoze and Solaris; Tornado 3 supports Linux too, but I don't believe that there is a prototyper for Tornado 3. HTH, John... "David" wrote in message news:... > Hi,All; > Please tell me if is possible to installed Tornado 2.0 Prototyper in Linux. > Wind River Tornado 2.0 Prototyper can be installed in Solaris OS, I have no > idea about Solaris OS. > Thanks > David > davidyang_us@yahoo.com --------------------------- Newsgroups: comp.os.vxworks Subject: Re: WindNet PPP、PPPoE Date: 19 Jan 2002 09:45:03 -0800 From: john_94501@yahoo.com (John) Organization: http://groups.google.com/ Message-ID: <488e459a.0201190945.56d3a175@posting.google.com> References: Hello, Yes, I am sure there are people reading this group who have that experience. However, what did you expect of them? You didn't ask anything about the products... John... sardine_sz@163.net (sardine) wrote in message news:... > hi gurus, > > is there have anyone with experence about WindNet PPP or PPPoE ? > > in high appreciate ! --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Checksum do not match [read carefully] Date: Sat, 19 Jan 2002 21:45:23 +0200 From: "Leonid Rosenboim" Organization: Verio Message-ID: References: You can ignore the checksum error safely, as long as you trust yourself to never mixup the image file and the actual target image. Still, I beleive you might find a better solution if you perform tha patch directly on WRS object library - use arXX to dearchive all objects in libCPUgnuvx.a into ojbCPUgnuvx/ directory, run a program that finds and patches the instructions you need patched, then rename the directory - give it a new CPU name and a new library. Of course if it turns out that the patching is needed in only a few object files, there could be a more elegant soltuon there, depending on where you find these instructions. I would speculate that instructions you patch would not be compiler generated, hence always found i assembly source files ... HTH, - - Leonid "Bill Pringlemeir" wrote in message news:uk7ufrslf.fsf@yahoo.com... > > I have this error, > > Connecting to target agent... succeeded. > Attaching C++ interface... succeeded. > Attaching coff OMF reader for ARM CPU family... succeeded. > Warning: Target checksum: 0xb781 (computed from 0x1000 to 0x201400). > Host checksum: 0x5e3a (computed from 0xc70118 to 0xe70518). > Warning: Core file checksums do not match. > > The reason is that my processor is an ARM 710A and the vxWorks image > has been compiled with Tornado 2.1.1. In order to run the image, I > must translate the hex code of some instruction to another that will > cause the undefined opcode exception to be raised. I then emulate the > instruction there. > > So, the checksums do not match, they shouldn't match... how can I make > them match? Is there some way to notify the target server that the > BSP has patched the image or can I modify the checksum routine? > > regards, > Bill Pringlemeir. > -- > I think you once were untainted three-quarters of the time. Could be.. > vxWorks FAQ, "http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html" --------------------------- Newsgroups: comp.os.vxworks Subject: Re: BootRom and MMU Date: Sat, 19 Jan 2002 19:48:37 GMT From: Ephraim Gadsby Organization: - minimal Message-ID: References: <0y%_7.14882$8e.753501@news> <3C43E9C9.8630FBE3@europem01.nt.com> On Tue, 15 Jan 2002 09:35:21 +0100, Sebastien Fauris wrote: >Hi vxWorkers, > >what I find strange is that in a bootrom >the PPC processor runs in real mode and so >the MMU is disabled. The MMU will be enabled >only in the kernel downloaded afterwards. If, for whatever reason, the system does not have memory physically mapped from zero, then the MMU is almost the first thing that needs to be set-up. --------------------------- Newsgroups: comp.os.vxworks Subject: errno 226 Date: 19 Jan 2002 21:21:19 -0800 From: bhorowit@cs.berkeley.edu (Benjamin Horowitz) Organization: http://groups.google.com/ Message-ID: Hi, When trying to close a file, I encountered errno 226. Could anyone please point me to where this errno is defined? I can't find it in errno.h. The error message I get is: HELLO.TXT: The system cannot find the file specified. HELLO.TXT: Access is denied. I'm using VxWorks 5.4, Tornado 2. Thanks, Ben --------------------------- Newsgroups: comp.os.vxworks,comp.sys.arm,comp.sys.m68k,comp.sys.powerpc.tech,de.comp.os.unix.bsd Subject: Senior Embedded Systems Developer Date: Sun, 20 Jan 2002 06:35:25 GMT From: "Mark Lane" Organization: AT&T Broadband Message-ID: Senior Embedded Systems Developer looking for contract opportunities ... - - pSOS, VxWorks, VRTX, Nucleus, QNX, OSE, INTEGRITY - - C/C++, Assembly, Microcode - - Object-Oriented, UML, Rational Rose, SASD - - PowerPC, Pentium, ARM, 683XX, 68K, TI OMAP - - ClearCase, PVCS, CVS, RCS, SEI/ISO process - - NT, Unix, GNU, Perl, ShellScripting - - TCP/IP, ATM, CAC, QOS, GSM/GPRS, SNMP - - USB, PCI, VME, ISA, DSL, JTAG, I2C, SPI - - BSP Development, Debug GREEN hardware - - Complex software design at any level Mark Lane Time Domain Systems, Inc. 214-232-4091 marklane5@attbi.com --------------------------- Newsgroups: comp.os.vxworks Subject: Booting from ATA Compact Flash Date: 20 Jan 2002 00:34:03 -0800 From: bhorowit@cs.berkeley.edu (Benjamin Horowitz) Organization: http://groups.google.com/ Message-ID: Hi, Please forgive this naive question: I'm trying to boot from an ATA Compact Flash disk. The disk is control 0, drive 0. I create a new bootrom_uncmp using Tornado II (and VxWorks 5.4). I copy bootrom_uncmp to a floppy. I invoke the mkbootAta function from the Tornado shell, as follows: mkbootAta 0, 0, "/fd0/BOOTRO~1 (BOOTRO~1 is the DOS file name.) I can now see bootrom.sys on the ATA drive. Now I reboot, this time without my normal boot floppy. I see the following messages: Searching for Boot Record from Floppy..Not Found Searching for Boot Record from IDE-0..OK But then the machine hangs. Normally I would see: VxLd........ [etc.] Any suggestions about where to look? I'm stuck. Thanks a lot, Ben --------------------------- Newsgroups: comp.os.vxworks Subject: VxWorks, Walnut 405GP tftp download error - Thank you!!! Date: 20 Jan 2002 02:55:32 -0800 From: arun4vxworks@indiatimes.com (Arun Prasad) Organization: http://groups.google.com/ Message-ID: References: <488e459a.0201171316.26b1891d@posting.google.com> Dear Tim, Thank you so much for your valuable Tips. Now our system has come up and we are able to download now. And this is what our project manager says :-) < To: Arun Prasad Date: Sun, 20 Jan 2002 15:12:54 +0530 This tip worked for us, and Walnut seems to be able to download the code now. Please convey our greatful thanks to Tim (moeller@niti.com) Thanks. Bhyrav "Tim" wrote in message news:... > I had to do the same thing for our Walnut board, even though > the WRS documentation suggests that they supply a ROM > already programmed. > > Incidentally, the file I used was BOOTROM_UNCMP.HEX, > which is supplied in the installed Tornado\target\config\walnut > directory. Also note that they recommend patching that > file with the correct start address; this can be done easily > with a simple text editor, prior to programming the part. > > -- Tim > > > "John" wrote in message > news:488e459a.0201171316.26b1891d@posting.google.com... > > Hello, > > > > You need to replace the boot ROM on the target system. Perhaps you > > were sent an actual VxWorks boot ROM with Tornado, if not, you will > > need to program an AMD29F040 flash device with the image that was on > > the CD. I just used the one that was on the board, and replaced the > > IBM monitor, but you may want to keep that somewhere safe and use a > > new part. > > > > To program the boot ROM (actually flash), you will need either an > > external programmer, or one of the JTAG debug tools that IBM has for > > the board and a copy of RISCwatch. See the BSP's documentation for > > more information. > > > > HTH, > > > > John... > > --------------------------- End of New-News digest ********************** From vxwexplo-errs@csg.lbl.gov Sun Jan 20 22:44:24 2002 From: surya lohith Date: Sun Jan 20 22:44:26 PST 2002 Subject: Regarding Small footprint for Vxworks and pSOS Hai i would like to know small footprint of - vxWorks and pSos in terms of (memory size) footprint ? ? +communcation ? ? +shell and graphics ? ? +browser support ? ? i.e when communcation is added what is the foot print like wise for graphics,shell and browser support i would like to know in terms of memory size Thanks in advance Lohith __________________________________________________ Do You Yahoo!? Send FREE video emails in Yahoo! Mail! http://promo.yahoo.com/videomail/ From vxwexplo-errs@csg.lbl.gov Mon Jan 21 04:03:13 2002 From: Vxworks Exploder Date: Mon Jan 21 04:03:15 PST 2002 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Mon Jan 21 04:03:09 PST 2002 Subject: Re: errno 226 Subject: Re: errno 226 Subject: Re: WindNet PPP、PPPoE Subject: Re: Machine Check Exception Subject: Is BPF or raw socket (PF_PACKET, SOCK_RAW,..) for receiving raw ethernet frames available? Subject: Re: Machine Check Exception Subject: Re: How to allocate memory for PCI driver? Subject: Re: Machine Check Exception Subject: vx works for set top box Subject: Why I can not boot through network? Subject: Re: Problem with font in Tornado 2 - Part 2 Subject: Re: Is BPF or raw socket (PF_PACKET, SOCK_RAW,..) for receiving raw ethernet frames available? Subject: windNet pppOEServiceNameAdd( ) ? Subject: Key handling Subject: Is malloc() atomic? Subject: bootpMsgSend() can not wakes up. Subject: Re: porting dbm onto vxworks Subject: Re: vx works for set top box Subject: Re: Is malloc() atomic? ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: errno 226 Date: Sun, 20 Jan 2002 18:04:00 +0200 From: "Leonid Rosenboim" Organization: Verio Message-ID: References: Benny, to make any sense at all, you need to provide more details: what you are trying to achieve, what device / file system you are using etc. "Benjamin Horowitz" wrote in message news:ab1f8a59.0201192121.222e37b2@posting.google.com... > Hi, > > When trying to close a file, I encountered errno 226. Could anyone > please point me to where this errno is defined? I can't find it in > errno.h. > > The error message I get is: > HELLO.TXT: The system cannot find the file specified. > HELLO.TXT: Access is denied. > > I'm using VxWorks 5.4, Tornado 2. > > Thanks, > Ben --------------------------- Newsgroups: comp.os.vxworks Subject: Re: errno 226 Date: 20 Jan 2002 15:07:53 -0800 From: john_94501@yahoo.com (John) Organization: http://groups.google.com/ Message-ID: <488e459a.0201201507.29177ca1@posting.google.com> References: Hello, You are probably using the FTP based netDrv file system. That would explain both the error number, and the fact that it only failed when you closed the file. Firstly, the reason for failing only when you close is that netDrv (whether using ftp or rsh), will only actually write a file on the host file system when you close it. Up until then, the file has been held in memory. You can tell which protocol you are using very easily: if you entered a password in the bootline for the taregt, you'll be using ftp, otherwise you'll be using rsh. Of course, if your host is a Windoze box, you only have the choice of ftp. Now, the error number. This is a little quirk of the ftp client library in VxWorks. It will place any FTP protocol error in the errno field so that you can see it. So, error 226 is probably an FTP error code. Looking this up at http://hpcf.nersc.gov/help/access/ftp_error_codes.html you'll find the following: 226 Closing data connection. Requested file action successful (for example, file transfer or file abort). The clue as to what is happening though is in the messages; "Access Denied" suggests that the ftp user you specified in the bootline does not have write permission. Check the config. of the ftp server (for Windoze), or check whether the user has write access to the directory you are using for Unix. HTH, John... bhorowit@cs.berkeley.edu (Benjamin Horowitz) wrote in message news:... > Hi, > > When trying to close a file, I encountered errno 226. Could anyone > please point me to where this errno is defined? I can't find it in > errno.h. > > The error message I get is: > HELLO.TXT: The system cannot find the file specified. > HELLO.TXT: Access is denied. > > I'm using VxWorks 5.4, Tornado 2. > > Thanks, > Ben --------------------------- Newsgroups: comp.os.vxworks Subject: Re: WindNet PPP、PPPoE Date: 20 Jan 2002 16:43:30 -0800 From: sardine_sz@163.net (sardine) Organization: http://groups.google.com/ Message-ID: References: <488e459a.0201190945.56d3a175@posting.google.com> sorry, i just want to know,if i select WindNet PPP,PPPoE , how much programs i need to code ? thx! --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Machine Check Exception Date: Sun, 20 Jan 2002 21:00:07 -0500 From: Yves Organization: Bell Sympatico Message-ID: <3C4B7627.B304755F@sympatico.ca> References: <488e459a.0201180048.23f0ed9d@posting.google.com> John wrote: > > Hello, > > Which type of PPC are you using, and what version of VxWorks? Also, > what else is running on your target at the time, or is it just the > serial output? In particular, are you using things like watchdogs or > POSIX timers? > > Rgds, > > John... > > "martin brook" wrote in message news:... > > Hi, > > > > I'm having problems with my application getting random machine check > > exceptions (0x200). The exception address reported does not correspond to > > any code which may cause a machine check. I understand that the exception > > reporting can be delayed so giving no clue as to where the exception is > > occuring. > > > > Has anyone got a method of debugging these exceptions ? or a way of turning > > the machine check off ? > > > > My app is doing a lot of serial I/O on ports 0/1 which are set to > > 115200baud. > > > > TIA martin > > > > > > > > --- > > Outgoing mail is certified Virus Free. > > Checked by AVG anti-virus system (http://www.grisoft.com). > > Version: 6.0.313 / Virus Database: 174 - Release Date: 1/3/02 Hi, we are having a similar problem with our system. A machine check exception is happening ... in a streaming method. It runs through the code on a number of occasion, than one time it produces an exception. This problem comes and goes with re-compiles... We have created a ring buffer with which all tasks check in on a task switch, to try and identify the task causing the problem, but have yet to succeed... the machine check exception can be caused by parity, invalid instruction and some bus errors (I don't have my PPC manual with me). Any suggestions on finding (what I assumed to be) the stray pointer would be welcomed at this point. PPC 750, Tornado II (with TCP2) on NT. We are using watchdogs. thanks Yves --------------------------- Newsgroups: comp.os.vxworks Subject: Is BPF or raw socket (PF_PACKET, SOCK_RAW,..) for receiving raw ethernet frames available? Date: 20 Jan 2002 22:12:16 -0800 From: sitalakshmi.gautham@wipro.com (Sitalakshmi Gautham) Organization: http://groups.google.com/ Message-ID: <2fe901df.0201202212.95cabd3@posting.google.com> Is BPF or raw socket(PF_PACKET, SOCK_RAW,..) for receiving raw ethernet frames by the application, available in VxWorks? Any info on alternatives to this will be appreciated. thanks, Sita --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Machine Check Exception Date: Mon, 21 Jan 2002 06:44:26 GMT From: "Gary M" Organization: AT&T Broadband Message-ID: References: <488e459a.0201180048.23f0ed9d@posting.google.com> <3C4B7627.B304755F@sympatico.ca> On the MPC8xx family at least, machine checks will also happen if you reference a physical address for which the memory controller (not the MMU) is configured but the target responds incorrectly or is unresponsive at that particular address. Additionally, bus targets which provide their own TA (Transfer Acknowledge) to the core but fail to do so before the core times out will cause a machine check. "Yves" wrote in message news:3C4B7627.B304755F@sympatico.ca... > > Hi, > > we are having a similar problem with our system. A machine check > exception is happening ... in a streaming method. It runs through the > code on a number of occasion, than one time it produces an exception. > This problem comes and goes with re-compiles... We have created a ring > buffer with which all tasks check in on a task switch, to try and > identify the task causing the problem, but have yet to succeed... > > the machine check exception can be caused by parity, invalid instruction > and some bus errors (I don't have my PPC manual with me). > > Any suggestions on finding (what I assumed to be) the stray pointer > would be welcomed at this point. > > PPC 750, Tornado II (with TCP2) on NT. We are using watchdogs. > > thanks > > Yves --------------------------- Newsgroups: comp.os.vxworks Subject: Re: How to allocate memory for PCI driver? Date: Mon, 21 Jan 2002 09:16:33 +0100 From: "Michael Lawnick" Organization: Buergernetz Dillingen Message-ID: References: <9ec53d11.0201180848.30bc26f0@posting.google.com> Reply-To: "Michael Lawnick" Sender: mlawnick@pd9e6542e.dip.t-dialin.net Hi Ignacio, you have got USER_RESERVED_MEMORY wrong! It's the amount of memory to be subtracted from automatically detected (LOCAL_MEM_AUTOSIZE) or defined (LOCAL_MEM_SIZE) RAM. This mem won't be cleared on reboot and may be used e.g. for WindView's post mortem debug. Memory partition creation is the second bad way. If you do that, system will add the space to the memory used for programms and variables. Just setup your device's PCI-configuration (should already be done by BIOS) and check that the memory spaces of the config-header (those with their BAR addresses ending with least bit 0, use pciDeviceShow and pciHeaderShow ) lay within your PCI-memory configuration of your sysPhysMemDesc in sysLib.c HTH - -- Mit freundlichen Grüßen, Michael Lawnick ============================================== SOFTEC GmbH Tel +49-731-96600-0 Promenade 17 Fax +49-731-96600-23 D-89073 Ulm Michael Lawnick Germany lawnick@softec.de ============================================== "Ignacio González" schrieb im Newsbeitrag news:9ec53d11.0201180848.30bc26f0@posting.google.com... > This should be simple, but I cannot get it working. > > I have a PC Pentium with the standard pcPentium BSP (with little > modifications) and it works fine. The PC has 128 MB, so I set the > following BSP Memory Configuration Parameters: > > LOCAL_MEM_AUTOSIZE = FALSE > LOCAL_MEM_SIZE = 0x07000000 (112 MB) > USER_RESERVED_MEMORY = 0x01000000 (16 MB, for special purposes) > > Now, I'm developing a driver for a PCI based frame-grabber (the only > PCI board in the PC, by the way), so I set this in config.h: > > #define INCLUDE_PCI > > This let me use the PCI library routines. Everything continues to > work, but I cannot access the grabber's memory-mapped registers. > > The frame-grabber registers are located in 0xF0804008 - 0xF0805008 > (almost at the top of the 4 GB address space), so I tried to create a > partition (memPartCreate) and get memory from it (memPartAlloc) but I > can create only a partition of 38 bytes or less (if I try more, I get > an exception), and, if I use this small partition, any memPartAlloc > (even as low as 10 bytes) returns me 0 reserved bytes. > > So I changed the above parameters to these: > > LOCAL_MEM_SIZE = 0x07000000 (112 MB) > USER_RESERVED_MEMORY = 0xF0000000 (3,75 GB, enough to reach the above > addresses when added to LOCAL_MEM_SIZE) > > This time, the VxWorks image won't get control after loading it and > jumping to its starting address (the console simply get frozen after > the jump.) > > So perhaps it's time to ask the experts. By the way, I'm not using the > PCI configuration parameters from the PCI configuration component (I > don't know if if must use them.) > > Any kind of pointer <> NULL would be much appreciated. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Machine Check Exception Date: 21 Jan 2002 00:47:53 -0800 From: john_94501@yahoo.com (John) Organization: http://groups.google.com/ Message-ID: <488e459a.0201210047.2915df0@posting.google.com> References: <488e459a.0201180048.23f0ed9d@posting.google.com> <3C4B7627.B304755F@sympatico.ca> Hello, > the machine check exception can be caused by parity, invalid instruction > and some bus errors (I don't have my PPC manual with me). Could be a stray pointer, but some other things to check on: 1) Stack sizes for all your tasks; make sure none have overflowed. 2) Make sure that you set VX_FP_TASK when needed. You might need it for other cases than just FP when using PPC - the GNU compiler for PPC has some "interesting" optimisations that cause it to borrow FP regs! Search the archives here for info on this, or just set VX_FP_TASK on all tasks and see if it magically fixes the problem. 3) Check your watchdog routines carefully; they are interrupt handlers and must follow the rules for ISRs. Those are my general guidelines for trapping these sorts of things. The other thing to look at is the address that caused the crash - is it an address that corresponds to something else? HTH, John... > Any suggestions on finding (what I assumed to be) the stray pointer > would be welcomed at this point. > > PPC 750, Tornado II (with TCP2) on NT. We are using watchdogs. > > thanks > > Yves --------------------------- Newsgroups: comp.os.vxworks Subject: vx works for set top box Date: 21 Jan 2002 00:56:19 -0800 From: akpkum@yahoo.com (Anil Kumar) Organization: http://groups.google.com/ Message-ID: <36827350.0201210056.419501d5@posting.google.com> hi All, Can VxWorks be used as an OS for a Set Top Box ? Any spl. Advice on this sunject. bye Anil Kumar.P akpkum@yahoo.com --------------------------- Newsgroups: comp.os.vxworks Subject: Why I can not boot through network? Date: 21 Jan 2002 01:07:07 -0800 From: wxh@founder.com.cn (Wen Xiaohui) Organization: http://groups.google.com/ Message-ID: <8fc342d0.0201210107.6eb9ecc4@posting.google.com> I have developed an END driver using DP83902A. The END driver works well when I use it in a VxWorks_ROM image. But when I want to make a boot_rom and try to boot through the network, the target server can not connect to the board, and I can not ping to the board, either. Are there any problems in my project configuration? I have undefine the INCLUDE_SERIAL macro in my "config.h". Thanks a lot! --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Problem with font in Tornado 2 - Part 2 Date: Mon, 21 Jan 2002 12:21:05 +0300 From: "Shrike" Organization: MTU-Intel ISP Message-ID: References: Thanks! It work. :) "Adam Ciszewski" ÓÏÏÂÝÉÌ/ÓÏÏÂÝÉÌÁ × ÎÏ×ÏÓÔÑÈ ÓÌÅÄÕÀÝÅÅ: news:a26klh$8q1$1@zeus.man.szczecin.pl... > > "Shrike" wrote: > > > I know about Font/Size tab in Options, but there is a blank space in size > > selector. And nothing can be written there. > > Please, download Service Pack 3 for Tornado 2 > from ftp.windriver.com > /pub/updates/tornado/SPR/sprT2CP3.tar > > Regards > AC > > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Is BPF or raw socket (PF_PACKET, SOCK_RAW,..) for receiving raw ethernet frames available? Date: 21 Jan 2002 01:25:01 -0800 From: john_94501@yahoo.com (John) Organization: http://groups.google.com/ Message-ID: <488e459a.0201210125.625aa82e@posting.google.com> References: <2fe901df.0201202212.95cabd3@posting.google.com> Hello, There are a number of ways to get raw ethernet frames, including BPF in the newer releases of VxWorks (and VxWorks AE). Check out the network programmer's guide for more information on this. HTH, John... sitalakshmi.gautham@wipro.com (Sitalakshmi Gautham) wrote in message news:<2fe901df.0201202212.95cabd3@posting.google.com>... > Is BPF or raw socket(PF_PACKET, SOCK_RAW,..) for receiving raw > ethernet frames by the application, available in VxWorks? Any info on > alternatives to this will be appreciated. > > thanks, > Sita --------------------------- Newsgroups: comp.os.vxworks Subject: windNet pppOEServiceNameAdd( ) ? Date: 21 Jan 2002 01:35:32 -0800 From: sardine_sz@163.net (sardine) Organization: http://groups.google.com/ Message-ID: hi folks, now i am debug windNet PPP&PPPOE(V1.2+ForV1.2Patch1.0) i add ppp and pppoe to vxworks, when i start PPPOE service use API pppOEServiceNameAdd there have follow error: 1. if invoke like this: pppOEServiceNameAdd(pppOeSysProfile,"",&applCallbacks,NULL) i can query the PPPOE service (name is pppOeAC:-1) with EnterNet300. 2. if invoke like this: pppOEServiceNameAdd(pppOeSysProfile,"gold",&applCallbacks,NULL) i cannot query any PPPOE service,why ? other: how i config ipcp_addr in ppp ? any info appreciate ! --------------------------- Newsgroups: comp.os.vxworks Subject: Key handling Date: 21 Jan 2002 01:38:11 -0800 From: patrick.molenaar@fcs-cs.com (Patrick) Organization: http://groups.google.com/ Message-ID: <986c860f.0201210138.7f41ba5e@posting.google.com> Hi vxWork-ers, I have a problem, i tried to make an application that uses the function keys F1 till F12, but i encountered a problem with the F3 key. In my main window i have a event that checks which key is pressed, but this function is not called when the F3 function key is pressed. As soon as the F3 is pressed the application quits without entering the event function. How can i get all function keys to work in my application? I have put a copy of the event function in this e-mail: ZafEventType MyMainWnd::Event(const ZafEventStruct &event) { ZafEventType ccode = LogicalEvent(event); ZafEventType eventType = event.InputType(); // Keyboard handling if (eventType == E_KEY) { // On key-down event if (event.modifiers & S_KEYDOWN) { // Get pressed key information const ZafKeyStruct *key = event.Key(); // Output the pressed key to the screen char mytext[25]; sprintf(mytext, "R=%d M=%d V=%d S=%d",event.rawCode, event.modifiers, key->value, key->shiftState); text[count++]->SetText(mytext); if (count >= MAX) count = 0; } } ccode = ZafWindow::Event(event); return (ccode); } Another problem i have is that the functionality of this function differs between windows and VxWorks. When i compile and run my application using Zinc 6.0 for windows then every function key does work correct, but i will receive TWO events for each key, one wit key value zero (indicating a special key is pressed and the next is the value of the key. When i run exactly the same program under VxWorks i will only receive one keyboard event but with a completely differen Key().rawCode. Does this mean my code is not portable, or is there a solution to make the functionality equal for windows and VxWorks. The program should work under VxWorks eventually, so my first problem is the most important one. I hope someone knows an answer to my questions, Best regards, Patrick Molenaar --------------------------- Newsgroups: comp.os.vxworks Subject: Is malloc() atomic? Date: 21 Jan 2002 01:54:00 -0800 From: arun4vxworks@indiatimes.com (Arun Prasad) Organization: http://groups.google.com/ Message-ID: Hi, Can anyone tell me about malloc() implementation in VxWorks? I'm using IBM PPC 405GP board. I want to know whether the entire implementation of malloc is within tasklock (I mean is it atomic?). Or, Do we need to task lock before malloc() and then task unlock as task switches may occur while the task is within malloc as the entire implementation of malloc may not be within taskLock? Hoping for a clarification on the implementation of malloc Have a nice time. Regards Arun --------------------------- Newsgroups: comp.os.vxworks Subject: bootpMsgSend() can not wakes up. Date: 21 Jan 2002 02:30:17 -0800 From: ren_sir@hotmail.com (rensir) Organization: http://groups.google.com/ Message-ID: <3cebd70f.0201210230.1f675412@posting.google.com> Has anyone used with a successful result ? I encountered a problem when I use bootpMsgSend(). It looks that BOOTP request being sent away to the server, and a reply is being sent back, but the bootpMsgSend() routine never wakes up. Instead it just re-transmits the bootp request. What can I do? Would you please give some advice? --------------------------- Newsgroups: comp.os.vxworks Subject: Re: porting dbm onto vxworks Date: 21 Jan 2002 02:51:58 -0800 From: tut_pharoh@yahoo.com (tutankhamun) Organization: http://groups.google.com/ Message-ID: <88c8dc81.0201210251.7fff27f1@posting.google.com> References: <4e22fcd6.0201181617.7100d150@posting.google.com> There should be a configuration file by name autoconf.h with gdbm distribution. Here, look for the following line: /* Define if you have the header file. */ #define HAVE_SYS_FILE_H 1 change it to /* Define if you have the header file. */ #define HAVE_SYS_FILE_H 0 Should fix your problem. tutankhamun =================================================================== ganesh_kumarg@rediffmail.com (ganesh godavari) wrote in message news:<4e22fcd6.0201181617.7100d150@posting.google.com>... > hai everyone, > i am new to VxWorks and also this group. > > i am trying to port gdbm-1.8.0 on to VxWorks. i need sys/file.h for > compiling. i am not able to find this file in my VxWorks header file > list. > > Does VxWorks have this header file?. if so from where can i download > it? > if VxWorks doesn't have it? how can i overcome this problem?. > > this problem must have been encountered by a lot of you previously. so > please help me out. > > Thanks in advance, > gkgodava --------------------------- Newsgroups: comp.os.vxworks Subject: Re: vx works for set top box Date: Mon, 21 Jan 2002 13:27:28 +0200 From: "Leonid Rosenboim" Organization: Verio Message-ID: References: <36827350.0201210056.419501d5@posting.google.com> VxWorks can be and is used on set tops, perhaps it is the MOST POPULAR OS for STBs. Look at Liberate Technoligies, they sell "middlewear" which icludes VxWorks to several of the largest STB vendors for Digital cable. On the other hand I know a few STB vendors who use Linux, which has some of its own merits. Here are some sample points: With VxWorks you save RAM, smapper OS footprint. With Linux time-to-prototype is by order of magnitude shorter then with VxWorks. Would I recommend VxWorks (over Linux) for a new STB is a totally different question, it depends on many factors, both technical is economical, for an intelligent and objective advise. - - Leonid Rosenboim "Anil Kumar" wrote in message news:36827350.0201210056.419501d5@posting.google.com... > hi All, > Can VxWorks be used as an OS for a Set Top Box ? > Any spl. Advice on this sunject. > bye > Anil Kumar.P > akpkum@yahoo.com --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Is malloc() atomic? Date: Mon, 21 Jan 2002 13:30:18 +0200 From: "Leonid Rosenboim" Organization: Verio Message-ID: References: Yes, malloc(), actually memPartAlloc() is atomic, and you can call these functions from your tasks without any additional locking. No, malloc() DOES NOT do taskLock or intLock, the protection is dome using Mutex specific to each memory partition. Note, you CAN'T CALL malloc() from within an interrupt service function, including WatchDog callback. Advise: RTFM - - Leonid "Arun Prasad" wrote in message news:e38d912c.0201210153.2de8ca1d@posting.google.com... > Hi, > > Can anyone tell me about malloc() implementation in VxWorks? > I'm using IBM PPC 405GP board. > > I want to know whether the entire implementation of malloc is within > tasklock (I mean is it atomic?). > > Or, > Do we need to task lock before malloc() and then task unlock as task > switches may occur while the task is within malloc as the entire > implementation of malloc may not be within taskLock? > > Hoping for a clarification on the implementation of malloc > > Have a nice time. > Regards > Arun --------------------------- End of New-News digest ********************** From vxwexplo-errs@csg.lbl.gov Tue Jan 22 04:03:17 2002 From: Vxworks Exploder Date: Tue Jan 22 04:03:19 PST 2002 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Tue Jan 22 04:03:12 PST 2002 Subject: Conflict of WDB and http server Subject: Re: Is malloc() atomic? Subject: Re: vx works for set top box Subject: Re: Conflict of WDB and http server Subject: Re: Machine Check Exception Subject: ppp with multicast Subject: Multicast over PPP Subject: Re: Using a PCI ethernet adapter on PCPentium BSP Subject: Re: Support for 8 bits + even parity in ttyDrv? Subject: Re: Checksum do not match [read carefully] Subject: Linker map file description Subject: tffs, tffsBootPutImage and booting from fallow area Subject: Re: Linker map file description Subject: Re: vx works for set top box Subject: Re: Using a PCI ethernet adapter on PCPentium BSP Subject: Re: BootRom and MMU Subject: Re: BootRom and MMU Subject: Re: BootRom and MMU Subject: Re: BootRom and MMU Subject: compiling WTX application on WinNT, VC++ 6 Subject: Re: Why I can not boot through network? Subject: Re: tffs, tffsBootPutImage and booting from fallow area Subject: Re: Booting from ATA Compact Flash Subject: Switching between network interfaces Subject: Re: Interrupt Latency Problem!!! Subject: Re: VxWorks, Walnut 405GP tftp download error Subject: Re: torVars and make Subject: Vxworks beginner Subject: SCSI Performance in vxWorks Subject: Re: Switching between network interfaces Subject: Re: Interrupt Latency Problem!!! Subject: Re: VxWorks, Walnut 405GP tftp download error Subject: Re: Vxworks beginner ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Conflict of WDB and http server Date: Mon, 21 Jan 2002 06:57:54 -0800 From: "Boris" Organization: N/A Message-ID: SGksIHdlIG5vdyBlbmNvdW50ZXIgYSBuZXcgcHJvYmxlbSwgd2hlbiB3ZSBpbnN0YWxsDQpodHRw IHNlcnZlciBvbiBvdXIgc3lzdGVtLCB0aGUgV0RCIGNhbid0IHdvcmsgd2VsbCwNCk9uIGNvbnNv bGUgd2Ugc2VlIGZvbGxvd2luZyBtZXNzYWdlOg0KMHg2ZWY1ODgodFdkYlRhc2spOldEQiBleGNl cHRpb24scmVzdGFydGluZyBhZ2VudCBpbiAxMCBzZWNvbmRzLi4uDQoweDZlZjU4OCh0V2RiVGFz ayk6V0RCIGV4Y2VwdGlvbixyZXN0YXJ0aW5nIGFnZW50IGluIDEwIHNlY29uZHMuLi4NCjB4NmVm NTg4KHRXZGJUYXNrKTpXREIgZXhjZXB0aW9uLHJlc3RhcnRpbmcgYWdlbnQgaW4gMTAgc2Vjb25k cy4uLg0KMHg2ZWY1ODgodFdkYlRhc2spOldEQiBleGNlcHRpb24scmVzdGFydGluZyBhZ2VudCBp biAxMCBzZWNvbmRzLi4uDQpJbXBsZW1lbnRhdGlvbiBEZXBlbmRlbnQgU29mdHdhcmUgRW11bGF0 aW9uDQpFeGNlcHRpb24gY3VycmVudCBpbnN0cnVjdGlvbiBhZGRyZXNzIDogMHgwMDAwMDAwMA0K TWFjaGluZSBTdGF0dXMgUmVnaXN0ZXIgOiAweDAwMDA5MDMwDQpDb25kaXRpb24gUmVnaXN0ZXIg OiAweDIyMjAwMDQwDQpUYXNrIDogMHg2ZWY1ODggInRXZGJUYXNrIg0KT3VyIHN5c3RlbSB1c2Ug TXBjODYwVC4NCkFueSBzdWdnZXN0aW9uIGlzIHdlbGNvbWVkLHRoYW5rIHlvdSBpbiBhZHZhbmNl IQ0KDQpTaW5jZXJlbHkgeW91cnMsDQpYaWFvDQo= --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Is malloc() atomic? Date: 21 Jan 2002 05:20:19 -0800 From: arun4vxworks@indiatimes.com (Arun Prasad) Organization: http://groups.google.com/ Message-ID: References: Dear Lenoid, <>Thank you for your kind information/clarification. Regards Arun "Leonid Rosenboim" wrote in message news:... > Yes, malloc(), actually memPartAlloc() is atomic, and you can call these > functions from your tasks without any additional locking. > No, malloc() DOES NOT do taskLock or intLock, the protection is dome using > Mutex specific to each memory partition. > Note, you CAN'T CALL malloc() from within an interrupt service function, > including WatchDog callback. > Advise: RTFM > > - Leonid > > "Arun Prasad" wrote in message > news:e38d912c.0201210153.2de8ca1d@posting.google.com... > > Hi, > > > > Can anyone tell me about malloc() implementation in VxWorks? > > I'm using IBM PPC 405GP board. > > > > I want to know whether the entire implementation of malloc is within > > tasklock (I mean is it atomic?). > > > > Or, > > Do we need to task lock before malloc() and then task unlock as task > > switches may occur while the task is within malloc as the entire > > implementation of malloc may not be within taskLock? > > > > Hoping for a clarification on the implementation of malloc > > > > Have a nice time. > > Regards > > Arun --------------------------- Newsgroups: comp.os.vxworks Subject: Re: vx works for set top box Date: Mon, 21 Jan 2002 13:33:20 +0000 (UTC) From: David Laight Organization: BT Openworld Message-ID: <3C4C18F4.D9CD5D16@l8s.co.uk> References: <36827350.0201210056.419501d5@posting.google.com> > > Would I recommend VxWorks (over Linux) for a new STB is a totally different > question, it depends on many factors, both technical is economical, for an > intelligent and objective advise. IMHO The actual choice of OS for a given system is often much more likely to be a political descision, and not based on the technical merits of the OS concerned. David --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Conflict of WDB and http server Date: Mon, 21 Jan 2002 14:49:43 +0100 From: "Michael Lawnick" Organization: Buergernetz Dillingen Message-ID: References: Reply-To: "Michael Lawnick" Sender: mlawnick@pd9e6542e.dip.t-dialin.net Hi Boris, >Exception current instruction address : 0x00000000 is the line: you called a function that was not resolved on linking or download. Look to the output on download, there is a message about unresolved symbols ! - -- Mit freundlichen Gr¨¹?en, Michael Lawnick ============================================== SOFTEC GmbH Tel +49-731-96600-0 Promenade 17 Fax +49-731-96600-23 D-89073 Ulm Michael Lawnick Germany lawnick@softec.de ============================================== "Boris" schrieb im Newsbeitrag news:a2h1ms$k90$1@mail.cn99.com... > Hi, we now encounter a new problem, when we install > http server on our system, the WDB can't work well, > On console we see following message: > 0x6ef588(tWdbTask):WDB exception,restarting agent in 10 seconds... > 0x6ef588(tWdbTask):WDB exception,restarting agent in 10 seconds... > 0x6ef588(tWdbTask):WDB exception,restarting agent in 10 seconds... > 0x6ef588(tWdbTask):WDB exception,restarting agent in 10 seconds... > Implementation Dependent Software Emulation > Exception current instruction address : 0x00000000 > Machine Status Register : 0x00009030 > Condition Register : 0x22200040 > Task : 0x6ef588 "tWdbTask" > Our system use Mpc860T. > Any suggestion is welcomed,thank you in advance! > > Sincerely yours, > Xiao > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Machine Check Exception Date: 21 Jan 2002 06:25:22 -0800 From: kannan@comneti.com (kannan) Organization: http://groups.google.com/ Message-ID: <2ede7703.0201210625.666f9918@posting.google.com> References: <488e459a.0201180048.23f0ed9d@posting.google.com> <3C4B7627.B304755F@sympatico.ca> Hi Yves, > we are having a similar problem with our system. A machine check > exception is happening ... in a streaming method. It runs through the > code on a number of occasion, than one time it produces an exception. > This problem comes and goes with re-compiles... We have created a ring > buffer with which all tasks check in on a task switch, to try and > identify the task causing the problem, but have yet to succeed... The contents that are printed in the console after getting the exception are: 1).Exception current instruction address 2).Machine Status Register 3).Condition Register 4).TaskId 5).TaskName so probably you will get that particular taskName and taskId which causes the Exception .Suppose if you are going a hanged state then probably you can write your own handler to logMsg the taskName and taskId in the console. So you will isolate the code which creates those exception.Check whether your default handler(VxWorks) makes that particular task which creates exception to suspended state? If so you can only check the task which goes to the suspended state. > the machine check exception can be caused by parity, invalid instruction > and some bus errors (I don't have my PPC manual with me). > > Any suggestions on finding (what I assumed to be) the stray pointer > would be welcomed at this point. > > PPC 750, Tornado II (with TCP2) on NT. We are using watchdogs. check whether your code has any invalid memory access. check this code : void machineChecktest (void) { printf("\n Test Task"); d (0xffffffff); /*some Invalid address*/ printf("\n Machine check exception test code over\n"); } - ->sp machineChecktest test this code and see whether the machine check exception is happening ... in a streaming method or you are able to get the machine check exception only once and also check whether you get that task machineChecktest to suspended state and the taskName and taskId are printed in your console. Rgds, kannan Yves wrote in message news:<3C4B7627.B304755F@sympatico.ca>... > John wrote: > > > > Hello, > > > > Which type of PPC are you using, and what version of VxWorks? Also, > > what else is running on your target at the time, or is it just the > > serial output? In particular, are you using things like watchdogs or > > POSIX timers? > > > > Rgds, > > > > John... > > > > "martin brook" wrote in message news:... > > > Hi, > > > > > > I'm having problems with my application getting random machine check > > > exceptions (0x200). The exception address reported does not correspond to > > > any code which may cause a machine check. I understand that the exception > > > reporting can be delayed so giving no clue as to where the exception is > > > occuring. > > > > > > Has anyone got a method of debugging these exceptions ? or a way of turning > > > the machine check off ? > > > > > > My app is doing a lot of serial I/O on ports 0/1 which are set to > > > 115200baud. > > > > > > TIA martin > > > > > > > > > > > > --- > > > Outgoing mail is certified Virus Free. > > > Checked by AVG anti-virus system (http://www.grisoft.com). > > > Version: 6.0.313 / Virus Database: 174 - Release Date: 1/3/02 > > Hi, > > we are having a similar problem with our system. A machine check > exception is happening ... in a streaming method. It runs through the > code on a number of occasion, than one time it produces an exception. > This problem comes and goes with re-compiles... We have created a ring > buffer with which all tasks check in on a task switch, to try and > identify the task causing the problem, but have yet to succeed... > > the machine check exception can be caused by parity, invalid instruction > and some bus errors (I don't have my PPC manual with me). > > Any suggestions on finding (what I assumed to be) the stray pointer > would be welcomed at this point. > > PPC 750, Tornado II (with TCP2) on NT. We are using watchdogs. > > thanks > > Yves --------------------------- Newsgroups: comp.os.vxworks Subject: ppp with multicast Date: Mon, 21 Jan 2002 15:25:58 GMT From: "alain tamdem" Organization: Rauland Borg Corporation Message-ID: Reply-To: "alain tamdem" Nobody here is dealing with ppp and multicast?? --------------------------- Newsgroups: comp.os.vxworks Subject: Multicast over PPP Date: Mon, 21 Jan 2002 15:28:47 GMT From: "alain tamdem" Organization: Rauland Borg Corporation Message-ID: Reply-To: "alain tamdem" Hi all, I am experiencing problem now to connect my configurtaion rpogram running on my PC to a target under Vxworks using serial port over PPP. In fact my application on my PC is sending the multicast ( 224.1.0.1) over the ppp link but this multicast does not seem to be received on the target side ( the target has also a Network card) Can anybody let me know why the multicast is not received on the PPP0 interface on the target ( on my target I have a socket with is bind to be listen for a multicast message on the serial port and a hook routine as describer in vxworks network programmer guide)?? any hint ?? thank in advance PS : just to complete my question by saying that the multicast flag is enabled on my ppp interface ( I saw that by doing a ifShow) Alain Alain TAMDEM Intern student RAULAND - BORG CORPORATION --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Using a PCI ethernet adapter on PCPentium BSP Date: Mon, 21 Jan 2002 16:45:35 +0100 From: "Werner Schiendl" Message-ID: <3c4c37a1$1@brateggebdc5.br-automation.co.at> References: <2002117-101355-104040@foorum.com> <3c46ee95$1@brateggebdc5.br-automation.co.at> <2002118-133844-311193@foorum.com> Hi, did you disable the Plug&Play function on the 3c509 board?? You NEED do so, because the driver does not work otherwise. A tool for that can be downloaded from 3com support web site (it is a MS-DOS tool). I think it was part of the full driver package for that board, its name is 3c5x9cfg.exe. Then, set proper resources manually and setup config.h appropriately. (look for the macros ending in _ELT, like INT_LVL_ELT in config.h and fix them for your config). rebuild the bootrom and recreate the setup disks hth Werner "Charly" wrote in message news:2002118-133844-311193@foorum.com... > > Thanks Verner, > > I leaved "EDIMAX" ( which seems NE2000 compatible) to "3C905C-TX" > But no success > > Now I use a simple ISA adapter "3C509TP ": no success. > Perhaps someone could correct me. > > Test1: > I use a direct cross cable bitween VxWorks and a PC Windows + WFTPD. > bootline for a boot on network: > "elt(0,0)host:vxWorks h=192.0.1.1 b=192.0.1.2 e=192.0.1.2 tn=target u=vxworks > pw=vxworks f=0x80 o=elt" > => Get VxWorks by TFTP timout > > Test2:Boot on floppy > "fd=0,0(0,0)host:/fd0/vxWorks h=192.0.1.1 b=192.0.1.2 u=target tn=target > u=vxworks pw=vxworks o=elt" > => start ok > in the Shell, I type "ifShow" > => > Flags 0x8069 UP LOOPBACK MULTICAST ARP RUNNING > Type SOFTWARE LOOPBACK > Internet address: 127.0.0.1 > netmask 0xff00000000 subnet 0xff00000000 > 0 packets receive 0 packet sent > .. > > I appriciate your suggestion to start my network > Thanks > Charly > > > > > > > > > > > > > > > > -- > Use our news server 'news.foorum.com' from anywhere. > More details at: http://nnrpinfo.go.foorum.com/ --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Support for 8 bits + even parity in ttyDrv? Date: Mon, 21 Jan 2002 16:49:31 +0100 From: "Werner Schiendl" Message-ID: <3c4c388d$1@brateggebdc5.br-automation.co.at> References: <3C45D4D8.5FE5F442@bitworkssystemsDELETETHIS.com> <488e459a.0201161606.1b6c2db@posting.google.com> <3C4756DC.79DBB9D7@bitworkssystemsDELETETHIS.com> Hi Jon, Be sure to always search the header files and the HTML files of the documentation as well, if you have a clue like SIO_HW_OPTS_SET. The search 'features' of the WRS documentation (Tornado II, Windows host) are pretty poor in my opinion. regards Werner "Jon" wrote in message news:3C4756DC.79DBB9D7@bitworkssystemsDELETETHIS.com... > Thanks I did find it. A person really needs this group to do this since none of this is in the printed manuals or the > online documentation. Searching for SIO_HW_OPTS_SET returns "No topics found". Thanks again! > > John wrote: > > > Hello, > > > > Take a look in the driver for your serial chip. For the standard > > pcPentiumX BSPs this is i8250Sio.c (under target/src/drv/sio). You > > will see in there that there is a function called i8250OptsSet() which > > is called as a result of an ioctl() on the device with the operation > > code SIO_HW_OPTS_SET. The argument of the ioctl() is passed through to > > i8250OptsSet(). > > > > The options you are interested in are PARENB and PARODD. Since you > > want even parity, you should only set the PARENB (parity enable), and > > not PARODD (select odd). These values are defined in sioLib.h, so > > you'll want to include that into whatever is making the ioctl() call. > > > > HTH, > > > > John... > > > > Jon wrote in message news:<3C45D4D8.5FE5F442@bitworkssystemsDELETETHIS.com>... > > > I have a serial device that uses 8 bits plus even parity and at first > > > glance didn't see away to set this in the tyLib ioctl options. Is there > > > a way to get the ttyDrv to transmit and receive 1 start bit, 8 data > > > bits, 1 even parity bit and 1 stop bit? Thats 11 bits total. > > > > > > Platform is Pentium running VxWorks AE 3.1 > > > > > > Jon Newbill > > > Bitworks Systems Inc. > > > jon @ bitworkssystemsNOSPAM.com > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Checksum do not match [read carefully] Date: 21 Jan 2002 11:38:43 -0500 From: Bill Pringlemeir Organization: Bell Sympatico Message-ID: References: Sender: bpringlemeir@DeadDuck >>>>> "Leonid" == Leonid Rosenboim writes: Leonid> You can ignore the checksum error safely, as long as you Leonid> trust yourself to never mixup the image file and the actual Leonid> target image. I trust myself most of the time [and I probably shouldn't]. I don't trust other people ;-) Leonid> Still, I beleive you might find a better solution if you Leonid> perform tha patch directly on WRS object library - use arXX No, I have multiple processors (on different boards). Some of them do support the ARM instructions that are being used (SMULL, MULL, LDRH, STRH, etc). However others do not. I wish to identify the CPU at boot up and then patch the code if needed. This is very similar to emulating a Floating Point CPU; except patching the code isn't needed. Leonid> Of course if it turns out that the patching is needed in only Leonid> a few object files, there could be a more elegant soltuon Leonid> there, depending on where you find these instructions. Such as? I don't have the source. The object files contain instructions that can not be executed on some CPUs. A patch at compile/link time would work, but I wish to do this at run time for the reason noted. Is there something better given those restrictions? Leonid> I would speculate that instructions you patch would not be Leonid> compiler generated, hence always found i assembly source Leonid> files ... No, most of the instructions are generated by a compiler. Except possibly the `LDRSB' instructions which I could not get gcc to generate. This seems to be found in `COFF' module processing and the file system. I don't think WRS code these in assembler. The memory copy routines use some `weird' instruction variants that aren't found anywhere else [bcopyWords, etc]. These are most likely coded in assembler. My hope was that there was some `wdbConfigChecksum.c' file that I could modify. However, since I couldn't find it myself I guess that it is impossible. regards, Bill Pringlemeir. - -- ANNOYED BY DESTROYER. SINKING. U-308. vxWorks FAQ, "http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html" --------------------------- Newsgroups: comp.os.vxworks Subject: Linker map file description Date: 21 Jan 2002 17:56:30 +0100 From: Klaus Zeitler Organization: Lucent Technologies Message-ID: Sender: kzeitler@sfsw51 tried to find some docu for the linker map file, but found only more than sparse hints in Tornado and also in GNU ld docu. If I've overlooked it I'd appreciate any pointers. Here's an excerpt of a PPC linker map file, maybe some kind soul can explain that to me. the files section contains e.g.: **FILES** cp.o 00000000 0050f9d8 2** 2d .text 00000000 00085aac 2** 3d .rodata I guess this is start address and size but what do the '2**' and the '2d', '3d' columns mean? and then there's the section called **LINK EDITOR MEMORY MAP** with e.g. the text section starting with: .text 00100000 8b66f8 0 2**2 load alloc contents from *(.text) this is probably an overall section info with start address and size followed by "0 2**2" which escapes me and the section flags. Then all the object files with their symbols are listed like: .text 00100000 50f9d8 50f9d8 2**2 elf32-powerpc cp.o(overhead 3995928 bytes) 001a25a8 chain6__27aRadioResourceManager_Actor 0029b178 setIE_Extensions__29dDCH_AddItem_R again I'd say this is the start address (00100000) and the size (50f9d8), then most of the time the 3rd number is the same as the 2nd but sometimes a zero instead. What does the 3rd number mean? And finally there's an overhead listed, so what's in there? Thanx for any help. Klaus - -- ------------------------------------------ | Klaus Zeitler Lucent Technologies | | Email: kzeitler@lucent.com | ------------------------------------------ - --- Your conscience never stops you from doing anything. It just stops you from enjoying it. --------------------------- Newsgroups: comp.os.vxworks Subject: tffs, tffsBootPutImage and booting from fallow area Date: Mon, 21 Jan 2002 17:27:28 +0000 From: Owain Phillips Organization: Siemens Inc. Message-ID: <3C4C4F80.3F96AD2C@siemenscomms.co.uk> This is a multi-part message in MIME format. - --------------1C0382EB86F19CFEEAC12C46 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Hi, Does tffsBootPutImage work? I do a tffsUnmount on the drive, then tffsBootPutImage( 0, 0, "vxWorks"). The first Meg of the image is burn to flash before the call falls over with "Physical write failed at address 0x100000". I acutally try to program the elf image into ROM rather than a binary image, but at this stage should this really matter? I can write my file image to the fallow region using one of my own knocked up routines, so the flash is OK; but does tffsBootPutImage do some other house keeping? What should my bootline be to boot from this fallow region? Any responses greatfully accepted, Kind regards, Owain - --------------1C0382EB86F19CFEEAC12C46 Content-Type: text/x-vcard; charset=us-ascii; name="owain.phillips.vcf" Content-Transfer-Encoding: 7bit Content-Description: Card for Owain Phillips Content-Disposition: attachment; filename="owain.phillips.vcf" begin:vcard n:Phillips;Owain tel;fax:+44 115 943 4969 tel;work:+44 115 943 2167 x-mozilla-html:TRUE url:http://www.siemenscomms.co.uk org:Siemens Comunications Ltd version:2.1 email;internet:owain.phillips@siemenscomms.co.uk title:Software Engineer adr;quoted-printable:;;Technology Drive=0D=0ABeeston;Nottingham;;NG9 1LA;United Kingdom fn:Owain Phillips end:vcard - --------------1C0382EB86F19CFEEAC12C46-- --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Linker map file description Date: 21 Jan 2002 12:30:35 -0500 From: Bill Pringlemeir Organization: Bell Sympatico Message-ID: References: Sender: bpringlemeir@DeadDuck >>>>> "Klaus" == Klaus Zeitler writes: Klaus> tried to find some docu for the linker map file, but found Klaus> only more than sparse hints in Tornado and also in GNU ld [snip] Klaus> cp.o 00000000 0050f9d8 2** 2d .text 00000000 00085aac 2** 3d Klaus> .rodata Klaus> I guess this is start address and size but what do the '2**' Klaus> and the '2d', '3d' columns mean? Klaus> and then there's the section called **LINK EDITOR MEMORY MAP** Klaus> with e.g. the text section starting with: Klaus> .text 00100000 8b66f8 0 2**2 load alloc contents from *(.text) Klaus> this is probably an overall section info with start address Klaus> and size followed by "0 2**2" which escapes me and the section Klaus> flags. These numbers are usually dependant on the object format that you are creating. For example, a `COFF' map file will be quite different than an `ELF' or `A.OUT' map file. I would guess that the "2**2" are an internal notation about the section properties. Ie, does it contain debug info. Is it `loadable', should it be `allocated', should it be `relocated', etc. Look at the ELF-stabs (or ELF-dwarf) sections generated when you have debug information. They generally won't be allocated as they don't reside on your target. A BSS (block storage segment) or the `all zero' section will be `allocated', but it won't have anything to `load'. Text, .rodata (ROM data) and .data will all be allocated and loaded. They may also need to be `reallocated'. You will also have an '.ini' and `.fini' or `.ctors' and `.dtors' sections if you use C++; these are for static constructors and destructors. Similar quandaries for documentation will be found if you use `objdump$(ARCH)'. This is because the GNU people can write documentation for the front end. (Objdump, and Ld arguments), but the backend (object format) can be very different depending on the way that `binutils', etc have been configured. I think that the GNU people call this `BFD' and there are documents on `BFD' internals. However, I don't think that you will find documentation on the MAP files (maybe a FAQ) or objdump output. It is suppose to be self-explanatory. hth, Bill Pringlemeir. - -- Have you ever had a small midget share intimate feelings while drinking Sterno straight from the can? Or been called by your friend while living on a planet where several Conspiracy members are following you? Or been given six cars made out of your neighbor? You will. And the company that will bring it to you: AT&T. vxWorks FAQ, "http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html" --------------------------- Newsgroups: comp.os.vxworks Subject: Re: vx works for set top box Date: Mon, 21 Jan 2002 20:31:59 GMT From: Alex Cellarius Organization: The South African Internet Exchange Message-ID: <1103_1011645119@pentiumii> References: <36827350.0201210056.419501d5@posting.google.com> <3C4C18F4.D9CD5D16@l8s.co.uk> On Mon, 21 Jan 2002 13:33:20 +0000 (UTC), David Laight wrote: > > > > > Would I recommend VxWorks (over Linux) for a new STB is a totally different > > question, it depends on many factors, both technical is economical, for an > > intelligent and objective advise. > > IMHO The actual choice of OS for a given system is often much more > likely to be a political descision, and not based on the technical > merits of the OS concerned. Amen to that!! --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Using a PCI ethernet adapter on PCPentium BSP Date: 21 Jan 2002 17:49:56 GMT From: Charly Message-ID: <2002121-184956-427038@foorum.com> References: <2002117-101355-104040@foorum.com> <3c46ee95$1@brateggebdc5.br-automation.co.at> <2002118-133844-311193@foorum.com> <3c4c37a1$1@brateggebdc5.br-automation.co.at> I don't see a possibility to "disable the Plug&Play" in this utility(3c5x9cfg. exe). But just change IT and I/O. I am now in contact with WindRiver. They already used the Realtek RTL 8029. It is a PCI adapter compatible with NE2000 driver. Regards Charly - -- Use our news server 'news.foorum.com' from anywhere. More details at: http://nnrpinfo.go.foorum.com/ --------------------------- Newsgroups: comp.os.vxworks Subject: Re: BootRom and MMU Date: 21 Jan 2002 11:44:50 -0800 From: vloscomp@yahoo.com (Vinh) Organization: http://groups.google.com/ Message-ID: References: <0y%_7.14882$8e.753501@news> <3C43E9C9.8630FBE3@europem01.nt.com> Hi, MMU in PPC does not mean that flat memory model can't be used. The PPC MMU can be set to run flat and cache enable, as for virtual, it's a sepatate configuration for the MMU. However, i think the original question is: Can anyone used the VisionClick to step through their VxWorks apps? Thank You. Ephraim Gadsby wrote in message news:... > On Tue, 15 Jan 2002 09:35:21 +0100, Sebastien Fauris > wrote: > > >Hi vxWorkers, > > > >what I find strange is that in a bootrom > >the PPC processor runs in real mode and so > >the MMU is disabled. The MMU will be enabled > >only in the kernel downloaded afterwards. > > > > If, for whatever reason, the system does not have memory physically > mapped from zero, then the MMU is almost the first thing that needs to > be set-up. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: BootRom and MMU Date: Mon, 21 Jan 2002 20:51:29 GMT From: "Jim" Organization: Magma Communications Ltd. Message-ID: References: Hey vinH, just out of curiousity the tlbie command ... if you add it to your romInit.s then step through this new code does it cause the same grief ... tlbie is a pretty important command it is unlikely it doesn't work ... my guess is there is a problem before this instruction not during ... Another $0.002, Jim "Vinh" wrote in message news:c129947c.0201081333.5d35caae@posting.google.com... > Hi all, > I am trying to debug our VxWorks boot for PPC, and i can't seem to > used VisionClick to step over an assembly instruction (tlbie) in the > '../target/templatePpc/sysALib.s'. > This is the sniped of code. > > sysALoop: > tlbie p0 > addi p0,p0,0x1000 /* increment bits 15-19 */ > bdnz sysALoop /* decrement CTR, branch if CTR != 0 */ > sync /* sync instr req'd after tlbie */ > > > Can any one explain to me why we need to invalidates the TLB in > _sysInit? > I though that when VxWorks comes up, it will used it's MMU lib to > initialized the PPC MMU. Some how the instruction caused the > VisionProbe to disconnect. > > > Any help is greatly appreciated. > > Thank you, > Vinh Lam --------------------------- Newsgroups: comp.os.vxworks Subject: Re: BootRom and MMU Date: 21 Jan 2002 17:46:59 -0500 From: Bill Pringlemeir Organization: Bell Sympatico Message-ID: References: Sender: bpringlemeir@DeadDuck >>>>> "Vinh" == Vinh writes: Vinh> Hi all, I am trying to debug our VxWorks boot for PPC, and i Vinh> can't seem to used VisionClick to step over an assembly Vinh> instruction (tlbie) in the '../target/templatePpc/sysALib.s'. Some instructions will disable debug support momentarily. Ie, this particular instruction might be relying on the fact that code will be fetch in the pipeline. By halting the processor (single stepping), you are interrupting the flow to the pipeline, cache, etc. Vinh> Can any one explain to me why we need to invalidates the TLB in Vinh> _sysInit? I though that when VxWorks comes up, it will used Vinh> it's MMU lib to initialized the PPC MMU. Some how the Vinh> instruction caused the VisionProbe to disconnect. The code might not be called from `reset'. It might be called through the vxWorks function `reboot()'. In such case, the TLB (translation look-aside buffer) can be full of data. It may or may not correspond to the code that it should be looking up. To ensure that the code fetches will come from physical memory instead of stale cache entries, you should flush the cache. My main suggestion here would be for you to "leave the code alone" and try to set break points far beyond this one. If you have flash and can't set breakpoints, I have made special images with a `BKPT' (or what ever) inserted before the called to `usrInit()'. You can always try this. You may need to alter some of the startup code to enable the BDM properly. My guess is that you are using the MPC860 or something. I haven't used PPC in a while. You have to set some bits in different registers to enable a debug mode. SIU or MCR or some TLA... If you are interested, look it up! If this is not MPC8xx related, then I am not really sure what the debug mechanism is (JTAG?), but a lot of this information should be relevant. hth, Bill Pringlemeir. - -- .. I'm IMAGINING a sensuous GIRAFFE, CAVORTING in the BACK ROOM of a KOSHER DELI -- vxWorks FAQ, "http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html" --------------------------- Newsgroups: comp.os.vxworks Subject: Re: BootRom and MMU Date: Mon, 21 Jan 2002 23:06:43 +0000 (UTC) From: David Laight Organization: BT Openworld Message-ID: <3C4C9F4F.685049A5@l8s.co.uk> References: Bill Pringlemeir wrote: > > Vinh> Hi all, I am trying to debug our VxWorks boot for PPC, and i > Vinh> can't seem to used VisionClick to step over an assembly > Vinh> instruction (tlbie) in the '../target/templatePpc/sysALib.s'. [snip] > > My main suggestion here would be for you to "leave the code alone" and > try to set break points far beyond this one. I've debugged that section of code by initialising a UART and writing bytes to it, just loop until the fifo is empty You normally have a reasonable number of registers free for argument passing etc - even if you have no DRAM Hoping to be able to single step through it is asking a bit much. David --------------------------- Newsgroups: comp.os.vxworks Subject: compiling WTX application on WinNT, VC++ 6 Date: Mon, 21 Jan 2002 18:31:36 -0500 From: "Cedric Le Roux" Message-ID: <3c4ca498$1_1@nopics.sjc> Anybody know a good place to get some detailed instructions on how to compile a WTX application on Windows NT and Visual C++ 6? I've gone through the Tornado API Guide. The sample code will not compile, and the configuration instructions don't work. It seems there are problems with include files. It's trying to use some target includes. (vxWorks.h). I'm not sure which "signals.h" it should use. I get lots of errors from that one. I suspect that it's probably something simple in the way I'm configuring VC++, but I'm totally at a loss... My configuration is pretty much default. Tornado in default location, VC++ in default location. I followed instructions in Tornado API Guide, those are the only changes I've made. Thanks for any help. Cedric --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Why I can not boot through network? Date: 21 Jan 2002 17:26:55 -0800 From: sardine_sz@163.net (sardine) Organization: http://groups.google.com/ Message-ID: References: <8fc342d0.0201210107.6eb9ecc4@posting.google.com> step by step: 1> config what's dev that your target boot network device name in file .../BSP/confing.h for example: (intel express 10/100M network adapter) #define DEFAULT_BOOT_LINE \ "fei(0,0) host:vxWorks h=192.168.0.12 e=192.168.0.112 u=anonymous pw=abc@ f=0x0" 2> include the network adapter macro in file ../BSP/config for example: change line: #undef INCLUDE_FEI as : #define INCLUDE_FEI 3>select WDB connect mode in your project facility WDB_END OR WDB_NETWORK 4>select network adapter type in your project facility network device->BSD network driver->(your network adapter type) ok, i think it should have a little help for u. other: you can reference tornado help: <>---> boot from network wxh@founder.com.cn (Wen Xiaohui) wrote in message news:<8fc342d0.0201210107.6eb9ecc4@posting.google.com>... > I have developed an END driver using DP83902A. The END driver works > well when I use it in a VxWorks_ROM image. But when I want to make a > boot_rom and try to boot through the network, the target server can > not connect to the board, and I can not ping to the board, either. Are > there any problems in my project configuration? I have undefine the > INCLUDE_SERIAL macro in my "config.h". > Thanks a lot! --------------------------- Newsgroups: comp.os.vxworks Subject: Re: tffs, tffsBootPutImage and booting from fallow area Date: Tue, 22 Jan 2002 01:41:46 GMT From: Andray Kaganovsky Organization: Primus Canada Message-ID: References: <3C4C4F80.3F96AD2C@siemenscomms.co.uk> Owain Phillips wrote in news:3C4C4F80.3F96AD2C@siemenscomms.co.uk: > Hi, > > Does tffsBootPutImage work? > > I do a tffsUnmount on the drive, then tffsBootPutImage( 0, 0, > "vxWorks"). > The first Meg of the image is burn to flash before the call falls over > with > "Physical write failed at address 0x100000". [AndrayK]. When you formatted flash array with tffsDevFormat(), what was the value of tffsFormatParams.bootImageLen ? Do you have enough room in the "raw" flash area that resides before FTL-formatted part of flash array ? > I acutally try to program the elf image into ROM rather than a binary > image, but at this stage should this really matter? [AndrayK]. As far as I know, it doesn't matter. > I can write my file image to the fallow region using one of my own > knocked up routines, so the flash is OK; but does tffsBootPutImage do > some other house keeping? > What should my bootline be to boot from this fallow region? [AndrayK]. I don't think you can use TrueFFS for booting from "raw" flash. My understanding is that you need to configure your board to place beginning of "raw" flash area at CPU reset vector, and that should do the trick. regards, Andray > Any responses greatfully accepted, > Kind regards, > > Owain > > --------------1C0382EB86F19CFEEAC12C46 > begin:vcard > n:Phillips;Owain > tel;fax:+44 115 943 4969 > tel;work:+44 115 943 2167 > x-mozilla-html:TRUE > url:http://www.siemenscomms.co.uk > org:Siemens Comunications Ltd > version:2.1 > email;internet:owain.phillips@siemenscomms.co.uk > title:Software Engineer > adr;quoted-printable:;;Technology Drive=0D=0ABeeston;Nottingham;;NG9 1LA;United Kingdom > fn:Owain Phillips > end:vcard > > Attachment decoded: owain.phillips.vcf > --------------1C0382EB86F19CFEEAC12C46-- --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Booting from ATA Compact Flash Date: Tue, 22 Jan 2002 01:50:27 GMT From: Andray Kaganovsky Organization: Primus Canada Message-ID: References: bhorowit@cs.berkeley.edu (Benjamin Horowitz) wrote in news:ab1f8a59.0201200034.2ef1e541@posting.google.com: > Hi, > > Please forgive this naive question: > > I'm trying to boot from an ATA Compact Flash disk. The disk is > control 0, drive 0. I create a new bootrom_uncmp using Tornado II > (and VxWorks 5.4). I copy bootrom_uncmp to a floppy. I invoke the > mkbootAta function from the Tornado shell, as follows: > > mkbootAta 0, 0, "/fd0/BOOTRO~1 > > (BOOTRO~1 is the DOS file name.) I can now see bootrom.sys on the ATA > drive. Now I reboot, this time without my normal boot floppy. I see > the following messages: > > Searching for Boot Record from Floppy..Not Found > Searching for Boot Record from IDE-0..OK > > But then the machine hangs. Normally I would see: > > VxLd........ [etc.] > > Any suggestions about where to look? I'm stuck. > > Thanks a lot, > Ben > [AndrayK]. Are you sure you have written vxWorks boot sector to your ATA disk ? Use WindRiver's VXSYS.COM utility to do it under DOS. Note: VXSYS.COM probably will not work under Windows, you need to actually boot DOS. regards, Andray --------------------------- Newsgroups: comp.os.vxworks Subject: Switching between network interfaces Date: Mon, 21 Jan 2002 17:57:54 -0800 From: "Randy Schafer" Organization: Northwest Link Message-ID: <3c4cc682_1@news.nwlink.com> Hello Vxperts, We have a custom MPC860 based board with two ethernet interfaces configured in two different subnets. We only allow 1 interface to be active (ifnet flag IFF_UP set) at any given time. When we switch interfaces, the arp table is flushed and the routing table is adjusted to direct both subnets to the active interface. This all works except for broadcasts. For some reason, it appears that the stack is still trying to route a broadcast for the inactive port to the inactive port yielding a ENETDOWN error. Configuration is as follows: Our MPC860 system has two network interfaces A and B. A configured as nameA= "cpm0" IP ADDRESS: 10.129.33.6 BROADCAST ADDRESS: 10.191.255.255 SUBNET MASK: 255.192.0.0 (0xffc00000) B configured nameB="cpm1" IP: 10.193.33.6 BROADCAST ADDRESS: 10.255.255.255 SUBNET MASK: 255.192.0.0 (ffc00000) ONLY ONE interface ( A or B ) is active (up) at any one time. When A is up: routeShow returns ROUTE NET TABLE destination gateway flags Refcnt Use Interface - ---------------------------------------------------------------------------- 10.128.0.0 10.129.33.6 101 0 0 cpm0 10.192.0.0 10.129.33.6 101 0 0 cpm0 - ---------------------------------------------------------------------------- ROUTE HOST TABLE destination gateway flags Refcnt Use Interface - ---------------------------------------------------------------------------- 127.0.0.1 127.0.0.1 5 0 0 lo0 - ---------------------------------------------------------------------------- When B is up: routeShow returns ROUTE NET TABLE destination gateway flags Refcnt Use Interface - ---------------------------------------------------------------------------- 10.128.0.0 10.193.33.6 101 0 0 cpm1 10.192.0.0 10.193.33.6 101 0 0 cpm1 - ---------------------------------------------------------------------------- ROUTE HOST TABLE destination gateway flags Refcnt Use Interface - ---------------------------------------------------------------------------- 127.0.0.1 127.0.0.1 5 0 0 lo0 - ---------------------------------------------------------------------------- The system switches from A to B by switching algorithm: if_down( ifunit( nameA ) ); arpFlush(); if_up( ifunit( nameB ) ); routeAdd.. The system periodically sends broadcast packets to address 10.191.255.255 and to broadcast address 10.255.255.255. This transmission usually works okay unless the system switches from one active interface to the other using the switching algorithm ( above ). After the switch, when the active interface changes from say A to B, we get S_errno_ENETDOWN error from the sendto call when the target broadcast address 10.255.255.255 is used. Any ideas??? Thanks - -- Randy Schafer Sr.Software Engineer Cinta Networks Corporation randy.schafer@CintaNetworks.com www.CintaNetworks.com --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Interrupt Latency Problem!!! Date: 22 Jan 2002 02:11:50 GMT From: Chris Smith Organization: Concentric Internet Services Message-ID: References: <488e459a.0201171332.553345cd@posting.google.com> Thanks for the ideas. We have some new information regarding this problem now that we are getting more familiar with Windview. It seems that we only see this problem when our ISR interrupts the tail end of the tWdbTask. Looking at the Windview log, it seems that this bug only occurs if we interrupt tWdbTask in a specific window (~30 - ~10 us before the end). If our ISR happens in the last ~10us of the tWdbTask, there is no problem, same if it occurs up to ~30us before the end. In our BSP, we have tried using both END driver and ethernet driver (our card has the DEC21140 chip). We are currently using the END driver since Windview requires it. > In particular, the fact that while interrupts may be nested, this does > require some careful setup of the priorities to get the correct > behaviour. In our ISR we are generating wvEvents at the beginning and end so we know that we are calling our ISR once per servo cycle by looking at Windview. > It would be worth trying to find out whether the > problem is your interrupt being delayed by a lock, or your ISR being > interrupted (perhaps before it actually starts executing your code). In WindView, we only see 2 interrupts, the ethernet and our external interrupt. Our interrupt is never interrupted by anything else. Lastly, we are fighting with Windriver support over this issue. They are VERY reluctant to admit that this is even a bug. They have issued a SPR number, but currently it is very low priority. They are telling us that their drivers are non-optimized and any latencies are just a result of that. We have been trying to get them to tell us how much of a latency is considered a bug, but have not heard an answer. Does anyone have anything else to add? Has anyone else seen similar results? Please let us know so we can provide more evidence to Wind River and hopefully get this bug acknowledged and fixed. --------------------------- Newsgroups: comp.os.vxworks,comp.sys.powerpc.tech Subject: Re: VxWorks, Walnut 405GP tftp download error Date: Mon, 21 Jan 2002 21:35:49 -0500 From: Mark Organization: BELLSOUTH.net Message-ID: <3C4CD004.6C961D58@bellsouth.net> References: Ar you running the IBM Bios or the VxWorks Bios? The IBM Bios will not boot a VxWorks image. The VxWorks comes with all you need to build a VxWorks bootrom. I forget how to do it. I have not used VxWorks in a while but I know you have to use their bootrom and their package and documentation comes with all you need.. Mark Arun Prasad wrote: > Hi, > > We are using Vxworks on IBM's Walnut PPC 405GP board. > The bootrom on it is the same one which came with this board. > We are trying to boot the system(walnut) using TFTP and download the > vxWorks image. > > The following error message is displayed: > >Not a valid boot image file > > We got this sequence a couple of times too: > Loading file "/tftpboot/mydir1/vxWorks" ... > Sending tftp boot request ... > Unknown destination, block 1 > Unknown destination, block 1 > Unknown destination, block 1 > Unknown destination, block 1 > Unknown destination, block 1 > Unknown destination, block 1 > Unknown destination, block 1 > Unknown destination, block 1 > Unknown destination, block 1 > TFTP boot failed. > > I am looking for info on how to build a tftp downloadable vxworks > image for the Walnut 405GP system. > > Is it looking for some thing specific at some specific location ... > Does the file type i.e., Binary or Hex matter? > Do we need use elfHex or linker with any specific options, > or any changes to the config? > > Looking forward for an early reply. > > Thanks in advance, > > -- --------------------------- Newsgroups: comp.os.vxworks Subject: Re: torVars and make Date: Tue, 22 Jan 2002 13:47:48 +0300 From: "Zhao Yandong" Organization: Bentium Ltd. (CN99) Message-ID: References: I also encounted similar questions. I added a line in autoexec.bat like this SHEL=C:\command.com /E:4096 /P and it solved,you can try it yuhang wrote in message news:bffd1323.0201161823.3471df94@posting.google.com... > I have one question. > when i use the command "c:\tornado\host\x86-win32\bin\torVars.bat",it > appears > " > c:\tornado\host\x86-win32\bin\torvars > c:\tornado\host\x86-win32\bin\rem Command Line Build Environments > c:\tornado\host\x86-win32\bin\set WIND_HOST_TYPE=x86-win32 > c:\tornado\host\x86-win32\bin\set WIND_BASE_=C:\Tornado > out of the environment space > c:\tornado\host\x86-win32\bin\set PATH=\host\\bin;D:\PROGRA~1\UEDIT;C > :\WINDOWS;c:\WINDOWS\COMMAND > out of the environment space > and I wonder if any way to change it ,and in > "c:\tornado\target\src\snmpv1\agent",use the command"make ", > in' c:\tornado\target\src\snmpv1\agent make CPU=ARM7TDMI TOOL=gnu > clean'", > it does not go , > "bad command or file name",why? > > Thankyou in advance! --------------------------- Newsgroups: comp.os.vxworks Subject: Vxworks beginner Date: 21 Jan 2002 22:26:17 -0800 From: sridhar@cmcltd.com (Sridhar CV) Organization: http://groups.google.com/ Message-ID: <8207c357.0201212226.3b5a9a55@posting.google.com> Hi, can any one of you suggest me the best way to master Vxworks application development in very short time. Regards Sridhar CV --------------------------- Newsgroups: comp.os.vxworks Subject: SCSI Performance in vxWorks Date: Tue, 22 Jan 2002 01:58:18 -0500 From: "Jared Maguire" Message-ID: <20020122.015817.1346811305.3143@analogic.com> Has anyone measured SCSI performnace greater than 10MB/s under vxWorks? What card/driver? We're using an LSI Logic 53C1010 controller, and couple of Seagate Cheetahs. We've also tried it on a sym895, and an adaptec 7880, and we can't seem to get a transfer to happen at greater than 10MB/s, sustained. What gives? - -J --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Switching between network interfaces Date: 22 Jan 2002 08:27:42 GMT From: Johan Borkhuis Organization: Agere Systems Message-ID: References: <3c4cc682_1@news.nwlink.com> "Randy Schafer" wrote: > Our MPC860 system has two network interfaces A and B. > > A configured as > nameA= "cpm0" > IP ADDRESS: 10.129.33.6 > BROADCAST ADDRESS: 10.191.255.255 > SUBNET MASK: 255.192.0.0 (0xffc00000) > > B configured > nameB="cpm1" > IP: 10.193.33.6 > BROADCAST ADDRESS: 10.255.255.255 > SUBNET MASK: 255.192.0.0 (ffc00000) > The system periodically sends broadcast packets to address > 10.191.255.255 and to broadcast address 10.255.255.255. > This transmission usually works okay unless the system > switches from one active interface to the other using the > switching algorithm ( above ). After the switch, when > the active interface changes from say A to B, we get > S_errno_ENETDOWN error from the sendto call when > the target broadcast address 10.255.255.255 is used. Just my 2 cents: The first interface has the address 10.129.33.6 with netmask ffc00000. This gives a network address of 10.128.0.0, with a broadcast address of 10.191.255.255. The second interface uses 10.193.33.6, ffc00000, 10.192.0.0 and 10.255.255.255. Just looking at the broadcast addresses: when sending to the broadcast address of the wrong interface the address does not match the netmask, and will probably be transmitted to the gateway address. As this is not specified (or not given in this message) the stack might return an error. Groeten, Johan - -- o o o o o o o . . . _____________________________ o _____ || Johan Borkhuis | .][__n_n_|DD[ ====_____ | borkhuis@agere.com | >(________|__|_[_________]_|__________________________| _/oo OOOOO oo` ooo ooo 'o!o!o o!o!o` === VxWorks FAQ: http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html === --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Interrupt Latency Problem!!! Date: 22 Jan 2002 00:33:18 -0800 From: john_94501@yahoo.com (John) Organization: http://groups.google.com/ Message-ID: <488e459a.0201220033.5b4135d4@posting.google.com> References: <488e459a.0201171332.553345cd@posting.google.com> Hello, > We have some new information regarding this problem now that we are > getting more familiar with Windview. It seems that we only see this > problem when our ISR interrupts the tail end of the tWdbTask. Looking > at the Windview log, it seems that this bug only occurs if we interrupt > tWdbTask in a specific window (~30 - ~10 us before the end). If our ISR > happens in the last ~10us of the tWdbTask, there is no problem, same if > it occurs up to ~30us before the end. I take it that what you are saying is that you see the ISR delayed if the hardware generates the interrupt in that window, rather than WindView showing the interrupt occurring in that window, but tWdbTask still running. Or did you mean the latter? Also, have you tried to find out what the WDB task is up to at the time? Can you make your ISR get hold of the PC from the WDB task's TCB perhaps... That might give you an idea of what it is doing at the time your interrupt arrives. You might also want to try the measurements without a target server attached. > In our BSP, we have tried using both END driver and ethernet driver (our > card has the DEC21140 chip). We are currently using the END driver > since Windview requires it. WindView shouldn't care; just switch it to the standard socket based upload. WDB will care, but only if you need system mode (you can switch WDB to WDB_NETWORK even with the END driver if you like - actually, might be a good idea anyway - it'll stop WDB scanning every incoming packet for ones it should handle). > In WindView, we only see 2 interrupts, the ethernet and our external > interrupt. Our interrupt is never interrupted by anything else. OK, so the sequence that causes the problem, if I've understood you, is a net IRQ, then tWdbTask running, then your interrupt occurring, but tWdbTask still running. That would require tWdbTask to have locked out interrupts, which it does seem to do, but only when it is generating events for the host, or VIO transactions between the host & target. HTH, John... > Lastly, we are fighting with Windriver support over this issue. They > are VERY reluctant to admit that this is even a bug. They have issued a > SPR number, but currently it is very low priority. They are telling us > that their drivers are non-optimized and any latencies are just a result > of that. We have been trying to get them to tell us how much of a > latency is considered a bug, but have not heard an answer. > > Does anyone have anything else to add? Has anyone else seen similar > results? Please let us know so we can provide more evidence to Wind > River and hopefully get this bug acknowledged and fixed. --------------------------- Newsgroups: comp.os.vxworks,comp.sys.powerpc.tech Subject: Re: VxWorks, Walnut 405GP tftp download error Date: 22 Jan 2002 03:27:54 -0800 From: arun4vxworks@indiatimes.com (Arun Prasad) Organization: http://groups.google.com/ Message-ID: References: <3C4CD004.6C961D58@bellsouth.net> Hi, Thanks for extending your hands. Our system came up and we are able to download image now. Tim's (Tim Moeller ) tips was of great help for us in this regard. Thank you all for extending your hands. Regards Arun Mark wrote in message news:<3C4CD004.6C961D58@bellsouth.net>... > Ar you running the IBM Bios or the VxWorks Bios? The IBM Bios will not > boot a VxWorks image. The VxWorks comes with all you need to build a > VxWorks bootrom. I forget how to do it. I have not used VxWorks in a > while but I know you have to use their bootrom and their package and > documentation comes with all you need.. > Mark > > > > Arun Prasad wrote: > > > Hi, > > > > We are using Vxworks on IBM's Walnut PPC 405GP board. > > The bootrom on it is the same one which came with this board. > > We are trying to boot the system(walnut) using TFTP and download the > > vxWorks image. > > > > The following error message is displayed: > > >Not a valid boot image file > > > > We got this sequence a couple of times too: > > Loading file "/tftpboot/mydir1/vxWorks" ... > > Sending tftp boot request ... > > Unknown destination, block 1 > > Unknown destination, block 1 > > Unknown destination, block 1 > > Unknown destination, block 1 > > Unknown destination, block 1 > > Unknown destination, block 1 > > Unknown destination, block 1 > > Unknown destination, block 1 > > Unknown destination, block 1 > > TFTP boot failed. > > > > I am looking for info on how to build a tftp downloadable vxworks > > image for the Walnut 405GP system. > > > > Is it looking for some thing specific at some specific location ... > > Does the file type i.e., Binary or Hex matter? > > Do we need use elfHex or linker with any specific options, > > or any changes to the config? > > > > Looking forward for an early reply. > > > > Thanks in advance, > > > > -- --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Vxworks beginner Date: Tue, 22 Jan 2002 06:46:13 -0500 From: joe durusau Organization: Bellsouth.Net Message-ID: <3C4D5105.6EB0BBAC@bellsouth.net> References: <8207c357.0201212226.3b5a9a55@posting.google.com> Spend a bunch of money on vxworks training, assuming that you already know real-time programming. Speaking only for myself, Joe Durusau Sridhar CV wrote: > Hi, > can any one of you suggest me the best way to master Vxworks > application development in very short time. > > Regards > Sridhar CV --------------------------- End of New-News digest ********************** From vxwexplo-errs@csg.lbl.gov Tue Jan 22 06:09:44 2002 From: mmathers@landis-lund.co.uk Date: Tue Jan 22 06:09:47 PST 2002 Subject: Universe II PCI/VME bridge Hello all, Not a question directly about vxWorks, but something that someone in the group may have experience of. We are using a Pentium III based VME card that uses the Universe II PCI to VME bridge. When accessing the VME memory space I have found that I get a big endian/ little endian problem. I've looked in the Universe II manual and have come across this statement in appendix E: "The Universe always performs Address Invariant translation between the PCI and VMEbus ports. Address Invariant mapping preserves the byte ordering of a data structure in a littleendian memory map and a big-endian memory map." The tables in the appendix indicate that the Universe device should perform the big/little endian translation (if I understand the table correctly). So why do I need to twist the bytes in my data? Does anybody else using the Universe II device have this problem. TIA Mick Mathers ======================================================= Mick Mathers Cranfield Precision, 90 Central Avenue, Wharley End, Cranfield, Bedfordshire MK43 0JR Tel: 01234 754075 Fax: 01234 754223 From vxwexplo-errs@csg.lbl.gov Tue Jan 22 06:26:15 2002 From: "Goldmann, Andreas" Date: Tue Jan 22 06:26:18 PST 2002 Subject: usage the serial interface after PPP stopping Hi All VxWorks-Experts, After starting ( pppInit() ) and stopping PPP ( pppDelete() ) on VxWorks the serial interface doesn't respond for a time. Have You an idea, what is going on and how can I determine when the serial interface is available for AT-Commands? (Disconnect-Hooks doesn't work if the connection is terminated on the local side). Regards, Andreas Goldmann From vxwexplo-errs@csg.lbl.gov Tue Jan 22 06:46:01 2002 From: "amanullah khan" Date: Tue Jan 22 06:46:03 PST 2002 Subject: Tornado Project : Build Specifications

Hi

I have some queries related to Tornado project facility.Can anyone tell, from which files  Tornado 2.0 project facilities gets the properties of the default BUILD SPECIFICATIONS.

Regards

Aman

 

 

 



 


Send and receive Hotmail on your mobile device: Click Here
From vxwexplo-errs@csg.lbl.gov Wed Jan 23 01:38:17 2002 From: "Goldmann, Andreas" Date: Wed Jan 23 01:38:20 PST 2002 Subject: Trouble with PPP-connection terminating Hi All VxWorks-Experts, If PPP-connection is established and a modem reset is done, the PPP link isn't terminated. If You issue pppstatShow or ifShow You see all PPP-infos as if PPP-connection were okay!? Is this a bug of PPP-implementation under VxWorks? Regards, Andreas Goldmann From vxwexplo-errs@csg.lbl.gov Wed Jan 23 04:03:25 2002 From: Vxworks Exploder Date: Wed Jan 23 04:03:29 PST 2002 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Wed Jan 23 04:03:20 PST 2002 Subject: Re: Using a PCI ethernet adapter on PCPentium BSP Subject: How to get MAC address of ethernet card? Subject: spurious interrupt + cache Subject: intConnect() vs pciIntConnect() Subject: Re: SCSI Performance in vxWorks Subject: Re: linking problem Subject: wind web/access control Subject: MVME2604 Subject: Re: tffs, tffsBootPutImage and booting from fallow area Subject: Vxworks crash on deleting a memory pointer Subject: Re: intConnect() vs pciIntConnect() Subject: Re: Vxworks crash on deleting a memory pointer Subject: Re: tffs, tffsBootPutImage and booting from fallow area Subject: Re: spurious interrupt + cache Subject: Re: MVME2604 Subject: I2C driver Subject: Re: How to get MAC address of ethernet card? Subject: Re: I2C driver Subject: Kernel WorkQ Overflow Subject: Re: I2C driver Subject: Re: intConnect() vs pciIntConnect() Subject: Retrives a buffer Subject: New WindSerf, TSR's and SPR's. Subject: problem downloading Tornado Prototyper Plus Subject: Acurate, high-precision and stable timing solutions, Need Help... Subject: Target license Fees? Subject: Re: MVME2604 Subject: Re: Acurate, high-precision and stable timing solutions, Need Help... Subject: Re: Acurate, high-precision and stable timing solutions, Need Help... Subject: Re: _protoSwIndex and inetsw[ ] used in our OSPF implementation Subject: Re: errno 226 Subject: Re: intConnect() vs pciIntConnect() Subject: Re: Caching, I/O Space memory mapping and DMA Subject: Re: pppInit() (IPCP address negotiation with PPP) Subject: Re: problem downloading Tornado Prototyper Plus Subject: Re: New WindSerf, TSR's and SPR's. Subject: Re: Detecting Memory Leaks Subject: routing problem ??? Subject: WorkQPanic???? Subject: SMC driver Subject: Error while downloading project on target??? Subject: Re: spurious interrupt + cache Subject: Re: intConnect() vs pciIntConnect() Subject: Re: intConnect() vs pciIntConnect() Subject: Re: spurious interrupt + cache Subject: Re: Kernel WorkQ Overflow Subject: Re: WorkQPanic???? Subject: Signal handler executions Subject: listing link modules Subject: Re: intConnect() vs pciIntConnect() Subject: Re: I2C driver Subject: Re: Retrives a buffer Subject: Re: How to allocate memory for PCI driver? Subject: STL compilation problem Subject: Re: Symbol table errors compiling an image when using STL Subject: Re: SMC driver Subject: Wind Technical Notes Subject: Re: spurious interrupt + cache Subject: Re: tffs, tffsBootPutImage and booting from fallow area Subject: !!!!!!!!!!!!!!!!!!!! SOFTWARE FOR SALE !!!!!!!!!!!!!!!!!!!!                                                                                                                                                            ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: Using a PCI ethernet adapter on PCPentium BSP Date: Tue, 22 Jan 2002 13:24:01 +0100 From: "Werner Schiendl" Message-ID: <3c4d59e3$1@brateggebdc5.br-automation.co.at> References: <2002117-101355-104040@foorum.com> <3c46ee95$1@brateggebdc5.br-automation.co.at> <2002118-133844-311193@foorum.com> <3c4c37a1$1@brateggebdc5.br-automation.co.at> <2002121-184956-427038@foorum.com> Hi, if you have a BIOS that enables the network board at startup, you can just define INCLUDE_END and setup IO_ADDR_ENE, etc. macros accordingly to the boards settings. If you want to use the PCI mechanism to find out the ressources, WRS should be able to help you get started. (There are sample files for that in your BSP directory, e. g. sysEl3c90xEnd.c) For the 3c5x9cfg.exe, I'd be _very_ surprised if you cannot disable PnP support. I think it was in the menus, so give it a second try to find it. This board works quite well _if_ you disabled ISA PnP support and setup proper resources manually. hth Werner "Charly" wrote in message news:2002121-184956-427038@foorum.com... > > I don't see a possibility to "disable the Plug&Play" in this utility(3c5x9cfg. > exe). But just change IT and I/O. > > I am now in contact with WindRiver. They already used the Realtek RTL 8029. It > is a PCI adapter compatible with NE2000 driver. > > Regards > Charly > > > -- > Use our news server 'news.foorum.com' from anywhere. > More details at: http://nnrpinfo.go.foorum.com/ --------------------------- Newsgroups: comp.os.vxworks Subject: How to get MAC address of ethernet card? Date: 22 Jan 2002 04:27:06 -0800 From: sitalakshmi.gautham@wipro.com (Sitalakshmi Gautham) Organization: http://groups.google.com/ Message-ID: <2fe901df.0201220427.111923b1@posting.google.com> How do i get the MAC address of ethernet interface with only the interface name. etherAddressResolve() requires the IP address which i dont have in my program. --------------------------- Newsgroups: comp.os.vxworks Subject: spurious interrupt + cache Date: 22 Jan 2002 04:41:56 -0800 From: ruchama@septier.com (Ruchama Goldshmidt) Organization: http://groups.google.com/ Message-ID: <5df340e0.0201220441.6b3b3999@posting.google.com> Hello vxWorkers!!! I develop a board based on the ppc8260 processor and using also the shared memory driver using as a master an MPC750 board. When enable the Data and Instruction cache I see that when I get an interrupt in the shared memory (IRQ5 interrupt ) I also get a spurious interrupt at interrupt number 0 and get many "UNINITIALIZED INTERRUPT" logs, which is very strange since interrupt number 0 is the error interrupt so why do I get it? when the caches are disabled I get only the IRQ5 and no other ,as excepted. does anyone have an advise?? TIA, Ruchama Goldshmidt Septier Communication Ltd. --------------------------- Newsgroups: comp.os.vxworks Subject: intConnect() vs pciIntConnect() Date: 22 Jan 2002 04:55:17 -0800 From: arun4vxworks@indiatimes.com (Arun Prasad) Organization: http://groups.google.com/ Message-ID: Hi All, I couldn't able to find the exact differences between intconnect() and pciIntConnect(). Can anyone list out the exact differences of intConnect and pciIntConnect and their usage? If intConnect is able to connect a PCI based ISR then why do we need a separate call (pciIntConnect)? Are there there any specific necessity for pciIntConnect where we can't use intConnect? I'm writing an ISR for a PCI ethernet Card on a IBM PPC405 board. Which call I need to use intconnect or pciConnect? Which is better and whey? Please do clarify these. Thanks in advance. Have a nice time. Regards Arun --------------------------- Newsgroups: comp.os.vxworks Subject: Re: SCSI Performance in vxWorks Date: 22 Jan 2002 13:01:55 GMT From: ac151@FreeNet.Carleton.CA (David Clarke) Organization: The National Capital FreeNet Message-ID: References: <20020122.015817.1346811305.3143@analogic.com> Reply-To: ac151@FreeNet.Carleton.CA (David Clarke) "Jared Maguire" (jmaguire@analogic.com) writes: > Has anyone measured SCSI performnace greater than 10MB/s under vxWorks? > What card/driver? The controller/drive is only part of the equation - the other part is going to be the type of connection/interface. If the board or drive only supports Fast SCSI, then 10MB/s is going to be fastest you can go. If you can do Fast Wide (16 bit), then you can get up to 20MB/s - and Wide Ultra will get you up to a max. possible of 40MB/s. Using an Ultra Wide Western Digital drive with a DY 4 SVME/DMV-179 SBC configured for 16-bit SCSI, we've gotten > 30MB/s (read/write). - -- - ------------------------------- /''' -------------------------------- David Clarke c-00 ac151@Freenet.carleton.ca > - ------------------------------- - -------------------------------- --------------------------- Newsgroups: comp.os.vxworks Subject: Re: linking problem Date: Tue, 22 Jan 2002 09:20:03 -0500 From: david lindauer Message-ID: <3C4D7513.1090501@notifier-is.net> References: <3C46D729.70802@notifier-is.net> - --------------040001080703030807090608 Content-Type: text/plain; charset=us-ascii; format=flowed Content-Transfer-Encoding: 7bit thanks! Joe Hagen wrote: >David, > >If this was "C" code, then the linkage would work. > >However, in C++, "const" will cause the variable to >have internal linkage, so the behavior you're seeing >is because the internal linkage "hides" the variable >from other modules. > >Joe > >"david lindauer" wrote in message >news:3C46D729.70802@notifier-is.net... > >>In one C++ file I have: >> >>typedef char * sz ; >>sz const monthlist[12] = { "jan","feb",... } ; >> >>and in another: >> >>typedef char *sz ; >>extern sz const monthlist[12] ; >> >>For some reason LD insists that the reference in the second file can not >>be resolved. I have used nm on the object files and both list >>'_monthlist', unadorned. One is type 't' and one is type 'U'. So it >>should work. Both files are definitely in the project. >> >>I have tried changing the name of the variable with no success. Anyone >>know what I am doing wrong? >> >>David >> > > - --------------040001080703030807090608 Content-Type: text/html; charset=us-ascii Content-Transfer-Encoding: 7bit thanks!

Joe Hagen wrote:
David,

If this was "C" code, then the linkage would work.

However, in C++, "const" will cause the variable to
have internal linkage, so the behavior you're seeing
is because the internal linkage "hides" the variable
from other modules.

Joe

"david lindauer" <dlindauer@notifier-is.net> wrote in message
news:3C46D729.70802@notifier-is.net...
In one C++ file I have:

typedef char * sz ;
sz const monthlist[12] = { "jan","feb",... } ;

and in another:

typedef char *sz ;
extern sz const monthlist[12] ;

For some reason LD insists that the reference in the second file can not
be resolved.  I have used nm on the object files and both list
'_monthlist', unadorned.  One is type 't' and one is type 'U'.  So it
should work.  Both files are definitely in the project.

I have tried changing th
e name of the variable with no success.  Anyone
know what I am doing wrong?

David




- --------------040001080703030807090608-- --------------------------- Newsgroups: comp.os.vxworks Subject: wind web/access control Date: Tue, 22 Jan 2002 09:26:44 -0500 From: david lindauer Message-ID: <3C4D76A4.8090608@notifier-is.net> We took a working WindWeb project and added access control (password checking and access control modules). We added appropriate configuration for the access control at the start of the httpd function in httpconf.c. At this point the browser asks for a password and then the server usually serves up a page 'server error 501, invalid RPM, unknown combination of URL and Method' and it lists the correct page. Sometimes the server will serve up the content we want, but this is rare. Taking access control back out makes it work properly. Anyone have any idea what would make it do this? We are using wind web 2.0. Thanks, David --------------------------- Newsgroups: comp.os.vxworks Subject: MVME2604 Date: Tue, 22 Jan 2002 15:47:28 +0100 From: Mark Koennecke Message-ID: High there, I have got a Motorolla MVME2604 board (from 1998) which I cannot get to boot. I built a boot image from the distribution and it loads alright but then does nothing when I boot. We triple checked the loading parameters and they seem fine. Due to organisational reasons I have only access to version 1.0.1 of the wind river software. My questions: - Has anybody managed to use such a board before and if so, with which version of vxWorks? - If it works with the version I have, what do I have to do? I assume a fix is required somewhere in the wind river source code. Thanks for any help Mark Koennecke --------------------------- Newsgroups: comp.os.vxworks Subject: Re: tffs, tffsBootPutImage and booting from fallow area Date: Tue, 22 Jan 2002 14:54:55 +0000 From: Owain Phillips Organization: Siemens Inc. Message-ID: <3C4D7D3F.FCEBE820@siemenscomms.co.uk> References: <3C4C4F80.3F96AD2C@siemenscomms.co.uk> This is a multi-part message in MIME format. - --------------341519C35C666286AC294347 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Hi Andray, Andray Kaganovsky wrote: > Owain Phillips wrote in > news:3C4C4F80.3F96AD2C@siemenscomms.co.uk: > > > Hi, > > > > Does tffsBootPutImage work? > > > > I do a tffsUnmount on the drive, then tffsBootPutImage( 0, 0, > > "vxWorks"). > > The first Meg of the image is burn to flash before the call falls over > > with > > "Physical write failed at address 0x100000". > > [AndrayK]. When you formatted flash array with tffsDevFormat(), what > was the value of tffsFormatParams.bootImageLen ? Do you have enough > room in the "raw" flash area that resides before FTL-formatted part > of flash array ? > Device was formated with a bootImageLen of 0x520000; so doesn't explain the fall over at the 1M limit. > > > I acutally try to program the elf image into ROM rather than a binary > > image, but at this stage should this really matter? > > [AndrayK]. As far as I know, it doesn't matter. > > > I can write my file image to the fallow region using one of my own > > knocked up routines, so the flash is OK; but does tffsBootPutImage do > > some other house keeping? > > > What should my bootline be to boot from this fallow region? > > [AndrayK]. I don't think you can use TrueFFS for booting from "raw" > flash. My understanding is that you need to configure your board to > place beginning of "raw" flash area at CPU reset vector, and that should > do the trick. > I shall try to create a memDrv device to treat an area of raw flash as a device from which I can boot. Thanks for suggestions, Owain - --------------341519C35C666286AC294347 Content-Type: text/x-vcard; charset=us-ascii; name="owain.phillips.vcf" Content-Transfer-Encoding: 7bit Content-Description: Card for Owain Phillips Content-Disposition: attachment; filename="owain.phillips.vcf" begin:vcard n:Phillips;Owain tel;fax:+44 115 943 4969 tel;work:+44 115 943 2167 x-mozilla-html:TRUE url:http://www.siemenscomms.co.uk org:Siemens Comunications Ltd version:2.1 email;internet:owain.phillips@siemenscomms.co.uk title:Software Engineer adr;quoted-printable:;;Technology Drive=0D=0ABeeston;Nottingham;;NG9 1LA;United Kingdom fn:Owain Phillips end:vcard - --------------341519C35C666286AC294347-- --------------------------- Newsgroups: comp.os.vxworks Subject: Vxworks crash on deleting a memory pointer Date: 22 Jan 2002 07:02:45 -0800 From: Sajith.Satheesan@geind.ge.com (Sajith Satheesan) Organization: http://groups.google.com/ Message-ID: <9b5debcd.0201220702.292df15d@posting.google.com> Hi All, I am having a page fault beging generated when I delete a pointer to an object in Vxworks. The operation on which the delete occurs is delete aiqreq; (where aiqreq is a pointer to an object) The page fault is occuring in the dllRemove function which is called as follows destructor -> builtin_delete -> free -> dllRemove -> page fault. other details are Page Fault Program Counter: 0x0022bcf8 Status Register: 0x00010286 Error Code: 0x00000002 Can anyone tell me what is causing this pagefault ? Thanks, Sajith --------------------------- Newsgroups: comp.os.vxworks Subject: Re: intConnect() vs pciIntConnect() Date: Tue, 22 Jan 2002 15:29:08 GMT From: "H. E. Taylor" Organization: MTS Internet Message-ID: <3C4DA04D.4E96@despam.autobahn.mb.ca> References: In article , Arun Prasad wrote: > > I couldn't able to find the exact differences between intconnect() > and pciIntConnect(). > > Can anyone list out the exact differences of intConnect and > pciIntConnect and their usage? > intConnect modifies the interrupt vector table. pciIntConnect adds your ISR to a collection. The essential fact to note is that there are only _4_ PCI interrupt vectors and there can be many PCI devices. So you need a way to share. >[...] > I'm writing an ISR for a PCI ethernet Card on a IBM PPC405 board. > Which call I need to use intconnect or pciConnect? Which is better > and whey? > pciIntConnect for the reasons above. - -het - -- "progress in software has not followed Moore's law." -John Holland Energy Alternatives: http://www.autobahn.mb.ca/~het/energy.html H.E. Taylor http://www.autobahn.mb.ca/~het/ --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Vxworks crash on deleting a memory pointer Date: 22 Jan 2002 10:21:45 -0500 From: Bill Pringlemeir Organization: Bell Sympatico Message-ID: References: <9b5debcd.0201220702.292df15d@posting.google.com> Sender: bpringlemeir@DeadDuck >>>>> "Sajith" == Sajith Satheesan writes: Sajith> Hi All, I am having a page fault beging generated when I Sajith> delete a pointer to an object in Vxworks. [snip] Sajith> destructor -> builtin_delete -> free -> dllRemove -> page Sajith> fault. Sajith> other details are Sajith> Page Fault Program Counter: 0x0022bcf8 Status Register: Sajith> 0x00010286 Error Code: 0x00000002 Sajith> Can anyone tell me what is causing this pagefault ? If you pass the destructor a bad pointer, then this will happen. It is most likely some sort of over-write situation. Unfortunately people often think that `bottom dwellers' like dllRemove() are responsible. I wrote a custom function for vxMemProbe() and now every time anyone in my company uses a bad pointer, I get a call... dllRemove() is just trying to take the memory from a list. Since it is crashing, I would guess that the memory before the object had been over-written. The allocator stores its own data there. If possible, set a breakpoint on the destructor call and look at the pointer values and memory before the object. Is it sane? Ie, does the pointer point to RAM. Are the 16 bytes before the pointer the same as when allocated (in the constructor)? You may also wish to tell us more of your system (BSP, Tornado/vxWorks version, and CPU) if you need further help. hth, Bill Pringlemeir. - -- How can you say that all stupid dudes should get ten-inchers ? I mean, you make me sick. vxWorks FAQ, "http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html" --------------------------- Newsgroups: comp.os.vxworks Subject: Re: tffs, tffsBootPutImage and booting from fallow area Date: Tue, 22 Jan 2002 11:50:44 GMT From: Andray Kaganovsky Organization: Primus Canada Message-ID: References: <3C4C4F80.3F96AD2C@siemenscomms.co.uk> <3C4D7D3F.FCEBE820@siemenscomms.co.uk> Owain Phillips wrote in news:3C4D7D3F.FCEBE820@siemenscomms.co.uk: > Hi Andray, > > Andray Kaganovsky wrote: > >> Owain Phillips wrote in >> news:3C4C4F80.3F96AD2C@siemenscomms.co.uk: >> >> > Hi, >> > >> > Does tffsBootPutImage work? >> > >> > I do a tffsUnmount on the drive, then tffsBootPutImage( 0, 0, >> > "vxWorks"). The first Meg of the image is burn to flash before the >> > call falls over with "Physical write failed at address 0x100000". >> >> [AndrayK]. When you formatted flash array with tffsDevFormat(), what >> was the value of tffsFormatParams.bootImageLen ? Do you have enough >> room in the "raw" flash area that resides before FTL-formatted part >> of flash array ? >> > > Device was formated with a bootImageLen of 0x520000; so doesn't explain the > fall over at the 1M limit. [snip] [AndrayK 2]. Did you set breakpoint in MTD's 'write' routine, to see if MTD actually receives 'write' requests with offsets higher then 0x100000 ? regards, Andray --------------------------- Newsgroups: comp.os.vxworks Subject: Re: spurious interrupt + cache Date: Tue, 22 Jan 2002 16:56:04 +0100 From: "Michael Lawnick" Organization: Buergernetz Dillingen Message-ID: References: <5df340e0.0201220441.6b3b3999@posting.google.com> Reply-To: "Michael Lawnick" Sender: mlawnick@pd9e65a7c.dip.t-dialin.net Hi Ruchama sounds for the 'your CPU is faster than your bus' problem: When you acknowledge the interrupt on the slave hardware, your actual write-access seems to be done by some 'middle-ware', e.g. PCI-bridge, VME-bridge,... Your CPU has already left ISR and reenabled its interrupts, before the real access on your slave is done. This explains, why uncached acknowledges succeed: your CPU is slower now. Do a read-back of the register you write your ACK to and all should be fine. Another problem could be, that your device is mapped as cacheable, but that I beleave would lead to complete malfunction. - -- Mit freundlichen Grüßen, Michael Lawnick ============================================== SOFTEC GmbH Tel +49-731-96600-0 Promenade 17 Fax +49-731-96600-23 D-89073 Ulm Michael Lawnick Germany lawnick@softec.de ============================================== "Ruchama Goldshmidt" schrieb im Newsbeitrag news:5df340e0.0201220441.6b3b3999@posting.google.com... > Hello vxWorkers!!! > > I develop a board based on the ppc8260 processor and using also the > shared memory driver using as a master an MPC750 board. > > When enable the Data and Instruction cache I see that when I get an > interrupt in the shared memory (IRQ5 interrupt ) I also get a spurious > interrupt at interrupt number 0 and get many "UNINITIALIZED INTERRUPT" > logs, > which is very strange since interrupt number 0 is the error interrupt > so why do I get it? > > when the caches are disabled I get only the IRQ5 and no other ,as > excepted. > > does anyone have an advise?? > > TIA, > > Ruchama Goldshmidt > Septier Communication Ltd. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: MVME2604 Date: Tue, 22 Jan 2002 17:03:55 +0100 From: "Michael Lawnick" Organization: Buergernetz Dillingen Message-ID: References: Reply-To: "Michael Lawnick" Sender: mlawnick@pd9e65a7c.dip.t-dialin.net Hi Mark, I have worked with that board under T101 and T2, all was fine. Minimum todo is the newest cummulative patch, but then it should work. Build a bootrom.hex, convert it with elfToBin and download it with PPCBug via TFTP (a TFTP server for Windows is part of all newer distributions). Proceed as described in target.nr in your BSP. Have you checked, you see the WRS Copyright string at 0xFF0001xx after pflush'ing the image ? - -- Mit freundlichen Grüßen, Michael Lawnick ============================================== SOFTEC GmbH Tel +49-731-96600-0 Promenade 17 Fax +49-731-96600-23 D-89073 Ulm Michael Lawnick Germany lawnick@softec.de ============================================== "Mark Koennecke" schrieb im Newsbeitrag news:Pine.OSF.4.20.0201221538440.25605-100000@lnsa17.psi.ch... > > High there, > > I have got a Motorolla MVME2604 board (from 1998) which I cannot get to > boot. I built a boot image from the distribution and it loads alright > but then does nothing when I boot. We triple checked the loading > parameters and they seem fine. > > Due to organisational reasons I have only access to version 1.0.1 of > the wind river software. > > My questions: > - Has anybody managed to use such a board before and if so, with which > version of vxWorks? > - If it works with the version I have, what do I have to do? I assume > a fix is required somewhere in the wind river source code. > > Thanks for any help > > > Mark Koennecke > --------------------------- Newsgroups: comp.os.vxworks Subject: I2C driver Date: Tue, 22 Jan 2002 16:30:18 GMT From: Tom Flynn Organization: AT&T Broadband Message-ID: <3C4D9347.43AA0E67@attbi.com> Reply-To: tomflynn2@attbi.com The I2C driver in the BSP that I'm using is only written for use during boot. It is not multitasking safe (according to the author's comments). We need to access the I2C during normal run time on the board that we are using. I have two questions. First, what would need to be done to make it usable during run time? Second, is there driver code somewhere that is multitasking safe that I can look at? I have access to other BSP code, so if someone knows of a BSP that includes an I2C driver that is multitasking safe, I would be interested in knowing which BSP that is. Thanks in advance. - -Tom Flynn --------------------------- Newsgroups: comp.os.vxworks Subject: Re: How to get MAC address of ethernet card? Date: Tue, 22 Jan 2002 17:39:20 +0100 From: "Werner Schiendl" Message-ID: <3c4d95b9$1@brateggebdc5.br-automation.co.at> References: <2fe901df.0201220427.111923b1@posting.google.com> Hi, in the VxWorks FAQ see http://www.xs4all.nl/~borkhuis/vxworks/vxw_pt4.html#4.8 hth Werner "Sitalakshmi Gautham" wrote in message news:2fe901df.0201220427.111923b1@posting.google.com... > How do i get the MAC address of ethernet interface with only the > interface name. etherAddressResolve() requires the IP address which i > dont have in my program. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: I2C driver Date: Tue, 22 Jan 2002 16:45:59 GMT From: "cfk@pacbell.net" Organization: Prodigy Internet http://www.prodigy.com Message-ID: References: <3C4D9347.43AA0E67@attbi.com> Dear Tom: I2C is a very simple two wire protocol with just data (SDA) and clock (SCL). It is certainly possible to interrupt an I2C driver in the midst of transmitting a byte. AFAIR, this needs to be done when SCL is in the low state, but you can read the I2C specification available from Philips under the paragraph "Pulse Stretching". The key thing to making an I2C driver multitasking safe is to ensure that the task switch occurs when the clock is in the appropriate state. I suspect the author of this driver merely didn't put the appropriate logic in for that. It is fairly simple to implement, it just takes a day or so of head scratching and studying other implementations of "bit banging" I2C software. Charles "Tom Flynn" wrote in message news:3C4D9347.43AA0E67@attbi.com... > > The I2C driver in the BSP that I'm using is only written for use during > boot. It is not multitasking safe (according to the author's comments). > We need to access the I2C during normal run time on the board that we > are using. > > I have two questions. First, what would need to be done to make it > usable during run time? > > Second, is there driver code somewhere that is multitasking safe that I > can look at? I have access to other BSP code, so if someone knows of a > BSP that includes an I2C driver that is multitasking safe, I would be > interested in knowing which BSP that is. > > Thanks in advance. > > -Tom Flynn > --------------------------- Newsgroups: comp.os.vxworks Subject: Kernel WorkQ Overflow Date: Tue, 22 Jan 2002 12:16:07 -0500 From: "nitin kumar" Organization: Marconi Message-ID: Hi, Is there a scenario where repeated msgQSend from a routine cause Kernel WorkQ over flow. msgQSend() is NOT being called from an ISR. My ISR just does a semGive() and wakes up another task that does the send to the Q. The Message Q is 256 entries large and I envision it will be full quite often so that msgQSend() shall be failing frequently. ( But that should not be a problem as the send is done with NO_TIMEOUT ). The board just hangs after a while, and sometimes I see the KernelworkQ overflow message. Please help. thanks nitin --------------------------- Newsgroups: comp.os.vxworks Subject: Re: I2C driver Date: Tue, 22 Jan 2002 18:22:59 GMT From: "Jim" Organization: Magma Communications Ltd. Message-ID: <76i38.17309$8e.1027293@news> References: <3C4D9347.43AA0E67@attbi.com> This is a stupid question ... but does wrapping calls to i2c with ioctl provide concurency solution? I have not RTM ... just thought i'd ask ... Jim "cfk@pacbell.net" wrote in message news:bHg38.35006$Fn6.1618022548@newssvr21.news.prodigy.com... > Dear Tom: > I2C is a very simple two wire protocol with just data (SDA) and clock > (SCL). It is certainly possible to interrupt an I2C driver in the midst of > transmitting a byte. AFAIR, this needs to be done when SCL is in the low > state, but you can read the I2C specification available from Philips under > the paragraph "Pulse Stretching". The key thing to making an I2C driver > multitasking safe is to ensure that the task switch occurs when the clock is > in the appropriate state. I suspect the author of this driver merely didn't > put the appropriate logic in for that. It is fairly simple to implement, it > just takes a day or so of head scratching and studying other implementations > of "bit banging" I2C software. > > Charles > > "Tom Flynn" wrote in message > news:3C4D9347.43AA0E67@attbi.com... > > > > The I2C driver in the BSP that I'm using is only written for use during > > boot. It is not multitasking safe (according to the author's comments). > > We need to access the I2C during normal run time on the board that we > > are using. > > > > I have two questions. First, what would need to be done to make it > > usable during run time? > > > > Second, is there driver code somewhere that is multitasking safe that I > > can look at? I have access to other BSP code, so if someone knows of a > > BSP that includes an I2C driver that is multitasking safe, I would be > > interested in knowing which BSP that is. > > > > Thanks in advance. > > > > -Tom Flynn > > > > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: intConnect() vs pciIntConnect() Date: 22 Jan 2002 10:27:41 -0800 From: arun4vxworks@indiatimes.com (Arun Prasad) Organization: http://groups.google.com/ Message-ID: References: <3C4DA04D.4E96@despam.autobahn.mb.ca> Dear Taylor, Thank you for your reply. << intConnect modifies the interrupt vector table. pciIntConnect adds your ISR to a collection. >> I got the ponit now. To use intConnect, I should have unique (unshared) interrupt for my device as it modifies the IVT. Otherwise I need to make it unique by setting a unique IRQ to a PCI Slot and putting our PCI based Card there and it simplifies our task in ISR as we don't need to check which device raised the interrupt as the IRQ has been already made as unique. Hope I'm right :-) If not please do let me know my misunderstanding. Regards Arun "H. E. Taylor" wrote in message news:<3C4DA04D.4E96@despam.autobahn.mb.ca>... > In article , > Arun Prasad wrote: > > > > I couldn't able to find the exact differences between intconnect() > > and pciIntConnect(). > > > > Can anyone list out the exact differences of intConnect and > > pciIntConnect and their usage? > > > > intConnect modifies the interrupt vector table. > pciIntConnect adds your ISR to a collection. > > The essential fact to note is that there are only _4_ PCI interrupt > vectors and there can be many PCI devices. So you need a way to share. > > >[...] > > I'm writing an ISR for a PCI ethernet Card on a IBM PPC405 board. > > Which call I need to use intconnect or pciConnect? Which is better > > and whey? > > > > pciIntConnect for the reasons above. > > > -het --------------------------- Newsgroups: comp.os.vxworks Subject: Retrives a buffer Date: 22 Jan 2002 10:54:26 -0800 From: jzhang@drs.ca (Jason) Organization: http://groups.google.com/ Message-ID: Hello, Can anybody tell me how to retrive a buffer from buffer poll and how to free the buffer on vxWorks. i.e. some function like this: buffGet (int BuffSize), buffFree () Cheers --------------------------- Newsgroups: comp.os.vxworks Subject: New WindSerf, TSR's and SPR's. Date: 22 Jan 2002 14:08:26 -0500 From: Bill Pringlemeir Organization: Bell Sympatico Message-ID: Sender: bpringlemeir@DeadDuck How does one access SPRs under the new WindSerf? It is hard to install shockwave plug-ins on lynx [in case the info is in the tutorial] ... regards, Bill Pringlemeir. - -- I feel like a wet parking meter on Darvon! vxWorks FAQ, "http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html" --------------------------- Newsgroups: comp.os.vxworks Subject: problem downloading Tornado Prototyper Plus Date: 22 Jan 2002 11:39:14 -0800 From: infovaan@lycos.com (Info Vaan) Organization: http://groups.google.com/ Message-ID: <71109c73.0201221139.14756619@posting.google.com> hi, i'm new to vxworks, and would have liked to wet my feet on vxworks using Tornado Prototyper plus, but on WindRiver website it seems all links pointing to it are broken. does anybody have any idea as to what's going on, and if they have any alternate links. have written to support and tornadoprototyper email id's but so far no result (waiting for last 5 days now). thanks, iw. --------------------------- Newsgroups: comp.os.vxworks Subject: Acurate, high-precision and stable timing solutions, Need Help... Date: 22 Jan 2002 11:46:05 -0800 From: gryffons@hotmail.com (Marc LeBlanc) Organization: http://groups.google.com/ Message-ID: <6087305d.0201221146.7cacd1fc@posting.google.com> I am currently working on implementing a TDMA protocol that requires from 1ms to 10ms timing difference between every nodes. It has to run on common Pentium PCs. Every node is completely separate from other nodes, they communicate through a radio link that is far from stable (or constant). So I need a realy stable timing solution to keep accurate timing between the nodes. The nodes will have access to GPS to get a common and constant 1 second interrupt. At this time, I'm simply using watchdog timers that I periodically synchronize to NTP. This provide about 2 to 4 system ticks of error between the nodes (at 100Hz clock rate ~ 30ms). This is too much. I need a higher precision timer than the system clock. And it has to be much more stable (the system clock drifts up to 8 seconds per day, this is bad). Are there any PCI cards that provide programmable high resolution constant interrupts? (say I want an interrupt every 30 seconds and I want it to be precise up to 1ms in the span of a day). This card should be easy to integrate in vxWorks. I'm also open to a software solution to this problem (cheaper is better) Thank you, Marc LeBlanc --------------------------- Newsgroups: comp.os.vxworks Subject: Target license Fees? Date: Tue, 22 Jan 2002 13:52:50 -0600 From: Andrew Johnson Organization: APS, Argonne National Laboratory Message-ID: <3C4DC312.E0C18DBE@aps.anl.gov> Can anyone confirm whether the following rumor that I've just heard could be true? I'm not sure I believe it, but it's feasible and would be an interesting development for scientific institutions. The rumor is that in the future WRS are only going to charge for target licenses that get sold on to other people/companies, not for targets that are only ever used in-house by the organization that owns the development license. Can anyone shed any light on this, or scotch it? My contact doesn't like asking WRS licensing questions that s/he doesn't already know the answer to... - - Andrew - -- Perfection is reached, not when there is no longer anything to add, but when there is no longer anything to take away. - - Antoine de Saint-Exupery --------------------------- Newsgroups: comp.os.vxworks Subject: Re: MVME2604 Date: 22 Jan 2002 12:05:24 -0800 From: john_94501@yahoo.com (John) Organization: http://groups.google.com/ Message-ID: <488e459a.0201221205.12b4276c@posting.google.com> References: Hello, Tornado 1.0.1 will work fine with the mv2600 series; you should make sure that you have an mv2604 BSP (there is one available for the 1.0.1 product - speak to your WRS sales person if you don't have it), but apart from that nothing special needs to be done to support this board. I would start by programming the bootrom_uncmp file that was installed by the CD rather than building your own, though it should not make any difference if you haven't changed any config files. How do you program the flash on the board? Are you using PPCBug to do it, or an external programmer? Make sure that you are giving whatever application you are using to burn the flash the correct format file. A common mistake is providing an S-records (hex) file when a binary file was expected. If all else fails, try calling your local FAE and see if they can help you out. HTH, John... Mark Koennecke wrote in message news:... > High there, > > I have got a Motorolla MVME2604 board (from 1998) which I cannot get to > boot. I built a boot image from the distribution and it loads alright > but then does nothing when I boot. We triple checked the loading > parameters and they seem fine. > > Due to organisational reasons I have only access to version 1.0.1 of > the wind river software. > > My questions: > - Has anybody managed to use such a board before and if so, with which > version of vxWorks? > - If it works with the version I have, what do I have to do? I assume > a fix is required somewhere in the wind river source code. > > Thanks for any help > > > Mark Koennecke --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Acurate, high-precision and stable timing solutions, Need Help... Date: 22 Jan 2002 15:02:16 -0500 From: Bill Pringlemeir Organization: Bell Sympatico Message-ID: References: <6087305d.0201221146.7cacd1fc@posting.google.com> Sender: bpringlemeir@DeadDuck >>>>> "Marc" == Marc LeBlanc writes: Marc> I am currently working on implementing a TDMA protocol that Marc> requires from 1ms to 10ms timing difference between every Marc> nodes. It has to run on common Pentium PCs. Every node is [snip] Marc> The nodes will have access to GPS to get a common and constant Marc> 1 second interrupt. At this time, I'm simply using watchdog [snip] The watchdog timers run off of the system clock. The default rate of the system clock is one `tick' every 60th of a second. You could re-program this to be 1 ms. See sysClkRateSet(), usrConfig.c. This might be problematic. I think that your problem is `clock skew' between the systems. The GPS source is universal. However, each node could receive the GPS tick when it is about to clock the system or just after it has clocked the system. With the standard vxWorks tick, this would give a skew of 16ms (at least). Cutting the system tick interval to 1ms would limit this. An alternative problem is that something is preventing the watchdogs routines from being called. This could be due to system loading at the interrupt level (or locked interrupts). If you do have a separate timer, you can avoid the problem of the `clock skew' by starting/resetting the timer with every GPS interrupt (assuming that the GPS interrupts are blocked somehow). hth, Bill Pringlemeir. - -- Not enough people play SKEE-BALL.. They're always thinking about COCAINE or and ALIEN BEINGS!! vxWorks FAQ, "http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html" --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Acurate, high-precision and stable timing solutions, Need Help... Date: Tue, 22 Jan 2002 20:09:12 GMT From: "cfk@pacbell.net" Organization: Prodigy Internet http://www.prodigy.com Message-ID: References: <6087305d.0201221146.7cacd1fc@posting.google.com> Dear Marc: If I had to solve this problem, I would try to obtain a GPS receiver that in itself was a PCI card. I did a search on google just now and the first link that came up was from http://www.truetime.com and they advertise a GPS card with a synchronized generator capable of interrupting with 3 microsecond accuracy. I suspect your problem with a system that depends on a communications link, be it USB, LAN, Serial, Parallel will always be a non-deterministic latency of the communications itself. Perhaps using the GPS more would help you. Charles "Marc LeBlanc" wrote in message news:6087305d.0201221146.7cacd1fc@posting.google.com... > I am currently working on implementing a TDMA protocol that requires > from 1ms to 10ms timing difference between every nodes. It has to run > on common Pentium PCs. Every node is completely separate from other > nodes, they communicate through a radio link that is far from stable > (or constant). So I need a realy stable timing solution to keep > accurate timing between the nodes. > > The nodes will have access to GPS to get a common and constant 1 > second interrupt. At this time, I'm simply using watchdog timers that > I periodically synchronize to NTP. This provide about 2 to 4 system > ticks of error between the nodes (at 100Hz clock rate ~ 30ms). This is > too much. > > I need a higher precision timer than the system clock. And it has to > be much more stable (the system clock drifts up to 8 seconds per day, > this is bad). > > Are there any PCI cards that provide programmable high resolution > constant interrupts? (say I want an interrupt every 30 seconds and I > want it to be precise up to 1ms in the span of a day). This card > should be easy to integrate in vxWorks. > > I'm also open to a software solution to this problem (cheaper is > better) > > Thank you, > Marc LeBlanc --------------------------- Newsgroups: comp.os.vxworks Subject: Re: _protoSwIndex and inetsw[ ] used in our OSPF implementation Date: 22 Jan 2002 20:12:04 GMT From: jefftkauff@aol.com (JeffTKauff) Organization: AOL http://www.aol.com Message-ID: <20020122151204.01346.00000273@mb-cg.aol.com> References: <1d411918.0201170053.b929e4e@posting.google.com> The easiest way is to use a raw socket (SOCK_RAW) and register for packets of type IPPROTO_OSPF. That way you don't have to mess with the networking internals. Jeff --------------------------- Newsgroups: comp.os.vxworks Subject: Re: errno 226 Date: 22 Jan 2002 12:35:39 -0800 From: bhorowit@cs.berkeley.edu (Benjamin Horowitz) Organization: http://groups.google.com/ Message-ID: References: <488e459a.0201201507.29177ca1@posting.google.com> You have diagnosed the problem -- amazing, given the fact that you had so little information to go on. The root cause was indeed that my ftp server is configured so that clients do not have write permission. Thanks a lot! Ben john_94501@yahoo.com (John) wrote in message news:<488e459a.0201201507.29177ca1@posting.google.com>... > > > You are probably using the FTP based netDrv file system. That would > explain both the error number, and the fact that it only failed when > you closed the file. > > ... --------------------------- Newsgroups: comp.os.vxworks Subject: Re: intConnect() vs pciIntConnect() Date: Tue, 22 Jan 2002 20:45:02 GMT From: "H. E. Taylor" Organization: MTS Internet Message-ID: <3C4DEA57.6992@despam.autobahn.mb.ca> References: <3C4DA04D.4E96@despam.autobahn.mb.ca> In article , Arun Prasad wrote: > Dear Taylor, > > Thank you for your reply. > >> intConnect modifies the interrupt vector table. >> pciIntConnect adds your ISR to a collection. > > I got the ponit now. > To use intConnect, I should have unique (unshared) interrupt for my > device as it modifies the IVT. Otherwise I need to make it unique by > setting a unique IRQ to a PCI Slot and putting our PCI based Card > there and it simplifies our task in ISR as we don't need to check > which device raised the interrupt as the IRQ has been already made as > unique. > Hope I'm right :-) > If not please do let me know my misunderstanding. > Sounds like you got it about right, although I would always check to make sure my device raised the interrupt. Weirdness happens. (Usually when you least can afford it.) I cannot over emphasize the importance of developing a sense of play in these matters. Experiment, play with various combinations, see what happens. Think of it as on the job training if you have to rationalize. The more you learn about the way your system works, the better off you will be in the crunch. Btw, you should have source for pciIntConnect(). - -het - -- "If a packet hits a pocket on a socket on a port, And the bus is interrupted as a very last resort, And the address of the memory makes your floppy disk abort, Then the socket packet pocket has an error to report!" - -Gene Ziegler (with a nod to Dr. Suess) Global Warming: http://www.autobahn.mb.ca/~het/globalwarming.html H.E. Taylor http://www.autobahn.mb.ca/~het/ --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Caching, I/O Space memory mapping and DMA Date: 22 Jan 2002 13:15:05 -0800 From: arun4vxworks@indiatimes.com (Arun Prasad) Organization: http://groups.google.com/ Message-ID: References: <8bG_7.327$qS6.799090@newsr2.u-net.net> Hi Micheal, Thank you for your comments/reply/clarifications. <>direct EMail? No. It's not. And yes, I am looking for additional/different opinions for my questions/doubts. And sorry for replying at this much delay. <>And now I got how to interpret it. Thanks. << This seems to be suspicious - any confusion possible ? Either DMA or local buffer, both would be redundant (surely still possible, but who wants to pay?) >>No. that's correct only. Our ethernet card does some other functionalities also and its bit costly too. <>I. See. We are enabling the MMU and we use cacheDmaMalloc only. So one of our problem (cache validation/flush) solved. Is it? And what happens if we write code with cacheDmaMalloc and if we disable MMU? And still I have a doubt on I/O access routines (sysInXXX and sysOutXXX). We use IBM PPC405GP which is Big Endian and our ethernet NIC card is PCI based (as you know PCI is little- Endian). As I already mentioned I find only simple assignment statements and a EIOIOSYNC macro in sysInByte and sysOutByte routines like this UCHAR data = (*(UCHAR *)address); EIEIO_SYNC; return data; Here if I change the assignment statement like the following one for sysInDWord () ULONG data = (*(ULONG *) address); Will it work? This is where we feel, we have to swap the bytes before making assignments to make it Endian-safe. (I mean endian conversion required here.) Hoping for a clarification on this Endian Issue. Have a nice time. Regards Arun "Michael Lawnick" wrote in message news:... > Hi Arund, > > as you posted your direct EMails with Dave here, I assume you want > different/additional opinions to your question. Here my 2 Cents: > > As mentioned correctly by Dave, PPC always see's PCI as memory mapped, but > this does not mean, that a PPC can't do PCI-I/O. It means that there is a > memory space, where the PCI-bridge of PPC will translate all accesses into > I/O-accesses and there are other spaces, that are interpreted in other ways. > > Your excerpt from sysPhysMemDesc is just _one_ entry and its correct: > > VM_STATE_MASK_VALID | VM_STATE_MASK_WRITABLE |VM_STATE_MASK_CACHEABLE > |VM_STATE_MASK_GUARDED > tells which flags you want to change (_MASK_) > > VM_STATE_VALID | VM_STATE_WRITABLE | VM_STATE_CACHEABLE_NOT > |VM_STATE_GUARDED > enables the entry and sets it writable, uncached and guarded - just as you > need it. > > You write: > > >>I'm sorry for the confusion. Our Card has its own DMA controller and > > its own Buffer/memory. > This seems to be suspicious - any confusion possible ? > Either DMA or local buffer, both would be redundant (surely still possible, > but who wants to pay?) > On our PPC405, our scndNIC 's registers have to be set up through PCI, data > is written to CPU-memory via DMA (by the NIC). As we have shut off MMU, I > must always do invalidate before data reading and flush before telling the > NIC that there's a new packet to transmit. > If our MMU would be enabled, I could allocate memory with cacheDmaMalloc and > wouldn't need to care. > > To complete your confusion about sysInByte, sysOutByte, ..., there are more > commands: > sysPciInByte, sysPciOutByte,... > The trick is: > the sysInXXX variant will write data as is, typically found on Intel arch. > sysPciInXXX will make a swap before access. This is needed, as PPC is > big-endian, but PCI is little-endian. Second issue are the instruction pipes > of PPC. As you might do two successive accesses to different PCI-registers, > but the result of the second one might be influenced by the first (e.g. > first enable device via control register, then read from status register), > it must be ensured, that the first instruction completed before the second. > This is done by a sync instruction as you might see in sysALib.s > > -- > Mit freundlichen Grüßen, > Michael Lawnick > ============================================== > SOFTEC GmbH Tel +49-731-96600-0 > Promenade 17 Fax +49-731-96600-23 > D-89073 Ulm Michael Lawnick > Germany lawnick@softec.de > ============================================== > > "Arun Prasad" schrieb im Newsbeitrag > news:e38d912c.0201091154.76631ff5@posting.google.com... > > Dear Dave, > > > > Thank you for your detailed reply/clarifications. > > > > < > was an x86 chip. > > >>Yes. You are absolutely right. I DO NOT HAVE EXPERIENCE other than > > x86, though I READ a bit on Z80 and Motorola 68000 series chips. > > > > < > space and memory space on PPC: everything is mapped to memory space, > > and there is no I/O port space. > > < > >>Is it so? I thought it's device/controller/card specific. If my > > memory is correct, I've read from Goankar's Microprocessor book that > > it is device-specific. But again that book dealt with 8085 and 8086 > > processors. And previously I worked on Linux Driver for SCSI Raid > > Controllers with i960 processors in them and I found I/O mapped and > > memory mapped cards separately and accordingly all the device access > > functions varied. > > It raises one more question from me. Hope you can/will answer that > > too. > > What happens if I have a ethernet controller which has a processor > > like intel's i960 (Hope this supports I/O mapped I/O) which supports > > I/O mapped I/O and if that controller is fixed on PPC system? > > Will PPC still map the device in to it's memory? > > > > << PCI space should indeed always be non-cacheable. But the system > > cannot do this for you. You must set up the BATs or PTEs (see manual > > re: sysBatDesc and sysPhysMemDesc) that cover the address range used > > by the PCI bus to have the G (guarded) and I (cache-inhibited) flags. > > Once you have done that, every card will be non-cacheable no matter > > where in PCI space the system decides to auto config it. > > >>Oh. Thanks for this information. But I haven't found sysBatDesc. But > > I was able to find sysPhysMemDesc. Here I'm giving the snap shot of > > it. > > > > (void *) PCI_MEMORY_START, > > (void *) PCI_MEMORY_START, > > PCI_MEMORY_MAP_END - PCI_MEMORY_START + 1, > > VM_STATE_MASK_VALID | VM_STATE_MASK_WRITABLE | > > VM_STATE_MASK_CACHEABLE | > > VM_STATE_MASK_GUARDED, > > VM_STATE_VALID | VM_STATE_WRITABLE | > > VM_STATE_CACHEABLE_NOT | > > VM_STATE_GUARDED > > > > Can you write me on how to interpret this? Because there are two > > entries are there two records are there. In this example, there are > > two PCI_MEMORY_START are there and correspondingly two different sets > > of flags are associated with them, one with Guarded and and non > > cacheable and the other one with cacheable and unguarded. I don't know > > why there are two different entries are there. Can you let me know > > this? > > > > <<> > If we directly access the device registers (sysInbyte() > > sysOutByte()), do we need to invalidate the cache as we are accessing > > HW registers? > > < > > > >>Oops... This is confusing again. > > From the file /* sysLib.c - IBM 405GP Walnut system-dependent routines > > */ > > I could able to find sysInByte() and sysOutByte() functions and > > similar PCI access routines to read/write one byte/Word/Long. > > But within the definitions sysInByte() and sysOutByte() everything > > treated as memory mapped only.I could able to see simple assignment > > statements and EIEIO_SYNC macros only. > > I guess these are only wrapper functions which simulates I/O mapped > > IO. Please do clarify me if I'm wong. > > > > << However I am very surprised to hear that you have an ethernet card > > that has no DMA. > > >>I'm sorry for the confusion. Our Card has its own DMA controller and > > its own Buffer/memory. > > > > < > the commands to the device registers that would cause it to access the > > system RAM. > > >>I've made a note of it. > > > > I have one more question. This is nothing to do with VxWorks or PPC. > > Just having a curiosity to know whether the e-mail ID you have given > > () is a valid one? If I send a mail, will > > it reach you? > > > > Thank you so much for your reply. > > > > I would also like to thank Michael Lawnick (Lawnick@softec.de) for > > his reply mail. > > > > < > actively > > fetch it, you shouldn't have any cache troubles when disabling cache > > for PCI > > address space of your NIC. > > >>Oh. I. See. Yes. This is the scenario with our controller (It has > > its own memory and it buffers.) > > > > Regards > > Arun > > -- > > Arun Prasad > > Software Engineer > > Adaptec India Pvt. Limited > > (A wholly owned subsidiary of Adaptec Inc., USA) > > 6-3-1086, IV Floor, Vista Grand Towers, > > Rajbhavan Road, Somajiguda, > > Hydereabad - 500082 > > Phone: (O):91-40-6661555/56/57/58 > > (Ext):234 > > Email: arun_prasad@adaptecin.com > > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: pppInit() (IPCP address negotiation with PPP) Date: Tue, 22 Jan 2002 22:55:12 +0100 From: hlb-NRW Email-Dienst Message-ID: <3C4DDFC0.FDDB4CE@fh-gelsenkirchen.de> References: <3C47420C.C591228F@fh-gelsenkirchen.de> <3C48040E.D2871B13@nohope.l8s.co.uk> Thank you for the information. Consequence: we moved with our Linux (Embedded Linux). Here everythink works fine. Gerhard David Laight schrieb: > Gerhard Juen wrote: > > > > Hi all, > > > > I want to connect a target ( a standard 80486 PC ) over the serial > > interface and a modem to an internet provider. In the > > IPCP address negotiation phase, our provider expects us > > to send a "0.0.0.0" as recommendated local and remote IP adsress. > > > > We use the command PPPInit( .. ). Any IP-Adress in the > > corresponding parameter fields which is not "0.0.0.0" is passed > > correctly as it is to our provider. If we take "0.0.0.0" or a NULL pointer > > (as proposed in the vxworks documentation for sending "0.0.0.0") > > our program ignores our parameter setting and passes the > > IP-address of the Ethernet network card which is also installed in the PC. > > > > Does anyone know how to pass a "0.0.0.0" to the provider during > > address negotiation? > > Don't think you can without patching the WRS code. > > The whole thing is stuck deeper into the mire than you can possibly > image. If you pass 0.0.0.0 (or null) the hostname is grabbed and a name > to address lookup done. The function used will try to use DNS (if > configured) even when passed a numeric address. If a route to your DNS > server exists (and an NFS mount might leave one lurking) it will have to > timeout the DNS request. > > David - -- Prof. Dr. Gerhard Juen FH Gelsenkirchen, Abt. Bocholt Münsterstr. 265 D-46397 Bocholt email: mailto:gerhard.juen@fh-gelsenkirchen.de web: http://www-et.bocholt.fh-gelsenkirchen.de/~juen Fon: 02871-2155 830 Fax: 02871-2155 800 --------------------------- Newsgroups: comp.os.vxworks Subject: Re: problem downloading Tornado Prototyper Plus Date: 22 Jan 2002 15:17:16 -0800 From: infovaan@lycos.com (Info Vaan) Organization: http://groups.google.com/ Message-ID: <71109c73.0201221517.239caa6a@posting.google.com> References: <71109c73.0201221139.14756619@posting.google.com> Figured it out!! DO NOT use the http://www.wrs.com site, instead use the, http://www.windriver.com site, and everything should work. - ---- infovaan@lycos.com (Info Vaan) wrote in message news:<71109c73.0201221139.14756619@posting.google.com>... > hi, > > i'm new to vxworks, and would have liked to wet my feet on vxworks > using Tornado Prototyper plus, but on WindRiver website it seems all > links pointing to it are broken. does anybody have any idea as to > what's going on, and if they have any alternate links. have written to > support and tornadoprototyper email id's but so far no result (waiting > for last 5 days now). > > thanks, > iw. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: New WindSerf, TSR's and SPR's. Date: 22 Jan 2002 15:31:14 -0800 From: john_94501@yahoo.com (John) Organization: http://groups.google.com/ Message-ID: <488e459a.0201221531.41030518@posting.google.com> References: Hello, You need to go through the Java applet thing on the left side of the main page - it is labelled Knowledge Base. That gives you an interactive folder tree to open and lets you get at the SPRs. For the TSRs, there is a link in the problem solving panel that will let you get the status of one of your TSRs. Hope that helps, John... Bill Pringlemeir wrote in message news:... > How does one access SPRs under the new WindSerf? It is hard to > install shockwave plug-ins on lynx [in case the info is in the > tutorial] ... > > regards, > Bill Pringlemeir. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Detecting Memory Leaks Date: 22 Jan 2002 16:08:41 -0800 From: vdalal@hotmail.com (Vasu Dalal) Organization: http://groups.google.com/ Message-ID: <363d071b.0201221608.7fbdd08d@posting.google.com> References: <3c2c51e7$0$12221$4d4efb8e@news.be.uu.net> Hello Mukul : Look at the VxWorks Reference Manual (Libraries) for the memShow library. The HTML version may also be available in the Vxworks/Tornado install dir under :: tornado/docs/vxworks/ref/memShow.html VxWorks/Tornado FAQ is at :: http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html HTH, vasu ref: > Hi all, > I am doing some tests( which I code in C) on VxWorks AE 1.1 and in my > tests I need to get the available physical memory( the RAM space left, > because my target is PC-Pentium 200MHz). I need this because I load the > system with a lot of threads and synchronization objects, and I test to see > if there are any memory leaks. Does anyone know if there is an API function > which > - returns the available total free space in the system( the space left on > the RAM). > Thanks, > Mukul --------------------------- Newsgroups: comp.os.vxworks Subject: routing problem ??? Date: Tue, 22 Jan 2002 22:14:44 +0900 From: "it2xpert" Organization: ONSE Telecom Message-ID: Hi , vxworks seniors !!! I desinged the mpc850sr board for router... I implemented the scc2,scc3 for ethernet mode... I use the T2, vxworks 5.4... I want to implement the Layer 3 router..... 192.168.1.0 -----------------192.168.1.1 (scc3) (scc2) 210.118.156.240 -------------------210.118.156.0 I used the RIP v.1 Routing table is following.... - -> routeShow ROUTE NET TABLE destination gateway flags Refcnt Use Interface - ---------------------------------------------------------------------------- 0.0.0.0 210.118.156.254 3 0 21 cpm0 192.168.1.0 192.168.1.1 101 0 0 cpm1 210.118.156.0 210.118.156.240 101 0 0 cpm0 - ---------------------------------------------------------------------------- ROUTE HOST TABLE destination gateway flags Refcnt Use Interface - ---------------------------------------------------------------------------- 127.0.0.1 127.0.0.1 5 1 65 lo0 - ---------------------------------------------------------------------------- value = 77 = 0x4d = 'M' - -> I trid the ping test !!! My pc is connected the 192.168.1.0 network (192.168.1.3) ping 192.168.1.1 ----- OK, ping 210.118.156.240 ---- OK but i did't reach the 210.118.156.xxx point.... I tried ping to the target server console.... - -> ping "210.118.156.xxx" this is OK.... what's wrong ???? --------------------------- Newsgroups: comp.os.vxworks Subject: WorkQPanic???? Date: Wed, 23 Jan 2002 02:24:24 GMT From: "Nejhdeh Ghevondian" Organization: Integrated Spectronics Message-ID: Reply-To: "Nejhdeh Ghevondian" Dear Anyone! Latrely when I add the intEnable() to my routine I get workQPanic error. There is a lot said about the multiple causes of this error. The problem is that I dont have access to the ISR of this application, since its a VxWorks libray. My question is, HOW DO U CHECK THE STATUS OF THE KERNEL'S WORK QUEUE. Regards Nejhdeh Ghevondian --------------------------- Newsgroups: comp.os.vxworks Subject: SMC driver Date: 22 Jan 2002 18:30:33 -0800 From: ntjiang@yahoo.com (Jason Jiang) Organization: http://groups.google.com/ Message-ID: <12116a88.0201221830.591bea57@posting.google.com> I have a question when reading SMC driver in my system. We use ppc860Sio.c (included by sysSerial.c) as the serial driver. It's working fine. But I didn't see any place to install the two callback functions. They are supposed to be installed by ppc860CallbackInstall() in ppc860Sio.c. But I didn't find any place to call this function. Pls help. Thanks. Jason --------------------------- Newsgroups: comp.os.vxworks Subject: Error while downloading project on target??? Date: 22 Jan 2002 20:12:14 -0800 From: kumar_amit@delhi.tcs.co.in (Amit) Organization: http://groups.google.com/ Message-ID: <7421c9fd.0201222012.335b07bb@posting.google.com> I have a target with memory 128 MB. My final image size is coming around 76 MB. So while downloading the image on target, system hangs up. I want to know whether there are certain size restrictions on image size which you can download . Thanks Ajay --------------------------- Newsgroups: comp.os.vxworks Subject: Re: spurious interrupt + cache Date: 22 Jan 2002 21:30:03 -0800 From: ruchama@septier.com (Ruchama Goldshmidt) Organization: http://groups.google.com/ Message-ID: <5df340e0.0201222130.1b5121b1@posting.google.com> References: <5df340e0.0201220441.6b3b3999@posting.google.com> Hello Michael! Thanks for your answer, sorry ,but what do you meen a "read-back of the register you write your ACK to" in the interrupt handeler or in the shared memory driver? TIA, best regards, Ruchama. "Michael Lawnick" wrote in message news:... > Hi Ruchama > > sounds for the 'your CPU is faster than your bus' problem: > > When you acknowledge the interrupt on the slave hardware, your actual > write-access seems to be done by some 'middle-ware', e.g. PCI-bridge, > VME-bridge,... > Your CPU has already left ISR and reenabled its interrupts, before the real > access on your slave is done. This explains, why uncached acknowledges > succeed: your CPU is slower now. > Do a read-back of the register you write your ACK to and all should be fine. > > Another problem could be, that your device is mapped as cacheable, but that > I beleave would lead to complete malfunction. > > -- > Mit freundlichen Grüßen, > Michael Lawnick > ============================================== > SOFTEC GmbH Tel +49-731-96600-0 > Promenade 17 Fax +49-731-96600-23 > D-89073 Ulm Michael Lawnick > Germany lawnick@softec.de --------------------------- Newsgroups: comp.os.vxworks Subject: Re: intConnect() vs pciIntConnect() Date: Tue, 22 Jan 2002 22:15:10 -0800 From: "drdiags" Organization: laserlink.net Message-ID: References: Arun, Depends on your BSP, but if it includes the target/src/drv/pciLib.c file you should be able to see that pciIntConnect creates a link list of ISRs to call. This, I guess was to support Interrupt sharing for PCI. intConnect() would only allow a single ISR per Interrupt line. You can then use pciIntDisconnect() to pop off your unique ISR from the link list. You should have this file. Some BSPs, like the HSI ppmc74xx have their own local version of pciIntConnect and handle this a little differently. From what I see, after inserting your ISR, there is a global pciInt ISR that will call all ISRs associated with the interrupt in question. "Arun Prasad" wrote in message news:e38d912c.0201220455.30a696b1@posting.google.com... > Hi All, > > I couldn't able to find the exact differences between intconnect() > and pciIntConnect(). > > Can anyone list out the exact differences of intConnect and > pciIntConnect and their usage? > > If intConnect is able to connect a PCI based ISR then why do we need > a separate call (pciIntConnect)? > > Are there there any specific necessity for pciIntConnect where we > can't use intConnect? > > I'm writing an ISR for a PCI ethernet Card on a IBM PPC405 board. > Which call I need to use intconnect or pciConnect? Which is better > and whey? > Please do clarify these. > Thanks in advance. > > Have a nice time. > Regards > Arun --------------------------- Newsgroups: comp.os.vxworks Subject: Re: intConnect() vs pciIntConnect() Date: Tue, 22 Jan 2002 22:25:49 -0800 From: "drdiags" Organization: laserlink.net Message-ID: References: Darn!! The true path and name of this file is %WIND_BASE%/target/src/drv/pci/pciIntLib.c <....SNIP....> > > "Arun Prasad" wrote in message > news:e38d912c.0201220455.30a696b1@posting.google.com... > > Hi All, > > > > I couldn't able to find the exact differences between intconnect() > > and pciIntConnect(). --------------------------- Newsgroups: comp.os.vxworks Subject: Re: spurious interrupt + cache Date: 22 Jan 2002 23:38:25 -0800 From: ruchama@septier.com (Ruchama Goldshmidt) Organization: http://groups.google.com/ Message-ID: <5df340e0.0201222338.62b47bac@posting.google.com> References: <5df340e0.0201220441.6b3b3999@posting.google.com> Hello!! In addition to the last letter I want to mention that the spurious interrupt I get is even before the handler routine when I printout the interrupts I get I see that I get vector number 23 (IRQ5) and right after that the vector number 0 and only after that there are the printouts from the interrupt handlers. thanks!!! best regards Ruchama "Michael Lawnick" wrote in message news:... > Hi Ruchama > > sounds for the 'your CPU is faster than your bus' problem: > > When you acknowledge the interrupt on the slave hardware, your actual > write-access seems to be done by some 'middle-ware', e.g. PCI-bridge, > VME-bridge,... > Your CPU has already left ISR and reenabled its interrupts, before the real > access on your slave is done. This explains, why uncached acknowledges > succeed: your CPU is slower now. > Do a read-back of the register you write your ACK to and all should be fine. > > Another problem could be, that your device is mapped as cacheable, but that > I beleave would lead to complete malfunction. > > -- > Mit freundlichen Grüßen, > Michael Lawnick > ============================================== > SOFTEC GmbH Tel +49-731-96600-0 > Promenade 17 Fax +49-731-96600-23 > D-89073 Ulm Michael Lawnick > Germany lawnick@softec.de > ============================================== > > "Ruchama Goldshmidt" schrieb im Newsbeitrag > news:5df340e0.0201220441.6b3b3999@posting.google.com... > > Hello vxWorkers!!! > > > > I develop a board based on the ppc8260 processor and using also the > > shared memory driver using as a master an MPC750 board. > > > > When enable the Data and Instruction cache I see that when I get an > > interrupt in the shared memory (IRQ5 interrupt ) I also get a spurious > > interrupt at interrupt number 0 and get many "UNINITIALIZED INTERRUPT" > > logs, > > which is very strange since interrupt number 0 is the error interrupt > > so why do I get it? > > > > when the caches are disabled I get only the IRQ5 and no other ,as > > excepted. > > > > does anyone have an advise?? > > > > TIA, > > > > Ruchama Goldshmidt > > Septier Communication Ltd. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Kernel WorkQ Overflow Date: 23 Jan 2002 08:26:13 GMT From: Johan Borkhuis Organization: Agere Systems Message-ID: References: "nitin kumar" wrote: > Is there a scenario where repeated msgQSend from a routine cause > Kernel WorkQ over flow. msgQSend() is NOT being called from an > ISR. My ISR just does a semGive() and wakes up another task that > does the send to the Q. > The Message Q is 256 entries large and I envision it will be full > quite > often so that msgQSend() shall be failing frequently. ( But that > should not be a problem as the send is done with NO_TIMEOUT ). This message about the kernel work-queue overflow has nothing to do with message queue's. The problem is that your interrupt is called so often that the kernel does not has time left. Most often it is an indication that your ISR does not clear the interrupt. Something that you could check is to make an empty ISR with only the function to clear the interrupt, and see if you still have the error. After that you can add your functionality. Groeten, Johan - -- o o o o o o o . . . _____________________________ o _____ || Johan Borkhuis | .][__n_n_|DD[ ====_____ | borkhuis@agere.com | >(________|__|_[_________]_|__________________________| _/oo OOOOO oo` ooo ooo 'o!o!o o!o!o` === VxWorks FAQ: http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html === --------------------------- Newsgroups: comp.os.vxworks Subject: Re: WorkQPanic???? Date: 23 Jan 2002 08:38:44 GMT From: Johan Borkhuis Organization: Agere Systems Message-ID: References: "Nejhdeh Ghevondian" wrote: > Latrely when I add the intEnable() to my routine I get workQPanic > error. There is a lot said about the multiple causes of this error. > The problem is that > I dont have access to the ISR of this application, since its a VxWorks > libray. > > My question is, HOW DO U CHECK THE STATUS OF THE KERNEL'S WORK QUEUE. As far as I know you cannot. As it is an overflow of the interrupt handling you don't have processing time to do this anyway. If you don't have access to the ISR (is it a driver that is only supplied in object code?) you should ask your FAE for assistance, eventually sending you the source code of the driver. Or you could modify your HW, as this driver is probably designed for some specific HW design. Groeten, Johan - -- o o o o o o o . . . _____________________________ o _____ || Johan Borkhuis | .][__n_n_|DD[ ====_____ | borkhuis@agere.com | >(________|__|_[_________]_|__________________________| _/oo OOOOO oo` ooo ooo 'o!o!o o!o!o` === VxWorks FAQ: http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html === --------------------------- Newsgroups: comp.os.vxworks Subject: Signal handler executions Date: 23 Jan 2002 01:01:09 -0800 From: rashanmu@npd.hcltech.com (Ramesh) Organization: http://groups.google.com/ Message-ID: <15a87607.0201230101.670b6e87@posting.google.com> Hi, I am reading the VxWorks Programmers Guide. It is given that "when a task receives a signal and if that task registers a handler for that signal the signal handler is executed on the task context". Also it is mentioned even the task is blocked the handler will get executed like that. If so what is the difference between a ISR and a signal handler. Suppose if a task of pri 100 receives a signal say SIGALARM and while receiving it is in the pend queue ( assume ) coz of some task of pri 99 pre-empts the task 100 and task 99 is using the CPU resource. Now since the task receives teh signal does the handler for that signal will get executed or the task 99 will execute. If handler gets executed means.. then under which task context will it gets execute. Kindly clear me. Thanks, Ramesh --------------------------- Newsgroups: comp.os.vxworks Subject: listing link modules Date: 23 Jan 2002 01:27:56 -0800 From: tony.pedley@powerconv.alstom.com (TonyP) Organization: http://groups.google.com/ Message-ID: Hi, Our system is designed such that we have a common base system onto which 3rd parties can develop and add drivers to add functionality. We want to keep our base system as lean as possible, however sometimes we get requests for new functionality which means adding files via the bsp. This then links in objects from the gnu library file What we require is for the driver writer to check whether his compiled code has any undefined sysmbols which will not be satisfied by the base system when the driver is added. The driver makefile can then extract those files from the library achive and link them in. I can access the symbols in both the driver and library achive. However what I have'nt found a way to do is generate a list of files which contain all the undefined symbols. I'm willing to write a perl script to do this, but I have a feeling that it may be difficult. Does anyone know of any tool or do anything similar to acheive this? Tony --------------------------- Newsgroups: comp.os.vxworks Subject: Re: intConnect() vs pciIntConnect() Date: Wed, 23 Jan 2002 10:27:06 +0100 From: "Werner Schiendl" Message-ID: <3c4e81ec$1@brateggebdc5.br-automation.co.at> References: <3C4DA04D.4E96@despam.autobahn.mb.ca> Hi, in my opinion, you should always: - - use pciIntConnect - - make sure that your device really raised the interrupt Then your driver is ready for sharing interrups if the need arises (one never knows). BTW: pciIntConnect is included in source IIRC and is based on intConnect() internally. hth Werner "Arun Prasad" wrote in message news:e38d912c.0201221027.29d019ff@posting.google.com... > Dear Taylor, > > Thank you for your reply. > > << intConnect modifies the interrupt vector table. > pciIntConnect adds your ISR to a collection. > >> I got the ponit now. > To use intConnect, I should have unique (unshared) interrupt for my > device as it modifies the IVT. Otherwise I need to make it unique by > setting a unique IRQ to a PCI Slot and putting our PCI based Card > there and it simplifies our task in ISR as we don't need to check > which device raised the interrupt as the IRQ has been already made as > unique. > Hope I'm right :-) > If not please do let me know my misunderstanding. > > Regards > Arun > > "H. E. Taylor" wrote in message news:<3C4DA04D.4E96@despam.autobahn.mb.ca>... > > In article , > > Arun Prasad wrote: > > > > > > I couldn't able to find the exact differences between intconnect() > > > and pciIntConnect(). > > > > > > Can anyone list out the exact differences of intConnect and > > > pciIntConnect and their usage? > > > > > > > intConnect modifies the interrupt vector table. > > pciIntConnect adds your ISR to a collection. > > > > The essential fact to note is that there are only _4_ PCI interrupt > > vectors and there can be many PCI devices. So you need a way to share. > > > > >[...] > > > I'm writing an ISR for a PCI ethernet Card on a IBM PPC405 board. > > > Which call I need to use intconnect or pciConnect? Which is better > > > and whey? > > > > > > > pciIntConnect for the reasons above. > > > > > > -het --------------------------- Newsgroups: comp.os.vxworks Subject: Re: I2C driver Date: Wed, 23 Jan 2002 10:34:34 +0100 From: "Werner Schiendl" Message-ID: <3c4e83ac$1@brateggebdc5.br-automation.co.at> References: <3C4D9347.43AA0E67@attbi.com> <76i38.17309$8e.1027293@news> Hi, No, VxWorks IO system does not prevent concurrent ioctl's on the same driver or device. You must care for that in your driver. Moreover, it does not solve the issue with the timing requirements. If the driver is doing it's work synchronously in task context, interrupting it would disturbe the protocol on the wire. The latter problem could probably being worked around using proper taskLock()/taskUnlock() pairs. hth Werner "Jim" wrote in message news:76i38.17309$8e.1027293@news... > This is a stupid question ... but does wrapping calls to i2c with ioctl > provide concurency solution? > > I have not RTM ... just thought i'd ask ... > > Jim > > "cfk@pacbell.net" wrote in message > news:bHg38.35006$Fn6.1618022548@newssvr21.news.prodigy.com... > > Dear Tom: > > I2C is a very simple two wire protocol with just data (SDA) and clock > > (SCL). It is certainly possible to interrupt an I2C driver in the midst of > > transmitting a byte. AFAIR, this needs to be done when SCL is in the low > > state, but you can read the I2C specification available from Philips under > > the paragraph "Pulse Stretching". The key thing to making an I2C driver > > multitasking safe is to ensure that the task switch occurs when the clock > is > > in the appropriate state. I suspect the author of this driver merely > didn't > > put the appropriate logic in for that. It is fairly simple to implement, > it > > just takes a day or so of head scratching and studying other > implementations > > of "bit banging" I2C software. > > > > Charles > > > > "Tom Flynn" wrote in message > > news:3C4D9347.43AA0E67@attbi.com... > > > > > > The I2C driver in the BSP that I'm using is only written for use during > > > boot. It is not multitasking safe (according to the author's comments). > > > We need to access the I2C during normal run time on the board that we > > > are using. > > > > > > I have two questions. First, what would need to be done to make it > > > usable during run time? > > > > > > Second, is there driver code somewhere that is multitasking safe that I > > > can look at? I have access to other BSP code, so if someone knows of a > > > BSP that includes an I2C driver that is multitasking safe, I would be > > > interested in knowing which BSP that is. > > > > > > Thanks in advance. > > > > > > -Tom Flynn > > > > > > > > > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Retrives a buffer Date: Wed, 23 Jan 2002 10:38:50 +0100 From: "Werner Schiendl" Message-ID: <3c4e84ac$1@brateggebdc5.br-automation.co.at> References: Hi, what do you mean with 'buffer'? a bunch of memory? should poll mean pool? you can use malloc() / free() to get memory from the system partition (= system memory pool). You can use memPartCreate() memPartAlloc() and memPartFree() to manage your own memory pool. hth Werner "Jason" wrote in message news:e555a185.0201221054.826a6d8@posting.google.com... > Hello, > > Can anybody tell me how to retrive a buffer from buffer poll and how > to free the buffer on vxWorks. > i.e. some function like this: > buffGet (int BuffSize), > buffFree () > > Cheers --------------------------- Newsgroups: comp.os.vxworks Subject: Re: How to allocate memory for PCI driver? Date: 23 Jan 2002 02:23:56 -0800 From: igtorque@eliop.es (=?ISO-8859-1?Q?Ignacio_Gonz=E1lez?=) Organization: http://groups.google.com/ Message-ID: <9ec53d11.0201230223.9654db2@posting.google.com> References: <9ec53d11.0201180848.30bc26f0@posting.google.com> Thank you, Iosif and Michael. I followed the guidelines of Iosif and they seem to work. And, Michael, you were right about USER_RESERVED_MEMORY. My wrong assumptions about its functionality didn't crash our previous code because some "smart" guru had patched bootConfig, etc. for his use, and the assumption and the patch had survived him (I killed him yesterday:-) --------------------------- Newsgroups: comp.os.vxworks Subject: STL compilation problem Date: 23 Jan 2002 02:29:48 -0800 From: ilkras@yahoo.com (Ilya Krasavin) Organization: http://groups.google.com/ Message-ID: <9ca84f0f.0201230229.4aab90d1@posting.google.com> Dear gurus. I am porting some prog. from NT to vxworks. There are some STL used. When I compile the file I get the error : no match for `__rb_tree_iterator,pair &,pair *> & = void' candidates are: __rb_tree_iterator,pair &,pair *>::operator =(const __rb_tree_iterator ,pair &,pair *> &) I run Tornado 2.0.2 with toolchain for PPC603. the compiler is g++(gcc) and reports version as "cygnus-2.7.2-960126 egcs-971225 tornado 2.0". What is this error? How can I cope with it? Are there any workarounds? Any ideas? Hope there is someone who succeded in the same job. thanks Ilya --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Symbol table errors compiling an image when using STL Date: 23 Jan 2002 02:41:14 -0800 From: sasacoh@europe.com (Sasa) Organization: http://groups.google.com/ Message-ID: <769ac895.0201230241.7e6993d9@posting.google.com> References: <3c29bbec$1@news.barak.net.il> <3c3833dd$1@news.barak.net.il> <488e459a.0201071346.16dad6f2@posting.google.com> <769ac895.0201171412.346dc387@posting.google.com> <488e459a.0201172059.1eeb7d1d@posting.google.com> john_94501@yahoo.com (John) wrote in message news:<488e459a.0201172059.1eeb7d1d@posting.google.com>... > sasacoh@europe.com (Sasa) wrote in message news:<769ac895.0201171412.346dc387@posting.google.com>... > > [history deleted to save space] > > > Hello! > > We're having exactly the same problems (on ARM9 board) with STL > > compiled together with symbol table (INCLUDE_SYM_TBL_INIT) and we > > don't use namespaces. If INCLUDE_SYM_TBL_INIT macro is excluded it > > works well but without possibility to execute commands from shell. But > > we need this! Any suggestions? > > The problem looks to be related to the generation of '.' characters in > the mangled names. I believe that this can be turned off for GNU, but > I cannot remember how. As a workaround, you can run sed over the > symTbl.c file to remove all the lines that contain identifiers with a > '.' in them. This can be added to the make rules (under target/h/make) > so that you don't need to do it manually each time. > > This will just make them unavailable to the target symbol table, which > will be OK provided you don't need to download anything that binds to > them (or call them directly from the shell - which is unlikely). If > you do need them, then you need to find a way of persuading the > compiler not to generate '.' in symbols. > > HTH, > > John... Yes, the problem is related to the generation of '.' characters in the mangled names. Unfortunately, I also don't know how to turn this off for GNU, so I tried with sed. But now I have another problem. I added sed command (sed "/\./d" SymTbl.c) to the make rules to remove all the lines with '.'. But Tornado build window convert all '/' to '\', so the sed command looked like this (sed "\\.\d" ...) NOT OK! Please help... Lp, Sasa --------------------------- Newsgroups: comp.os.vxworks Subject: Re: SMC driver Date: 23 Jan 2002 02:41:41 -0800 From: raj_rashid@hotmail.com (Rashid) Organization: http://groups.google.com/ Message-ID: <2a6118da.0201230241.1604435d@posting.google.com> References: <12116a88.0201221830.591bea57@posting.google.com> Hi Jason, In sysLib.c we call.......sysSerialHwInit() function....... ....Now this function is implemented in sysSerial.c of BSP in which structures for channels like ppc860Chan is defined. In sysSerialHwInit() some parameters of the structure ppc860Chan are initialized and a reference of ppc860Chan is passed to the function ppc860SioDevInit() in which the ppc860Chan structure is initialized with ppc860SioDrvFuncs of which ppc860CallbackInstall() is a part. Now the sioLib library provides functions like sioCallbackInstall() to install ISR callbacks to get and put chars. So when u make a call to sioCallbackInstall the corresponding function(ppc860SioCallbackInstall) registered with your serial channel(ppc860Chan) is called. You can find usage of sioCallbackInstall() in file WIND_BASE\target\src\drv\wdb\wdbSlipPktDrv.c regards, Rashid ntjiang@yahoo.com (Jason Jiang) wrote in message news:<12116a88.0201221830.591bea57@posting.google.com>... > I have a question when reading SMC driver in my system. We use > ppc860Sio.c (included by sysSerial.c) as the serial driver. It's > working fine. But I didn't see any place to install the two callback > functions. They are supposed to be installed by > ppc860CallbackInstall() in ppc860Sio.c. But I didn't find any place to > call this function. Pls help. Thanks. > Jason --------------------------- Newsgroups: comp.os.vxworks Subject: Wind Technical Notes Date: 23 Jan 2002 03:01:51 -0800 From: igtorque@eliop.es (=?ISO-8859-1?Q?Ignacio_Gonz=E1lez?=) Organization: http://groups.google.com/ Message-ID: <9ec53d11.0201230301.3678a448@posting.google.com> Too late I noticed there are some references in VxWorks' manuals to Wind Technical Notes that I don't have (e.g. WTN 41) Alas, my maintainance contract has expired, and I'd like to consult a couple of those WTN (41 at least). I know this can be considered "politically" incorrect but, can anybody tell me where can I find them (appart from WindSurf)? I've alreay found WTN 49. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: spurious interrupt + cache Date: Wed, 23 Jan 2002 11:06:45 +0000 From: Graham Baxter Organization: Graham Baxter (Software) Limited Message-ID: <3C4E9945.572AB6D6@NOSPAM.bcs.org.uk> References: <5df340e0.0201220441.6b3b3999@posting.google.com> Ruchama, This solution obtained from Wind River worked for me with exactly the same problem. I hope it helps. Regards, Graham Baxter (VxWorks and pSOS BSP'S) Freelance Software Engineer gbaxter@NOSPAM.bcs.org.uk - ------------------------------- The solution to this problem was to modify the m8260IntrCtl.c source file in the following way: In the m8260IntrCtl.c file, in the function m8260IntrDeMux(), right after it enables only higher level priorities in the interrupt mask registers it calls the interrupt unlock which causes an unknown interrupt to be generated because the values in the registers haven't settled yet. By adding a call to CACHE_PIPE_FLUSH() the error message disappears and everything seems to work OK. Here's a snippet of the code: ... /* enable only the higher priority interrupts */ * M8260_SIMR_L(immrVal)&= enableHighPriInts[IVEC_TO_INUM((VOIDFUNCPTR *)intVec)].simr_l; * M8260_SIMR_H(immrVal)&= enableHighPriInts[IVEC_TO_INUM((VOIDFUNCPTR *)intVec)].simr_h; CACHE_PIPE_FLUSH (); /* unlock the interrupt */ intUnlock (_PPC_MSR_EE); /* call the Interrupt Handler */ --------------------------- Newsgroups: comp.os.vxworks Subject: Re: tffs, tffsBootPutImage and booting from fallow area Date: Wed, 23 Jan 2002 11:34:16 +0000 From: Owain Phillips Organization: Siemens Inc. Message-ID: <3C4E9FB8.A98B303D@siemenscomms.co.uk> References: <3C4C4F80.3F96AD2C@siemenscomms.co.uk> <3C4D7D3F.FCEBE820@siemenscomms.co.uk> This is a multi-part message in MIME format. - --------------AEDE032764FD543193C69C7C Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Hi Andray, > > [AndrayK 2]. Did you set breakpoint in MTD's 'write' routine, to see > if MTD actually receives 'write' requests with offsets higher then 0x100000 ? > > regards, Andray No; as I didn't 'have' to use the tffsPutBootImage routine; I have my own routines that don't rely on tffs or cfisc driver. I have given up on the tffs for putting my boot images in ROM and for booting from the fallow region. I have installed a memDrv and can boot from the created memory device......must now implement the next Vxworks workaround; the fact the network device doesn't get initialised if you boot from certain named (not tffs) device..... Oh well after working with WRS and tornado for the last year I don't expect anything to run by the book. All the best, Owain - --------------AEDE032764FD543193C69C7C Content-Type: text/x-vcard; charset=us-ascii; name="owain.phillips.vcf" Content-Transfer-Encoding: 7bit Content-Description: Card for Owain Phillips Content-Disposition: attachment; filename="owain.phillips.vcf" begin:vcard n:Phillips;Owain tel;fax:+44 115 943 4969 tel;work:+44 115 943 2167 x-mozilla-html:TRUE url:http://www.siemenscomms.co.uk org:Siemens Comunications Ltd version:2.1 email;internet:owain.phillips@siemenscomms.co.uk title:Software Engineer adr;quoted-printable:;;Technology Drive=0D=0ABeeston;Nottingham;;NG9 1LA;United Kingdom fn:Owain Phillips end:vcard - --------------AEDE032764FD543193C69C7C-- --------------------------- Newsgroups: comp.os.vxworks Subject: !!!!!!!!!!!!!!!!!!!! SOFTWARE FOR SALE !!!!!!!!!!!!!!!!!!!!                                                                                                                                                            Date: Wednesday, 23 Jan 2002 03:33:35 -0600 From: programz_4u@writeme.com Organization: Message-ID: <23010203.3335@writeme.com> Reply-To: david4programz@yahoo.com PC and Mac software available. $10US or €10 per CD. Worldwide shipping. E-mail programz_d@consultant.com OR programz_4u@writeme.com for a full list.6_V/3ju=K9 --------------------------- End of New-News digest ********************** From vxwexplo-errs@csg.lbl.gov Wed Jan 23 07:44:52 2002 From: "Garner, David (N-Progeny)" Date: Wed Jan 23 07:44:55 PST 2002 Subject: Cumulative Patch 4 Problems I have installed CP4 on Tornado 2.00. The target is a PPC based MV2308 and contains a PMC hard drive with a device name of /pcfsA. After installing CP4 I noticed that Ethernet networking is extremely slow and when I nfsMount the hard drive from another VxWorks SBC the files can not be accessed. If I go back to CP3 everything works fine. Does anyone know if there are major issues with CP4? I have searched WindSurf and can find nothing relevant. Matter of fact I can't find much of anything on their new web-site. The patches area is now empty. Thanks for any help, Dave Garner daveg@progeny.net From vxwexplo-errs@csg.lbl.gov Wed Jan 23 16:30:19 2002 From: "CICON,JAMES (HP-Vancouver,ex1)" Date: Wed Jan 23 16:30:21 PST 2002 Subject: porting from diab compiler to gnu compiler The following code snippet compiles using DIAB tools, but it creates errors using the GNU tools. Could anyone demonstrate to me how to make this compile under GNU? UINT32 kram1; UINT32 setRambar(UINT32 rambar) { kram1 = rambar; asm(" move.l _kram1, d0"); asm(" movec.l d0, rambar1"); return kram1; } This is for the CF5407 BSP. I am using Tornado 2 tools. James Cicon Hewlett Packard Digital Printing Technologies keywords: VxWorks, Tornado From vxwexplo-errs@csg.lbl.gov Thu Jan 24 04:03:19 2002 From: Vxworks Exploder Date: Thu Jan 24 04:03:22 PST 2002 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Thu Jan 24 04:03:14 PST 2002 Subject: Re: routing problem ??? Subject: ipAttach returns error Subject: Help!! - Posix signals implementation Subject: Re: Acurate, high-precision and stable timing solutions, Need Help... Subject: Re: SMC driver Subject: Re: Symbol table errors compiling an image when using STL Subject: Re: Help!! - Posix signals implementation Subject: Re: Symbol table errors compiling an image when using STL Subject: Re: intConnect() vs pciIntConnect() Subject: TCP/IP Performance under VxWorks Subject: FTP connection ok but ... Subject: Creating Build Rules in Project Workspace Subject: Re: Error while downloading project on target??? Subject: Re: SCSI Performance in vxWorks Subject: Re: Booting from ATA Compact Flash Subject: Re: Vxworks beginner Subject: Re: SCSI Performance in vxWorks Subject: Re: routing problem ??? Subject: Memory lost during open,read, close with NFS and netDrv Subject: Can I use i/o system "read()" to read a i/o port address? Subject: Re: SMC driver Subject: linking problem Subject: Re: Memory lost during open,read, close with NFS and netDrv Subject: Re: Can I use i/o system "read()" to read a i/o port address? Subject: Re: FTP connection ok but ... Subject: Re: TCP/IP Performance under VxWorks Subject: Re: Help!! - Posix signals implementation Subject: QMC HDLC driver for 860! Subject: WindNet PPP ----establish connection fail ? Subject: Re: FTP connection ok but ... Subject: NFS-problem Subject: Re: STL compilation problem Subject: infinite while loop at usrAppInit() Subject: Re: tffs, tffsBootPutImage and booting from fallow area Subject: Re: spurious interrupt + cache Subject: Re: infinite while loop at usrAppInit() ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: routing problem ??? Date: Wed, 23 Jan 2002 15:32:44 GMT From: "Jim" Organization: Magma Communications Ltd. Message-ID: References: can you use subnet 1.X? Jim "it2xpert" wrote in message news:a2jojc$6eu$1@hiline.shinbiro.com... > Hi , vxworks seniors !!! > I desinged the mpc850sr board for router... > I implemented the scc2,scc3 for ethernet mode... > > I use the T2, vxworks 5.4... > I want to implement the Layer 3 router..... > > 192.168.1.0 -----------------192.168.1.1 (scc3) > > (scc2) > 210.118.156.240 -------------------210.118.156.0 > > I used the RIP v.1 > Routing table is following.... > > -> routeShow > > ROUTE NET TABLE > destination gateway flags Refcnt Use Interface > -------------------------------------------------------------------------- - -- > 0.0.0.0 210.118.156.254 3 0 21 cpm0 > 192.168.1.0 192.168.1.1 101 0 0 cpm1 > 210.118.156.0 210.118.156.240 101 0 0 cpm0 > -------------------------------------------------------------------------- - -- > > ROUTE HOST TABLE > destination gateway flags Refcnt Use Interface > -------------------------------------------------------------------------- - -- > 127.0.0.1 127.0.0.1 5 1 65 lo0 > -------------------------------------------------------------------------- - -- > value = 77 = 0x4d = 'M' > -> > > I trid the ping test !!! > My pc is connected the 192.168.1.0 network (192.168.1.3) > ping 192.168.1.1 ----- OK, > ping 210.118.156.240 ---- OK > > but i did't reach the 210.118.156.xxx point.... > > I tried ping to the target server console.... > -> ping "210.118.156.xxx" > this is OK.... > > what's wrong ???? > > > > --------------------------- Newsgroups: comp.os.vxworks Subject: ipAttach returns error Date: Wed, 23 Jan 2002 10:47:06 -0500 From: "Harjinder Chayra" Organization: Storm Internet Services Message-ID: Sender: terayon@gw-terayon.storm.ca I am trying to get an END driver to be recognised by the TCP/IP stack. Although the END driver appears to initialised correctly when I do a muxShow, muxShow Device: Bridge Unit: 0 Description: Imedia Bridge END value = 0 = 0x0 when I excute the following code, pEnd = endFindByName( DeviceName, Unit ); if ( pEnd == NULL ) { printf("endFindByName could not find %s %d\n",DeviceName, Unit ); return (ERROR); } /* End if */ if ( muxIoctl(pEnd, EIOCGMIB2, (caddr_t) &endM2Tbl)== ERROR) { printf("muxIoctl returned ERROR\n"); return( ERROR ); } if ( ipAttach( Unit, DeviceName ) != OK ) { printf("ipAttach returned ERROR\n"); return( ERROR ); } the ipAttach call returns ERROR. I have set IP_MAX_UNITS 2 I do a command line build in T2/target/config/simpc Appreciate any insight. Thanks Harjinder --------------------------- Newsgroups: comp.os.vxworks Subject: Help!! - Posix signals implementation Date: Wed, 23 Jan 2002 16:53:40 +0100 From: Jacques Skubich Organization: =?iso-8859-1?Q?D=E9partement?= Informatique - INSA de Lyon Message-ID: <3C4EDC84.96E3D858@insa-lyon.fr> I am using the full VxSim simulator and I get some trouble in implementing SIGNALS. The following program is producing this output : ********************************* VxWorks Copyright 1984-1998 Wind River Systems, Inc. CPU: VxSim for Windows VxWorks: 5.4 BSP version: 1.1/1 Creation date: Apr 22 1999 WDB: Ready. Wait for Signal before kill after kill ************************************* What happens with the handler and logMsg??? ************************************** #include #include #include #include #include void SigHand(int no) { logMsg("in the handler\n",0,0,0,0,0,0); } /*----------------------------------*/ int P() { struct sigaction action; action.sa_handler= SigHand; sigemptyset(&action.sa_mask); sigaddset(&action.sa_mask,SIGUSR2); action.sa_flags= 0; sigaction(SIGUSR2,&action,NULL); for(;;) { printf("Wait for Signal \n"); sigsuspend(&action.sa_mask); printf("Signal arrived \n"); } return(0); } /* */ int S(int tId) { taskDelay(600); printf("before kill\n"); kill(tId, SIGUSR2); printf("after kill \n"); return(0); } void M() { int PId, SId; PId=taskSpawn("P",100,0,10000,P,0,0,0,0,0,0,0,0,0,0); SId=taskSpawn("S",100,0,10000,S,PId,0,0,0,0,0,0,0,0,0); exit(0); } /*==========================================================*/ /* */ - -- |// (o o) ___________________________oOO__(_)__OOo________________________________ IMPORTANT ** Nouveau N° de fax ** IMPORTANT ** New fax number Jacques J. Skubich Departement d'Informatique - Blaise Pascal INSA de Lyon F-69621 VILLEURBANNE CEDEX Jacques.Skubich@insa-lyon.fr Tel/Phone : +33 (0)4 72 43 84 86 Facsimile : +33 (0)4 37 47 90 82 Mobile : +33 (0)6 99 61 25 18 ___________________________________________________________________________ "Si nous avons chacun un objet et que nous les échangeons, nous avons chacun un objet. Si nous avons chacun une idée et que nous les échangeons, nous avons chacun deux idées." Proverbe chinois. ___________________________________________________________________________ --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Acurate, high-precision and stable timing solutions, Need Help... Date: Wed, 23 Jan 2002 16:50:57 GMT From: Craig Johnston Organization: The Boeing Company Message-ID: <3C4EE9F1.E52D7842@boeing.com> References: <6087305d.0201221146.7cacd1fc@posting.google.com> Sender: nntp@news.boeing.com (Boeing NNTP News Access) What might work for you is something like the Datum 635PCI card. This card has it's own disciplined oscillator that is synchronized to an external source (GPS(637PCI), IRIGB, or 1PPS). The card can then be programmed to generate an interrupt on a periodic basis or at a specific time. If your GPS receiver can output IRIGB that is best, but a 1PPS signal will work also (or you could go with the 637PCI model and combine your GPS/Timing function into one card). The card also lets you compensate for any latency or bias in your timing source. With a good time source (GPS), you should have no problem getting < 1ms of absolute accuracy out of your system. http://www.datum.com/test_blp.html Craig Marc LeBlanc wrote: > > I am currently working on implementing a TDMA protocol that requires > from 1ms to 10ms timing difference between every nodes. It has to run > on common Pentium PCs. Every node is completely separate from other > nodes, they communicate through a radio link that is far from stable > (or constant). So I need a realy stable timing solution to keep > accurate timing between the nodes. > > The nodes will have access to GPS to get a common and constant 1 > second interrupt. At this time, I'm simply using watchdog timers that > I periodically synchronize to NTP. This provide about 2 to 4 system > ticks of error between the nodes (at 100Hz clock rate ~ 30ms). This is > too much. > > I need a higher precision timer than the system clock. And it has to > be much more stable (the system clock drifts up to 8 seconds per day, > this is bad). > > Are there any PCI cards that provide programmable high resolution > constant interrupts? (say I want an interrupt every 30 seconds and I > want it to be precise up to 1ms in the span of a day). This card > should be easy to integrate in vxWorks. > > I'm also open to a software solution to this problem (cheaper is > better) > > Thank you, > Marc LeBlanc - -- Craig Johnston Did you know you can't steer a boat that isn't moving? Just like a life. -- Paul Lutus --------------------------- Newsgroups: comp.os.vxworks Subject: Re: SMC driver Date: 23 Jan 2002 09:56:29 -0800 From: ntjiang@yahoo.com (Jason Jiang) Organization: http://groups.google.com/ Message-ID: <12116a88.0201230956.1b276d27@posting.google.com> References: <12116a88.0201221830.591bea57@posting.google.com> <2a6118da.0201230241.1604435d@posting.google.com> Hi Rashid, Thank you very much for the reply. I found the definition #define sioCallbackInstall(pSioChan, callbackType, callback, callbackArg) \ ((pSioChan)->pDrvFuncs->callbackInstall (pSioChan, callbackType, \ in sioLib.h. So as you said, when I call sioCallbackInstall(), ppc860SioCallbackInstall() will be called indeed. I also found that sioCallbackInstall() is called by wdbSlipPktDevInit(), which is then called by usrRoot()->wdbConfig()->wdbCommIfInit() if INCLUDE_WDB is defined. I got two doubts: 1) I dumped out all the symbols in my vxworks image, but didn't find wdbSlipPktDevInit(); 2) If I don't define INCLUDE_WDB, does it mean I cannot get and put chars? Pls help. Thanks again. Have a good day! Jason raj_rashid@hotmail.com (Rashid) wrote in message news:<2a6118da.0201230241.1604435d@posting.google.com>... > Hi Jason, > > In sysLib.c we call.......sysSerialHwInit() function....... > ....Now this function is implemented in sysSerial.c of BSP in which > structures for channels like ppc860Chan is defined. > > In sysSerialHwInit() some parameters of the structure > ppc860Chan are initialized and a reference of ppc860Chan is passed to > the function ppc860SioDevInit() in which the ppc860Chan structure is > initialized with ppc860SioDrvFuncs of which ppc860CallbackInstall() is > a part. > > Now the sioLib library provides functions like > sioCallbackInstall() to install ISR callbacks to get and put chars. So > when u make a call to sioCallbackInstall the corresponding > function(ppc860SioCallbackInstall) registered with your serial > channel(ppc860Chan) is called. > > You can find usage of sioCallbackInstall() in file > WIND_BASE\target\src\drv\wdb\wdbSlipPktDrv.c > > regards, > Rashid > > > ntjiang@yahoo.com (Jason Jiang) wrote in message news:<12116a88.0201221830.591bea57@posting.google.com>... > > I have a question when reading SMC driver in my system. We use > > ppc860Sio.c (included by sysSerial.c) as the serial driver. It's > > working fine. But I didn't see any place to install the two callback > > functions. They are supposed to be installed by > > ppc860CallbackInstall() in ppc860Sio.c. But I didn't find any place to > > call this function. Pls help. Thanks. > > Jason --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Symbol table errors compiling an image when using STL Date: 23 Jan 2002 10:00:04 -0800 From: john_94501@yahoo.com (John) Organization: http://groups.google.com/ Message-ID: <488e459a.0201231000.73e3f50c@posting.google.com> References: <3c29bbec$1@news.barak.net.il> <3c3833dd$1@news.barak.net.il> <488e459a.0201071346.16dad6f2@posting.google.com> <769ac895.0201171412.346dc387@posting.google.com> <488e459a.0201172059.1eeb7d1d@posting.google.com> <769ac895.0201230241.7e6993d9@posting.google.com> Hello, > Yes, the problem is related to the generation of '.' characters in the > mangled names. Unfortunately, I also don't know how to turn this off > for GNU, so I tried with sed. But now I have another problem. I added > sed command (sed "/\./d" SymTbl.c) to the make rules to remove all the > lines with '.'. But Tornado build window convert all '/' to '\', so > the sed command looked like this (sed "\\.\d" ...) NOT OK! Please > help... Sorry, I don't use Windoze as a development platform for exactly these sorts of reasons. I imagine that the easiest way to work around the conversion would be to place the sed script in an external file rather than on the command line, then use the -f option to read that script. HTH, John... --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Help!! - Posix signals implementation Date: Wed, 23 Jan 2002 13:03:14 -0500 From: Doug Dotson Organization: Verio Message-ID: <3C4EFAE2.EE4829A8@digidata.com> References: <3C4EDC84.96E3D858@insa-lyon.fr> Does your application terminate before the log task has a change to print the message? logMsg() just queues up messages which ar then printed out by a log task. doug Jacques Skubich wrote: > I am using the full VxSim simulator and I get some trouble in > implementing SIGNALS. > The following program is producing this output : > ********************************* > > VxWorks > > Copyright 1984-1998 Wind River Systems, Inc. > > CPU: VxSim for Windows > VxWorks: 5.4 > BSP version: 1.1/1 > Creation date: Apr 22 1999 > WDB: Ready. > > Wait for Signal > before kill > after kill > > ************************************* > What happens with the handler and logMsg??? > ************************************** > > #include > #include > #include > #include > #include > > void SigHand(int no) > { > logMsg("in the handler\n",0,0,0,0,0,0); > } > > /*----------------------------------*/ > int P() > > { > struct sigaction action; > > action.sa_handler= SigHand; > sigemptyset(&action.sa_mask); > sigaddset(&action.sa_mask,SIGUSR2); > action.sa_flags= 0; > > sigaction(SIGUSR2,&action,NULL); > > for(;;) > { > printf("Wait for Signal \n"); > sigsuspend(&action.sa_mask); > printf("Signal arrived \n"); > } > return(0); > > } /* */ > > int S(int tId) > { > taskDelay(600); > printf("before kill\n"); > kill(tId, SIGUSR2); > printf("after kill \n"); > return(0); > } > > void M() > { > int PId, SId; > > PId=taskSpawn("P",100,0,10000,P,0,0,0,0,0,0,0,0,0,0); > SId=taskSpawn("S",100,0,10000,S,PId,0,0,0,0,0,0,0,0,0); > exit(0); > > } > /*==========================================================*/ > /* */ > > -- > |// > (o o) > ___________________________oOO__(_)__OOo________________________________ > > IMPORTANT ** Nouveau N° de fax ** IMPORTANT ** New fax number > > Jacques J. Skubich > Departement d'Informatique - Blaise Pascal > INSA de Lyon > F-69621 VILLEURBANNE CEDEX > > Jacques.Skubich@insa-lyon.fr > Tel/Phone : +33 (0)4 72 43 84 86 > Facsimile : +33 (0)4 37 47 90 82 > Mobile : +33 (0)6 99 61 25 18 > > ___________________________________________________________________________ > > "Si nous avons chacun un objet et que nous les échangeons, nous avons > chacun > un objet. Si nous avons chacun une idée et que nous les échangeons, nous > avons > chacun deux idées." Proverbe chinois. > ___________________________________________________________________________ --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Symbol table errors compiling an image when using STL Date: 23 Jan 2002 13:06:40 -0500 From: Bill Pringlemeir Organization: Bell Sympatico Message-ID: References: <3c29bbec$1@news.barak.net.il> <3c3833dd$1@news.barak.net.il> <488e459a.0201071346.16dad6f2@posting.google.com> <769ac895.0201171412.346dc387@posting.google.com> <488e459a.0201172059.1eeb7d1d@posting.google.com> <769ac895.0201230241.7e6993d9@posting.google.com> Sender: bpringlemeir@DeadDuck >>>>> "Sasa" == Sasa writes: Sasa> Yes, the problem is related to the generation of '.' characters Sasa> in the mangled names. Unfortunately, I also don't know how to Sasa> turn this off for GNU, so I tried with sed. But now I have Sasa> another problem. I added sed command (sed "/\./d" SymTbl.c) to Sasa> the make rules to remove all the lines with '.'. But Tornado Sasa> build window convert all '/' to '\', so the sed command looked Sasa> like this (sed "\\.\d" ...) NOT OK! Please help... Lp, Sasa You can use `find' on NT. I run the following when creating symbols, find "GLOBAL" tmp.2 | $(MUNCH) >ctdt.c On advice from this newsgroup. The find program is faster than the script than searching for "GLOBAL" in tcl or perl. Perhaps you could use "find /v ." to filter the entries with a period. This will remove all lines. hth, Bill Pringlemeir. - -- vxWorks FAQ, "http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html" --------------------------- Newsgroups: comp.os.vxworks Subject: Re: intConnect() vs pciIntConnect() Date: 23 Jan 2002 10:14:40 -0800 From: arun4vxworks@indiatimes.com (Arun Prasad) Organization: http://groups.google.com/ Message-ID: References: Hi All, Thank you so much for your replies/clarifications. But, as of now, I'm going with intConnect only having an assumption that there will be only one PCI card will be in my system as we have a tight schedule. Once we meet that schedue, I'll try to experiment all your suggestions. Thanks again. Regards Arun "drdiags" wrote in message news:... > Darn!! > > The true path and name of this file is > %WIND_BASE%/target/src/drv/pci/pciIntLib.c > > <....SNIP....> > > > > "Arun Prasad" wrote in message > > news:e38d912c.0201220455.30a696b1@posting.google.com... > > > Hi All, > > > > > > I couldn't able to find the exact differences between intconnect() > > > and pciIntConnect(). --------------------------- Newsgroups: comp.os.vxworks Subject: TCP/IP Performance under VxWorks Date: Wed, 23 Jan 2002 13:34:55 -0500 From: Paul Randall Organization: Posted via Supernews, http://www.supernews.com Message-ID: <3C4F024F.5324B28C@delta-info.com> Hi all, I have an application where I need to send or receive up to 30 Mbits/sec of sustained UDP data over Ethernet. I am evaluating different RTOSes and processors to come up with an architecture for our application. Does anyone have TCP/IP performance numbers for VxWorks (or other RTOSes for that matter) running on VxWorks supported processors? Also, does the ability to DMA data from the processor to an Ethernet controller make much of a difference in performance? I have been looking at the SMSC 91C111 ethernet controller which does not require a PCI bus but it also does not support DMA transfers. Thanks in advance, Paul Randall Delta Information Systems --------------------------- Newsgroups: comp.os.vxworks Subject: FTP connection ok but ... Date: 23 Jan 2002 15:42:11 GMT From: Charly Message-ID: <2002123-164211-214208@foorum.com> dear All, I use a direct point to point cross cable from VxWorks target and WIN98 the host ping works and the target ping works. but there is a FTP problem : On The Shell target, I type - ->ls After a 10 sec timout, the shell display value = -1 On the Host, the FTP server displays: [L 4610] 01/23/02 10:02:02 Connection accepted from 192.0.1.2 [C 4610] 01/23/02 10:02:02 Command "USER target" received [C 4610] 01/23/02 10:02:02 PASSword accepted [L 4610] 01/23/02 10:02:02 User target logged in. [C 4610] 01/23/02 10:02:02 Command "TYPE I" received [C 4610] 01/23/02 10:02:02 TYPE set to I N [C 4610] 01/23/02 10:02:02 Command "PORT 192,0,1,2,4,11" received [C 4610] 01/23/02 10:02:02 PORT set to 192.0.1.2 - 1035 (4,11) [C 4610] 01/23/02 10:02:02 Command "RETR intru" received [C 4610] 01/23/02 10:02:02 RETR failed - file does not exist [G 4610] 01/23/02 10:02:02 RETR failed - file does not exist [C 4610] 01/23/02 10:02:02 QUIT or close - user target logged out [L 4618] 01/23/02 16:35:30 Connection accepted from 192.0.1.2 [C 4618] 01/23/02 16:35:30 Command "USER target" received [C 4618] 01/23/02 16:35:30 PASSword accepted [L 4618] 01/23/02 16:35:30 User target logged in. [C 4618] 01/23/02 16:35:30 Command "TYPE I" received [C 4618] 01/23/02 16:35:30 TYPE set to I N [C 4618] 01/23/02 16:35:30 Command "PORT 192,0,1,2,4,4" received [C 4618] 01/23/02 16:35:30 PORT set to 192.0.1.2 - 1028 (4,4) [C 4618] 01/23/02 16:35:30 Command "NLST" received [C 4618] 01/23/02 16:35:30 NLST started successfully for path/wildcard [C 4618] 01/23/02 16:35:50 Transfer aborted [G 4618] 01/23/02 16:35:50 Got file listing unsuccessfully [! 4618] 01/23/02 16:35:50 Timeout while sending data [C 4618] 01/23/02 16:35:50 QUIT or close - user target logged out [! -001] 01/23/02 16:35:50 Received a command on socket 40 which I can't find It seems that the data sent on port 1028 is not seen by the target. Who could help me to solve this problem? Did it happen elsewhere? Charly - -- Use our news server 'news.foorum.com' from anywhere. More details at: http://nnrpinfo.go.foorum.com/ --------------------------- Newsgroups: comp.os.vxworks Subject: Creating Build Rules in Project Workspace Date: 23 Jan 2002 10:49:06 -0800 From: amanullah_khan@hotmail.com (Amanullah Khan) Organization: http://groups.google.com/ Message-ID: Hi I am working in Tornado 2.0 with vxWorks 5.4. I have some queries related to the project facility. Can you tell me, from where the properties of Build specification is got?. Acutally in the build specification RULES only vxWorks, vxWorks_rom and vxWorks_romCompress, vxWorks_romResident are listed. Why other builds is not listed?. If other rules has to be displayed then, what should be done for that. I tried creating a new rule in the Build Specifications. The newly created rule is coming as Custom Rule. How to make it as a normal rule ie like vxWorks. Thanx in advance Regards Khan --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Error while downloading project on target??? Date: 23 Jan 2002 10:59:42 -0800 From: dragonli@yahoo.com (June) Organization: http://groups.google.com/ Message-ID: References: <7421c9fd.0201222012.335b07bb@posting.google.com> I am not sure if mine suggestion is right. Anyway here it is, pls check the cache memory size you defined for your target. by default it is 1M defined as following: tgtsvr.exe -V -m 1048576 good luck kumar_amit@delhi.tcs.co.in (Amit) wrote in message news:<7421c9fd.0201222012.335b07bb@posting.google.com>... > I have a target with memory 128 MB. My final image size is coming > around 76 MB. So while downloading the image on target, system hangs > up. I want to know whether there are certain size restrictions on > image size which you can download . > > > Thanks > > Ajay --------------------------- Newsgroups: comp.os.vxworks Subject: Re: SCSI Performance in vxWorks Date: Wed, 23 Jan 2002 09:15:38 +1030 From: Tim Shaw Organization: Defence Science and Technology Organisation Message-ID: <3C4DEB92.3DFD5FD7@dsto.defence.gov.au> References: <20020122.015817.1346811305.3143@analogic.com> We've had up to about 17MBytes/sec on a Synergy VGM5 (the driver had been modified by Synergy), and about the same on a MVME2307 with SYM875 (we had to get hold of the source code and modify it to get that performance). I've never used the 895, but if it is using the same driver as the wr$ driver for the mv2307 there's a lot of tweaking possible to get big performance improvements (both increasing data transfer rate and reducing pci usage). HTH Tim Jared Maguire wrote: > Has anyone measured SCSI performnace greater than 10MB/s under vxWorks? > What card/driver? > > We're using an LSI Logic 53C1010 controller, and couple of Seagate > Cheetahs. We've also tried it on a sym895, and an adaptec 7880, and we > can't seem to get a transfer to happen at greater than 10MB/s, sustained. > What gives? > > -J --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Booting from ATA Compact Flash Date: Wed, 23 Jan 2002 22:12:44 +0100 From: "Werner Schiendl" Message-ID: <3c4f274d$1@brateggebdc5.br-automation.co.at> References: Hi, verify that the file BOOTROM.SYS is not fragmented from MS-DOS's point of view. You can verify this with chkdsk C:\bootrom.sys (assuming your flash disk has drive letter C) There will be a line at the end of the output, that tells you whether it is or not. If it is fragmented, VxLd loads garbage from disk into memory, 'cause it only searches for the start of BOOTROM.SYS and reads the following physical sectors into memory. It's completely random if the system just hangs or crashes immediately. hth Werner "Benjamin Horowitz" wrote in message news:ab1f8a59.0201200034.2ef1e541@posting.google.com... > Hi, > > Please forgive this naive question: > > I'm trying to boot from an ATA Compact Flash disk. The disk is > control 0, drive 0. I create a new bootrom_uncmp using Tornado II > (and VxWorks 5.4). I copy bootrom_uncmp to a floppy. I invoke the > mkbootAta function from the Tornado shell, as follows: > > mkbootAta 0, 0, "/fd0/BOOTRO~1 > > (BOOTRO~1 is the DOS file name.) I can now see bootrom.sys on the ATA > drive. Now I reboot, this time without my normal boot floppy. I see > the following messages: > > Searching for Boot Record from Floppy..Not Found > Searching for Boot Record from IDE-0..OK > > But then the machine hangs. Normally I would see: > > VxLd........ [etc.] > > Any suggestions about where to look? I'm stuck. > > Thanks a lot, > Ben --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Vxworks beginner Date: Wed, 23 Jan 2002 22:09:55 GMT From: "Vikas Anand" Organization: ATT Broadband Message-ID: References: <8207c357.0201212226.3b5a9a55@posting.google.com> Go to www.wrs.com "Sridhar CV" wrote in message news:8207c357.0201212226.3b5a9a55@posting.google.com... > Hi, > can any one of you suggest me the best way to master Vxworks > application development in very short time. > > Regards > Sridhar CV --------------------------- Newsgroups: comp.os.vxworks Subject: Re: SCSI Performance in vxWorks Date: Wed, 23 Jan 2002 19:13:04 -0500 From: "Jared Maguire" Message-ID: <20020123.191303.1389867269.5196@analogic.com> References: <20020122.015817.1346811305.3143@analogic.com> <3C4DEB92.3DFD5FD7@dsto.defence.gov.au> In article <3C4DEB92.3DFD5FD7@dsto.defence.gov.au>, "Tim Shaw" wrote: > We've had up to about 17MBytes/sec on a Synergy VGM5 (the driver had > been modified by Synergy), and about the same on a MVME2307 with SYM875 > (we had to get hold of the source code and modify it to get that > performance). I've never used the 895, but if it is using the same > driver as the wr$ driver for the mv2307 there's a lot of tweaking > possible to get big performance improvements (both increasing data > transfer rate and reducing pci usage). . Just now I timed scsi performance under both Win2k and vxWorks on the same hardware, and Win2k performed it's 4MB write in about 100ms, which is pretty decent. VxWorks performed the same write in 400ms. I dove in with windview to try to see what's going on, and found that the SCSI driver's interrupt handler gets called a few times at each of THREE points during a write-operation. These points are at t=0ms, t=100ms and t=400ms. The t=100ms point is interesting, as that's EXACTLY how long windows takes to write the data. I'm starting to suspect that the scsi driver is actually writing the data at a reasonable speed, but then is doing SOMETHING else before it returns successfully. *shrug* That's a bummer that we might need to get the code for the driver and tweak it. Oh, well. Thanks for the reality-check, - -Jared --------------------------- Newsgroups: comp.os.vxworks Subject: Re: routing problem ??? Date: Thu, 24 Jan 2002 10:53:03 +1030 From: Tim Shaw Organization: Defence Science and Technology Organisation Message-ID: <3C4F53E7.B4E3ABEF@dsto.defence.gov.au> References: Have you configured the route on the pc so that it knows to send to 192.168.1.0 when it wants to send to 210.118.156.0? it2xpert wrote: > Hi , vxworks seniors !!! > I desinged the mpc850sr board for router... > I implemented the scc2,scc3 for ethernet mode... > > I use the T2, vxworks 5.4... > I want to implement the Layer 3 router..... > > 192.168.1.0 -----------------192.168.1.1 (scc3) > > (scc2) > 210.118.156.240 -------------------210.118.156.0 > > I used the RIP v.1 > Routing table is following.... > > -> routeShow > > ROUTE NET TABLE > destination gateway flags Refcnt Use Interface > ---------------------------------------------------------------------------- > 0.0.0.0 210.118.156.254 3 0 21 cpm0 > 192.168.1.0 192.168.1.1 101 0 0 cpm1 > 210.118.156.0 210.118.156.240 101 0 0 cpm0 > ---------------------------------------------------------------------------- > > ROUTE HOST TABLE > destination gateway flags Refcnt Use Interface > ---------------------------------------------------------------------------- > 127.0.0.1 127.0.0.1 5 1 65 lo0 > ---------------------------------------------------------------------------- > value = 77 = 0x4d = 'M' > -> > > I trid the ping test !!! > My pc is connected the 192.168.1.0 network (192.168.1.3) > ping 192.168.1.1 ----- OK, > ping 210.118.156.240 ---- OK > > but i did't reach the 210.118.156.xxx point.... > > I tried ping to the target server console.... > -> ping "210.118.156.xxx" > this is OK.... > > what's wrong ???? --------------------------- Newsgroups: comp.os.vxworks Subject: Memory lost during open,read, close with NFS and netDrv Date: 23 Jan 2002 16:32:42 -0800 From: vdalal@hotmail.com (Vasu Dalal) Organization: http://groups.google.com/ Message-ID: <363d071b.0201231632.48875909@posting.google.com> Hello all : I am trying to understand memory allocation/fragmentation wrt netDrv/NFS in particular and with VxWorks in general. Can someone please explain after looking at the sequence of file open's, read's and close's with both netDrv and NFS given below as to what happened to the "48" bytes ? I believe they are NOT lost but i cannot understand how they can be accounted for. This will also help my understanding of the memLib and some of its data structures. Also when using netDrv why is the size of the memory allocated for a 2767 byte file 3424 bytes ? In the sequence below : /CF => is a DOS FS device on a Vxworks target anotherVxWorksHost: => is the host name for a VxWorks card used as a host cfg.txt => is a file containing some text I am using VxWorks version 5.4, Kernel Wind version 2.5 and Processor is 8260 ( PPC arch ). Sorry for the long post and thank you for your time. Any input is appreciated. TIA. vasu PS: [snip1] - -> memShow status bytes blocks avg block max block ------ --------- -------- ---------- ---------- current free 11190848 18 621713 6815808 alloc 7940016 3027 2623 - cumulative alloc 30594200 11345 2696 - value = 0 = 0x0 - -> open "anotherVxWorksHost:/CF/cfg.txt",2 <-- open using netDrv value = 15 = 0xf - -> memShow status bytes blocks avg block max block ------ --------- -------- ---------- ---------- current free 11187424 18 621523 6815808 alloc 7943440 3036 2616 - <-- memory is fragmented by 9 more blocks cumulative alloc 30603776 11361 2693 - value = 0 = 0x0 - -> 11190848-11187424 value = 3424 = 0xd60 <-- memory allocated is 3424 while the size of the file "cfg.txt" is 2767 bytes - -> close 15 value = 0 = 0x0 - -> memShow status bytes blocks avg block max block ------ --------- -------- ---------- ---------- current free 11190800 18 621711 6815808 alloc 7940064 3028 2622 - cumulative alloc 30615840 11384 2689 - value = 0 = 0x0 - -> 11190848-11190800 value = 48 = 0x30 = '0' <-- Lost "48" bytes during the whole sequence of open and close using netDrv - -> [snip1] [snip2] - -> memShow status bytes blocks avg block max block ------ --------- -------- ---------- ---------- current free 11190344 19 588965 6815808 alloc 7940520 3033 2618 - cumulative alloc 30812704 11813 2608 - value = 0 = 0x0 - -> open "anotherVxWorksHost:/CF/cfgip.txt",0 <-- open using netDrv value = 15 = 0xf - -> memShow status bytes blocks avg block max block ------ --------- -------- ---------- ---------- current free 11186920 19 588785 6815808 alloc 7943944 3042 2611 - <-- memory is fragmented by 9 more blocks cumulative alloc 30822240 11828 2605 - value = 0 = 0x0 - -> - -> 11190344-11186920 value = 3424 = 0xd60 <-- memory allocated is 3424 while the size of the file "cfg.txt" is 2767 bytes - -> - -> buf=malloc(1024) buf = 0x4215a0: value = 25152736 = 0x17fcce0 - -> memShow status bytes blocks avg block max block ------ --------- -------- ---------- ---------- current free 11185888 19 588730 6815808 alloc 7944976 3043 2610 - <-- memory is fragmented by 1 more block cumulative alloc 30829424 11836 2604 - value = 0 = 0x0 - -> - -> 11186920-11185888 value = 1032 = 0x408 <-- 1024 bytes+8 bytes for memLib/memPartLib data structures - -> read 15,buf,1024 value = 1024 = 0x400 - -> memShow status bytes blocks avg block max block ------ --------- -------- ---------- ---------- current free 11185888 19 588730 6815808 <-- No change as expected alloc 7944976 3043 2610 - <-- No change in # of blocks as expected cumulative alloc 30838552 11844 2603 - value = 0 = 0x0 - -> free buf value = 0 = 0x0 - -> memShow status bytes blocks avg block max block ------ --------- -------- ---------- ---------- current free 11186920 19 588785 6815808 alloc 7943944 3042 2611 - cumulative alloc 30845832 11850 2603 - value = 0 = 0x0 - -> 11186920-11185888 value = 1032 = 0x408 <-- Recovered the 1032 bytes previously malloc'ed - -> close 15 value = 0 = 0x0 - -> memShow status bytes blocks avg block max block ------ --------- -------- ---------- ---------- current free 11190296 19 588962 6815808 alloc 7940568 3034 2617 - cumulative alloc 30856128 11858 2602 - value = 0 = 0x0 - -> - -> 11190344-11190296 value = 48 = 0x30 = '0' <-- Lost "48" bytes during the whole sequence of open, read and close using netDrv [snip2] [snip3] - -> memShow status bytes blocks avg block max block ------ --------- -------- ---------- ---------- current free 11190296 19 588962 6815808 alloc 7940568 3034 2617 - cumulative alloc 30862480 11870 2600 - value = 0 = 0x0 - -> - -> open "/CF/cfgip.txt",0 <-- open using NFS value = 15 = 0xf - -> memShow status bytes blocks avg block max block ------ --------- -------- ---------- ---------- current free 11181808 19 588516 6815808 alloc 7949056 3037 2617 - <-- memory is fragmented by 3 more blocks cumulative alloc 30877080 11879 2599 - value = 0 = 0x0 - -> 11190296-11181808 value = 8488 = 0x2128 <-- 8K memory+extra for NFS since NFS works in 8192 byte(8K) chunks - -> buf=malloc(1024) buf = 0x4215a0: value = 25146488 = 0x17fb478 - -> memShow status bytes blocks avg block max block ------ --------- -------- ---------- ---------- current free 11180776 19 588461 6815808 alloc 7950088 3038 2616 - cumulative alloc 30887400 11892 2597 - value = 0 = 0x0 - -> 11181808-11180776 value = 1032 = 0x408 <-- 1024 bytes+8 bytes for memLib/memPartLib data structures - -> read 15,buf,1024 value = 1024 = 0x400 - -> memShow status bytes blocks avg block max block ------ --------- -------- ---------- ---------- current free 11180776 19 588461 6815808 <-- No change as expected alloc 7950088 3038 2616 - <-- No change in # of blocks as expected cumulative alloc 30896488 11899 2596 - value = 0 = 0x0 - -> free buf value = 0 = 0x0 - -> memShow status bytes blocks avg block max block ------ --------- -------- ---------- ---------- current free 11181808 19 588516 6815808 alloc 7949056 3037 2617 - cumulative alloc 30902560 11904 2595 - value = 0 = 0x0 - -> - -> 11181808-11181808 <-- Recovered the 1032 bytes previously malloc'ed value = 0 = 0x0 - -> close 15 value = 0 = 0x0 - -> memShow status bytes blocks avg block max block ------ --------- -------- ---------- ---------- current free 11190248 19 588960 6815808 alloc 7940616 3035 2616 - cumulative alloc 30908592 11908 2595 - value = 0 = 0x0 - -> - -> 11190296-11190248 value = 48 = 0x30 = '0' <-- Lost "48" bytes during the whole sequence of open, read and close using NFS - -> [snip3] Keywords : malloc, free, memory fragmentation, memory alignment, NFS, netDrv --------------------------- Newsgroups: comp.os.vxworks Subject: Can I use i/o system "read()" to read a i/o port address? Date: 23 Jan 2002 19:47:05 -0800 From: yong_bvcom@sina.com (bigtornado) Organization: http://groups.google.com/ Message-ID: <5624c1bf.0201231947.614d58a5@posting.google.com> hi, I want add a i/0 port on my firmware,it's have a fixed port address,we can read data from it use: data = *addr; but it's efficiency too low and speed "crazy" slow,so I want use read() to access it,what should I do? Thanks! --------------------------- Newsgroups: comp.os.vxworks Subject: Re: SMC driver Date: 23 Jan 2002 21:19:26 -0800 From: raj_rashid@hotmail.com (Rashid) Organization: http://groups.google.com/ Message-ID: <2a6118da.0201232119.5dd8368c@posting.google.com> References: <12116a88.0201221830.591bea57@posting.google.com> <2a6118da.0201230241.1604435d@posting.google.com> <12116a88.0201230956.1b276d27@posting.google.com> Hello Jason, > 1) I dumped out all the symbols in my vxworks image, but didn't find > wdbSlipPktDevInit(); This symbol is included when you define INCLUDE_WDB_COMM_SERIAL. The routine wdbSlipPktDevInit( ) initializes a SLIP packet device for WDB agent. You should include this device if you want to WDB serial connection. > 2) If I don't define INCLUDE_WDB, does it mean I cannot get and put > chars? INCLUDE_WDB includes a software agent in your vxWorks image to support Tornado tools. Now u can choose the type o0f connection u want between target server on Host and WDB agent on target. For ex. INCLUDE_WDB_COMM_END for WDB END driver connection. INCLUDE_WDB_COMM_NETWORK for WDB network connection. INCLUDE_WDB_COMM_SERIAL for WDB serial connection.............etc. > Pls help. Thanks again. > Have a good day! > Jason Hope this helps. regards, Rashid --------------------------- Newsgroups: comp.os.vxworks Subject: linking problem Date: Thu, 24 Jan 2002 07:31:54 -0000 From: "Tommy Fallsen" Organization: Kongsberg Defence & Aerospace, Norway Message-ID: Hi Anyone had this error before? Cannot allocate 0x00383a41 bytes os memory for 'group .begin,.text,.init (and so on) Memory area 'mem1' is full. (server does'nt run out of mem or swap during compiling and linking) Anyone had similar problem? What might cause this? It only happens when user use debugging options. Are there flags i should take a closer look at. Software:Solaris 2.6 ,diab compiler 4.4d Hardware: Enterprise 450. Tommy --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Memory lost during open,read, close with NFS and netDrv Date: 23 Jan 2002 23:05:01 -0800 From: john_94501@yahoo.com (John) Organization: http://groups.google.com/ Message-ID: <488e459a.0201232305.7b102374@posting.google.com> References: <363d071b.0201231632.48875909@posting.google.com> Hello, This is such a frequently asked question; when you use literal strings from the shell (such as those filenames you are passing to open()), the shell allocates some memory to store the string in. Such it doesn't know when you've finished with the string, it cannot free the memory. To prove this to yourself, try allocating the string to a variable before the initial memShow() call: - -> filename="myFileName" - -> memShow - -> open filename, 2 HTH, John... vdalal@hotmail.com (Vasu Dalal) wrote in message news:<363d071b.0201231632.48875909@posting.google.com>... > Hello all : > > I am trying to understand memory allocation/fragmentation wrt > netDrv/NFS in particular and with VxWorks in general. > > Can someone please explain after looking at the sequence of > file open's, read's and close's with both netDrv and NFS given > below as to what happened to the "48" bytes ? > I believe they are NOT lost but i cannot understand how > they can be accounted for. This will also help my understanding > of the memLib and some of its data structures. > > Also when using netDrv why is the size of the memory allocated > for a 2767 byte file 3424 bytes ? > > In the sequence below : > /CF => is a DOS FS device on a Vxworks target > anotherVxWorksHost: => is the host name for a VxWorks card > used as a host > cfg.txt => is a file containing some text > > I am using VxWorks version 5.4, Kernel Wind version 2.5 and > Processor is 8260 ( PPC arch ). > > Sorry for the long post and thank you for your time. > Any input is appreciated. > TIA. > > vasu > > PS: > > [snip1] > -> memShow > status bytes blocks avg block max block > ------ --------- -------- ---------- ---------- > current > free 11190848 18 621713 6815808 > alloc 7940016 3027 2623 - > cumulative > alloc 30594200 11345 2696 - > value = 0 = 0x0 > > -> open "anotherVxWorksHost:/CF/cfg.txt",2 <-- open using netDrv > value = 15 = 0xf > > -> memShow > status bytes blocks avg block max block > ------ --------- -------- ---------- ---------- > current > free 11187424 18 621523 6815808 > alloc 7943440 3036 2616 - <-- memory is > fragmented > by 9 more > blocks > cumulative > alloc 30603776 11361 2693 - > value = 0 = 0x0 > > -> 11190848-11187424 > value = 3424 = 0xd60 <-- memory allocated > is 3424 while the size of the file "cfg.txt" is 2767 > bytes > -> close 15 > value = 0 = 0x0 > > -> memShow > status bytes blocks avg block max block > ------ --------- -------- ---------- ---------- > current > free 11190800 18 621711 6815808 > alloc 7940064 3028 2622 - > cumulative > alloc 30615840 11384 2689 - > value = 0 = 0x0 > > -> 11190848-11190800 > value = 48 = 0x30 = '0' <-- Lost "48" bytes > during the whole sequence of open and close using > netDrv > -> > [snip1] > > > > [snip2] > -> memShow > status bytes blocks avg block max block > ------ --------- -------- ---------- ---------- > current > free 11190344 19 588965 6815808 > alloc 7940520 3033 2618 - > cumulative > alloc 30812704 11813 2608 - > value = 0 = 0x0 > > -> open "anotherVxWorksHost:/CF/cfgip.txt",0 <-- open using netDrv > value = 15 = 0xf > > -> memShow > status bytes blocks avg block max block > ------ --------- -------- ---------- ---------- > current > free 11186920 19 588785 6815808 > alloc 7943944 3042 2611 - <-- memory is > fragmented by > 9 more blocks > cumulative > alloc 30822240 11828 2605 - > value = 0 = 0x0 > -> > > -> 11190344-11186920 > value = 3424 = 0xd60 <-- memory allocated > is 3424 > while the size of the > file > "cfg.txt" is 2767 > bytes > -> > -> buf=malloc(1024) > buf = 0x4215a0: value = 25152736 = 0x17fcce0 > > -> memShow > status bytes blocks avg block max block > ------ --------- -------- ---------- ---------- > current > free 11185888 19 588730 6815808 > alloc 7944976 3043 2610 - <-- memory is > fragmented by > 1 more block > cumulative > alloc 30829424 11836 2604 - > value = 0 = 0x0 > -> > > -> 11186920-11185888 > value = 1032 = 0x408 <-- 1024 bytes+8 bytes > for > memLib/memPartLib data > structures > -> read 15,buf,1024 > value = 1024 = 0x400 > > -> memShow > status bytes blocks avg block max block > ------ --------- -------- ---------- ---------- > current > free 11185888 19 588730 6815808 <-- No change as > expected > alloc 7944976 3043 2610 - <-- No change in # of > blocks > as expected > cumulative > alloc 30838552 11844 2603 - > value = 0 = 0x0 > > -> free buf > value = 0 = 0x0 > > -> memShow > status bytes blocks avg block max block > ------ --------- -------- ---------- ---------- > current > free 11186920 19 588785 6815808 > alloc 7943944 3042 2611 - > cumulative > alloc 30845832 11850 2603 - > value = 0 = 0x0 > > -> 11186920-11185888 > value = 1032 = 0x408 <-- Recovered the 1032 > bytes > previously > malloc'ed > > -> close 15 > value = 0 = 0x0 > > -> memShow > status bytes blocks avg block max block > ------ --------- -------- ---------- ---------- > current > free 11190296 19 588962 6815808 > alloc 7940568 3034 2617 - > cumulative > alloc 30856128 11858 2602 - > value = 0 = 0x0 > -> > > -> 11190344-11190296 > value = 48 = 0x30 = '0' <-- Lost "48" bytes > during the > whole sequence of > open, read > and close using netDrv > > > > [snip2] > > > [snip3] > -> memShow > status bytes blocks avg block max block > ------ --------- -------- ---------- ---------- > current > free 11190296 19 588962 6815808 > alloc 7940568 3034 2617 - > cumulative > alloc 30862480 11870 2600 - > value = 0 = 0x0 > -> > -> open "/CF/cfgip.txt",0 <-- open using NFS > value = 15 = 0xf > > -> memShow > status bytes blocks avg block max block > ------ --------- -------- ---------- ---------- > current > free 11181808 19 588516 6815808 > alloc 7949056 3037 2617 - <-- memory is > fragmented > by 3 more > blocks > cumulative > alloc 30877080 11879 2599 - > value = 0 = 0x0 > > -> 11190296-11181808 > value = 8488 = 0x2128 <-- 8K memory+extra > for NFS since NFS works in 8192 > byte(8K) chunks > > -> buf=malloc(1024) > buf = 0x4215a0: value = 25146488 = 0x17fb478 > > -> memShow > status bytes blocks avg block max block > ------ --------- -------- ---------- ---------- > current > free 11180776 19 588461 6815808 > alloc 7950088 3038 2616 - > cumulative > alloc 30887400 11892 2597 - > value = 0 = 0x0 > > -> 11181808-11180776 > value = 1032 = 0x408 <-- 1024 bytes+8 bytes > for > memLib/memPartLib data > structures > -> read 15,buf,1024 > value = 1024 = 0x400 > > -> memShow > status bytes blocks avg block max block > ------ --------- -------- ---------- ---------- > current > free 11180776 19 588461 6815808 <-- No change as > expected > alloc 7950088 3038 2616 - <-- No change in # of > blocks as expected > cumulative > alloc 30896488 11899 2596 - > value = 0 = 0x0 > > -> free buf > value = 0 = 0x0 > > -> memShow > status bytes blocks avg block max block > ------ --------- -------- ---------- ---------- > current > free 11181808 19 588516 6815808 > alloc 7949056 3037 2617 - > cumulative > alloc 30902560 11904 2595 - > value = 0 = 0x0 > -> > -> 11181808-11181808 <-- Recovered the 1032 > bytes > previously > malloc'ed > value = 0 = 0x0 > > -> close 15 > value = 0 = 0x0 > > -> memShow > status bytes blocks avg block max block > ------ --------- -------- ---------- ---------- > current > free 11190248 19 588960 6815808 > alloc 7940616 3035 2616 - > cumulative > alloc 30908592 11908 2595 - > value = 0 = 0x0 > -> > -> 11190296-11190248 > value = 48 = 0x30 = '0' <-- Lost "48" bytes > during the > whole sequence of > open, read > and close using NFS > > -> > > > [snip3] > > > Keywords : > malloc, free, memory fragmentation, memory alignment, NFS, netDrv --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Can I use i/o system "read()" to read a i/o port address? Date: Thu, 24 Jan 2002 09:28:47 +0200 From: "Leonid Rosenboim" Organization: Verio Message-ID: References: <5624c1bf.0201231947.614d58a5@posting.google.com> The only disadvantage of accessing your I/O port from application using apointer is portability. if the port address changes application code will nee to be changed. If you want to use read(), you will need to write a simple device driver (see xWorks Programmers Guide on I/O subsystem). Using a driver will not improve performance, because the driver will need to use the same kind of access you are using now, plus the I/O subsystem overhead will slow things down a bit. "bigtornado" wrote in message news:5624c1bf.0201231947.614d58a5@posting.google.com... > hi, > I want add a i/0 port on my firmware,it's have a fixed port > address,we can read data from it use: > data = *addr; > but it's efficiency too low and speed "crazy" slow,so I want use > read() to access it,what should I do? > > Thanks! --------------------------- Newsgroups: comp.os.vxworks Subject: Re: FTP connection ok but ... Date: Thu, 24 Jan 2002 09:32:06 +0200 From: "Leonid Rosenboim" Organization: Verio Message-ID: References: <2002123-164211-214208@foorum.com> I suggest you try ping with different packet sizes to check the driver. The long timeout suggests that the FTP data transfer suffers from packet loss. "Charly" wrote in message news:2002123-164211-214208@foorum.com... > > dear All, > > > I use a direct point to point cross cable from VxWorks target and WIN98 > the host ping works and the target ping works. > but there is a FTP problem : > > On The Shell target, I type > ->ls > After a 10 sec timout, the shell display > value = -1 > On the Host, the FTP server displays: > > [L 4610] 01/23/02 10:02:02 Connection accepted from 192.0.1.2 > [C 4610] 01/23/02 10:02:02 Command "USER target" received > [C 4610] 01/23/02 10:02:02 PASSword accepted > [L 4610] 01/23/02 10:02:02 User target logged in. > [C 4610] 01/23/02 10:02:02 Command "TYPE I" received > [C 4610] 01/23/02 10:02:02 TYPE set to I N > [C 4610] 01/23/02 10:02:02 Command "PORT 192,0,1,2,4,11" received > [C 4610] 01/23/02 10:02:02 PORT set to 192.0.1.2 - 1035 (4,11) > [C 4610] 01/23/02 10:02:02 Command "RETR intru" received > [C 4610] 01/23/02 10:02:02 RETR failed - file does not exist > [G 4610] 01/23/02 10:02:02 RETR failed - file does not exist > [C 4610] 01/23/02 10:02:02 QUIT or close - user target logged out > [L 4618] 01/23/02 16:35:30 Connection accepted from 192.0.1.2 > [C 4618] 01/23/02 16:35:30 Command "USER target" received > [C 4618] 01/23/02 16:35:30 PASSword accepted > [L 4618] 01/23/02 16:35:30 User target logged in. > [C 4618] 01/23/02 16:35:30 Command "TYPE I" received > [C 4618] 01/23/02 16:35:30 TYPE set to I N > [C 4618] 01/23/02 16:35:30 Command "PORT 192,0,1,2,4,4" received > [C 4618] 01/23/02 16:35:30 PORT set to 192.0.1.2 - 1028 (4,4) > [C 4618] 01/23/02 16:35:30 Command "NLST" received > [C 4618] 01/23/02 16:35:30 NLST started successfully for path/wildcard > [C 4618] 01/23/02 16:35:50 Transfer aborted > [G 4618] 01/23/02 16:35:50 Got file listing unsuccessfully > [! 4618] 01/23/02 16:35:50 Timeout while sending data > [C 4618] 01/23/02 16:35:50 QUIT or close - user target logged out > [! -001] 01/23/02 16:35:50 Received a command on socket 40 which I can't find > > It seems that the data sent on port 1028 is not seen by the target. > Who could help me to solve this problem? > Did it happen elsewhere? > Charly > > -- > Use our news server 'news.foorum.com' from anywhere. > More details at: http://nnrpinfo.go.foorum.com/ --------------------------- Newsgroups: comp.os.vxworks Subject: Re: TCP/IP Performance under VxWorks Date: Thu, 24 Jan 2002 09:36:26 +0200 From: "Leonid Rosenboim" Organization: Verio Message-ID: References: <3C4F024F.5324B28C@delta-info.com> In my previous compny - BitBand, we have acheive much higher performance with both UDP and TCP over FastEthernet, using VxWorks. It's an issue of the Ethernet hardware and driver more then anything else. Yes, the DMA, or to be precise Bus Mastership which makes a great deal of a difference. Without DMA, the CPU will be busy transferring data between memory and MAC. CPU and system bus performance is yet another issue which can be of importance, although I recon 30 Mbps can be done on almost anything with a good MAC and a good driver. - - Leonid "Paul Randall" wrote in message news:3C4F024F.5324B28C@delta-info.com... > Hi all, > > I have an application where I need to send or receive up to 30 Mbits/sec > of sustained UDP data over Ethernet. I am evaluating different RTOSes > and processors to come up with an architecture for our application. > > Does anyone have TCP/IP performance numbers for VxWorks (or other RTOSes > for that matter) running on VxWorks supported processors? > > Also, does the ability to DMA data from the processor to an Ethernet > controller make much of a difference in performance? I have been > looking at the SMSC 91C111 ethernet controller which does not require a > PCI bus but it also does not support DMA transfers. > > Thanks in advance, > > Paul Randall > Delta Information Systems > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Help!! - Posix signals implementation Date: Thu, 24 Jan 2002 09:23:26 +0100 From: Jacques Skubich Organization: =?iso-8859-1?Q?D=E9partement?= Informatique - INSA de Lyon Message-ID: <3C4FC47E.FE647085@insa-lyon.fr> References: <3C4EDC84.96E3D858@insa-lyon.fr> <3C4EFAE2.EE4829A8@digidata.com> Hi Doug, Task P is still pending even if tasks M and S are finished. And if the "sigsuspend(&action.sa_mask);" is replaced by a basic "pause" then I got the same output plus an "Interrupt 13....." > Does your application terminate before the log task has > a change to print the message? logMsg() just queues up > messages which ar then printed out by a log task. > > doug > > Jacques Skubich wrote: > > > I am using the full VxSim simulator and I get some trouble in > > implementing SIGNALS. > > The following program is producing this output : > > ********************************* > > > > VxWorks > > > > Copyright 1984-1998 Wind River Systems, Inc. > > > > CPU: VxSim for Windows > > VxWorks: 5.4 > > BSP version: 1.1/1 > > Creation date: Apr 22 1999 > > WDB: Ready. > > > > Wait for Signal > > before kill > > after kill > > > > ************************************* > > What happens with the handler and logMsg??? > > ************************************** > > > > #include > > #include > > #include > > #include > > #include > > > > void SigHand(int no) > > { > > logMsg("in the handler\n",0,0,0,0,0,0); > > } > > > > /*----------------------------------*/ > > int P() > > > > { > > struct sigaction action; > > > > action.sa_handler= SigHand; > > sigemptyset(&action.sa_mask); > > sigaddset(&action.sa_mask,SIGUSR2); > > action.sa_flags= 0; > > > > sigaction(SIGUSR2,&action,NULL); > > > > for(;;) > > { > > printf("Wait for Signal \n"); > > sigsuspend(&action.sa_mask); > > printf("Signal arrived \n"); > > } > > return(0); > > > > } /* */ > > > > int S(int tId) > > { > > taskDelay(600); > > printf("before kill\n"); > > kill(tId, SIGUSR2); > > printf("after kill \n"); > > return(0); > > } > > > > void M() > > { > > int PId, SId; > > > > PId=taskSpawn("P",100,0,10000,P,0,0,0,0,0,0,0,0,0,0); > > SId=taskSpawn("S",100,0,10000,S,PId,0,0,0,0,0,0,0,0,0); > > exit(0); > > > > } > > /*==========================================================*/ > > /* */ > > > > -- > > |// > > (o o) > > ___________________________oOO__(_)__OOo________________________________ > > > > IMPORTANT ** Nouveau N° de fax ** IMPORTANT ** New fax number > > > > Jacques J. Skubich > > Departement d'Informatique - Blaise Pascal > > INSA de Lyon > > F-69621 VILLEURBANNE CEDEX > > > > Jacques.Skubich@insa-lyon.fr > > Tel/Phone : +33 (0)4 72 43 84 86 > > Facsimile : +33 (0)4 37 47 90 82 > > Mobile : +33 (0)6 99 61 25 18 > > > > ___________________________________________________________________________ > > > > "Si nous avons chacun un objet et que nous les échangeons, nous avons > > chacun > > un objet. Si nous avons chacun une idée et que nous les échangeons, nous > > avons > > chacun deux idées." Proverbe chinois. > > ___________________________________________________________________________ - -- |// (o o) ___________________________oOO__(_)__OOo________________________________ IMPORTANT ** Nouveau N° de fax ** IMPORTANT ** New fax number Jacques J. Skubich Departement d'Informatique - Blaise Pascal INSA de Lyon F-69621 VILLEURBANNE CEDEX Jacques.Skubich@insa-lyon.fr Tel/Phone : +33 (0)4 72 43 84 86 Facsimile : +33 (0)4 37 47 90 82 Mobile : +33 (0)6 99 61 25 18 ___________________________________________________________________________ "Si nous avons chacun un objet et que nous les échangeons, nous avons chacun un objet. Si nous avons chacun une idée et que nous les échangeons, nous avons chacun deux idées." Proverbe chinois. ___________________________________________________________________________ --------------------------- Newsgroups: comp.os.vxworks Subject: QMC HDLC driver for 860! Date: Thu, 24 Jan 2002 16:32:11 +0800 From: "kola" Organization: Changzhou Telecom Message-ID: Hi, Did anybody implement the QMC HDLC driver for 860? And can you tell me the maximum SS7 channels or LAPD channels you had implemented? I had some questions about it! Thanks a lot! Best Regards! KolaYe --------------------------- Newsgroups: comp.os.vxworks Subject: WindNet PPP ----establish connection fail ? Date: 24 Jan 2002 01:01:00 -0800 From: sardine_sz@163.net (sardine) Organization: http://groups.google.com/ Message-ID: hi folks, now i config WindNet PPP&PPPOE, after i finished it,i use EnterNet300 dial to PPPOE AC, but there is error as follow: - -------------------------------------------------------------- 01/24/2002 14:40:31 - Beginning Negotiation 01/24/2002 14:40:31 - E0xxx: ACNAME: [MyAC] 01/24/2002 14:40:31 - D0XXX: PPPUP: Phase ESTABLISH 01/24/2002 14:41:01 - D0XXX: PPPUP: [Out of the loop] fDone = [0] 01/24/2002 14:41:45 - E0038P: PPP Establish Failed - -------------------------------------------------------------- what's wrong ? any advice will be appriecrate !!! --------------------------- Newsgroups: comp.os.vxworks Subject: Re: FTP connection ok but ... Date: Thu, 24 Jan 2002 17:15:46 +0800 From: "wei.yang" Organization: Changzhou Telecom Message-ID: References: <2002123-164211-214208@foorum.com> Hi, It may be the problem of your FTP server. Change a host or FTP server and try. Good Luck - -- Wei.Yang "Leonid Rosenboim" wrote in message news:newscache$pznfqg$t9p$1@lnews.actcom.co.il... > I suggest you try ping with different packet sizes to check the driver. > The long timeout suggests that the FTP data transfer suffers from packet > loss. > > "Charly" wrote in message > news:2002123-164211-214208@foorum.com... > > > > dear All, > > > > > > I use a direct point to point cross cable from VxWorks target and WIN98 > > the host ping works and the target ping works. > > but there is a FTP problem : > > > > On The Shell target, I type > > ->ls > > After a 10 sec timout, the shell display > > value = -1 > > On the Host, the FTP server displays: > > > > [L 4610] 01/23/02 10:02:02 Connection accepted from 192.0.1.2 > > [C 4610] 01/23/02 10:02:02 Command "USER target" received > > [C 4610] 01/23/02 10:02:02 PASSword accepted > > [L 4610] 01/23/02 10:02:02 User target logged in. > > [C 4610] 01/23/02 10:02:02 Command "TYPE I" received > > [C 4610] 01/23/02 10:02:02 TYPE set to I N > > [C 4610] 01/23/02 10:02:02 Command "PORT 192,0,1,2,4,11" received > > [C 4610] 01/23/02 10:02:02 PORT set to 192.0.1.2 - 1035 (4,11) > > [C 4610] 01/23/02 10:02:02 Command "RETR intru" received > > [C 4610] 01/23/02 10:02:02 RETR failed - file does not exist > > [G 4610] 01/23/02 10:02:02 RETR failed - file does not exist > > [C 4610] 01/23/02 10:02:02 QUIT or close - user target logged out > > [L 4618] 01/23/02 16:35:30 Connection accepted from 192.0.1.2 > > [C 4618] 01/23/02 16:35:30 Command "USER target" received > > [C 4618] 01/23/02 16:35:30 PASSword accepted > > [L 4618] 01/23/02 16:35:30 User target logged in. > > [C 4618] 01/23/02 16:35:30 Command "TYPE I" received > > [C 4618] 01/23/02 16:35:30 TYPE set to I N > > [C 4618] 01/23/02 16:35:30 Command "PORT 192,0,1,2,4,4" received > > [C 4618] 01/23/02 16:35:30 PORT set to 192.0.1.2 - 1028 (4,4) > > [C 4618] 01/23/02 16:35:30 Command "NLST" received > > [C 4618] 01/23/02 16:35:30 NLST started successfully for path/wildcard > > [C 4618] 01/23/02 16:35:50 Transfer aborted > > [G 4618] 01/23/02 16:35:50 Got file listing unsuccessfully > > [! 4618] 01/23/02 16:35:50 Timeout while sending data > > [C 4618] 01/23/02 16:35:50 QUIT or close - user target logged out > > [! -001] 01/23/02 16:35:50 Received a command on socket 40 which I can't > find > > > > It seems that the data sent on port 1028 is not seen by the target. > > Who could help me to solve this problem? > > Did it happen elsewhere? > > Charly > > > > -- > > Use our news server 'news.foorum.com' from anywhere. > > More details at: http://nnrpinfo.go.foorum.com/ > > --------------------------- Newsgroups: comp.os.vxworks Subject: NFS-problem Date: Thu, 24 Jan 2002 10:32:23 +0100 From: "Johan Bostedt" Organization: Mailgate.ORG Server - http://www.Mailgate.ORG Message-ID: Sender: johan.bostedt@racomna.se Hello VxWorkers, I have a problem with NFS (or more likely dosFs 1.0). My target system has a flash file system (TFFS) with dosFs 1.0 on top. The file system is exported to NFS. My target is a NFS server. From a Unix client I can mount the target file system through NFS and I can 'ls' the directory information. If I add a file on the TFFS file system from the target side and perform 'ls' from the client side the new file will appear and all is well. But if I remove a file on the TFFS file system from the target side and perform 'ls' from the client side the file seems to be there still... I can come around the problem if I disable NFS client caching. But I really don't like this since it degrades performance and has to be made on every client that will access the target TFFS. Any suggestions? Regards, Johan Bostedt - ---------------------------------------------------------------------- Johan Bostedt mailto:johan@racomna.se Phone: +46(0)18- 12 40 70 Racomna AB http://www.racomna.se Direct: +46(0)18- 12 40 95 Salagatan 16 A Mobile: +46(0)70-657 64 70 753 30 Uppsala Fax: +46(0)18- 12 40 74 SWEDEN --------------------------- Newsgroups: comp.os.vxworks Subject: Re: STL compilation problem Date: Thu, 24 Jan 2002 10:45:13 +0100 From: "Geurt Vos" Organization: Rohill Technologies BV Message-ID: References: <9ca84f0f.0201230229.4aab90d1@posting.google.com> > Dear gurus. > I am porting some prog. from NT to vxworks. There are some STL used. When I > compile the file I get the error : > no match for `__rb_tree_iterator,pair int,TerminalState *> &,pair *> & = void' > candidates are: __rb_tree_iterator *>,pair &,pair > *>::operator =(const __rb_tree_iterator > ,pair &,pair *> &) > > I run Tornado 2.0.2 with toolchain for PPC603. the compiler is g++(gcc) and > reports version as "cygnus-2.7.2-960126 egcs-971225 tornado 2.0". > > > What is this error? How can I cope with it? Are there any workarounds? Any > ideas? > Could you post (part of) the code that use the STL? I guess you're using something like std::map, correct? Obviously some assignment is (according to GCC) invalid, but other than that this error message is rather useless without the code... Geurt --------------------------- Newsgroups: comp.os.vxworks Subject: infinite while loop at usrAppInit() Date: Thu, 24 Jan 2002 19:13:51 +0900 From: "jnick" Organization: ONSE Telecom Message-ID: Hi all. I have added infinite while loop at usrAppInit(void) under the vxWorks. My target board has ethernet and tcp/ip protocol stack by vxWorks so it has been answering for pinging from PC. When I add while(1); the target board does not work for ping. What is wrong? I guess that vxWorks kernel managing tasks in order to avoid infinite loop hanging. void usrAppInit (void) { #ifdef USER_APPL_INIT USER_APPL_INIT; /* for backwards compatibility */ #endif /* add application specific code here */ while(1); } from Jnick --------------------------- Newsgroups: comp.os.vxworks Subject: Re: tffs, tffsBootPutImage and booting from fallow area Date: Thu, 24 Jan 2002 11:30:03 +0100 From: "Michael Lawnick" Organization: Buergernetz Dillingen Message-ID: References: <3C4C4F80.3F96AD2C@siemenscomms.co.uk> <3C4D7D3F.FCEBE820@siemenscomms.co.uk> <3C4E9FB8.A98B303D@siemenscomms.co.uk> Reply-To: "Michael Lawnick" Sender: mlawnick@p5088e937.dip.t-dialin.net Hi Owain, > I have installed a memDrv and can boot from the created memory device......must > now implement the next Vxworks workaround; the fact the network device doesn't > get initialised if you boot from certain named (not tffs) device..... have you set the 'others' parameter with the name of your network device ? - -- Mit freundlichen Grüßen, Michael Lawnick ============================================== SOFTEC GmbH Tel +49-731-96600-0 Promenade 17 Fax +49-731-96600-23 D-89073 Ulm Michael Lawnick Germany lawnick@softec.de ============================================== "Owain Phillips" schrieb im Newsbeitrag news:3C4E9FB8.A98B303D@siemenscomms.co.uk... > Hi Andray, > > > > > > [AndrayK 2]. Did you set breakpoint in MTD's 'write' routine, to see > > if MTD actually receives 'write' requests with offsets higher then 0x100000 ? > > > > regards, Andray > > No; as I didn't 'have' to use the tffsPutBootImage routine; I have my own > routines that don't rely on tffs or cfisc driver. I have given up on the tffs > for putting my boot images in ROM and for booting from the fallow region. > > I have installed a memDrv and can boot from the created memory device......must > now implement the next Vxworks workaround; the fact the network device doesn't > get initialised if you boot from certain named (not tffs) device..... > > Oh well after working with WRS and tornado for the last year I don't expect > anything to run by the book. > > All the best, > Owain > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: spurious interrupt + cache Date: Thu, 24 Jan 2002 11:49:06 +0100 From: "Michael Lawnick" Organization: Buergernetz Dillingen Message-ID: References: <5df340e0.0201220441.6b3b3999@posting.google.com> <5df340e0.0201222130.1b5121b1@posting.google.com> Reply-To: "Michael Lawnick" Sender: mlawnick@p5088e937.dip.t-dialin.net Hi Ruchama, to switch off interrupt line, the machine must do two things: - - acknowledge the interrupt on the external H/W - - acknowledge the interrupt on the onboard/onchip interrupt controller. While the second action should be done already by your BSP, the first one must be coded in the driver (as BSP can't know what to do). What is actually to be done depends on the particular H/W: some need a simple read of a register, others a write of a 1bit, etc.(all on your external H/W). Your problem could be, that your writing is actually done after the acknowledgement of interrupt controller (often occurs on MV2xxx via VME), as some middleware delays. The problem is solved by writing the acknowledge value to the H/W and read the value back again. This way you 'force' the write, as it has to be finished before the re-read can be done. Your CPU can't proceed before the read cycle is done (in contrast to a write) and will have to wait. - -- Mit freundlichen Grüßen, Michael Lawnick ============================================== SOFTEC GmbH Tel +49-731-96600-0 Promenade 17 Fax +49-731-96600-23 D-89073 Ulm Michael Lawnick Germany lawnick@softec.de ============================================== "Ruchama Goldshmidt" schrieb im Newsbeitrag news:5df340e0.0201222130.1b5121b1@posting.google.com... > Hello Michael! > > Thanks for your answer, > sorry ,but what do you meen a "read-back of the register you write > your ACK to" in the interrupt handeler or in the shared memory driver? > > TIA, > best regards, > Ruchama. > > > "Michael Lawnick" wrote in message news:... > > Hi Ruchama > > > > sounds for the 'your CPU is faster than your bus' problem: > > > > When you acknowledge the interrupt on the slave hardware, your actual > > write-access seems to be done by some 'middle-ware', e.g. PCI-bridge, > > VME-bridge,... > > Your CPU has already left ISR and reenabled its interrupts, before the real > > access on your slave is done. This explains, why uncached acknowledges > > succeed: your CPU is slower now. > > Do a read-back of the register you write your ACK to and all should be fine. > > > > Another problem could be, that your device is mapped as cacheable, but that > > I beleave would lead to complete malfunction. > > > > -- > > Mit freundlichen Grüßen, > > Michael Lawnick > > ============================================== > > SOFTEC GmbH Tel +49-731-96600-0 > > Promenade 17 Fax +49-731-96600-23 > > D-89073 Ulm Michael Lawnick > > Germany lawnick@softec.de --------------------------- Newsgroups: comp.os.vxworks Subject: Re: infinite while loop at usrAppInit() Date: Thu, 24 Jan 2002 06:42:58 -0500 From: joe durusau Organization: Bellsouth.Net Message-ID: <3C4FF341.10D806CE@bellsouth.net> References: I don't remember the details, but the task in question probably has very high priority, and is therefore preventing tnettask from running. Why on earth would you add an empty infinite while loop there?? Try: while(1) { taskDelay(2); } and see if the system runs OK. Speaking only for myself, Joe Durusau jnick wrote: > Hi all. > > I have added infinite while loop at usrAppInit(void) under the vxWorks. > My target board has ethernet and tcp/ip protocol stack by vxWorks so > it has been answering for pinging from PC. > > When I add while(1); the target board does not work for ping. > What is wrong? I guess that vxWorks kernel managing tasks in order to > avoid infinite loop hanging. > > void usrAppInit (void) > { > #ifdef USER_APPL_INIT > USER_APPL_INIT; /* for backwards compatibility */ > #endif > > /* add application specific code here */ > > while(1); > > } > > from Jnick --------------------------- End of New-News digest ********************** From vxwexplo-errs@csg.lbl.gov Thu Jan 24 10:53:32 2002 From: "Downs,Stephen" Date: Thu Jan 24 10:53:34 PST 2002 Subject: Linking to 3rd Party Archive File I am trying to get a third-party-supplied archive file (.a) to link with my VxWorks downloadable app files. I have tried using settings, e.g. LIBS macro and linker options in the Build properties dialog box. They are ignored. How do I use a third-party library with Tornado? My environment is Pentium BSP / VxWorks 5.4 / Tornado 2.02 / Windows 2000 TIA, Steve Downs Avtec Inc. From vxwexplo-errs@csg.lbl.gov Fri Jan 25 04:03:16 2002 From: Vxworks Exploder Date: Fri Jan 25 04:03:19 PST 2002 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Fri Jan 25 04:03:12 PST 2002 Subject: Re: Caching, I/O Space memory mapping and DMA Subject: Re: FTP connection ok but ... Subject: Re: Signal handler executions Subject: Re: Error while downloading project on target??? Subject: Re: NFS-problem Subject: Re: infinite while loop at usrAppInit() Subject: Simulate Wind Web Server on PC Subject: How to configure RAM size Subject: Re: Acurate, high-precision and stable timing solutions, Need Help... Subject: Web server and TIME_WAIT Subject: Re: I2C driver Subject: Re: routing problem ??? Subject: Re: SMC driver Subject: Re: Web server and TIME_WAIT Subject: while(1) statement on usrAppInit() Subject: infinite loop with while statement? Subject: Why sysLib.c is not added into the project files set automatically when creating a new project Subject: Questions on NPT device driver or bugs in NPT IP ? Subject: Re: How to configure RAM size Subject: Re: Vxworks crash on deleting a memory pointer Subject: Re: NFS-problem Subject: Re: while(1) statement on usrAppInit() Subject: Re: TCP/IP Performance under VxWorks Subject: Re: while(1) statement on usrAppInit() Subject: Re: Creating Build Rules in Project Workspace Subject: Re: Why sysLib.c is not added into the project files set automatically when creating a new project Subject: Re: tffs, tffsBootPutImage and booting from fallow area Subject: Re: Vxworks crash on deleting a memory pointer Subject: Re: Questions on NPT device driver or bugs in NPT IP ? Subject: Re: I2C driver Subject: Re: infinite loop with while statement? ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: Caching, I/O Space memory mapping and DMA Date: Thu, 24 Jan 2002 12:20:49 +0100 From: "Michael Lawnick" Organization: Buergernetz Dillingen Message-ID: References: <8bG_7.327$qS6.799090@newsr2.u-net.net> Reply-To: "Michael Lawnick" Sender: mlawnick@p5088e937.dip.t-dialin.net Hi Arun, > >>I. See. We are enabling the MMU and we use cacheDmaMalloc only. So > one of our problem (cache validation/flush) solved. Is it? > > And what happens if we write code with cacheDmaMalloc and if we > disable MMU? look into docs/vxworks/ref/cacheLib.html how to write a drive that should work with all cache/MMU settings. > ULONG data = (*(ULONG *) address); > > Will it work? > No, use sysPciOutLong(), defined in sysALib.s It will do the swapping within the write command. HTH - -- Mit freundlichen Grüßen, Michael Lawnick ============================================== SOFTEC GmbH Tel +49-731-96600-0 Promenade 17 Fax +49-731-96600-23 D-89073 Ulm Michael Lawnick Germany lawnick@softec.de ============================================== "Arun Prasad" schrieb im Newsbeitrag news:e38d912c.0201221315.688e264d@posting.google.com... > Hi Micheal, > > Thank you for your comments/reply/clarifications. > > < different/additional opinions to your question. Here my 2 Cents: > >>direct EMail? No. It's not. > And yes, I am looking for additional/different opinions for my > questions/doubts. And sorry for replying at this much delay. > > < correct: > >>And now I got how to interpret it. Thanks. > > << This seems to be suspicious - any confusion possible ? > Either DMA or local buffer, both would be redundant (surely still > possible, but who wants to pay?) > >>No. that's correct only. Our ethernet card does some other > functionalities also and its bit costly too. > > < cacheDmaMalloc and wouldn't need to care. > >>I. See. We are enabling the MMU and we use cacheDmaMalloc only. So > one of our problem (cache validation/flush) solved. Is it? > > And what happens if we write code with cacheDmaMalloc and if we > disable MMU? > > And still I have a doubt on I/O access routines (sysInXXX and > sysOutXXX). > We use IBM PPC405GP which is Big Endian and our ethernet NIC card is > PCI based (as you know PCI is little- Endian). As I already mentioned > I find only simple assignment statements and a EIOIOSYNC macro in > sysInByte and sysOutByte routines like this > > > UCHAR data = (*(UCHAR *)address); > > EIEIO_SYNC; > > return data; > > Here if I change the assignment statement like the following one for > sysInDWord () > > ULONG data = (*(ULONG *) address); > > Will it work? > > This is where we feel, we have to swap the bytes before making > assignments to make it Endian-safe. (I mean endian conversion required > here.) > > Hoping for a clarification on this Endian Issue. > > Have a nice time. > Regards > Arun > > "Michael Lawnick" wrote in message news:... > > Hi Arund, > > > > as you posted your direct EMails with Dave here, I assume you want > > different/additional opinions to your question. Here my 2 Cents: > > > > As mentioned correctly by Dave, PPC always see's PCI as memory mapped, but > > this does not mean, that a PPC can't do PCI-I/O. It means that there is a > > memory space, where the PCI-bridge of PPC will translate all accesses into > > I/O-accesses and there are other spaces, that are interpreted in other ways. > > > > Your excerpt from sysPhysMemDesc is just _one_ entry and its correct: > > > > VM_STATE_MASK_VALID | VM_STATE_MASK_WRITABLE |VM_STATE_MASK_CACHEABLE > > |VM_STATE_MASK_GUARDED > > tells which flags you want to change (_MASK_) > > > > VM_STATE_VALID | VM_STATE_WRITABLE | VM_STATE_CACHEABLE_NOT > > |VM_STATE_GUARDED > > enables the entry and sets it writable, uncached and guarded - just as you > > need it. > > > > You write: > > > >>I'm sorry for the confusion. Our Card has its own DMA controller and > > > its own Buffer/memory. > > This seems to be suspicious - any confusion possible ? > > Either DMA or local buffer, both would be redundant (surely still possible, > > but who wants to pay?) > > On our PPC405, our scndNIC 's registers have to be set up through PCI, data > > is written to CPU-memory via DMA (by the NIC). As we have shut off MMU, I > > must always do invalidate before data reading and flush before telling the > > NIC that there's a new packet to transmit. > > If our MMU would be enabled, I could allocate memory with cacheDmaMalloc and > > wouldn't need to care. > > > > To complete your confusion about sysInByte, sysOutByte, ..., there are more > > commands: > > sysPciInByte, sysPciOutByte,... > > The trick is: > > the sysInXXX variant will write data as is, typically found on Intel arch. > > sysPciInXXX will make a swap before access. This is needed, as PPC is > > big-endian, but PCI is little-endian. Second issue are the instruction pipes > > of PPC. As you might do two successive accesses to different PCI-registers, > > but the result of the second one might be influenced by the first (e.g. > > first enable device via control register, then read from status register), > > it must be ensured, that the first instruction completed before the second. > > This is done by a sync instruction as you might see in sysALib.s > > > > -- > > Mit freundlichen Grüßen, > > Michael Lawnick > > ============================================== > > SOFTEC GmbH Tel +49-731-96600-0 > > Promenade 17 Fax +49-731-96600-23 > > D-89073 Ulm Michael Lawnick > > Germany lawnick@softec.de > > ============================================== > > > > "Arun Prasad" schrieb im Newsbeitrag > > news:e38d912c.0201091154.76631ff5@posting.google.com... > > > Dear Dave, > > > > > > Thank you for your detailed reply/clarifications. > > > > > > < > > was an x86 chip. > > > >>Yes. You are absolutely right. I DO NOT HAVE EXPERIENCE other than > > > x86, though I READ a bit on Z80 and Motorola 68000 series chips. > > > > > > < > > space and memory space on PPC: everything is mapped to memory space, > > > and there is no I/O port space. > > > < > > >>Is it so? I thought it's device/controller/card specific. If my > > > memory is correct, I've read from Goankar's Microprocessor book that > > > it is device-specific. But again that book dealt with 8085 and 8086 > > > processors. And previously I worked on Linux Driver for SCSI Raid > > > Controllers with i960 processors in them and I found I/O mapped and > > > memory mapped cards separately and accordingly all the device access > > > functions varied. > > > It raises one more question from me. Hope you can/will answer that > > > too. > > > What happens if I have a ethernet controller which has a processor > > > like intel's i960 (Hope this supports I/O mapped I/O) which supports > > > I/O mapped I/O and if that controller is fixed on PPC system? > > > Will PPC still map the device in to it's memory? > > > > > > << PCI space should indeed always be non-cacheable. But the system > > > cannot do this for you. You must set up the BATs or PTEs (see manual > > > re: sysBatDesc and sysPhysMemDesc) that cover the address range used > > > by the PCI bus to have the G (guarded) and I (cache-inhibited) flags. > > > Once you have done that, every card will be non-cacheable no matter > > > where in PCI space the system decides to auto config it. > > > >>Oh. Thanks for this information. But I haven't found sysBatDesc. But > > > I was able to find sysPhysMemDesc. Here I'm giving the snap shot of > > > it. > > > > > > (void *) PCI_MEMORY_START, > > > (void *) PCI_MEMORY_START, > > > PCI_MEMORY_MAP_END - PCI_MEMORY_START + 1, > > > VM_STATE_MASK_VALID | VM_STATE_MASK_WRITABLE | > > > VM_STATE_MASK_CACHEABLE | > > > VM_STATE_MASK_GUARDED, > > > VM_STATE_VALID | VM_STATE_WRITABLE | > > > VM_STATE_CACHEABLE_NOT | > > > VM_STATE_GUARDED > > > > > > Can you write me on how to interpret this? Because there are two > > > entries are there two records are there. In this example, there are > > > two PCI_MEMORY_START are there and correspondingly two different sets > > > of flags are associated with them, one with Guarded and and non > > > cacheable and the other one with cacheable and unguarded. I don't know > > > why there are two different entries are there. Can you let me know > > > this? > > > > > > <<> > If we directly access the device registers (sysInbyte() > > > sysOutByte()), do we need to invalidate the cache as we are accessing > > > HW registers? > > > < > > > > > >>Oops... This is confusing again. > > > From the file /* sysLib.c - IBM 405GP Walnut system-dependent routines > > > */ > > > I could able to find sysInByte() and sysOutByte() functions and > > > similar PCI access routines to read/write one byte/Word/Long. > > > But within the definitions sysInByte() and sysOutByte() everything > > > treated as memory mapped only.I could able to see simple assignment > > > statements and EIEIO_SYNC macros only. > > > I guess these are only wrapper functions which simulates I/O mapped > > > IO. Please do clarify me if I'm wong. > > > > > > << However I am very surprised to hear that you have an ethernet card > > > that has no DMA. > > > >>I'm sorry for the confusion. Our Card has its own DMA controller and > > > its own Buffer/memory. > > > > > > < > > the commands to the device registers that would cause it to access the > > > system RAM. > > > >>I've made a note of it. > > > > > > I have one more question. This is nothing to do with VxWorks or PPC. > > > Just having a curiosity to know whether the e-mail ID you have given > > > () is a valid one? If I send a mail, will > > > it reach you? > > > > > > Thank you so much for your reply. > > > > > > I would also like to thank Michael Lawnick (Lawnick@softec.de) for > > > his reply mail. > > > > > > < > > actively > > > fetch it, you shouldn't have any cache troubles when disabling cache > > > for PCI > > > address space of your NIC. > > > >>Oh. I. See. Yes. This is the scenario with our controller (It has > > > its own memory and it buffers.) > > > > > > Regards > > > Arun > > > -- > > > Arun Prasad > > > Software Engineer > > > Adaptec India Pvt. Limited > > > (A wholly owned subsidiary of Adaptec Inc., USA) > > > 6-3-1086, IV Floor, Vista Grand Towers, > > > Rajbhavan Road, Somajiguda, > > > Hydereabad - 500082 > > > Phone: (O):91-40-6661555/56/57/58 > > > (Ext):234 > > > Email: arun_prasad@adaptecin.com > > > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: FTP connection ok but ... Date: 24 Jan 2002 09:16:28 GMT From: Charly Message-ID: <2002124-101628-161535@foorum.com> References: <2002123-164211-214208@foorum.com> Thanks for this answer. I use the Tornado2 WFTPD on Win98 SE. I assume that FTP uses UDP and TCP. I now look for the parameters to increase in the PCPentium BSP. but which one? Regards Charly - -- Use our news server 'news.foorum.com' from anywhere. More details at: http://nnrpinfo.go.foorum.com/ --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Signal handler executions Date: Thu, 24 Jan 2002 13:19:18 +0100 From: "Michael Lawnick" Organization: Buergernetz Dillingen Message-ID: References: <15a87607.0201230101.670b6e87@posting.google.com> Reply-To: "Michael Lawnick" Sender: mlawnick@p5088e937.dip.t-dialin.net The signal will be executed if the task's priority allows it. It will be executed even if the task is pending, so the signal routine can do a semGive/msQSend/... on the semaphore/msgQ/... the task is pending and free it this way. HTH - -- Mit freundlichen Grüßen, Michael Lawnick ============================================== SOFTEC GmbH Tel +49-731-96600-0 Promenade 17 Fax +49-731-96600-23 D-89073 Ulm Michael Lawnick Germany lawnick@softec.de ============================================== "Ramesh" schrieb im Newsbeitrag news:15a87607.0201230101.670b6e87@posting.google.com... > Hi, > > I am reading the VxWorks Programmers Guide. It is given that "when a > task receives a signal and if that task registers a handler for that > signal the signal handler is executed on the task context". Also it is > mentioned even the task is blocked the handler will get executed like > that. > > If so what is the difference between a ISR and a signal handler. > > Suppose if a task of pri 100 receives a signal say SIGALARM and while > receiving it is in the pend queue ( assume ) coz of some task of pri > 99 pre-empts the task 100 and task 99 is using the CPU resource. Now > since the task receives teh signal does the handler for that signal > will get executed or the task 99 will execute. If handler gets > executed means.. then under which task context will it gets execute. > Kindly clear me. > > Thanks, > Ramesh --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Error while downloading project on target??? Date: Thu, 24 Jan 2002 13:15:31 +0100 From: "Michael Lawnick" Organization: Buergernetz Dillingen Message-ID: References: <7421c9fd.0201222012.335b07bb@posting.google.com> Reply-To: "Michael Lawnick" Sender: mlawnick@p5088e937.dip.t-dialin.net Hi Amit, if you download your image via FTP, you have to take in account that the complete transfer is done before the image is converted from ELF to the final binary. In short terms: when loading via FTP, the image may only be half of remaining memory, in your case about 60 - 63MB. - -- Mit freundlichen Grüßen, Michael Lawnick ============================================== SOFTEC GmbH Tel +49-731-96600-0 Promenade 17 Fax +49-731-96600-23 D-89073 Ulm Michael Lawnick Germany lawnick@softec.de ============================================== "Amit" schrieb im Newsbeitrag news:7421c9fd.0201222012.335b07bb@posting.google.com... > I have a target with memory 128 MB. My final image size is coming > around 76 MB. So while downloading the image on target, system hangs > up. I want to know whether there are certain size restrictions on > image size which you can download . > > > Thanks > > Ajay --------------------------- Newsgroups: comp.os.vxworks Subject: Re: NFS-problem Date: Thu, 24 Jan 2002 12:44:05 +0000 (UTC) From: David Laight Organization: BT Openworld Message-ID: <3C5001EA.CACF7132@l8s.co.uk> References: > From a Unix client I can mount the target file system through NFS and I can > 'ls' the directory information. If I add a file on the TFFS file system from > the target side and perform 'ls' from the client side the new file will > appear and all is well. But if I remove a file on the TFFS file system from > the target side and perform 'ls' from the client side the file seems to be > there still... Make ABSOLUTELY sure the two system clocks are synchronised. NFS probably uses the 'mtime' of the directory to determine whether it has changed. However it is likely to be comparing client and server times. David --------------------------- Newsgroups: comp.os.vxworks Subject: Re: infinite while loop at usrAppInit() Date: 24 Jan 2002 13:50:02 +0100 From: Reinhold Wirth Organization: WEB.DE Message-ID: <3c5002fa@netnews.web.de> References: <3C4FF341.10D806CE@bellsouth.net> Hi if you finish your usrAppInit() with "while(1);" your bootTask is never finished. It seems, that the priority of the bootTask is higher than the netTask. The netTask gets no time to respond. If you change the loop to: while(1)taskDelay(2); the netTask may run during the delay-time. joe durusau wrote: > I don't remember the details, but the task in question probably has >very high priority, and is therefore preventing tnettask from running. >Why on earth would you add an empty infinite while loop there?? >Try: > >while(1) >{ >taskDelay(2); >} > >and see if the system runs OK. > >Speaking only for myself, > >Joe Durusau > > >jnick wrote: > >> Hi all. >> >> I have added infinite while loop at usrAppInit(void) under the vxWorks. >> My target board has ethernet and tcp/ip protocol stack by vxWorks so >> it has been answering for pinging from PC. >> >> When I add while(1); the target board does not work for ping. >> What is wrong? I guess that vxWorks kernel managing tasks in order to >> avoid infinite loop hanging. >> >> void usrAppInit (void) >> { >> #ifdef USER_APPL_INIT >> USER_APPL_INIT; /* for backwards compatibility */ >> #endif >> >> /* add application specific code here */ >> >> while(1); >> >> } >> >> from Jnick > > - -- __________________________________________________________ News suchen, lesen, schreiben mit http://newsgroups.web.de --------------------------- Newsgroups: comp.os.vxworks Subject: Simulate Wind Web Server on PC Date: Thu, 24 Jan 2002 16:36:08 +0200 From: "Yaron Cohen" Message-ID: <3c501c91@news.012.net.il> Hi, I would like to ask for your help. I am building a web site using Wind Web Server 2.0 and I am new for Vxworks and Tornado. I would like to test it and work with it first on my PC. If this is possible ?, installing the web server on the PC. Can I use Tornado Prototyper for this purpose ? Thanks and have a nice day, Yaron --------------------------- Newsgroups: comp.os.vxworks Subject: How to configure RAM size Date: 24 Jan 2002 07:03:41 -0800 From: bryanramesh@yahoo.com (Bryan) Organization: http://groups.google.com/ Message-ID: Hi, I am using VxSIM on NT 4.0. THe Sim is taking 1.8MB as RAM capacity while running as a image. I want to have more than that capacity. Since it is running on top of Win NT i can take some more memory ( from original RAM 128k ). Is that configurable.. how to acheive it. Thanks Bryan --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Acurate, high-precision and stable timing solutions, Need Help... Date: 24 Jan 2002 08:22:56 -0800 From: gryffons@hotmail.com (Marc LeBlanc) Organization: http://groups.google.com/ Message-ID: <6087305d.0201240822.70e20a11@posting.google.com> References: <6087305d.0201221146.7cacd1fc@posting.google.com> <3C4EE9F1.E52D7842@boeing.com> At first sight, this is exacly what I need. And plus they seem to have vxWorks drivers. I'm going to need some more information about its interfaces and modes of operation... But I'll email them to get that information. Thanks for the reply, Marc > What might work for you is something like the Datum 635PCI card. This card has > it's own disciplined oscillator that is synchronized to an external source > (GPS(637PCI), IRIGB, or 1PPS). The card can then be programmed to generate an > interrupt on a periodic basis or at a specific time. If your GPS receiver can > output IRIGB that is best, but a 1PPS signal will work also (or you could go with > the 637PCI model and combine your GPS/Timing function into one card). > > The card also lets you compensate for any latency or bias in your timing source. > With a good time source (GPS), you should have no problem getting < 1ms of > absolute accuracy out of your system. > > http://www.datum.com/test_blp.html > > Craig > --------------------------- Newsgroups: comp.os.vxworks Subject: Web server and TIME_WAIT Date: Thu, 24 Jan 2002 11:34:37 -0600 From: Dinker Charak Organization: Fermi National Accelerator Laboratory Message-ID: <3C5045AD.3D30BE92@fnal.gov> Hi, We have a small web server. When we access it thru a browser, and hit reload/refresh 14 times, it freezes for some time and then does a refresh. On the vxworks side, if I do inetstatShow, I see many sockets in TIME_WAIT state. The allocated memory is used up and it waits for the previous sockets to close and then able to open more. Now, WindSurf tells me that TIME_WAIT ~ 1 min which is better than stringent 2 min recommended by BSD4.4. So what is the way out? Turning so_linger.l_onoff = 1; and so_linger.l_linger = 0; does not help as it seems then it is unable to collect data and closes tself too fast. Did anyone also face this problem? Any other work around? Can we avoid this freezing? Wonder why is it not so apparent over internet? apache it seems, has TIME_WAIT ~ 30 secs. Any pointer, info or help will be much appreciated. Regards, Dinker --------------------------- Newsgroups: comp.os.vxworks Subject: Re: I2C driver Date: 24 Jan 2002 12:12:45 -0800 From: vloscomp@yahoo.com (Vinh) Organization: http://groups.google.com/ Message-ID: References: <3C4D9347.43AA0E67@attbi.com> <76i38.17309$8e.1027293@news> <3c4e83ac$1@brateggebdc5.br-automation.co.at> "Werner Schiendl" wrote in message news:<3c4e83ac$1@brateggebdc5.br-automation.co.at>... > Hi, > > No, VxWorks IO system does not prevent concurrent ioctl's on the same driver > or device. > You must care for that in your driver. > > Moreover, it does not solve the issue with the timing requirements. > If the driver is doing it's work synchronously in task context, interrupting > it would disturbe the protocol on the wire. > > The latter problem could probably being worked around using proper > taskLock()/taskUnlock() pairs. taskLock() and taskUnlock() is a poor man choice for solving concurrency problem. The best way is to designed your driver model to handle concurrency. Lots of driver does not used taskLock and taskUnlock to solved concurrency, so you may want to take a close look at their structure. > > hth > Werner > > "Jim" wrote in message > news:76i38.17309$8e.1027293@news... > > This is a stupid question ... but does wrapping calls to i2c with ioctl > > provide concurency solution? > > > > I have not RTM ... just thought i'd ask ... > > > > Jim > > > > "cfk@pacbell.net" wrote in message > > news:bHg38.35006$Fn6.1618022548@newssvr21.news.prodigy.com... > > > Dear Tom: > > > I2C is a very simple two wire protocol with just data (SDA) and > clock > > > (SCL). It is certainly possible to interrupt an I2C driver in the midst > of > > > transmitting a byte. AFAIR, this needs to be done when SCL is in the low > > > state, but you can read the I2C specification available from Philips > under > > > the paragraph "Pulse Stretching". The key thing to making an I2C driver > > > multitasking safe is to ensure that the task switch occurs when the > clock > is > > > in the appropriate state. I suspect the author of this driver merely > didn't > > > put the appropriate logic in for that. It is fairly simple to implement, > it > > > just takes a day or so of head scratching and studying other > implementations > > > of "bit banging" I2C software. > > > > > > Charles > > > > > > "Tom Flynn" wrote in message > > > news:3C4D9347.43AA0E67@attbi.com... > > > > > > > > The I2C driver in the BSP that I'm using is only written for use > during > > > > boot. It is not multitasking safe (according to the author's > comments). > > > > We need to access the I2C during normal run time on the board that we > > > > are using. > > > > > > > > I have two questions. First, what would need to be done to make it > > > > usable during run time? > > > > > > > > Second, is there driver code somewhere that is multitasking safe that > I > > > > can look at? I have access to other BSP code, so if someone knows of > a > > > > BSP that includes an I2C driver that is multitasking safe, I would be > > > > interested in knowing which BSP that is. > > > > > > > > Thanks in advance. > > > > > > > > -Tom Flynn > > > > > > > > > > > > > > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: routing problem ??? Date: 24 Jan 2002 13:31:55 -0800 From: jzhang@drs.ca (Jason) Organization: http://groups.google.com/ Message-ID: References: <3C4F53E7.B4E3ABEF@dsto.defence.gov.au> You must configue the PC to make sure it is able to send packets back to your target. If you are running window, try the following command to add routes on your PC: route ADD 192.168.1.0 MASK x.x.x.x 192.168.1.3 METRIC 1 Regards, Tim Shaw wrote in message news:<3C4F53E7.B4E3ABEF@dsto.defence.gov.au>... > Have you configured the route on the pc so that it knows to send to 192.168.1.0 > when it wants to send to 210.118.156.0? > > it2xpert wrote: > > > Hi , vxworks seniors !!! > > I desinged the mpc850sr board for router... > > I implemented the scc2,scc3 for ethernet mode... > > > > I use the T2, vxworks 5.4... > > I want to implement the Layer 3 router..... > > > > 192.168.1.0 -----------------192.168.1.1 (scc3) > > > > (scc2) > > 210.118.156.240 -------------------210.118.156.0 > > > > I used the RIP v.1 > > Routing table is following.... > > > > -> routeShow > > > > ROUTE NET TABLE > > destination gateway flags Refcnt Use Interface > > ---------------------------------------------------------------------------- > > 0.0.0.0 210.118.156.254 3 0 21 cpm0 > > 192.168.1.0 192.168.1.1 101 0 0 cpm1 > > 210.118.156.0 210.118.156.240 101 0 0 cpm0 > > ---------------------------------------------------------------------------- > > > > ROUTE HOST TABLE > > destination gateway flags Refcnt Use Interface > > ---------------------------------------------------------------------------- > > 127.0.0.1 127.0.0.1 5 1 65 lo0 > > ---------------------------------------------------------------------------- > > value = 77 = 0x4d = 'M' > > -> > > > > I trid the ping test !!! > > My pc is connected the 192.168.1.0 network (192.168.1.3) > > ping 192.168.1.1 ----- OK, > > ping 210.118.156.240 ---- OK > > > > but i did't reach the 210.118.156.xxx point.... > > > > I tried ping to the target server console.... > > -> ping "210.118.156.xxx" > > this is OK.... > > > > what's wrong ???? --------------------------- Newsgroups: comp.os.vxworks Subject: Re: SMC driver Date: 24 Jan 2002 14:11:38 -0800 From: ntjiang@yahoo.com (Jason Jiang) Organization: http://groups.google.com/ Message-ID: <12116a88.0201241411.705dd3f5@posting.google.com> References: <12116a88.0201221830.591bea57@posting.google.com> <2a6118da.0201230241.1604435d@posting.google.com> <12116a88.0201230956.1b276d27@posting.google.com> <2a6118da.0201232119.5dd8368c@posting.google.com> Hi Rashid, > This symbol is included when you define INCLUDE_WDB_COMM_SERIAL. The > routine wdbSlipPktDevInit( ) initializes a SLIP packet device for WDB > agent. You should include this device if you want to WDB serial > connection. I checked out that I didn't define INCLUDE_WDB_COMM_SERIAL. I used Network for my WDB connection, instead of serial port. So I think that's why I couldn't find the symbol in my image. But the question is, if wdbSlipPktDevInit(), seems to me the only place to install the callback functions, is not in the image, how can the callback functions be installed? > > > 2) If I don't define INCLUDE_WDB, does it mean I cannot get and put > > chars? > > INCLUDE_WDB includes a software agent in your vxWorks image to support > Tornado tools. Now u can choose the type o0f connection u want between > target server on Host and WDB agent on target. For ex. > > INCLUDE_WDB_COMM_END for WDB END driver connection. > INCLUDE_WDB_COMM_NETWORK for WDB network connection. > INCLUDE_WDB_COMM_SERIAL for WDB serial connection.............etc. > Got you. But why callback function installation (part of SMC driver initialization) has something to do with WDB. I mean, if the vxworks image needs not to support WDB (Tornado), it may still needs to init the driver. Due to the two points, is there anywhere else to do the installation? Thanks a lot. Regards, Jason --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Web server and TIME_WAIT Date: Fri, 25 Jan 2002 09:16:20 +1030 From: Tim Shaw Organization: Defence Science and Technology Organisation Message-ID: <3C508EBB.B753F6DF@dsto.defence.gov.au> References: <3C5045AD.3D30BE92@fnal.gov> We had a similar problem with an FTP-like server. We found that sockets could stay in the TIME_WAIT state for > 15 minutes, no-one could say why. So we just up-ed the number of sockets that we could have open at a time, which sorked for our situation - it may not work for yours. Dinker Charak wrote: > Hi, > > We have a small web server. When we access it thru a browser, and hit > reload/refresh 14 times, it freezes for some time and then does a > refresh. > > On the vxworks side, if I do inetstatShow, I see many sockets in > TIME_WAIT state. The allocated memory is used up and it waits for the > previous sockets to close and then able to open more. > > Now, WindSurf tells me that TIME_WAIT ~ 1 min which is better than > stringent 2 min recommended by BSD4.4. > > So what is the way out? Turning so_linger.l_onoff = 1; and > so_linger.l_linger = 0; does not help as it seems then it is unable to > collect data and closes tself too fast. > > Did anyone also face this problem? Any other work around? Can we avoid > this freezing? Wonder why is it not so apparent over internet? apache it > seems, has TIME_WAIT ~ 30 secs. > > Any pointer, info or help will be much appreciated. > > Regards, > Dinker --------------------------- Newsgroups: comp.os.vxworks Subject: while(1) statement on usrAppInit() Date: Thu, 24 Jan 2002 15:15:25 +0900 From: "jnick" Organization: ONSE Telecom Message-ID: Hi all I have a question for while statement at userAppInit function. void usrAppInit (void) { #ifdef USER_APPL_INIT USER_APPL_INIT; /* for backwards compatibility */ #endif /* add application specific code here */ printf("start usrAppInit\n"); while(1); } My target board has ethernet and TCP/IP protocol stack using vxWorks. Without while(1); my target board well answering my pinging from host PC. But with while(1); target board is not answering for pining. I guess that the kernel was broken or hanging at while(1); statment. Teach me please. From Nick --------------------------- Newsgroups: comp.os.vxworks Subject: infinite loop with while statement? Date: Thu, 24 Jan 2002 15:33:03 +0900 From: "jnick" Organization: ONSE Telecom Message-ID: Hi all~! I have added while loop on function usrAppInit(). Unfortunately my target board could not answering for pinging. Of course my target board has ethernet and tcp/ip protocol stacks using vxWorks. I guess that while(1); could not affect to vxWorks kernel's task scheduling but it can not answering for pining from PC. What is wrong? from nick /*************************************************************************** *** * * usrAppInit - initialize the users application */ void usrAppInit (void) { #ifdef USER_APPL_INIT USER_APPL_INIT; /* for backwards compatibility */ #endif /* add application specific code here */ printf("start usrAppInit\n"); goApp(); while(1); } --------------------------- Newsgroups: comp.os.vxworks Subject: Why sysLib.c is not added into the project files set automatically when creating a new project Date: 24 Jan 2002 19:02:32 -0800 From: lihuapeng@21cn.com (DragonSpring) Organization: http://groups.google.com/ Message-ID: <1d411918.0201241902.5625e4cb@posting.google.com> We use Tornado II/VxWorks 5.4 for MPC860. The host OS is NT 4.0 + patch 6. After we modified the original ADS860 BSP, when we create a new bootable project, the sysLib.c is not added into the project file list automatically, we have to add it manually. (right click and select Add Files...) But if we use the original ADS860 BSP, sysLib.c will be added automatically. Why? --------------------------- Newsgroups: comp.os.vxworks Subject: Questions on NPT device driver or bugs in NPT IP ? Date: 24 Jan 2002 19:38:13 -0800 From: houcheng@ms17.hinet.net (houcheng) Organization: http://groups.google.com/ Message-ID: Hi all: My development environment is as following: Tornado 2.0 for PPC Tornado 2.0 BSP ads860 Cumulative Patch T2CP4 I am writing a NPT driver for my ATM network card. The device driver is a IP over ATM network card, and its upper layer is IP protocol. Its protocol stack is as following: - -> muxShow Device: cpm Unit: 0 Description: MPC860 Power-QUICC Enhanced Network Driver Protocol: IP 4.4 ARP Type: 2054 Recv 0xa3cb0 Shutdown 0xa40a8 Protocol: IP 4.4 TCP/IP Type: 2048 Recv 0xa3cb0 Shutdown 0xa3fc0 Device: sar Unit: 0 Description: END SAR Driver. Protocol: rtATMARP Type: 2085 Recv 0x24e48 Shutdown 0x24fa4 Protocol: IP 4.4 TCP/IP Type: 2048 Recv 0xa3ecc Shutdown 0xa4150 I found that the tNetwork task and my ARP task has errno 0x3d0001: S_objLib_OBJ_ID_ERROR. - ->i tNetTask netTask 762bc8 50 PEND 7ec60 762b10 3d0001 0 tTelnetd telnetd 74a108 55 PEND 7ec60 749f68 0 0 tFtpdTask 8bee8 748b18 55 PEND 7ec60 7489b8 0 0 tPortmapd portmapd 74b6d0 100 PEND 7ec60 74b550 16 0 ATMARP_TASKAtmarpTask 3097d8 150 DELAY 82f30 309758 3d0001 9 I start to ping the "sar0" interface with 30 sessions concurrently, all reponses ok. But after one hour latter, the ping timeouts suddenly. 1. ping to "cpm0" does not response. ( cpm0 is vxworks provide END driver ) 2. arp to "cpm0" do response arp packet back. 3. ping to "sar0" does not response. 4. arp to "sar0" < my rtATMARP protocol > does reponse arp packet back. 5. ping to "sar0" does not response, "sar0" receiving packet well and pass it to MUX, but IP layer does not send packet back. 6. The END's network pool still works fine. When "sar0" receiving a IP frame, a cluster is acquired to store the frame. After pass the mBlk to MUX (IP layer), this cluster is freed to END's network pool. Is anything wrong with my END driver or any problems in IP layer ? Any helps or comments are welcome and thanks for your help. Lin Houcheng houcheng@turbonetworks.com.tw --------------------------- Newsgroups: comp.os.vxworks Subject: Re: How to configure RAM size Date: 24 Jan 2002 23:00:48 -0800 From: tut_pharoh@yahoo.com (tutankhamun) Organization: http://groups.google.com/ Message-ID: <88c8dc81.0201242300.5a71ddee@posting.google.com> References: Start VxSim from a command line with /r parameter vxWorks.exe /r16777215 this will theoretically give your VxSim around 16Mb to play with. tutankhamun =========================================================================== bryanramesh@yahoo.com (Bryan) wrote in message news:... > Hi, > > I am using VxSIM on NT 4.0. THe Sim is taking 1.8MB as RAM capacity > while running as a image. I want to have more than that capacity. > Since it is running on top of Win NT i can take some more memory ( > from original RAM 128k ). Is that configurable.. how to acheive it. > > Thanks > Bryan --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Vxworks crash on deleting a memory pointer Date: 24 Jan 2002 23:24:46 -0800 From: Sajith.Satheesan@geind.ge.com (Sajith Satheesan) Organization: http://groups.google.com/ Message-ID: <9b5debcd.0201242324.761bb59a@posting.google.com> References: <9b5debcd.0201220702.292df15d@posting.google.com> Bill Pringlemeir wrote in message news:... > >>>>> "Sajith" == Sajith Satheesan writes: > > Sajith> Hi All, I am having a page fault beging generated when I > Sajith> delete a pointer to an object in Vxworks. > > [snip] > > Sajith> destructor -> builtin_delete -> free -> dllRemove -> page > Sajith> fault. > > Sajith> other details are > > Sajith> Page Fault Program Counter: 0x0022bcf8 Status Register: > Sajith> 0x00010286 Error Code: 0x00000002 > > Sajith> Can anyone tell me what is causing this pagefault ? > > If you pass the destructor a bad pointer, then this will happen. It > is most likely some sort of over-write situation. Unfortunately > people often think that `bottom dwellers' like dllRemove() are > responsible. I wrote a custom function for vxMemProbe() and now every > time anyone in my company uses a bad pointer, I get a call... > dllRemove() is just trying to take the memory from a list. Since it > is crashing, I would guess that the memory before the object had been > over-written. The allocator stores its own data there. > > If possible, set a breakpoint on the destructor call and look at the > pointer values and memory before the object. Is it sane? Ie, does > the pointer point to RAM. Are the 16 bytes before the pointer the > same as when allocated (in the constructor)? You may also wish to tell > us more of your system (BSP, Tornado/vxWorks version, and CPU) if you > need further help. > > hth, > Bill Pringlemeir. Hi Bill, Well, we use ACE Libraries for building our software. This crash did not occur with an earlier version of ACE which is ACE519 It has started to happen only with ACE5117. Which means there is nothing logically wrong with my code. I also had a look at the pointer values on construction and desctruction. There was nothing abnormal about it. Can u give me more details about working of the _dllRemove method. The details of my system are x86 Family 6 Model 8 Stepping 3. and Vxworks version is 5.4 Tornado version is 2.0 --------------------------- Newsgroups: comp.os.vxworks Subject: Re: NFS-problem Date: Fri, 25 Jan 2002 09:46:38 +0200 From: "Leonid Rosenboim" Organization: Verio Message-ID: References: David is right, the client will compare directory names (cache against server) before it will assume the directory is unchanged and use cached handles, but dosFs 1.0 does not update directory modification time when a file is deleted, In other words, dosFs 1.0 and its NFS server can offer no solution to this situation. - - Leonid "Johan Bostedt" wrote in message news:a2okbd$61o$1@newsreader.mailgate.org... > Hello VxWorkers, > > I have a problem with NFS (or more likely dosFs 1.0). My target system has a > flash file system (TFFS) with dosFs 1.0 on top. The file system is exported > to NFS. My target is a NFS server. > > From a Unix client I can mount the target file system through NFS and I can > 'ls' the directory information. If I add a file on the TFFS file system from > the target side and perform 'ls' from the client side the new file will > appear and all is well. But if I remove a file on the TFFS file system from > the target side and perform 'ls' from the client side the file seems to be > there still... > > I can come around the problem if I disable NFS client caching. But I really > don't like this since it degrades performance and has to be made on every > client that will access the target TFFS. > > Any suggestions? > > > Regards, > Johan Bostedt > ---------------------------------------------------------------------- > Johan Bostedt mailto:johan@racomna.se Phone: +46(0)18- 12 40 70 > Racomna AB http://www.racomna.se Direct: +46(0)18- 12 40 95 > Salagatan 16 A Mobile: +46(0)70-657 64 70 > 753 30 Uppsala Fax: +46(0)18- 12 40 74 > SWEDEN > > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: while(1) statement on usrAppInit() Date: Fri, 25 Jan 2002 09:48:37 +0200 From: "Leonid Rosenboim" Organization: Verio Message-ID: References: The usrAppInit() function must not enter infinite loops, because it would stop the remainder of system initialization process. Instead, you should put the loop info a separate function, and from usrAppInit() you spawn a task dedicated to that functuion. You can't monopolize the Root task. "jnick" wrote in message news:a2o904$p17$1@hiline.shinbiro.com... > Hi all > I have a question for while statement at userAppInit function. > > void usrAppInit (void) > { > #ifdef USER_APPL_INIT > USER_APPL_INIT; /* for backwards compatibility */ > #endif > > /* add application specific code here */ > printf("start usrAppInit\n"); > > while(1); > } > > > My target board has ethernet and TCP/IP protocol stack using vxWorks. > Without while(1); my target board well answering my pinging from host PC. > But with while(1); target board is not answering for pining. I guess that > the kernel was broken or hanging at while(1); statment. > Teach me please. > > From Nick > > > > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: TCP/IP Performance under VxWorks Date: Fri, 25 Jan 2002 08:14:24 GMT From: "Gary M" Organization: AT&T Broadband Message-ID: References: <3C4F024F.5324B28C@delta-info.com> A lot also depends on what you actually do with the data packets. 30 Mbits/sec isn't much until you consider what you will be doing with it. We looked at certain VOIP architectures using MPC8xx family parts, and decided on a different architecture based on performance of the VxWorks stack and the interface to the codec DSPs we were interested in using. "Paul Randall" wrote in message news:3C4F024F.5324B28C@delta-info.com... > Hi all, > > I have an application where I need to send or receive up to 30 Mbits/sec > of sustained UDP data over Ethernet. I am evaluating different RTOSes > and processors to come up with an architecture for our application. > > Does anyone have TCP/IP performance numbers for VxWorks (or other RTOSes > for that matter) running on VxWorks supported processors? > > Also, does the ability to DMA data from the processor to an Ethernet > controller make much of a difference in performance? I have been > looking at the SMSC 91C111 ethernet controller which does not require a > PCI bus but it also does not support DMA transfers. > > Thanks in advance, > > Paul Randall > Delta Information Systems > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: while(1) statement on usrAppInit() Date: Fri, 25 Jan 2002 08:28:07 GMT From: "Gary M" Organization: AT&T Broadband Message-ID: References: userAppInit() runs in the context of the root task. "while (1);" will consume all of the processor at the task processing level. So it is no surprise that your board won't respond to pings. Why would you have the "while (1);" statement anyway? It's not doing anything useful. "jnick" wrote in message news:a2o904$p17$1@hiline.shinbiro.com... > Hi all > I have a question for while statement at userAppInit function. > > void usrAppInit (void) > { > #ifdef USER_APPL_INIT > USER_APPL_INIT; /* for backwards compatibility */ > #endif > > /* add application specific code here */ > printf("start usrAppInit\n"); > > while(1); > } > > > My target board has ethernet and TCP/IP protocol stack using vxWorks. > Without while(1); my target board well answering my pinging from host PC. > But with while(1); target board is not answering for pining. I guess that > the kernel was broken or hanging at while(1); statment. > Teach me please. > > From Nick > > > > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Creating Build Rules in Project Workspace Date: 25 Jan 2002 08:32:13 GMT From: Johan Borkhuis Organization: Agere Systems Message-ID: References: amanullah_khan@hotmail.com (Amanullah Khan) wrote: > I am working in Tornado 2.0 with vxWorks 5.4. I have some queries > related to the project facility. Can you tell me, from where the > properties of Build specification is got?. Acutally in the build > specification RULES only vxWorks, vxWorks_rom and > vxWorks_romCompress, vxWorks_romResident are listed. Why other builds > is not listed?. If other rules has to be displayed then, what should > be done for that. I tried creating a new rule in the Build > Specifications. The newly created rule is coming as Custom Rule. How > to make it as a normal rule ie like vxWorks. The custon rules are just like the normal rules. You can create the actions to be taken "Create or Edit rule" window. You can enter all the dependencies and the commands, and they will appear in the Makefile just like you entered them. Groeten, Johan - -- o o o o o o o . . . _____________________________ o _____ || Johan Borkhuis | .][__n_n_|DD[ ====_____ | borkhuis@agere.com | >(________|__|_[_________]_|__________________________| _/oo OOOOO oo` ooo ooo 'o!o!o o!o!o` === VxWorks FAQ: http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html === --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Why sysLib.c is not added into the project files set automatically when creating a new project Date: Fri, 25 Jan 2002 08:42:34 GMT From: "Gary M" Organization: AT&T Broadband Message-ID: <_T848.3636$b14.13915@rwcrnsc54> References: <1d411918.0201241902.5625e4cb@posting.google.com> When you perform a clean build in the BSP directory, it should create a new "depend." file, which I believe is parsed by Tornado when the project is created. There should be an entry in that file starting with syslib.o and listing its dependencies. If a clean build doesn't accomplish this, there's a serious problem with your BSP which should be fixed before creating new projects. "DragonSpring" wrote in message news:1d411918.0201241902.5625e4cb@posting.google.com... > We use Tornado II/VxWorks 5.4 for MPC860. The host OS is NT 4.0 + > patch 6. > > After we modified the original ADS860 BSP, when we create a new > bootable project, the sysLib.c is not added into the project file list > automatically, we have to add it manually. (right click and select Add > Files...) > > But if we use the original ADS860 BSP, sysLib.c will be added > automatically. > > Why? --------------------------- Newsgroups: comp.os.vxworks Subject: Re: tffs, tffsBootPutImage and booting from fallow area Date: Fri, 25 Jan 2002 09:39:52 +0100 From: "Werner Schiendl" Message-ID: <3c5119da$1@brateggebdc5.br-automation.co.at> References: <3C4C4F80.3F96AD2C@siemenscomms.co.uk> <3C4D7D3F.FCEBE820@siemenscomms.co.uk> <3C4E9FB8.A98B303D@siemenscomms.co.uk> Hi, "Michael Lawnick" wrote in message news:a2onng$3gm$1@snoopy.bndlg.de... > Hi Owain, > > > I have installed a memDrv and can boot from the created memory > device......must > > now implement the next Vxworks workaround; the fact the network device > doesn't > > get initialised if you boot from certain named (not tffs) device..... > > have you set the 'others' parameter with the name of your network device ? > This does not help, if you use a boot device that is not one of (scsi | ide | ata | fd). You need to change $(WIND_BASE)/target/config/comps/src/net/usrNetBoot.c change the function void usrNetDevNameGet (void) as requrired. hth Werner --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Vxworks crash on deleting a memory pointer Date: Fri, 25 Jan 2002 09:45:33 +0100 From: "Werner Schiendl" Message-ID: <3c511b2f$1@brateggebdc5.br-automation.co.at> References: <9b5debcd.0201220702.292df15d@posting.google.com> <9b5debcd.0201242324.761bb59a@posting.google.com> Hi, Don't be fooled by the fact that it did work 'til now. Keep in mind, that memory allocation for x86 always rounds up (of course) to the next full 8 bytes. Maybe you had a structure with 12 bytes, but wrote 2 or so bytes past the end? Consider the structure has changes and is 16 bytes now, if you still continue to write 2 bytes past its end - guess what happens? Right: You overwrite the memory allocation header of the _subsequent_ block of memory. I'd recommend to set a memory breakpoint to the location of the memory block header that crashes when freed. Study the shell's bh command help for that. hth Werner "Sajith Satheesan" wrote in message news:9b5debcd.0201242324.761bb59a@posting.google.com... > Bill Pringlemeir wrote in message news:... > > >>>>> "Sajith" == Sajith Satheesan writes: > > > > Sajith> Hi All, I am having a page fault beging generated when I > > Sajith> delete a pointer to an object in Vxworks. > > > > [snip] > > > > Sajith> destructor -> builtin_delete -> free -> dllRemove -> page > > Sajith> fault. > > > > Sajith> other details are > > > > Sajith> Page Fault Program Counter: 0x0022bcf8 Status Register: > > Sajith> 0x00010286 Error Code: 0x00000002 > > > > Sajith> Can anyone tell me what is causing this pagefault ? > > > > If you pass the destructor a bad pointer, then this will happen. It > > is most likely some sort of over-write situation. Unfortunately > > people often think that `bottom dwellers' like dllRemove() are > > responsible. I wrote a custom function for vxMemProbe() and now every > > time anyone in my company uses a bad pointer, I get a call... > > dllRemove() is just trying to take the memory from a list. Since it > > is crashing, I would guess that the memory before the object had been > > over-written. The allocator stores its own data there. > > > > If possible, set a breakpoint on the destructor call and look at the > > pointer values and memory before the object. Is it sane? Ie, does > > the pointer point to RAM. Are the 16 bytes before the pointer the > > same as when allocated (in the constructor)? You may also wish to tell > > us more of your system (BSP, Tornado/vxWorks version, and CPU) if you > > need further help. > > > > hth, > > Bill Pringlemeir. > > > Hi Bill, > > Well, we use ACE Libraries for building our software. This crash did > not occur with an earlier version of ACE which is ACE519 It has > started to happen only with ACE5117. Which means there is nothing > logically wrong with my code. I also had a look at the pointer values > on construction and desctruction. There was nothing abnormal about it. > Can u give me more details about working of the _dllRemove method. The > details of my system are x86 Family 6 Model 8 Stepping 3. > and Vxworks version is 5.4 Tornado version is 2.0 --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Questions on NPT device driver or bugs in NPT IP ? Date: 25 Jan 2002 08:47:51 GMT From: Johan Borkhuis Organization: Agere Systems Message-ID: References: houcheng@ms17.hinet.net (houcheng) wrote: > I found that the tNetwork task and my ARP task has errno > 0x3d0001: S_objLib_OBJ_ID_ERROR. You should only look ar errno when you get an ERROR reply from a function that defines errno. In this case some internal function returned an error and the calling function forgot to clear it. > I start to ping the "sar0" interface with 30 sessions > concurrently, all > reponses ok. But after one hour latter, the ping timeouts suddenly. Is the netTask still running? Groeten, Johan - -- o o o o o o o . . . _____________________________ o _____ || Johan Borkhuis | .][__n_n_|DD[ ====_____ | borkhuis@agere.com | >(________|__|_[_________]_|__________________________| _/oo OOOOO oo` ooo ooo 'o!o!o o!o!o` === VxWorks FAQ: http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html === --------------------------- Newsgroups: comp.os.vxworks Subject: Re: I2C driver Date: Fri, 25 Jan 2002 09:51:05 +0100 From: "Werner Schiendl" Message-ID: <3c511c7b$1@brateggebdc5.br-automation.co.at> References: <3C4D9347.43AA0E67@attbi.com> <76i38.17309$8e.1027293@news> <3c4e83ac$1@brateggebdc5.br-automation.co.at> "Vinh" wrote in message news:c129947c.0201241212.7dcbedf0@posting.google.com... > > taskLock() and taskUnlock() is a poor man choice for solving > concurrency problem. > > The best way is to designed your driver model to handle concurrency. > Lots of driver does not used taskLock and taskUnlock to solved > concurrency, so you may want to take a close look at their structure. > I think this completely depends on what the purpose is. The OP did not state what he wants to do with the I2C bus. Maybe just initialize some hardware just once - but cannot be done for some reason before starting multitasking? Of course a fully multi-tasking-aware driver is more elegant - but more work, too. But it's probably working fine with just a pair of taskLock() / taskUnlock calls. Or maybe even intLock, for that matter. regards Werner --------------------------- Newsgroups: comp.os.vxworks Subject: Re: infinite loop with while statement? Date: Fri, 25 Jan 2002 06:36:08 -0500 From: joe durusau Organization: BELLSOUTH.net Message-ID: <3C514328.212F3E55@bellsouth.net> References: Are you able to access the messages being posted to the newsgroup. Several people have pointed out to you that while(1); is GUARANTEED to hang the system. The root task runs at highest priority and blocks all other activity. To repeat a question asked several times before: Why on earth do you want to do this anyway???? Speaking only for myself, Joe Durusau jnick wrote: > Hi all~! > > I have added while loop on function usrAppInit(). Unfortunately > my target board could not answering for pinging. > Of course my target board has ethernet and tcp/ip protocol stacks using > vxWorks. > I guess that while(1); could not affect to vxWorks kernel's task scheduling > but it can not answering for pining from PC. > What is wrong? > > from nick > > /*************************************************************************** > *** > * > * usrAppInit - initialize the users application > */ > > void usrAppInit (void) > { > #ifdef USER_APPL_INIT > USER_APPL_INIT; /* for backwards compatibility */ > #endif > > /* add application specific code here */ > printf("start usrAppInit\n"); > > goApp(); > > while(1); > > } --------------------------- End of New-News digest ********************** From vxwexplo-errs@csg.lbl.gov Fri Jan 25 10:56:43 2002 From: Jaya Prakash Date: Fri Jan 25 10:56:46 PST 2002 Subject: Re:Web server and TIME_WAIT ( VxWorks ) This message is in MIME format. Since your mail reader does not understand this format, some or all of this message may not be legible. ------_=_NextPart_001_01C1A5D2.01306AB0 Content-Type: text/plain; charset="iso-8859-1" > -----Original Message----- > From: Jaya Prakash > Sent: Friday, January 25, 2002 10:34 AM > To: 'dinker@fnal.gov' > Subject: Re:Web server and TIME_WAIT > > Hi > > I don't see the problem with U R web server.Could U look at the net Stack > Pool may be the net stack is running out of memory. > at shell do this see ,how many times the net stack failled/attempted for > free space. > > -> netStackDataPoolShow > > --> netStackSysPoolShow > > > I belive U R system ( Sys) Pool is running out of memory. > > Reconfigure U R net stack and come up with a optimum value for mblks and > cblks for the configuration. > > I hope this would solve U R problem. > > rgrds, > JP > ############################################################################ ########################## This email communication may contain CONFIDENTIAL INFORMATION and is intended only for the use of the intended recipients identified above. If you are not the intended recipient of this communication, you must not use, disclose, distribute, copy or print this email. If you have received this communication in error, please immediately notify the sender by reply email, delete the communication and destroy all copies. ############################################################################ ########################## ------_=_NextPart_001_01C1A5D2.01306AB0 Content-Type: text/html; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable Re:Web server and TIME_WAIT ( VxWorks )

 -----Original Message-----
From:   Jaya Prakash 
Sent:   = Friday, January 25, 2002 10:34 AM
To:     'dinker@fnal.gov'
Subject:        Re:Web server and TIME_WAIT

Hi

I don't see the problem with U R web = server.Could U look at the net Stack Pool may be the net stack is = running out of memory.

at shell do this see ,how many times = the net  stack failled/attempted for free space.

-> netStackDataPoolShow

--> netStackSysPoolShow


I belive U R system ( Sys) Pool is = running out of memory.

Reconfigure U R net stack and come up = with a optimum value for mblks and cblks for the configuration.

I hope this would solve U R = problem.

rgrds,
JP

#########################################################= ############################################# This email communication = may contain CONFIDENTIAL INFORMATION and is intended only for the use = of the intended recipients identified above.  If you are not the = intended recipient of this communication, you must not use, disclose, = distribute, copy or print this email. If you have received this = communication in error, please immediately notify the sender by reply = email, delete the communication and destroy all copies. = ########################################################################= ##############################

------_=_NextPart_001_01C1A5D2.01306AB0-- From vxwexplo-errs@csg.lbl.gov Sat Jan 26 04:03:19 2002 From: Vxworks Exploder Date: Sat Jan 26 04:03:21 PST 2002 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Sat Jan 26 04:03:14 PST 2002 Subject: Re: infinite loop with while statement? Subject: Re: NFS-problem Subject: Re: rlogin error - Help Please... Subject: Re: compiling WTX application on WinNT, VC++ 6 Subject: Re: Memory lost during open,read, close with NFS and netDrv Subject: Re: Getting UART errors from i8250Sio driver Subject: Re: while(1) statement on usrAppInit() Subject: Loading Modules without File System Subject: Re: compiling WTX application on WinNT, VC++ 6 Subject: Re: Loading Modules without File System Subject: mkdir with ftp Subject: How to print the contents of packet header Subject: Using SPY to measure ISR on 8260 Subject: Re: Questions on NPT device driver or bugs in NPT IP ? Subject: Re: while(1) statement on usrAppInit() Subject: Re: Loading Modules without File System Subject: Re: errno 226 Subject: SEMAPHORE PROTECTION Subject: Re: SEMAPHORE PROTECTION Subject: CLI Parser for VxWorks Subject: getting task information in tylib.c - possible? Subject: Re: mkdir with ftp Subject: Re: Loading Modules without File System Subject: TSA and HDLC for xDSL under vxWorks Subject: Re: CLI Parser for VxWorks Subject: Re: getting task information in tylib.c - possible? - FOUND A SOLUTION Subject: What can I do to get the size of my executable for powerPc down? Subject: readers writers problem Subject: missing ftime and alarm?? Subject: Re: porting dbm onto vxworks Subject: Re: What can I do to get the size of my executable for powerPc down? Subject: Re: mkdir with ftp Subject: Re: Why sysLib.c is not added into the project files set automatically when creating a new project Subject: Problem in memory which allocated by tRootTask and using it by application ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: infinite loop with while statement? Date: Fri, 25 Jan 2002 14:36:42 +0100 From: "Werner Schiendl" Message-ID: <3c515f6c$1@brateggebdc5.br-automation.co.at> References: Hi, "jnick" wrote in message news:a2oa16$pe1$1@hiline.shinbiro.com... > > Hi all~! > > I have added while loop on function usrAppInit(). Unfortunately > my target board could not answering for pinging. > Of course my target board has ethernet and tcp/ip protocol stacks using > vxWorks. > I guess that while(1); could not affect to vxWorks kernel's task scheduling > but it can not answering for pining from PC. > What is wrong? > I guess your understanding of VxWorks' scheduling mechanism. VxWorks works different than (most?) desktop OS. It has a purely priority based scheduling system. The higher the priority (where a smaller number means higher priority, 0 is the highest, 255 the lowest) the more likely the task will run. VxWorks always runs the highest priority task that is currently ready. If a high priority task (like yours) runs forever, lower priority tasks (like tNetTask, which is responsible for networking) will NOT run AT ALL. VxWorks has no round-robin scheduling by default. (You can enable round-robin scheduling, but ONLY for tasks of the SAME PRIORITY). Does this clarify the confusion? hth Werner P.S.: Please don't post the same question to the group so many times! P.P.S.: Study the VxWorks programmers guide, it's definitely worth the time. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: NFS-problem Date: Fri, 25 Jan 2002 13:50:05 -0000 From: "Richard Horton" Organization: Newport Networks Ltd. Message-ID: <1011966617.722511@azores.network-i.net> References: Could you just add a single file with a temporary name to the directory after deleting the selected file then delete the temporary file? - -- All opinions, unless otherwise stated are my own and not those of my employer. Richard Horton Newport Networks Ltd. "Users are like a virus, each causing a thousand tiny crises until the host finally dies" "Leonid Rosenboim" wrote in message news:newscache$9cjhqg$fo2$1@lnews.actcom.co.il... > David is right, the client will compare directory names (cache against > server) before it > will assume the directory is unchanged and use cached handles, but dosFs 1.0 > does not > update directory modification time when a file is deleted, > > In other words, dosFs 1.0 and its NFS server can offer no solution to this > situation. > > - Leonid > > "Johan Bostedt" wrote in message > news:a2okbd$61o$1@newsreader.mailgate.org... > > Hello VxWorkers, > > > > I have a problem with NFS (or more likely dosFs 1.0). My target system has > a > > flash file system (TFFS) with dosFs 1.0 on top. The file system is > exported > > to NFS. My target is a NFS server. > > > > From a Unix client I can mount the target file system through NFS and I > can > > 'ls' the directory information. If I add a file on the TFFS file system > from > > the target side and perform 'ls' from the client side the new file will > > appear and all is well. But if I remove a file on the TFFS file system > from > > the target side and perform 'ls' from the client side the file seems to be > > there still... > > > > I can come around the problem if I disable NFS client caching. But I > really > > don't like this since it degrades performance and has to be made on every > > client that will access the target TFFS. > > > > Any suggestions? > > > > > > Regards, > > Johan Bostedt > > ---------------------------------------------------------------------- > > Johan Bostedt mailto:johan@racomna.se Phone: +46(0)18- 12 40 70 > > Racomna AB http://www.racomna.se Direct: +46(0)18- 12 40 95 > > Salagatan 16 A Mobile: +46(0)70-657 64 70 > > 753 30 Uppsala Fax: +46(0)18- 12 40 74 > > SWEDEN > > > > > > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: rlogin error - Help Please... Date: Fri, 25 Jan 2002 15:30:42 +0100 From: "Friedrich Ensslin" Message-ID: References: <3c4f596c_4@news3.uncensored-news.com> Hi, the first thing, you should check, is whether "ping" works (of course, it must be configured), because I suppose, that the problem is not rlogin, but the network connection itself. If you have another network interface on your boards, which you use for booting, the shared memory might still be uninitialized. To add the shared memory network to the boot parameters as "other" might help then. Have you checked the routing tables of your boards ? hth, Friedrich Ensslin ________________________________________ AM3 AutoMotive MultiMedia AG friedrich.ensslin@am3.com ________________________________________ "Bigg K" schrieb im Newsbeitrag news:3c4f596c_4@news3.uncensored-news.com... > Using VxWorks 5.4... > > To rlogin from board 0 to board 1 through shared memory, must I have > > operating system components/kernel components/shared memory objects > > installed or is > > network components/network devices/shared memory network components/ > auto address setup > shared memory as second interface > shared memory bootline parsing > shared memory network initialization > > sufficient for the needed shared memory-related components? > > Do any constants in config.h or configAll.h need changing? > > I'm getting the following error: > > rlogin: could not connect to host. > value = -1 = 0xffffffff = end + 0xfff1e223 > > Thanks for helping... > > > > ______________________________________________________________________ > Posted Via Uncensored-News.Com - Still Only $9.95 - http://www.uncensored-news.com > With NINE Servers In California And Texas - The Worlds Uncensored News Source > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: compiling WTX application on WinNT, VC++ 6 Date: Fri, 25 Jan 2002 14:36:31 GMT From: bill@woola.qswtools.com (William C. Cox) Organization: EarthLink Inc. -- http://www.EarthLink.net Message-ID: References: <3c4ca498$1_1@nopics.sjc> In article <3c4ca498$1_1@nopics.sjc>, "Cedric Le Roux" writes: > Anybody know a good place to get some detailed instructions on how to > compile a WTX application on Windows NT and Visual C++ 6? > The "Getting Started examples are only intended to work with the compiler that shipped with your copy of Tornado. Nobody that I know of has built binaries with VC++ and run them under Tornado. VC++ generates PE-COFF object files, and the VxWorks loaders (except for the NT simulator loader) cannot read it. Well, the I980 and ARM loaders can read it, but VC++ is generating Pentium code, isn't it? > I've gone through the Tornado API Guide. The sample code will not compile, --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Memory lost during open,read, close with NFS and netDrv Date: Fri, 25 Jan 2002 15:39:32 +0100 From: "Friedrich Ensslin" Message-ID: References: <363d071b.0201231632.48875909@posting.google.com> Hi, AFAIK, the problem is the shell, or at least the way, how the shell treats strings - in our case the filenames. The shell allocates memory for these strings, to pass the pointer to the "open" routine. After that, the memory will never be freed by the shell - the shell has no clue, how long the memory has to be valid, so it does not free the string's memory at all. Try the same from inside a C-program, and you will see that there is no memory leak. hth, Friedrich Ensslin ________________________________________ AM3 AutoMotive MultiMedia AG friedrich.ensslin@am3.com ________________________________________ "Vasu Dalal" schrieb im Newsbeitrag news:363d071b.0201231632.48875909@posting.google.com... > Hello all : > > I am trying to understand memory allocation/fragmentation wrt > netDrv/NFS in particular and with VxWorks in general. > > Can someone please explain after looking at the sequence of > file open's, read's and close's with both netDrv and NFS given > below as to what happened to the "48" bytes ? > I believe they are NOT lost but i cannot understand how > they can be accounted for. This will also help my understanding > of the memLib and some of its data structures. > > Also when using netDrv why is the size of the memory allocated > for a 2767 byte file 3424 bytes ? > > In the sequence below : > /CF => is a DOS FS device on a Vxworks target > anotherVxWorksHost: => is the host name for a VxWorks card > used as a host > cfg.txt => is a file containing some text > > I am using VxWorks version 5.4, Kernel Wind version 2.5 and > Processor is 8260 ( PPC arch ). > > Sorry for the long post and thank you for your time. > Any input is appreciated. > TIA. > > vasu > > PS: > > [snip1] > -> memShow > status bytes blocks avg block max block > ------ --------- -------- ---------- ---------- > current > free 11190848 18 621713 6815808 > alloc 7940016 3027 2623 - > cumulative > alloc 30594200 11345 2696 - > value = 0 = 0x0 > > -> open "anotherVxWorksHost:/CF/cfg.txt",2 <-- open using netDrv > value = 15 = 0xf > > -> memShow > status bytes blocks avg block max block > ------ --------- -------- ---------- ---------- > current > free 11187424 18 621523 6815808 > alloc 7943440 3036 2616 - <-- memory is > fragmented > by 9 more > blocks > cumulative > alloc 30603776 11361 2693 - > value = 0 = 0x0 > > -> 11190848-11187424 > value = 3424 = 0xd60 <-- memory allocated > is 3424 while the size of the file "cfg.txt" is 2767 > bytes > -> close 15 > value = 0 = 0x0 > > -> memShow > status bytes blocks avg block max block > ------ --------- -------- ---------- ---------- > current > free 11190800 18 621711 6815808 > alloc 7940064 3028 2622 - > cumulative > alloc 30615840 11384 2689 - > value = 0 = 0x0 > > -> 11190848-11190800 > value = 48 = 0x30 = '0' <-- Lost "48" bytes > during the whole sequence of open and close using > netDrv > -> > [snip1] > > > > [snip2] > -> memShow > status bytes blocks avg block max block > ------ --------- -------- ---------- ---------- > current > free 11190344 19 588965 6815808 > alloc 7940520 3033 2618 - > cumulative > alloc 30812704 11813 2608 - > value = 0 = 0x0 > > -> open "anotherVxWorksHost:/CF/cfgip.txt",0 <-- open using netDrv > value = 15 = 0xf > > -> memShow > status bytes blocks avg block max block > ------ --------- -------- ---------- ---------- > current > free 11186920 19 588785 6815808 > alloc 7943944 3042 2611 - <-- memory is > fragmented by > 9 more blocks > cumulative > alloc 30822240 11828 2605 - > value = 0 = 0x0 > -> > > -> 11190344-11186920 > value = 3424 = 0xd60 <-- memory allocated > is 3424 > while the size of the > file > "cfg.txt" is 2767 > bytes > -> > -> buf=malloc(1024) > buf = 0x4215a0: value = 25152736 = 0x17fcce0 > > -> memShow > status bytes blocks avg block max block > ------ --------- -------- ---------- ---------- > current > free 11185888 19 588730 6815808 > alloc 7944976 3043 2610 - <-- memory is > fragmented by > 1 more block > cumulative > alloc 30829424 11836 2604 - > value = 0 = 0x0 > -> > > -> 11186920-11185888 > value = 1032 = 0x408 <-- 1024 bytes+8 bytes > for > memLib/memPartLib data > structures > -> read 15,buf,1024 > value = 1024 = 0x400 > > -> memShow > status bytes blocks avg block max block > ------ --------- -------- ---------- ---------- > current > free 11185888 19 588730 6815808 <-- No change as > expected > alloc 7944976 3043 2610 - <-- No change in # of > blocks > as expected > cumulative > alloc 30838552 11844 2603 - > value = 0 = 0x0 > > -> free buf > value = 0 = 0x0 > > -> memShow > status bytes blocks avg block max block > ------ --------- -------- ---------- ---------- > current > free 11186920 19 588785 6815808 > alloc 7943944 3042 2611 - > cumulative > alloc 30845832 11850 2603 - > value = 0 = 0x0 > > -> 11186920-11185888 > value = 1032 = 0x408 <-- Recovered the 1032 > bytes > previously > malloc'ed > > -> close 15 > value = 0 = 0x0 > > -> memShow > status bytes blocks avg block max block > ------ --------- -------- ---------- ---------- > current > free 11190296 19 588962 6815808 > alloc 7940568 3034 2617 - > cumulative > alloc 30856128 11858 2602 - > value = 0 = 0x0 > -> > > -> 11190344-11190296 > value = 48 = 0x30 = '0' <-- Lost "48" bytes > during the > whole sequence of > open, read > and close using netDrv > > > > [snip2] > > > [snip3] > -> memShow > status bytes blocks avg block max block > ------ --------- -------- ---------- ---------- > current > free 11190296 19 588962 6815808 > alloc 7940568 3034 2617 - > cumulative > alloc 30862480 11870 2600 - > value = 0 = 0x0 > -> > -> open "/CF/cfgip.txt",0 <-- open using NFS > value = 15 = 0xf > > -> memShow > status bytes blocks avg block max block > ------ --------- -------- ---------- ---------- > current > free 11181808 19 588516 6815808 > alloc 7949056 3037 2617 - <-- memory is > fragmented > by 3 more > blocks > cumulative > alloc 30877080 11879 2599 - > value = 0 = 0x0 > > -> 11190296-11181808 > value = 8488 = 0x2128 <-- 8K memory+extra > for NFS since NFS works in 8192 > byte(8K) chunks > > -> buf=malloc(1024) > buf = 0x4215a0: value = 25146488 = 0x17fb478 > > -> memShow > status bytes blocks avg block max block > ------ --------- -------- ---------- ---------- > current > free 11180776 19 588461 6815808 > alloc 7950088 3038 2616 - > cumulative > alloc 30887400 11892 2597 - > value = 0 = 0x0 > > -> 11181808-11180776 > value = 1032 = 0x408 <-- 1024 bytes+8 bytes > for > memLib/memPartLib data > structures > -> read 15,buf,1024 > value = 1024 = 0x400 > > -> memShow > status bytes blocks avg block max block > ------ --------- -------- ---------- ---------- > current > free 11180776 19 588461 6815808 <-- No change as > expected > alloc 7950088 3038 2616 - <-- No change in # of > blocks as expected > cumulative > alloc 30896488 11899 2596 - > value = 0 = 0x0 > > -> free buf > value = 0 = 0x0 > > -> memShow > status bytes blocks avg block max block > ------ --------- -------- ---------- ---------- > current > free 11181808 19 588516 6815808 > alloc 7949056 3037 2617 - > cumulative > alloc 30902560 11904 2595 - > value = 0 = 0x0 > -> > -> 11181808-11181808 <-- Recovered the 1032 > bytes > previously > malloc'ed > value = 0 = 0x0 > > -> close 15 > value = 0 = 0x0 > > -> memShow > status bytes blocks avg block max block > ------ --------- -------- ---------- ---------- > current > free 11190248 19 588960 6815808 > alloc 7940616 3035 2616 - > cumulative > alloc 30908592 11908 2595 - > value = 0 = 0x0 > -> > -> 11190296-11190248 > value = 48 = 0x30 = '0' <-- Lost "48" bytes > during the > whole sequence of > open, read > and close using NFS > > -> > > > [snip3] > > > Keywords : > malloc, free, memory fragmentation, memory alignment, NFS, netDrv --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Getting UART errors from i8250Sio driver Date: Fri, 25 Jan 2002 16:01:38 +0100 From: "Friedrich Ensslin" Message-ID: References: <3C475BD6.F5E5A512@bitworkssystemsDELETETHIS.com> Hi, you are right, the error handling in the serial drivers is rudimentary. For your needs, you might add an error flag to the device's data structure. This can be filled in the error branch of the driver's ISR and read by dedicated ioctl function. Friedrich Ensslin ________________________________________ AM3 AutoMotive MultiMedia AG friedrich.ensslin@am3.com ________________________________________ "Jon" schrieb im Newsbeitrag news:3C475BD6.F5E5A512@bitworkssystemsDELETETHIS.com... > Now that I've been able to get my serial port set to the appropriate > protocol (8+even+1 stop), now I also need to get UART error codes > perferable on a character by character basis. Looking at the driver > source i8250Sio.c it appears the line status register where things like > parity errors, framing errors and overrun errors are reported is read by > the driver but then ignored. Am I reading this correctly? Am I going > to have to modify the driver to get access to UART error codes? Thanks > in advance. > > Jon Newbill > Bitworks Systems Inc. > jon@bitworkssystemsSPAMOUT.com > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: while(1) statement on usrAppInit() Date: Fri, 25 Jan 2002 08:08:52 -0800 From: "Michael R. Kesti" Organization: MK Associates Message-ID: <3C518314.442D41BD@gv.net> References: Reply-To: mkesti@gv.net jnick wrote: >Hi all >I have a question for while statement at userAppInit function. > >void usrAppInit (void) > { >#ifdef USER_APPL_INIT > USER_APPL_INIT; /* for backwards compatibility */ >#endif > > /* add application specific code here */ > printf("start usrAppInit\n"); > > while(1); > } > >My target board has ethernet and TCP/IP protocol stack using vxWorks. >Without while(1); my target board well answering my pinging from host PC. >But with while(1); target board is not answering for pining. I guess that >the kernel was broken or hanging at while(1); statment. >Teach me please. > >From Nick As its name inplies, usrAppInit is the place to initialize one's application rather than the place to run that application. You are supposed to start your application's tasks or call a function that starts those tasks and allow usrAppInit to return. If usrAppInit never returns, then the task scheduler is never executed and the network tasks never run. Also, a task that never pends, such as one that consists of while(1), must be at a priority that is lower (higher value) than any other task that one expects to run. If such a task has a higher priority than any other task, it will "starve" the lower priority task(s). The following, therefore, should work: void appTask (void); void usrAppInit (void) { int appTaskId; /* spawn application task with lowest possible priority (255) */ appTaskId = taskSpawn("appTask",255,0,2000,appTask,0,0,0,0,0,0,0,0,0,0); } void appTask (void) { while(1); } - -- ======================================================================== Michael Kesti | "And like, one and one don't make | two, one and one make one." mkesti@gv.net | - The Who, Bargain --------------------------- Newsgroups: comp.os.vxworks Subject: Loading Modules without File System Date: Fri, 25 Jan 2002 10:40:41 -0600 From: "John W. Komp" Organization: Medtronic Inc. Message-ID: <3C518A88.F1F116CA@uscorp.net> When in the field my target system no longer has access to any normal serial or file system connection but I would like to still load new code modules. My target system's connection to the outside world is a rather odd, nonstandard section of dual ported memory. What I would like to do is let my target system code receive a new module coff file through this memory port and then submit it to the loadModule system. Unfortunately loadModule wants the module from a file system or stream, neither of which I have. Can I somehow feed the coff file into a lower level part of loadModule or is there a spec somewhere that would give me hints on how to create a compliant stream out of the new module coff file? Any ideas would help. Thanks, John - -- **************************************************** This morning I shot six holes in my freezer I think I've got cabin fever Somebody sound the alarm - J. Buffett **************************************************** --------------------------- Newsgroups: comp.os.vxworks Subject: Re: compiling WTX application on WinNT, VC++ 6 Date: 25 Jan 2002 12:45:07 -0500 From: Bill Pringlemeir Organization: Bell Sympatico Message-ID: References: <3c4ca498$1_1@nopics.sjc> Sender: bpringlemeir@DeadDuck [William" == William C Cox ] [Cedric" == "Cedric Le Roux" ] Cedric> Anybody know a good place to get some detailed instructions Cedric> on how to compile a WTX application on Windows NT and Visual Cedric> C++ 6? William> The "Getting Started examples are only intended to work with William> the compiler that shipped with your copy of Tornado. Nobody William> that I know of has built binaries with VC++ and run them William> under Tornado. William> VC++ generates PE-COFF object files, and the VxWorks loaders [snip] As I understand this, Cedric wishes to compile a Windows application that will talk to the target server (unless I haven't groked his intent). Look at the WTX API in the "Tornado API Guide" and you will see what is possible. ... Cedric> I've gone through the Tornado API Guide. The sample code Cedric> will not compile, Look at the compiler messages and try to figure out what is happening. I have done what you are talking about. I think it is fairly rare [compared to DosFs and TCP/IP] for people using vxWorks to do this. So there is no "detailed instructions" on WTX applications. You have to do some work unfortunately. You can, of course, submit instruction to the FAQ when you are done! I am pretty sure that the old WindSurf didn't have anything on this. I don't know how to find anything on the new WindSerf, so I cann't help you there... regards, Bill Pringlemeir. - -- The reason why can teach us how to breathe. vxWorks FAQ, "http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html" --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Loading Modules without File System Date: Fri, 25 Jan 2002 18:12:45 +0000 (UTC) From: David Laight Organization: BT Openworld Message-ID: <3C51A072.160BA0A3@l8s.co.uk> References: <3C518A88.F1F116CA@uscorp.net> > What I would like to do is let my target system code receive a new > module coff file through this memory port and then submit it to the > loadModule system. Unfortunately loadModule wants the module from a file > system or stream, neither of which I have. > > Can I somehow feed the coff file into a lower level part of loadModule > or is there a spec somewhere that would give me hints on how to create a > compliant stream out of the new module coff file? memDrv will probably do it, but you could write a driver of your own. That way you could free up the space afterwards, or request the file in fragments etc. David --------------------------- Newsgroups: comp.os.vxworks Subject: mkdir with ftp Date: 25 Jan 2002 10:16:27 -0800 From: schinnaraj@hotmail.com (saran) Organization: http://groups.google.com/ Message-ID: <384a07c2.0201251016.5bf7aed5@posting.google.com> Hi all, I'm having problem in using the unsupported ftpdLib, can any one help me in compiling and using it? Thanks in advance for your help. Saran --------------------------- Newsgroups: comp.os.vxworks Subject: How to print the contents of packet header Date: 25 Jan 2002 10:49:49 -0800 From: jzhang@drs.ca (Jason) Organization: http://groups.google.com/ Message-ID: Hello, I have received an udp packet and its in mbuf. Can somebody tell me how to print the contents of packet header. For example, the IP address in dot seperated format and the udp port numbers etc. Any sample code for functions available ? Regards, --------------------------- Newsgroups: comp.os.vxworks Subject: Using SPY to measure ISR on 8260 Date: 25 Jan 2002 11:20:56 -0800 From: ryeung@earthlink.net (Raymond Yeung) Organization: http://groups.google.com/ Message-ID: I want to measure CPU utilization on a 8260 board, the main function of which is to pass data between a Fast Ethernet interface and ATM interfaces (i.e. FCCs). I trust that the time info. for tasks are accurate. However, I've no confidence on the interrupt time. I know that SPY uses auxiliary clock. Our BSP assigned TIMER2 as auxiliary clock and has a lower priority than the FCCs. I explored the option of reassigning auxClk priority to the highest using SICR register (setting it to 0x0d000000). Is that the only thing (and is this correct) I need to do to get SPY measure ISRs as well? --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Questions on NPT device driver or bugs in NPT IP ? Date: 25 Jan 2002 11:46:43 -0800 From: houcheng@ms17.hinet.net (houcheng) Organization: http://groups.google.com/ Message-ID: References: Johan Borkhuis wrote in message news:... > houcheng@ms17.hinet.net (houcheng) wrote: > > I found that the tNetwork task and my ARP task has errno > > 0x3d0001: S_objLib_OBJ_ID_ERROR. > > You should only look ar errno when you get an ERROR reply from a function > that defines errno. In this case some internal function returned an error > and the calling function forgot to clear it. > > > I start to ping the "sar0" interface with 30 sessions > > concurrently, all > > reponses ok. But after one hour latter, the ping timeouts suddenly. > > Is the netTask still running? > > Groeten, > Johan Thanks for your reply Groeten, Today, I've modify my driver code slightly, and there is no errrno again. But still ping fail after one hour. Yes, the netTask is still running, and the "sar" driver is still receiving frames, but IP does not response. I obserbe that : The input packet count of sar device increases, but the input count of IP ( ipstatShow displays all IP counts ) does not increase. I guess there is something wrong with the IP layer. houcheng@NoSPAM.turbonetworks.com.tw --------------------------- Newsgroups: comp.os.vxworks Subject: Re: while(1) statement on usrAppInit() Date: 25 Jan 2002 11:58:47 -0800 From: houcheng@ms17.hinet.net (houcheng) Organization: http://groups.google.com/ Message-ID: References: "jnick" wrote in message news:... > Hi all > I have a question for while statement at userAppInit function. > > void usrAppInit (void) > { > #ifdef USER_APPL_INIT > USER_APPL_INIT; /* for backwards compatibility */ > #endif > > /* add application specific code here */ > printf("start usrAppInit\n"); > > while(1); > } > Specific speaking, the usrAppInit is run under Root task. The problem is that: tRootTask has higher priority than the tNetTask. So your IP won't response to the ping, Since the CPU is busy at looping, no time sharing to the lower priority task tNetTask. Lin Houcheng houcheng@NoSpamla.turbonetworks.com.tw --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Loading Modules without File System Date: Fri, 25 Jan 2002 13:58:55 -0600 From: "John W. Komp" Organization: Medtronic Inc. Message-ID: <3C51B8FF.6868602C@uscorp.net> References: <3C518A88.F1F116CA@uscorp.net> <3C51A072.160BA0A3@l8s.co.uk> David, Thanks for the pointer. I think iosLib is probably the route I'll take. - -John David Laight wrote: > > What I would like to do is let my target system code receive a new > > module coff file through this memory port and then submit it to the > > loadModule system. Unfortunately loadModule wants the module from a file > > system or stream, neither of which I have. > > > > Can I somehow feed the coff file into a lower level part of loadModule > > or is there a spec somewhere that would give me hints on how to create a > > compliant stream out of the new module coff file? > > memDrv will probably do it, but you could write a driver of your own. > That way you could free up the space afterwards, or request the file in > fragments etc. > > David - -- **************************************************** This morning I shot six holes in my freezer I think I've got cabin fever Somebody sound the alarm - J. Buffett **************************************************** --------------------------- Newsgroups: comp.os.vxworks Subject: Re: errno 226 Date: 25 Jan 2002 12:07:36 -0800 From: houcheng@ms17.hinet.net (houcheng) Organization: http://groups.google.com/ Message-ID: References: bhorowit@cs.berkeley.edu (Benjamin Horowitz) wrote in message news:... > Hi, > > When trying to close a file, I encountered errno 226. Could anyone > please point me to where this errno is defined? I can't find it in > errno.h. > > The error message I get is: > HELLO.TXT: The system cannot find the file specified. > HELLO.TXT: Access is denied. > > I'm using VxWorks 5.4, Tornado 2. > > Thanks, > Ben A tip to display the "standard " errno, First you have to include "error status Table" comp. in the vxWorks. In target shell, execute the commands: -> errnoSet 0x7 -> perror Then, the error string is printed. This only prints out standard errno, hmm... errno 226, as John said, that's FTP errno. --------------------------- Newsgroups: comp.os.vxworks Subject: SEMAPHORE PROTECTION Date: 25 Jan 2002 12:11:16 -0800 From: pepsi_000@hotmail.com (vxJunkie) Organization: http://groups.google.com/ Message-ID: <676eb679.0201251211.591eea4f@posting.google.com> hi, Here is a brief description of my problem. I have a task that protects access to a database using mutex semaphores. According to our design only task "X" can write to the Database. All other tasks only GET information from the database. All the tasks running in the system share the same semaphore. Is there a way i can allow multiple reads from the database when task "X" is not writing into the database. Currently all the tasks in the system have to take a semaphore to read information from the database. Does anybody have some kind of a wrapper around mutex semaphores to acheive this? Thanks in advance, sesh --------------------------- Newsgroups: comp.os.vxworks Subject: Re: SEMAPHORE PROTECTION Date: 25 Jan 2002 13:15:40 -0800 From: Joe Chung Message-ID: References: <676eb679.0201251211.591eea4f@posting.google.com> You've described the classic "Single Writer Multiple Reader" problem. Search google web/groups for that phrase or permutations thereof. There are many solutions to this problem. One easy solutions is to use a Mutex to protect a readerCount and a Binary Semaphore to block the writer. Readers inc (and takes the writer semaphore if this is the first reader) the readerCount when reading, and dec the readerCount when done. If it's the last reader (readerCount == 0 after dec), then give the writer semaphore. The writer simply attempts to take the writer semaphore before writing. This solution favors the reader, since any one of them may block out the writer, but solutions can be devised to favor the reader(s) or give equal preference (fairness) as much as possible. - -jc - -- (apply 'concat (reverse (list "com" (char-to-string 46) "yahoo" (char-to-string 64) "joechung"))) --------------------------- Newsgroups: comp.os.vxworks Subject: CLI Parser for VxWorks Date: 25 Jan 2002 13:47:53 -0800 From: dpraveen@hotmail.com (Praveen Dulam) Organization: http://groups.google.com/ Message-ID: <182ee52a.0201251347.2a1d1421@posting.google.com> Hi I am looking for CLI commands parser for VxWorks. TIA Praveen --------------------------- Newsgroups: comp.os.vxworks Subject: getting task information in tylib.c - possible? Date: 25 Jan 2002 15:21:57 -0800 From: mark_ck_chang@yahoo.com (Mark Chang) Organization: http://groups.google.com/ Message-ID: <9b6f8883.0201251521.39b24bdd@posting.google.com> Is it possible in tyIRd in tylib.c to get task information? What I mean is that a certain task installed the tyIRd via ptyDrv->iosDrvInstall and so I think there must be a link somehow even if this routine is called at the int level. I see that a stucture used for tyIRd is typedef struct /* TY_DEV - tty device descriptor */ { DEV_HDR devHdr; /* I/O device header */ RING_ID rdBuf; /* ring buffer for read */ SEMAPHORE rdSyncSem; /* reader synchronization semaphore */ SEMAPHORE mutexSem; /* mutual exclusion semaphore */ struct /* current state of the read channel */ { unsigned char xoff; /* input has been XOFF'd */ unsigned char pending; /* XON/XOFF will be sent when xmtr is free*/ unsigned char canceled; /* read has been canceled */ unsigned char flushingRdBuf; /* critical section marker */ } rdState; RING_ID wrtBuf; /* ring buffer for write */ SEMAPHORE wrtSyncSem; /* writer synchronization semaphore */ struct /* current state of the write channel */ { unsigned char busy; /* transmitter is busy sending character */ unsigned char xoff; /* output has been XOFF'd */ unsigned char cr; /* CR should be inserted next (after LF) */ unsigned char canceled; /* write has been canceled */ unsigned char flushingWrtBuf; /* critical section marker */ unsigned char wrtBufBusy; /* task level writing to buffer */ } wrtState; UINT8 lnNBytes; /* number of bytes in unfinished new line */ UINT8 lnBytesLeft; /* number of bytes left in incompletely dequeued line */ unsigned short options; /* options in effect for this channel */ FUNCPTR txStartup; /* pointer to routine to start xmitter */ FUNCPTR protoHook; /* protocol specific hook routine */ int protoArg; /* protocol specific argument */ SEL_WAKEUP_LIST selWakeupList;/* tasks that are selected on this dev */ } TY_DEV; And I wonder if selWakeupList has the taskID that installed this call back? The selWakeupList is typedef struct { SEMAPHORE listMutex; /* mutex semaphore for list */ SEL_WAKEUP_NODE firstNode; /* usually one deep, stash first one */ LIST wakeupList; /* list of SEL_WAKEUP_NODE's */ } SEL_WAKEUP_LIST; and SEL_WAKEUP_NODE is typedef struct selWkNode { NODE linkedListHooks;/* hooks for wakeup list */ SEM_ID wakeupSem; /* select'ed task synch. semaphore */ BOOL dontFree; /* first in free list isn't malloc'ed */ int taskId; /* taskId of select'ed task */ int fd; /* fd to set in fd_set on activity */ SELECT_TYPE type; /* activity task is interested in */ fd_set * pReadFds; /* select'ed task's read fd_set */ fd_set * pWriteFds; /* select'ed task's write fd_set */ fd_set * pExceptFds; /* select'ed task's exception fd_set */ /* the following are needed for safe task deletion */ fd_set * pOrigReadFds; /* ptr. to original read fd_set */ fd_set * pOrigWriteFds; /* ptr. to orignial write fd_set */ } SEL_WAKEUP_NODE; And there is a taskID. Can this be the one? If it is, will this be the only one? I think so, since SEL_WAKEUP_NODE firstNode; /* usually one deep, stash first one */ states just that "usually one deep" Any thoughts? Sorry if this message is choppy. TIA Mark --------------------------- Newsgroups: comp.os.vxworks Subject: Re: mkdir with ftp Date: 25 Jan 2002 15:57:17 -0800 From: mark_ck_chang@yahoo.com (Mark Chang) Organization: http://groups.google.com/ Message-ID: <9b6f8883.0201251557.665b9c42@posting.google.com> References: <384a07c2.0201251016.5bf7aed5@posting.google.com> Is the last revision entry coded as 03c,05jun00,dlr? It was pretty much drop and replace for me. I made changes to the session status, took out user limits and removed anonymous ftp. But other than, no problems. btw - what are your problems. Mark schinnaraj@hotmail.com (saran) wrote in message news:<384a07c2.0201251016.5bf7aed5@posting.google.com>... > Hi all, > > I'm having problem in using the unsupported ftpdLib, can any one help > me in compiling and using it? > > Thanks in advance for your help. > > Saran --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Loading Modules without File System Date: 25 Jan 2002 16:54:19 -0800 From: john_94501@yahoo.com (John) Organization: http://groups.google.com/ Message-ID: <488e459a.0201251654.78a7f562@posting.google.com> References: <3C518A88.F1F116CA@uscorp.net> <3C51A072.160BA0A3@l8s.co.uk> Hello, > memDrv will probably do it, but you could write a driver of your own. > That way you could free up the space afterwards, or request the file in > fragments etc. You can always free up the space from memDrv too if you are careful... just use memAddToPool() to donate the space to the heap once you're done with it. Just make sure you unmount it before doing this ;-) HTH, John... --------------------------- Newsgroups: comp.os.vxworks Subject: TSA and HDLC for xDSL under vxWorks Date: Fri, 25 Jan 2002 15:32:36 +0900 From: "jnick" Organization: ONSE Telecom Message-ID: Hi all! Does anybody has any information for the following issues? 1. on T1 or E1, the data has been sent over the line with HDLC protocol. In case of this, how do they make a decision about HDLC address each other? Any reserved address for that? 2. Does anybody has a MPC860 source code for TSA related (with HDLC protocol) for vxWorks? Jnick --------------------------- Newsgroups: comp.os.vxworks Subject: Re: CLI Parser for VxWorks Date: 25 Jan 2002 17:27:33 -0800 From: mark_ck_chang@yahoo.com (Mark Chang) Organization: http://groups.google.com/ Message-ID: <9b6f8883.0201251727.781eea1a@posting.google.com> References: <182ee52a.0201251347.2a1d1421@posting.google.com> If you need it free - use TCL (we use 7.3 in our embedded box) or use one of WindRiver's RapidControl for CLI. Mark dpraveen@hotmail.com (Praveen Dulam) wrote in message news:<182ee52a.0201251347.2a1d1421@posting.google.com>... > Hi > > I am looking for CLI commands parser for VxWorks. > > TIA > Praveen --------------------------- Newsgroups: comp.os.vxworks Subject: Re: getting task information in tylib.c - possible? - FOUND A SOLUTION Date: 25 Jan 2002 18:07:33 -0800 From: mark_ck_chang@yahoo.com (Mark Chang) Organization: http://groups.google.com/ Message-ID: <9b6f8883.0201251807.1b0d4bd@posting.google.com> References: <9b6f8883.0201251521.39b24bdd@posting.google.com> Well, I guess this works OK. The TY_DEV has a devHdr which has a devHdr.name which is /pty/telnet0.M or /tyco/0 so I know who is calling the service. This works for now. Mark --------------------------- Newsgroups: comp.os.vxworks Subject: What can I do to get the size of my executable for powerPc down? Date: Sat, 26 Jan 2002 02:42:56 GMT From: "Madhav Ancha" Organization: AT&T Broadband Message-ID: Hi, We use cpp and stl for our development and we now have a massive .0 that we are finding hard to fit in our system. What can I do to reduce the size of the executable? Would stripping symbols be helpful? Can I do anything else? Any help is appreciated What's the common way to strip symbols on a NT host. The stripPPC seems to be broken on NT. with thanks, Madhav. --------------------------- Newsgroups: comp.os.vxworks Subject: readers writers problem Date: 25 Jan 2002 19:30:32 -0800 From: ganesh_kumarg@rediffmail.com (ganesh godavari) Organization: http://groups.google.com/ Message-ID: <4e22fcd6.0201251930.1ad208c7@posting.google.com> hai everyone, i am trying to port gdbm-1.8.0 onto vxworks. dbm uses file locking to ensure data integrity. i am wondering if any one has written a good code for multiple readers and multiple writters problem using semaphores?. if so can u share the code. Thanks gkgodava --------------------------- Newsgroups: comp.os.vxworks Subject: missing ftime and alarm?? Date: 25 Jan 2002 19:41:59 -0800 From: ganesh_kumarg@rediffmail.com (ganesh godavari) Organization: http://groups.google.com/ Message-ID: <4e22fcd6.0201251941.4f274089@posting.google.com> hello everyone, i am trying to port openssl-0.96b onto VxWorks. i found the ftime and alarm functions missing in VxWorks. does VxWorks have related function for ftime and alarm?. if so can u tell me. has anyone ported openssl onto VxWorks? if so can u share the make files? Thanks, gkgodava --------------------------- Newsgroups: comp.os.vxworks Subject: Re: porting dbm onto vxworks Date: 25 Jan 2002 20:18:39 -0800 From: ganesh_kumarg@rediffmail.com (ganesh godavari) Organization: http://groups.google.com/ Message-ID: <4e22fcd6.0201252018.442b61c4@posting.google.com> References: <4e22fcd6.0201181617.7100d150@posting.google.com> <88c8dc81.0201210251.7fff27f1@posting.google.com> hello, Thanks for help. have u ported dbm libraries onto vxworks?. i am facing some problem with locking. VxWorks doesn't seem to have flock and fcntl file locking function. i am wondering if u have written wrap around functions for fcntl or flock. if so can u share it. Thanks gkgodava if i turn off HAVE_SYS_FILE_H then i have to turn off AVE_SYS_ tut_pharoh@yahoo.com (tutankhamun) wrote in message news:<88c8dc81.0201210251.7fff27f1@posting.google.com>... > There should be a configuration file by name autoconf.h with gdbm > distribution. Here, look for the following line: > > /* Define if you have the header file. */ > #define HAVE_SYS_FILE_H 1 > > change it to > > /* Define if you have the header file. */ > #define HAVE_SYS_FILE_H 0 > > Should fix your problem. > > tutankhamun > > =================================================================== > > ganesh_kumarg@rediffmail.com (ganesh godavari) wrote in message news:<4e22fcd6.0201181617.7100d150@posting.google.com>... > > hai everyone, > > i am new to VxWorks and also this group. > > > > i am trying to port gdbm-1.8.0 on to VxWorks. i need sys/file.h for > > compiling. i am not able to find this file in my VxWorks header file > > list. > > > > Does VxWorks have this header file?. if so from where can i download > > it? > > if VxWorks doesn't have it? how can i overcome this problem?. > > > > this problem must have been encountered by a lot of you previously. so > > please help me out. > > > > Thanks in advance, > > gkgodava --------------------------- Newsgroups: comp.os.vxworks Subject: Re: What can I do to get the size of my executable for powerPc down? Date: Sat, 26 Jan 2002 04:49:05 GMT From: "H. E. Taylor" Organization: MTS Internet Message-ID: <3C525044.16E@despam.autobahn.mb.ca> References: In article Madhav Ancha wrote: > > We use cpp and stl for our development and we now have a massive .0 > that we are finding hard to fit in our system. What can I do to reduce the > size of the executable? Would stripping symbols be helpful? Can I do > anything else? Any help is appreciated > > What's the common way to strip symbols on a NT host. The stripPPC seems > to be broken on NT. > This is a common perception with templates. One solution is to not use them. ;-))> Otherwise, have you looked at the -o compiler options? - -het - -- "If a packet hits a pocket on a socket on a port, And the bus is interrupted as a very last resort, And the address of the memory makes your floppy disk abort, Then the socket packet pocket has an error to report!" - -Gene Ziegler (with a nod to Dr. Suess) Global Warming: http://www.autobahn.mb.ca/~het/globalwarming.html H.E. Taylor http://www.autobahn.mb.ca/~het/ --------------------------- Newsgroups: comp.os.vxworks Subject: Re: mkdir with ftp Date: 25 Jan 2002 21:17:54 -0800 From: houcheng@ms17.hinet.net (houcheng) Organization: http://groups.google.com/ Message-ID: References: <384a07c2.0201251016.5bf7aed5@posting.google.com> <9b6f8883.0201251557.665b9c42@posting.google.com> > schinnaraj@hotmail.com (saran) wrote in message news:<384a07c2.0201251016.5bf7aed5@posting.google.com>... > > Hi all, > > > > I'm having problem in using the unsupported ftpdLib, can any one help > > me in compiling and using it? > > > > Thanks in advance for your help. > > > > Saran Hi Saran: Don't you got the source code of ftpdLib ? You can get this source code from you FAE. The DOSFS2 product also includes the ftpdLib source code, you can get it from your FAE. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Why sysLib.c is not added into the project files set automatically when creating a new project Date: 25 Jan 2002 21:44:25 -0800 From: lihuapeng@21cn.com (DragonSpring) Organization: http://groups.google.com/ Message-ID: <1d411918.0201252144.47956fb9@posting.google.com> References: <1d411918.0201241902.5625e4cb@posting.google.com> correct answer from Golden, Dave [Dave.Golden@compaq.com] Use the windows explorer and check out the file names in your bsp directory. Make sure that the filename for sysLib.c is in all lowercase except for the "L" in lib. Do the same for sysALib.c. We ran into this problem too. I suspect that the tcl files that the tornado IDE uses are case sensitive and since the NT filesystem supports mixed case filenames, these special files have to match exactly. Dave lihuapeng@21cn.com (DragonSpring) wrote in message news:<1d411918.0201241902.5625e4cb@posting.google.com>... > We use Tornado II/VxWorks 5.4 for MPC860. The host OS is NT 4.0 + > patch 6. > > After we modified the original ADS860 BSP, when we create a new > bootable project, the sysLib.c is not added into the project file list > automatically, we have to add it manually. (right click and select Add > Files...) > > But if we use the original ADS860 BSP, sysLib.c will be added > automatically. > > Why? --------------------------- Newsgroups: comp.os.vxworks Subject: Problem in memory which allocated by tRootTask and using it by application Date: 25 Jan 2002 23:00:27 -0800 From: roh_youngjin@hotmail.com (Youngjin Roh) Organization: http://groups.google.com/ Message-ID: <799f3d42.0201252300.a1cbcc@posting.google.com> Dear experts. I am using VxWorks 5.4 and Tornado 2.0. I have developed several applications which is spawned by usrAppInit. For some other reason, the initialization of some memory block is done by usrAppInit directly, and the spawned application use that memory block. I've faced the problem with memory. After spawned application, I found that "memShow" command returns "Invalid memory block ~~ blah blah blah" something like that. So I change the usrAppInit for initialization usr mem block into "task". so a temporary task initialize some memory block which will be used for application, and then the memShow problem disappearred. This is very simple question. The application can use the memory directly which tRoorTask malloed space? Thanks you Youngjin --------------------------- End of New-News digest ********************** From vxwexplo-errs@csg.lbl.gov Sun Jan 27 04:03:09 2002 From: Vxworks Exploder Date: Sun Jan 27 04:03:11 PST 2002 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Sun Jan 27 04:03:05 PST 2002 Subject: Re: DOS File Corruption Subject: Re: TCP/IP Performance under VxWorks Subject: s/w vs. h/w breakpoints Subject: Suggestions? Subject: Cache problem Subject: Re: s/w vs. h/w breakpoints Subject: Re: SCSI Performance in vxWorks ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: DOS File Corruption Date: Sat, 26 Jan 2002 19:05:54 +0200 From: "Leonid Rosenboim" Organization: Verio Message-ID: References: <3c0fdc4b@news.tce.com> Rarely I follow up on my own postings, but this one deserves the occasion: It came to my attention, that at least some customers receive DosFs 2.0 from WindRiver with bugs which where discovered and patched as early as 1999. Luckily, DosFs 2.0 does not have that many bugs in it, but this one is specifically important: SPR# 31480 VFAT directory entries disappear when crossing cluster boundary So I recommend you make sure you got this one fixed in your code. - - Leonid - -------------------------------------- ???? ?? ????? ?????? ?? ?????????? ? ????? ????????, ?????? ????? ? ??????. "Leonid Rosenboim" wrote in message news:newscache$pxxyng$95h$1@lnews.actcom.co.il... > There is insufficient details here, what version of DosFs ? What kind of PC > card ? > Using TFFS ? What kind of corruption ? > > Still, you may want to search on google groups for past discussions on Flash > disk corruption, there where discussions on Flash disk corruption issues. > > > "Fred" wrote in message > news:3c0fdc4b@news.tce.com... > > We're using PCMCIA flash cards to store our downloaded vxWorks, but have > > recently had problems with file corruption. Sometimes the file system is > > corrupted, sometimes the file is corrupted. Occassionally we can > reproduce > > the problem by booting the board during a FTP transfer. Does anybody have > > any ideas as to what is causing the problem and how to fix it? > > > > > > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: TCP/IP Performance under VxWorks Date: Sat, 26 Jan 2002 19:10:18 +0200 From: "Leonid Rosenboim" Organization: Verio Message-ID: References: <3C4F024F.5324B28C@delta-info.com> Gary, the gritty part in VoIP wrt performance is it uses very small packets in order to minimize efffect of lost packets and minimize delay. Maximum PPS (packets per second) is another critical performance issue in protocol stacks AND network drivers. Your experience of the PPC860 does not surprise me at all. - - Leonid - ---- - ---- Education is wher eyou learn the rules, Experience is when you learn about Exceptions. - ---- "Gary M" wrote in message news:At848.3587$a%4.15361@rwcrnsc51.ops.asp.att.net... > A lot also depends on what you actually do with the data packets. 30 > Mbits/sec isn't much until you consider what you will be doing with it. We > looked at certain VOIP architectures using MPC8xx family parts, and decided > on a different architecture based on performance of the VxWorks stack and > the interface to the codec DSPs we were interested in using. > > "Paul Randall" wrote in message > news:3C4F024F.5324B28C@delta-info.com... > > Hi all, > > > > I have an application where I need to send or receive up to 30 Mbits/sec > > of sustained UDP data over Ethernet. I am evaluating different RTOSes > > and processors to come up with an architecture for our application. > > > > Does anyone have TCP/IP performance numbers for VxWorks (or other RTOSes > > for that matter) running on VxWorks supported processors? > > > > Also, does the ability to DMA data from the processor to an Ethernet > > controller make much of a difference in performance? I have been > > looking at the SMSC 91C111 ethernet controller which does not require a > > PCI bus but it also does not support DMA transfers. > > > > Thanks in advance, > > > > Paul Randall > > Delta Information Systems > > > > --------------------------- Newsgroups: comp.os.vxworks Subject: s/w vs. h/w breakpoints Date: 26 Jan 2002 10:16:34 -0800 From: brightpolestar@yahoo.com (Daniel) Organization: http://groups.google.com/ Message-ID: <28ee7c97.0201261016.67543d98@posting.google.com> Gurus, Can anybody tell what a software breakpoints is vs. the hardware breakpoints? Thanks. --------------------------- Newsgroups: comp.realtime,comp.realtime.jobs,comp.os.vxworks,comp.os.qnx Subject: Suggestions? Date: 26 Jan 2002 10:35:26 -0800 From: amcheeseus@yahoo.com (amit choudhary) Organization: http://groups.google.com/ Message-ID: <4cd34f34.0201261035.4561ad4@posting.google.com> I have developed a real time kernel using XINU Operating system. What would be the right way of presenting it to the realtime community for evaluation, suggestions, etc.?. I am a fresh CS graduate and want to develop my profile in this field. Since, I do not have access to commercial RTOSes like Vxworks, PSOS, WinCE, etc., how do I progress?, Are there any realtime project groups open for volunteers?. amit. --------------------------- Newsgroups: comp.os.vxworks Subject: Cache problem Date: Sat, 26 Jan 2002 21:47:41 -0800 From: "Boris" Organization: N/A Message-ID: SGksYWxsLA0KICBPdXIgdGVzdCBmb3VuZCB0aGF0IG91ciBwcm9ncmFtIGNhbid0IHJ1biBmYXN0 ZXIgb24NCmEgODYwVCA4ME1IeiBjaGlwIHRoYW4gb24gYSA1ME1IeiBjaGlwcyx0aGUgT1Mgb24g b3VyDQp0YXJnZXQgYm9hcmQgaXMgdnhXb3JrcyA1LjQsV2UgZG91YnQgaWYgd2UgcHJvcGVybHkg c2V0DQp0aGUgY2FjaGUuQ2FuIGFueW9uZSBwcm92aWRlIHVzZWZ1bCBpbmZvcm1hdGlvbiBvbg0K aG93IHRvIGVuYWJsZSBjYWNoZSBpbiB2eFdvcmtzPw0KICBUaGFuayB5b3UgaW4gYWR2YW5jZS4N Cg0KU2luY2VyZWx5IHlvdXJzLA0KWGlhbw0K --------------------------- Newsgroups: comp.os.vxworks Subject: Re: s/w vs. h/w breakpoints Date: Sun, 27 Jan 2002 04:40:33 GMT From: dale Message-ID: References: <28ee7c97.0201261016.67543d98@posting.google.com> In article <28ee7c97.0201261016.67543d98@posting.google.com>, brightpolestar@yahoo.com says... > Gurus, > > Can anybody tell what a software breakpoints is vs. the hardware breakpoints? > > Thanks. > A software breakpoint is an instruction (trap, S/W Int, whatever depending on the processor) inserted by the debugger directly into code, such that the debugger will trap on that instruction. A hardware breakpoint is a feature that _may_ be provided by the processor--the debugger can set it with an address to cause a trap or exception when the processor encounters that address; no code modification is necessary (good for ROM debug). An unlimited number of soft breaks are possible; hard breakpoints are limited to what the processor's debug component can support (one, two or few). A hard breakpoint _may_ apply to data references as well as instruction addresses (again, this depends on the processor's architecture); soft breaks can only apply to instruction addresses. - ---- Dale --------------------------- Newsgroups: comp.os.vxworks Subject: Re: SCSI Performance in vxWorks Date: Sun, 27 Jan 2002 09:18:42 +0200 From: "Leonid Rosenboim" Organization: Verio Message-ID: References: <20020122.015817.1346811305.3143@analogic.com> <3C4DEB92.3DFD5FD7@dsto.defence.gov.au> <20020123.191303.1389867269.5196@analogic.com> One possible guess is that SOMETHING could be verifying data that just has been written, but I doubt that the driver does anything useful during it's extra time. Wind River has been allways kinda shaky in this area, if you read this group for a long time you'd notice that device driver is certainly not one of WRS's strong areas, and knowledge is much weaker then other driver areas. In other words, if you need a good driver, you'd be better off paying someone to develop o optimize it for you, e.g. one of former WRS developers now freelancing. - - Leonid "Jared Maguire" wrote in message news:20020123.191303.1389867269.5196@analogic.com... > In article <3C4DEB92.3DFD5FD7@dsto.defence.gov.au>, "Tim Shaw" > wrote: > > > We've had up to about 17MBytes/sec on a Synergy VGM5 (the driver had > > been modified by Synergy), and about the same on a MVME2307 with SYM875 > > (we had to get hold of the source code and modify it to get that > > performance). I've never used the 895, but if it is using the same > > driver as the wr$ driver for the mv2307 there's a lot of tweaking > > possible to get big performance improvements (both increasing data > > transfer rate and reducing pci usage). > > . Just now I timed scsi performance under both Win2k and vxWorks on > the same hardware, and Win2k performed it's 4MB write in about 100ms, > which is pretty decent. VxWorks performed the same write in 400ms. I dove > in with windview to try to see what's going on, and found that the SCSI > driver's interrupt handler gets called a few times at each of THREE points > during a write-operation. These points are at t=0ms, t=100ms and t=400ms. > The t=100ms point is interesting, as that's EXACTLY how long windows takes > to write the data. I'm starting to suspect that the scsi driver is > actually writing the data at a reasonable speed, but then is doing > SOMETHING else before it returns successfully. *shrug* > > That's a bummer that we might need to get the code for the driver and > tweak it. Oh, well. > > Thanks for the reality-check, > > -Jared --------------------------- End of New-News digest ********************** From vxwexplo-errs@csg.lbl.gov Mon Jan 28 04:03:12 2002 From: Vxworks Exploder Date: Mon Jan 28 04:03:14 PST 2002 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Mon Jan 28 04:03:08 PST 2002 Subject: Re: s/w vs. h/w breakpoints Subject: Re: Tornado 2.0 Prototyper for Linux? Subject: Resume, Embedded software engineer Subject: Freier C-Programmierer sucht neues Projekt ab April Subject: Re: compiling WTX application on WinNT, VC++ 6 Subject: Re: Error while downloading project on target??? Subject: Re: SEMAPHORE PROTECTION Subject: Re: Simulate Wind Web Server on PC Subject: How do i download the VxWorks image Subject: Re: SMC driver Subject: Re: Error while downloading project on target??? Subject: Re: SEMAPHORE PROTECTION Subject: Re: What can I do to get the size of my executable for powerPc down? Subject: problem with sprT2CP3 Subject: All tasks execute in the same address space Subject: Re: All tasks execute in the same address space ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: s/w vs. h/w breakpoints Date: Sun, 27 Jan 2002 12:29:11 GMT From: u913848876@spawnkill.ip-mobilphone.net Message-ID: References: <28ee7c97.0201261016.67543d98@posting.google.com> Sender: u913848876@spawnkill.ip-mobilphone.net > > Gurus, > > Can anybody tell what a software breakpoints is vs. the hardware breakpoints? > > Thanks. A software breakpoint is implemented by the debugger planting a trap instruction in the instruction stream (replacing the opcode that was at that location). The debugger gains control when such a breakpoint is hit, via the associated trap handler. It then can put back the original opcode. Obviously, this means that you can not put software breakpoints in ROM. On the other hand, the debugger can simultaneously maintain any number of active software breakpoints. A hardware breakpoint is implemented by employing the on-chip debugging facilities of the CPU (e.g., Intel 386 family, PowerPC and others). This allows putting a breakpoint at any location (including in ROM). On the other hand, the number of hardware breakpoints is very limited (e.g., the PowerPC family only supports one). Some processors support (hardware) data breakpoints in addition to instruction breakpoints, which allow tracing of variable modification (this can not be accomplished, in real-time, via software breakpoints). These facilities, if present, are also severely limited in quantity. HTH, Beni Falk - -- Spam protected message from: Sent by falk_beni from rada included in com Posted via http://www.usenet-replayer.com/cgi/content/new --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Tornado 2.0 Prototyper for Linux? Date: Sun, 27 Jan 2002 15:22:58 GMT From: bill@woola.qswtools.com (William C. Cox) Organization: EarthLink Inc. -- http://www.EarthLink.net Message-ID: References: <488e459a.0201190940.65a9756a@posting.google.com> In article <488e459a.0201190940.65a9756a@posting.google.com>, john_94501@yahoo.com (John) writes: > Hello, > > No, Tornado 2.0 only supports Windoze and Solaris; Tornado 3 supports > Linux too, but I don't believe that there is a prototyper for Tornado > 3. Yes, there is such a prototyper. The behavior of VxWorks AE, with memory protection domains, requires that the prototyper be different. > > HTH, > > John... > > "David" wrote in message news:... >> Hi,All; >> Please tell me if is possible to installed Tornado 2.0 Prototyper in Linux. >> Wind River Tornado 2.0 Prototyper can be installed in Solaris OS, I have no >> idea about Solaris OS. >> Thanks >> David >> davidyang_us@yahoo.com --------------------------- Newsgroups: comp.os.vxworks Subject: Resume, Embedded software engineer Date: 27 Jan 2002 12:26:33 -0800 From: bijannazem@hotmail.com (bijan nazem) Organization: http://groups.google.com/ Message-ID: <18ca8ef6.0201271226.40e163fc@posting.google.com> Bijan Nazem 325, Huntsmill Blvd. Scarborough, Ontario, M1W3C9 Tel:(416) 490-0420 BijanNazem@hotmail.com Objective: To obtain a position as an Embedded, real-time software engineer, in which I can utilize my present skills and further enhance my knowledge with diverse applications. Education: 9/93 to 2/96 M.Sc. Degree in Electrical Engineering (Major: Control Eng.)(Isfahan University of Technology, Isfahan- Iran) GPA 17.78 out of 20 9/88 to 2/93 B.Sc. Degree in Electrical Engineering (Major: Electronic Eng.)(Shiraz University, Shiraz- Iran) GPA 16.89 out of 20 Experiences and technical skills: Expert at: • C and C++ programming language 5 years • Object oriented programming 5 years • VxWorks operating systems (Real-time operating system) 4 years • Tornado development environment 4 years • TCP/IP network programming for Windows NT and VxWorks 4 years • Win32 multithread programming 3 years • VME bus and multiprocessors environment 4 years • Device driver for VxWorks (Low-level programming for hardware interfacing) 4 years • Serial port communication programming for Windows NT and VxWorks 4 years • Control system design and closed loop control systems 5 years • Real-time control systems 4 years • MATLAB and Simulink 5 years • MOORE DCS system 3 years good at: • Microsoft .NET framework and C# language • Microsoft Visual C++, MFC, ActiveX, XML and GUI design • ODBC, SQL and database access using Visual C++ • DSP (TMS 6000 series) assembly language and archetecture • Windows NT/2000 archetecture and configuration • Device driver for windows NT using DriverWorks and Softice • Digital signal processing and filtering • IEC 1131-3 standard programming language • Citect (A PC-based SCADA system) • LabView (G programming) Employment and working history: 4/96 to 2001 IRISA (Industrial Automation and MIS systems Company), Job title: Software Engineer. • MVME-162 embedded controller (A SBC board made by Motorola) configuration and system level programming for VxWorks running on that hardware. • Object oriented device driver for IP ADC, IP DAC and IP- Quadrature Modules to provide interface between hardware (I/O) and application. • Design and implementation of 3964 protocol (A character based protocol for communicating with Siemens mini-computers and PLCs in C++ language for VxWorks and Windows NT (Win32). • Design and implementation of Modbus on ethernet protocol (an industrial network protocol for communication with SCADA systems) in C++ language for VxWorks targets. • Design and development of some communication tasks for VxWorks on one side and LabView (running on Windows NT workstation) on the other side using TCP/IP protocol in order to enable computer room engineer to set parameters on embedded controllers and see real-time data with 1-msec resolution through LabView application. • Design and implementation of complicated closed loop control systems and digital filters using MATLAB/Simulink and coding them in C++ (VxWorks, VME, MVME-162) to implement AGC and thickness control system for the hot strip finishing mill of Mobarakeh Steel Company (A plant in Isfahan-Iran with 3.5 million tons per year productivity) • Writing device driver for VMIVME-5576 (A reflective memory card made by VMIC for VME bus)to reading and writing data and generating and catching interrupts for VxWorks . • Writing device driver for VMIPCI-5576 (A reflective memory card made by VMIC for PCI bus)to reading and writing data and generating and catching interrupts for Windows NT using DriverWorks. --------------------------- Newsgroups: comp.os.vxworks Subject: Freier C-Programmierer sucht neues Projekt ab April Date: Sun, 27 Jan 2002 21:28:01 +0100 From: "M.Thoemel" Organization: T-Online Message-ID: Followup-To: comp.os.vxworks Guten Tag, ich bin freier Sofwareentwickler mit einem sehr breiten Spektrum. Ab dem April 2002 stehe ich wieder zur Verfügung. Meine Schwerpunkte sind: - - C (aber auch andere Sprachen!) - - Realtimesysteme - - systemnahe Programmierung - - spezielle Hardware-Schnittstellen - - Netzwerke - - Treiber - - Streamsdriver - - Grafische Datenauswertung/Darstellung - - Software-Messungen - - Gruppenleitung - - Projektierung - - Projektleitung etc... Sie können sich im Detail informieren unter: http://matthias.thoemel.bei.t-online.de Ich würde mich über einen Kontakt mit Ihnen freuen. Ihr Matthias Thömel (matthias.thoemel@t-online.de) --------------------------- Newsgroups: comp.os.vxworks Subject: Re: compiling WTX application on WinNT, VC++ 6 Date: Sun, 27 Jan 2002 17:27:25 -0500 From: "Cedric Le Roux" Message-ID: <3c547e80_2@nopics.sjc> References: <3c4ca498$1_1@nopics.sjc> Thanks for the replies guys!! Well I was able to get it figured out. I found out that it is critical to have your kernel built with the right includes. Like symbol table synchronization and WDB agent. Also, Target server must be configured with TSFS enabled on read/write. Target Server must also be set to synchronize symbols. After a little more playing around with code... (Byte swapping.... ~fun~) I was able to set variables on the target from VC++. Now I'm struggling to set values contained in structures. But that will come in time. Later all, Cedric "Bill Pringlemeir" wrote in message news:uzo32s4t8.fsf@yahoo.com... > > [William" == William C Cox ] > [Cedric" == "Cedric Le Roux" ] > > Cedric> Anybody know a good place to get some detailed instructions > Cedric> on how to compile a WTX application on Windows NT and Visual > Cedric> C++ 6? > > William> The "Getting Started examples are only intended to work with > William> the compiler that shipped with your copy of Tornado. Nobody > William> that I know of has built binaries with VC++ and run them > William> under Tornado. > > William> VC++ generaes PE-COFF object files, and the VxWorks loaders > [snip] > > As I understand this, Cedric wishes to compile a Windows application > that will talk to the target server (unless I haven't groked his > intent). Look at the WTX API in the "Tornado API Guide" and you will > see what is possible. > > ... > > Cedric> I've gone through the Tornado API Guide. The sample code > Cedric> will not compile, > > Look at the compiler messages and try to figure out what is happening. > I have done what you are talking about. I think it is fairly rare > [compared to DosFs and TCP/IP] for people using vxWorks to do this. > So there is no "detailed instructions" on WTX applications. You have > to do some work unfortunately. You can, of course, submit instruction > to the FAQ when you are done! > > I am pretty sure that the old WindSurf didn't have anything on this. > I don't know how to find anything on the new WindSerf, so I cann't > help you there... > > regards, > Bill Pringlemeir. > > -- > The reason why can teach us how to breathe. > vxWorks FAQ, "http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html" --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Error while downloading project on target??? Date: Sun, 27 Jan 2002 17:29:32 -0500 From: "Cedric Le Roux" Message-ID: <3c547efe$1_1@nopics.sjc> References: <7421c9fd.0201222012.335b07bb@posting.google.com> If you are using a PowerPC, there is a 32M limit for loading. Hope that is not the case for you... Cedric "Amit" wrote in message news:7421c9fd.0201222012.335b07bb@posting.google.com... > I have a target with memory 128 MB. My final image size is coming > around 76 MB. So while downloading the image on target, system hangs > up. I want to know whether there are certain size restrictions on > image size which you can download . > > > Thanks > > Ajay --------------------------- Newsgroups: comp.os.vxworks Subject: Re: SEMAPHORE PROTECTION Date: 27 Jan 2002 17:06:32 -0800 From: pepsi_000@hotmail.com (vxJunkie) Organization: http://groups.google.com/ Message-ID: <676eb679.0201271706.3f40e89f@posting.google.com> References: <676eb679.0201251211.591eea4f@posting.google.com> Hi Joe, Your solution sounds very interesting. By using a mutex semaphore b/n readers we are serializing access to the database. I am looking for a solution where the readers need not wait for a semaphore. The only time a reader waits is when the DB is being modified by the writer. Thanks, sesh Joe Chung wrote in message news:... > You've described the classic "Single Writer Multiple Reader" problem. > Search google web/groups for that phrase or permutations thereof. > > There are many solutions to this problem. > > One easy solutions is to use a Mutex to protect a readerCount and a Binary > Semaphore to block the writer. Readers inc (and takes the writer semaphore > if this is the first reader) the readerCount when reading, and dec the > readerCount when done. If it's the last reader (readerCount == 0 after > dec), then give the writer semaphore. The writer simply attempts to take > the writer semaphore before writing. > > This solution favors the reader, since any one of them may block out the > writer, but solutions can be devised to favor the reader(s) or give equal > preference (fairness) as much as possible. > > -jc --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Simulate Wind Web Server on PC Date: 27 Jan 2002 19:05:25 -0800 From: chrismooremobile@yahoo.com (Chris Moore) Organization: http://groups.google.com/ Message-ID: References: <3c501c91@news.012.net.il> Use the free GoAhead web server from http://www.goahead.com You can probably get that to run under the prototyper if you can get your IP connection working through to the prototyper. "Yaron Cohen" wrote in message news:<3c501c91@news.012.net.il>... > Hi, > > I would like to ask for your help. > > I am building a web site using Wind Web Server 2.0 and I am new for Vxworks > and Tornado. > > I would like to test it and work with it first on my PC. > > If this is possible ?, installing the web server on the PC. > > Can I use Tornado Prototyper for this purpose ? > > > > Thanks and have a nice day, > > Yaron --------------------------- Newsgroups: comp.os.vxworks Subject: How do i download the VxWorks image Date: 27 Jan 2002 21:41:20 -0800 From: deepaks@sasken.com (Deepak Shetty) Organization: http://groups.google.com/ Message-ID: <421ece65.0201272141.306925c2@posting.google.com> Hi, folks I am currently working on an Intel IQ80310 IOP board and need to downlaod the VxWorks image bootrom.bin provided on to the on-board flash unfortunately the board manual doesnt describe the memory location / address where i need to put in the image for proper execution of the same. the board has redboot as the boot monitor, and i need to use the FRU ( flash recovery utility ) provided by the intel people to download the image ( redboot as well as any other, for eg: vxwrks ) onto the on board flash via the PCI interface of the PC to which the board is connected. but i guess i need to downlaod the image to a specified address of the flash as there is a rotary switch setting provided on the board if the image downlaoded is of vxworks, but the manual doesnt give the address of the same any idea where i need to download the image of vxworks ? in case any one has dealt with such similar prob before ? I am unable to get any support from wrs / intel on the same :( also usualy how is the memory mapped for the flash ...? redboot shows some info when i enter fis list & fis free, but i am confused ... can someone tell me how is the memory mapping present betn the flash and the onboard 32MB RAM ? Thanx, deepak [ "To steal ideas from one person is plagiarism, to steal ideas from many is research." ] -Anon. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: SMC driver Date: 27 Jan 2002 23:46:53 -0800 From: raj_rashid@hotmail.com (Rashid) Organization: http://groups.google.com/ Message-ID: <2a6118da.0201272346.74ae4b57@posting.google.com> References: <12116a88.0201221830.591bea57@posting.google.com> <2a6118da.0201230241.1604435d@posting.google.com> <12116a88.0201230956.1b276d27@posting.google.com> <2a6118da.0201232119.5dd8368c@posting.google.com> <12116a88.0201241411.705dd3f5@posting.google.com> Hi Jason, > I checked out that I didn't define INCLUDE_WDB_COMM_SERIAL. I used > Network for my WDB connection, instead of serial port. So I think > that's why I couldn't find the symbol in my image. But the question > is, if wdbSlipPktDevInit(), seems to me the only place to install the > callback functions, is not in the image, how can the callback > functions be installed? wdbSlipPktDevInit() is _NOT_ the only place to install callback functions. That i referred just as an example of how callbacks are used. > Got you. But why callback function installation (part of SMC driver > initialization) has something to do with WDB. I mean, if the vxworks > image needs not to support WDB (Tornado), it may still needs to init > the driver. A higher layer library that wants to use this driver (e.g., ttyDrv) will install its own callback routine using the SIO_INSTALL_CALLBACK ioctl command. In the example for communication between target server and tornado agent over serial link the callback functions are installed by wdbSlipPktDevInit(). You didn't tell for what application u r using the serial i/f. > Due to the two points, is there anywhere else to do the installation? The callback installation is not done in driver itself, it is done in the higher layer. regards, Rashid --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Error while downloading project on target??? Date: Mon, 28 Jan 2002 09:09:22 +0100 From: "Michael Lawnick" Organization: Buergernetz Dillingen Message-ID: References: <7421c9fd.0201222012.335b07bb@posting.google.com> Reply-To: "Michael Lawnick" Sender: mlawnick@p5088fdda.dip.t-dialin.net Hi Amit, in addition to my prior answer: don't forget to setup your bootrom with a high enough RAM_HIGH_ADR in config.h and makefile. Otherwise your kernel's download will overwrite boot runtime code/data ! - -- Mit freundlichen Grüßen, Michael Lawnick ============================================== SOFTEC GmbH Tel +49-731-96600-0 Promenade 17 Fax +49-731-96600-23 D-89073 Ulm Michael Lawnick Germany lawnick@softec.de ============================================== "Amit" schrieb im Newsbeitrag news:7421c9fd.0201222012.335b07bb@posting.google.com... > I have a target with memory 128 MB. My final image size is coming > around 76 MB. So while downloading the image on target, system hangs > up. I want to know whether there are certain size restrictions on > image size which you can download . > > > Thanks > > Ajay --------------------------- Newsgroups: comp.os.vxworks Subject: Re: SEMAPHORE PROTECTION Date: Mon, 28 Jan 2002 10:11:08 +0100 From: "Michael Lawnick" Organization: Buergernetz Dillingen Message-ID: References: <676eb679.0201251211.591eea4f@posting.google.com> <676eb679.0201271706.3f40e89f@posting.google.com> Reply-To: "Michael Lawnick" Sender: mlawnick@p5088fdda.dip.t-dialin.net Hi vxJunkie, Joe seemed to have answered your question with the right ideas. Don't expect others to solve your application problems, this NG in its main stream helps to solve problems with VxWorks/Tornado ! - -- Mit freundlichen Grüßen, Michael Lawnick ============================================== SOFTEC GmbH Tel +49-731-96600-0 Promenade 17 Fax +49-731-96600-23 D-89073 Ulm Michael Lawnick Germany lawnick@softec.de ============================================== "vxJunkie" schrieb im Newsbeitrag news:676eb679.0201271706.3f40e89f@posting.google.com... > Hi Joe, > > Your solution sounds very interesting. > By using a mutex semaphore b/n readers we are serializing access to > the database. I am looking for a solution where the readers need not > wait for a semaphore. > The only time a reader waits is when the DB is being modified by the > writer. > > Thanks, > sesh > > > > Joe Chung wrote in message news:... > > You've described the classic "Single Writer Multiple Reader" problem. > > Search google web/groups for that phrase or permutations thereof. > > > > There are many solutions to this problem. > > > > One easy solutions is to use a Mutex to protect a readerCount and a Binary > > Semaphore to block the writer. Readers inc (and takes the writer semaphore > > if this is the first reader) the readerCount when reading, and dec the > > readerCount when done. If it's the last reader (readerCount == 0 after > > dec), then give the writer semaphore. The writer simply attempts to take > > the writer semaphore before writing. > > > > This solution favors the reader, since any one of them may block out the > > writer, but solutions can be devised to favor the reader(s) or give equal > > preference (fairness) as much as possible. > > > > -jc --------------------------- Newsgroups: comp.os.vxworks Subject: Re: What can I do to get the size of my executable for powerPc down? Date: Mon, 28 Jan 2002 10:15:20 +0100 From: "Michael Lawnick" Organization: Buergernetz Dillingen Message-ID: References: Reply-To: "Michael Lawnick" Sender: mlawnick@p5088fdda.dip.t-dialin.net Hi Madhav, you are correct, stripPPC is boken, but you may use objCopyPPC instead. This should work as long as you have compiled without -g (debug information). - -- Mit freundlichen Grüßen, Michael Lawnick ============================================== SOFTEC GmbH Tel +49-731-96600-0 Promenade 17 Fax +49-731-96600-23 D-89073 Ulm Michael Lawnick Germany lawnick@softec.de ============================================== "Madhav Ancha" schrieb im Newsbeitrag news:QIo48.3277$Om2.5730@rwcrnsc53... > Hi, > > We use cpp and stl for our development and we now have a massive .0 > that we are finding hard to fit in our system. What can I do to reduce the > size of the executable? Would stripping symbols be helpful? Can I do > anything else? Any help is appreciated > > What's the common way to strip symbols on a NT host. The stripPPC seems > to be broken on NT. > > with thanks, > Madhav. > > > > --------------------------- Newsgroups: comp.os.vxworks Subject: problem with sprT2CP3 Date: Mon, 28 Jan 2002 12:58:04 +0300 From: "Shrike" Organization: MTU-Intel ISP Message-ID: Hello! I've installed sprT2CP3 (mainly, to adjust font size ;)) and now I experience the following problems: - - I can't add new project to the workspace if I base it on any BSP: I receive error "couldn't dublicate input handle: bad file number" - - I can't view and edit component list in VxWorks tab in Workspace window: + transforms to -, but I can't see anything ;-( Thanks in advance. --------------------------- Newsgroups: comp.os.vxworks Subject: All tasks execute in the same address space Date: 28 Jan 2002 02:56:31 -0800 From: sridhar@cmcltd.com (Sridhar CV) Organization: http://groups.google.com/ Message-ID: <8207c357.0201280256.4c534549@posting.google.com> Hi, can you please eloborate on the statement "in Vxworks all tasks execute in the same address space" Thanks in anticipation Sridhar CV Engineer R&D RealTime Systems CMC Limited,Gachibowli Hyderabad, INDIA-500019 --------------------------- Newsgroups: comp.os.vxworks Subject: Re: All tasks execute in the same address space Date: Mon, 28 Jan 2002 06:29:36 -0500 From: joe durusau Organization: BELLSOUTH.net Message-ID: <3C553620.F4FEB2B1@bellsouth.net> References: <8207c357.0201280256.4c534549@posting.google.com> It means that there is no 'wall of separation' provided between tasks. In some OS's, if a task tries to write to memory not assigned directly to it, a fault of some sort is detecte, and the write does not take place. This is not true in vxWorks, at least through 5.4. If you write to (for example) some critical location, like an interrupt vector, you will crash the entire system, because all tasks can access all memory in the entire system. Speaking only for myself, Joe Durusau Sridhar CV wrote: > Hi, > > can you please eloborate on the statement "in Vxworks all tasks > execute in the same address space" > > Thanks in anticipation > > Sridhar CV > Engineer R&D > RealTime Systems > CMC Limited,Gachibowli > Hyderabad, > INDIA-500019 --------------------------- End of New-News digest ********************** From vxwexplo-errs@csg.lbl.gov Mon Jan 28 10:32:25 2002 From: Jaya Prakash Date: Mon Jan 28 10:32:27 PST 2002 Subject: Re: Error while downloading project on target??? This message is in MIME format. Since your mail reader does not understand this format, some or all of this message may not be legible. ------_=_NextPart_001_01C1A82A.171ACE10 Content-Type: text/plain; charset="iso-8859-1" Hi Amit No Download Limittation for vxWorks image.BUT ftp and tftp has its own limitations say tftp 6MB max. Rgrds, JP ############################################################################ ########################## This email communication may contain CONFIDENTIAL INFORMATION and is intended only for the use of the intended recipients identified above. If you are not the intended recipient of this communication, you must not use, disclose, distribute, copy or print this email. If you have received this communication in error, please immediately notify the sender by reply email, delete the communication and destroy all copies. ############################################################################ ########################## ------_=_NextPart_001_01C1A82A.171ACE10 Content-Type: text/html; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable Re: Error while downloading project on target???

Hi Amit

No Download = Limittation for vxWorks image.BUT ftp and tftp has its own limitations = say tftp 6MB max.

Rgrds,
JP


#########################################################= ############################################# This email communication = may contain CONFIDENTIAL INFORMATION and is intended only for the use = of the intended recipients identified above.  If you are not the = intended recipient of this communication, you must not use, disclose, = distribute, copy or print this email. If you have received this = communication in error, please immediately notify the sender by reply = email, delete the communication and destroy all copies. = ########################################################################= ##############################

------_=_NextPart_001_01C1A82A.171ACE10-- From vxwexplo-errs@csg.lbl.gov Mon Jan 28 13:27:52 2002 From: "Katz, Joel" Date: Mon Jan 28 13:27:54 PST 2002 Subject: Booting from tffs and using DHCP to connect to network in vxWorks Hi All: I've spent the past few days trying to get a vxWorks PCPentium kernel to build which boots from an M-Systems Disk On Chip using tffs, but connecting to the network using DHCP to get an IP address. I was getting an error message saying that the system could not attach to tffs. Thanks to someone on this news group, I learned that I needed to modify usrNetBoot.c in tornado/target/config/comps/src/net to add tffs to the list of boot devices that need to use the other field to specify the network adapter, but that was insufficient to actually solve my problem. It turns out, in addition, that usrNetDhcpcBootSetup.c(also in tornado/target/config/comps/src/net) needs to be modified. The routine dhcpcConfigSet appears to use sysBootParams.bootDev as the network adapter. For hard disks, disk on chips, floppy drives, etc, it should be using the other field as the network adapter. I copied the sysBootParams.other field into sysBootParams.bootDev before the call to dhcpcConfigSet and restore the original setting afterwards and I can now reliably use DHCP to get an IP address while booting from disk on chip. There is an outstanding TSR 72355 covering this same issue. If anyone else is encountering similar problems and would like copies of my code mods, feel free to e-mail me at jmk@alum.mit.edu and I'll send them along. I'd rather not clutter the list with code. Joel Katz, Staff Engineer, Affymetrix Inc., 4G Crosby Drive, Bedford, MA 01730 (781)-280-1753 --- (781)-280-1777(FAX) From vxwexplo-errs@csg.lbl.gov Tue Jan 29 04:03:16 2002 From: Vxworks Exploder Date: Tue Jan 29 04:03:19 PST 2002 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Tue Jan 29 04:03:12 PST 2002 Subject: Calling "C" functions from WindSh via serial-line Subject: Re: Problem in memory which allocated by tRootTask and using it by application Subject: Re: Calling "C" functions from WindSh via serial-line Subject: Re: Problem in memory which allocated by tRootTask and using it by application Subject: what is af0?/arp requests missing in tornado 2.1.1 Subject: Re: s/w vs. h/w breakpoints Subject: VxWorks AE C++ Static constructor Subject: Re: DOS File Corruption Subject: Re: New WindSerf, TSR's and SPR's. Subject: How to force 10Base-T on a 10/100 NIC in vxworks? Subject: Re: New WindSerf, TSR's and SPR's. Subject: Re: How to force 10Base-T on a 10/100 NIC in vxworks? Subject: Re: SMC driver Subject: Re: Suggestions? Subject: texcTask exception 768 Subject: Re: Suggestions? Subject: Did anyone implement 4Ms HDLC channels ON MP860? Subject: About STLport Subject: problems and bursts in using pci+cache Subject: Re: Calling "C" functions from WindSh via serial-line Subject: Re: spurious interrupt + cache Subject: Re: Did anyone implement 4Ms HDLC channels ON MP860? Subject: Re: Uninitialized Interrupt! Vector number 47 Subject: Re: WWW.GIWERSWORLD.ORG Subject: Re: what is af0?/arp requests missing in tornado 2.1.1 Subject: Re: About STLport Subject: Curses lib under vxworks Subject: Re: New WindSerf, TSR's and SPR's. Subject: problem with -mlongcall Subject: [comp.os.vxworks] FAQ - index Subject: Why "dosFsDevInit failed"? Subject: Death of VxWorks when "sysClkConnect" Subject: 860SAR UTOPIA.... What is syncronization signals? ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Calling "C" functions from WindSh via serial-line Date: Mon, 28 Jan 2002 15:47:21 +0100 From: "Mads Bligaard Nielsen \(M@H\)" Organization: Ericsson Diax (DXD), Mobile@Home Message-ID: Anybody knows why I cannot call "C" functions from the shell that is available on the serial line (PPC target). If I start a shell via a TargetAgent (Ethernet) then it is possible - but it seems like the shell launced via the serial connection not is aware of the correct symbol table. I want to use the serial connection because my test functions is used to test some VLAN functionality that (currently) means that my TargetAgent host is unavailable via Ethernet... **Any ideas how to make the symbol table visible for this shell? ###Log - -->printf "Hello\n" undefined symbol: printf - -> vlanTestIoctl(0,0,0,0) undefined symbol: vlanTestIoctl - -- Mads Bligaard Nielsen System Engineer, PDU Mobile@Home Ericsson Diax, http://dxd.ericsson.se strip "NOJUNK." to reply... --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Problem in memory which allocated by tRootTask and using it by application Date: 28 Jan 2002 15:18:49 GMT From: Johan Borkhuis Organization: Agere Systems Message-ID: References: <799f3d42.0201252300.a1cbcc@posting.google.com> roh_youngjin@hotmail.com (Youngjin Roh) wrote: > Dear experts. > > I am using VxWorks 5.4 and Tornado 2.0. > I have developed several applications which is spawned by usrAppInit. > For some other reason, the initialization of some memory block is done > by usrAppInit directly, and the spawned application use that memory > block. > > I've faced the problem with memory. > After spawned application, I found that "memShow" command returns > "Invalid memory block ~~ blah blah blah" something like that. > > So I change the usrAppInit for initialization usr mem block into > "task". so a temporary task initialize some memory block which will be > used for application, and then the memShow problem disappearred. > > This is very simple question. > > The application can use the memory directly which tRoorTask malloed > space? Yes, it can. There is no owner of memory in VxWorks 5.4, AE is a different story. But it looks like your code has a problem. An error like this usually occurs when a memorystructure is owverwritten. So first check your code to see if it accesses memory outside of its memory block. Groeten, Johan - -- o o o o o o o . . . _____________________________ o _____ || Johan Borkhuis | .][__n_n_|DD[ ====_____ | borkhuis@agere.com | >(________|__|_[_________]_|__________________________| _/oo OOOOO oo` ooo ooo 'o!o!o o!o!o` === VxWorks FAQ: http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html === --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Calling "C" functions from WindSh via serial-line Date: 28 Jan 2002 15:26:31 GMT From: Johan Borkhuis Organization: Agere Systems Message-ID: References: "Mads Bligaard Nielsen \(M@H\)" wrote: > **Any ideas how to make the symbol table visible for this shell? Include the symbol table components (symbol table, initialize symbol table and build-in symbol table) and symbol table show routines. Groeten, Johan - -- o o o o o o o . . . _____________________________ o _____ || Johan Borkhuis | .][__n_n_|DD[ ====_____ | borkhuis@agere.com | >(________|__|_[_________]_|__________________________| _/oo OOOOO oo` ooo ooo 'o!o!o o!o!o` === VxWorks FAQ: http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html === --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Problem in memory which allocated by tRootTask and using it by application Date: Mon, 28 Jan 2002 17:01:31 +0100 From: "Werner Schiendl" Message-ID: <3c5575dd$1@brateggebdc5.br-automation.co.at> References: <799f3d42.0201252300.a1cbcc@posting.google.com> Hi, VxWorks 5.4 does not track or make a difference who allocates memory. So allocating memory in usrAppInit and using it later on will work. However, be sure that you do not free the memory at the end of usrAppInit (since it normally does/should not wait for the application to complete, you'd free the memory too early). I guess there is something wrong with your passing the memory block's address or you overwrite some bytes _before_ the allocated area. (The information for memory manager in VxWorks is stored here). hth Werner "Youngjin Roh" wrote in message news:799f3d42.0201252300.a1cbcc@posting.google.com... > Dear experts. > > I am using VxWorks 5.4 and Tornado 2.0. > I have developed several applications which is spawned by usrAppInit. > For some other reason, the initialization of some memory block is done by > usrAppInit directly, and the spawned application use that memory block. > > I've faced the problem with memory. > After spawned application, I found that "memShow" command returns > "Invalid memory block ~~ blah blah blah" something like that. > > So I change the usrAppInit for initialization usr mem block into "task". > so a temporary task initialize some memory block which will be used for > application, and then the memShow problem disappearred. > > This is very simple question. > > The application can use the memory directly which tRoorTask malloed space? > > > Thanks you > > Youngjin --------------------------- Newsgroups: comp.os.vxworks Subject: what is af0?/arp requests missing in tornado 2.1.1 Date: Mon, 28 Jan 2002 17:39:40 +0100 From: Markus Pietrek Organization: FS Forth Systeme GmbH Message-ID: Hi folks, I'm currently trying to port a BSP from Tornado 2.0 to Tornado 2.1.1 (NetARM 50), based on pid7t. My main problem is that now the network interface is not working any more. Circumstances are as following: After booting, bootrom tries to download the kernel with FTP and prints this message "Attached TCP/IP interface to ln0. 0x8fdac50 (tBoot): ln0: can't handle af0 0x8fdac50 (tBoot): ln0: can't handle af0 0x8fdac50 (tBoot): ln0: can't handle af0 0x8fdac50 (tBoot): ln0: can't handle af0 0x8fdac50 (tBoot): ln0: can't handle af0 Attaching network interface lo0... done. Loading... " Then it hangs in "Loading". With a lan analyzer I was able see that it tries to establish a FTP connection with correct IP addresses. So the ethernet drive should work fine. But the preliminary ARP request was missing, therefore the wrong MAC address was used. Manually trying to add the MAC address with arpAdd() didn't worked either as it wasn't stored in the table. So I tried to ping the target from my host, but this produced a lot of more "can't handle af0" messages, this time from the tNetTask and with no response on the network by the target. The message "can't handle af0" is being produced in /tornado/target/lib/obj*/if_loop.o by the command loioctl. That was all I could figure out. So my questions. What does that message "can't handle af0" means and how can I get rid of it? How am I'm able to enable ARP requests? Thanks in advance, Markus Pietrek --------------------------- Newsgroups: comp.os.vxworks Subject: Re: s/w vs. h/w breakpoints Date: 28 Jan 2002 11:53:40 -0500 From: Bill Pringlemeir Organization: Bell Sympatico Message-ID: References: <28ee7c97.0201261016.67543d98@posting.google.com> Sender: bpringlemeir@DeadDuck >>>>> "dale" == dale writes: [snip] dale> An unlimited number of soft breaks are possible; hard dale> breakpoints are limited to what the processor's debug component dale> can support (one, two or few). A hard breakpoint _may_ apply dale> to data references as well as instruction addresses (again, dale> this depends on the processor's architecture); soft breaks can dale> only apply to instruction addresses. I would like to add that many debuggers use the HW breakpoints to implement the step over and single step feature. This makes it simple to implement a step-over in ROM code. Some other set a `single step' interrupt and run the entire routine in single step mode (way slow dude/dudette!) In a custom RTOS (a long long time ago in a universe far far away), I implement this step over by modifying the return value on the stack (or link register on the ARM). The step-over function then runs at full speed and takes a return address to a routine in the image that has a break point instruction. This would allow the vxWorks debugger (or any debugger) to step over without a hardware break point. More over, it would let developers use the hardware breakpoint for other purposes, like the data breakpoint mentioned above. I found this feature disconcerting in the Jeeni tools and the WRS debugger (I am not sure which does this, it might have even been SDS or HMI?). Regards, Bill Pringlemeir. - -- Your programs will start making love. Tomorrow a small bomb would be a good thing to make a stab at. Play Car Bomb. Never get married this week without reading mail first. You'll often be unhappy. Your lucky numbers are 20, 30, and 40. vxWorks FAQ, "http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html" --------------------------- Newsgroups: comp.os.vxworks Subject: VxWorks AE C++ Static constructor Date: Mon, 28 Jan 2002 14:12:43 -0500 From: David Roberge Organization: Universite du Quebec Message-ID: <3C55A2AB.C6A504D0@ele.etsmtl.ca.NOSPAM> Hi, I have a question about C++ static constructors in VxWorks AE. Here is an example of the situation. I have two components, one user shared lib protection domain and one application domain. 1. component A downloaded in user shared lib protection domain 2. component B downloaded in user shared lib protection domain 3. application domain attached to user shared lib protection domain The static constructors of component A & B are called when the application domain attach to the shared library. However I need the constructors of component A to be executed before the constructors of component B. Is there a way to tell VxWorks the order of execution for static C++ constructors in this dynamic loading context ? Many thanks. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: DOS File Corruption Date: 28 Jan 2002 13:32:34 -0800 From: cjohnson@oresis.com (Charles Johnson) Organization: http://groups.google.com/ Message-ID: References: <3c0fdc4b@news.tce.com> SPR #31480 does appear to be visible through WindSurf. At least I couldn't find it. We had implement a change in the ATA device driver where if we get a read error on a sector on our Silicon Tech flash drive, we re-write the sector. While we loose the data, it appears to allow us to recover the sector so that it can be reused. - --Charles Johnson Oresis Communications, Inc. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: New WindSerf, TSR's and SPR's. Date: 28 Jan 2002 13:35:05 -0800 From: cjohnson@oresis.com (Charles Johnson) Organization: http://groups.google.com/ Message-ID: References: <488e459a.0201221531.41030518@posting.google.com> They don't appear to have much of a search mechanism though. - --Charles Johnson Oresis Communications, Inc. john_94501@yahoo.com (John) wrote in message news:<488e459a.0201221531.41030518@posting.google.com>... > Hello, > > You need to go through the Java applet thing on the left side of the > main page - it is labelled Knowledge Base. That gives you an > interactive folder tree to open and lets you get at the SPRs. > > For the TSRs, there is a link in the problem solving panel that will > let you get the status of one of your TSRs. > > Hope that helps, > > John... > > Bill Pringlemeir wrote in message news:... > > How does one access SPRs under the new WindSerf? It is hard to > > install shockwave plug-ins on lynx [in case the info is in the > > tutorial] ... > > > > regards, > > Bill Pringlemeir. --------------------------- Newsgroups: comp.os.vxworks Subject: How to force 10Base-T on a 10/100 NIC in vxworks? Date: 28 Jan 2002 13:45:55 -0800 From: jthomp01@harris.com (Thomp) Organization: http://groups.google.com/ Message-ID: <42438be1.0201281345.fa85268@posting.google.com> I am using an Ampro Little Board P5x equipped with an AMD PCnet+ 10/100 NIC controller with a Pentium configuration. In our chasis, a hardware filter is present on the ethernet line that prevents 100Base-T operation. However, the 100Base-T auto-negotiation signals make it out enough to attempt 100Bt operation. Data will NOT pass at the higher speed and must be limited to 10Base-T. The hardware cannot be changed at this point. So, is there a way to force 10Base-T operation through the driver? James --------------------------- Newsgroups: comp.os.vxworks Subject: Re: New WindSerf, TSR's and SPR's. Date: 28 Jan 2002 17:08:05 -0500 From: Bill Pringlemeir Organization: Bell Sympatico Message-ID: References: <488e459a.0201221531.41030518@posting.google.com> Sender: bpringlemeir@DeadDuck >>> Bill Pringlemeir wrote: >>> How does one access SPRs under the new WindSerf? It is hard to >>> install shockwave plug-ins on lynx [in case the info is in the >>> tutorial] ... >>> john_94501@yahoo.com (John) wrote, >> You need to go through the Java applet thing on the left side of >> the main page - it is labelled Knowledge Base. That gives you an >> interactive folder tree to open and lets you get at the SPRs. >> >> For the TSRs, there is a link in the problem solving panel that >> will let you get the status of one of your TSRs. >>> "Charles" == Charles Johnson writes: Charles> They don't appear to have much of a search mechanism though. I couldn't find it either. I sent an email about this. They told me to call tech support. I imagine tech support will love 100's of developers calling. Has anybody found TSR/SPRs using Netscape? I thought that maybe my contract had run out, but the email confirmed that it is active (as of today...). I can not access the TSR or SPRs as John said ;-( tia, Bill Pringlemeir. - -- Have you ever tried to make your dog generate ESP? Or been given lots of drugs by the government while operating on cauliflower? Or used a conference call to hum God Save the Queen madly with characters from Barney and Friends? You will. And the company that will bring it to you: AT&T. vxWorks FAQ, "http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html" --------------------------- Newsgroups: comp.os.vxworks Subject: Re: How to force 10Base-T on a 10/100 NIC in vxworks? Date: Mon, 28 Jan 2002 22:46:32 +0000 (UTC) From: David Laight Organization: BT Openworld Message-ID: <3C55D51E.97888335@l8s.co.uk> References: <42438be1.0201281345.fa85268@posting.google.com> Thomp wrote: > > I am using an Ampro Little Board P5x equipped with an AMD PCnet+ > 10/100 NIC controller with a Pentium configuration. In our chasis, a > hardware filter is present on the ethernet line that prevents > 100Base-T operation. However, the 100Base-T auto-negotiation signals > make it out enough to attempt 100Bt operation. Data will NOT pass at > the higher speed and must be limited to 10Base-T. The hardware cannot > be changed at this point. > > So, is there a way to force 10Base-T operation through the driver? Well, what you need to do is modify the transceiver's link advertise register through the MII interface. Unfortunately I don't have the PCnet+ book, not the WRS driver :-) David --------------------------- Newsgroups: comp.os.vxworks Subject: Re: SMC driver Date: 28 Jan 2002 17:21:54 -0800 From: ntjiang@yahoo.com (Jason Jiang) Organization: http://groups.google.com/ Message-ID: <12116a88.0201281721.3d47b4fa@posting.google.com> References: <12116a88.0201221830.591bea57@posting.google.com> <2a6118da.0201230241.1604435d@posting.google.com> <12116a88.0201230956.1b276d27@posting.google.com> <2a6118da.0201232119.5dd8368c@posting.google.com> <12116a88.0201241411.705dd3f5@posting.google.com> <2a6118da.0201272346.74ae4b57@posting.google.com> Hi Rashid, I think I'm clear now. Thank you very much. Best Regards, Jason --------------------------- Newsgroups: comp.realtime,comp.realtime.jobs,comp.os.vxworks,comp.os.qnx Subject: Re: Suggestions? Date: 28 Jan 2002 17:48:20 -0800 From: bsd_mike@hotmail.com (Mike) Organization: http://groups.google.com/ Message-ID: References: <4cd34f34.0201261035.4561ad4@posting.google.com> I owuld sure love to see it! - -Mike amcheeseus@yahoo.com (amit choudhary) wrote in message news:<4cd34f34.0201261035.4561ad4@posting.google.com>... > I have developed a real time kernel using XINU Operating system. What > would be the right way of presenting it to the realtime community for > evaluation, suggestions, etc.?. > > I am a fresh CS graduate and want to develop my profile in this field. > Since, I do not have access to commercial RTOSes like Vxworks, PSOS, > WinCE, etc., how do I progress?, Are there any realtime project groups > open for volunteers?. > > amit. --------------------------- Newsgroups: comp.os.vxworks Subject: texcTask exception 768 Date: 28 Jan 2002 18:22:24 -0800 From: Soiba.Mohammad@strixsys.com (Soiba Mohammad) Organization: http://groups.google.com/ Message-ID: Hello everyone, I am getting the following exception in the tExcTask. All our other tasks are running, while the tExtTask is suspended. We are running VxWorks 5.4 on the Motorola 860 platform. Has anyone ever seen this problem before? I would appreciate any help. Exception number 768: Task: 0x3ee418 (tExcTask) data access program counter: 0x947dc access address: 0xaaaba282 machine status: 0x9030 11eaec vxTaskEntry +5c : excTask ([]) 8acb4 excTask +a4 : selWakeupAll ([]) --------------------------- Newsgroups: comp.realtime,comp.realtime.jobs,comp.os.vxworks,comp.os.qnx Subject: Re: Suggestions? Date: 28 Jan 2002 21:45:09 -0500 From: Bill Pringlemeir Organization: Bell Sympatico Message-ID: References: <4cd34f34.0201261035.4561ad4@posting.google.com> Followup-To: comp.realtime Sender: bpringlemeir@DeadDuck [followups set to `comp.realtime' as cross-post is `generic']. >>>>> "amit" == amit choudhary writes: amit> I have developed a real time kernel using XINU Operating amit> system. What would be the right way of presenting it to the amit> realtime community for evaluation, suggestions, etc.?. "http://www.google.com/search?q=XINU" Appears that other people have done work on this. I guess this is the same system. You can setup a page on SourceForge, etc if you wish to make the software free and your additions are significant. amit> I am a fresh CS graduate and want to develop my profile in this amit> field. Since, I do not have access to commercial RTOSes like amit> Vxworks, PSOS, WinCE, etc., how do I progress?, Are there any amit> realtime project groups open for volunteers?. PSOS has/is being merged into vxWorks. vxWorks has a free `prototyper' that you can use to explore the API (see "www.wrs.com"). Most technical people would not hold the fact that you haven't programmed for a specific RTOS against you when applying for a job. You can buy a WinCE device and program for it. Doesn't sound nice? This is about what most real WinCE developers do. Actually many vxWorks developers do the same. It is time consuming to develop hardware and associated drivers. eCos, uCos, and uLinux (Linux and FreeBSD as well) would all be good free software alternatives. These have the added benefit of no royalties. vxWorks is better if you have people, who don't know what an RTOS is, programming for a device. I am unsure about QNX. I thought that they recently expanded beyond the x86 architecture, but I can not see this product info on their web sight. Hopefully this conversation will not degenerate. I have set followup as noted above. Regards, Bill Pringlemeir. - -- DESTROYER. TORPEDOED. SINKING. U-634. vxWorks FAQ, "http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html" --------------------------- Newsgroups: comp.os.vxworks Subject: Did anyone implement 4Ms HDLC channels ON MP860? Date: 28 Jan 2002 18:55:22 -0800 From: lionjackyfish@sina.com.cn (JackyLiu) Organization: http://groups.google.com/ Message-ID: <425faaf3.0201281855.5945e6d0@posting.google.com> Did anyone implement 4Ms HDLC channels ON MP860? PLEASE CONTACT WITH ME . THANKS. --------------------------- Newsgroups: comp.os.vxworks Subject: About STLport Date: Tue, 29 Jan 2002 13:41:36 +0900 From: Yasuyuki Kino Organization: OMRON SOFTWARE Co., Ltd. Message-ID: I'm trying to make vxWorks module with STLport 4.5.1 . However I got the error message. - ----------------------------------------------- d:/home/kino/vss/pg-x/develop/src/STLport-4.5.1/stlport/cstring:27: ../include/string.h: No such file or directory In file included from d:/home/kino/vss/pg-x/develop/src/STLport-4.5.1/stlport/stl/_algobase.h:47 : : - ----------------------------------------------- I put STLport files on my project directory, and added the compile option "-I./STLport-4.5.1/stlport". I'm using Tornado 2.1 , vxWorks 5.4, gcc 2.96 . Please help. Thanks. - -- kino --------------------------- Newsgroups: comp.os.vxworks Subject: problems and bursts in using pci+cache Date: 28 Jan 2002 22:14:43 -0800 From: ruchama@septier.com (Ruchama Goldshmidt) Organization: http://groups.google.com/ Message-ID: <5df340e0.0201282214.10605ee1@posting.google.com> hello vxWorkers! I develop a ppc8260 board, using a shared memory when the master on the bus is a MPC750 board. When enable the cache and the MMU I see that even though I marked the PCI region as cache inhibited the access to the bus is in bursts that means that the processor thinks that this area is cached. When I run a simple application which only used an UDP protocol to send packets from the 750 to the 8260 boards via the PCI bridge, this (or maybe another reason...)caused to many many (almost 80%!!) missed packets . If disable the cache there are no missed packets!!! my definitions in the BAT are: /* D BAT 2 - PCI */ ((PCI_MEM_ADRS & _MMU_UBAT_BEPI_MASK) | _MMU_UBAT_BL_256M | _MMU_UBAT_VS | _MMU_UBAT_VP), ((PCI_MEM_ADRS & _MMU_LBAT_BRPN_MASK) | _MMU_LBAT_PP_RW | _MMU_LBAT_CACHE_INHIBIT | _MMU_LBAT_GUARDED), and in the IBAT is the same and just the guarded bit is off. A. Is anyone know why the access is in bursts , (maybe my definitions are wrong?) B. why there are so many missed packets?? TIA!! regards, Ruchama --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Calling "C" functions from WindSh via serial-line Date: Tue, 29 Jan 2002 08:30:25 +0100 From: "Mads Bligaard Nielsen \(M@H\)" Organization: Ericsson Diax (DXD), Mobile@Home Message-ID: References: Thanks for all your replies - here is one from WindRiver: There is basically two ways to solve this. I guess you are referring to the target shell, which I am describing as solution number one. But in your case I guess suggestion number two is a better approach, please pick a choice ;) 1) Target Shell This shell is executed on the target (as the name implies). The problem you may face here is that you don't have any symbols available (probably the reason why you cannot call these functions). You need to use a so-called built-in symbol table (that is store all the symbol information on the target). This can easily be achieved using the Tornado Project Configurator Tool: Development tool components->symbol table components-> symbol table initialization components->select symbol table initialization-> built-in symbol table After this your image should be alot bigger :( but you can access all public functions / variables directly from the target shell :) 2) WindShell This shell is executed on the host. Most likely via the Tornado IDE. Today you are accessing the debug agent (WDB) on the target via ethernet, but as you describe you cannot do so as you are testing the VLAN implementation. I suggest that you try to configure the target agent to use the serial interface as physical media. However, before doing so you need to reconfigure your VxWorks image with the Project Configurator Tool changing WDB connection from END to serial. Development tool components->WDB agent components-> select WDB connection->WDB serial connection I think this is the only thing you need to change in your vxWorks configuration in order to debug everything using the serial channel instead of the network interface. "Mads Bligaard Nielsen (M@H)" wrote in message news:a33mt2$hlv$1@newstoo.ericsson.se... > Anybody knows why I cannot call "C" functions from the shell that is > available on the serial line (PPC target). > If I start a shell via a TargetAgent (Ethernet) then it is possible - but it > seems like the shell launced via the serial --------------------------- Newsgroups: comp.os.vxworks Subject: Re: spurious interrupt + cache Date: 28 Jan 2002 23:35:08 -0800 From: ruchama@septier.com (Ruchama Goldshmidt) Organization: http://groups.google.com/ Message-ID: <5df340e0.0201282335.1d8f6f3e@posting.google.com> References: <5df340e0.0201220441.6b3b3999@posting.google.com> <3C4E9945.572AB6D6@NOSPAM.bcs.org.uk> Thanks , it's really helped me solve the problem!!! Ruchama Graham Baxter wrote in message news:<3C4E9945.572AB6D6@NOSPAM.bcs.org.uk>... > Ruchama, > > This solution obtained from Wind River worked for me with exactly the > same problem. > > I hope it helps. > > > Regards, > > > Graham Baxter (VxWorks and pSOS BSP'S) > Freelance Software Engineer > gbaxter@NOSPAM.bcs.org.uk > > ------------------------------- > > The solution to this problem was to modify the m8260IntrCtl.c source > file > in the following way: > > In the m8260IntrCtl.c file, in the function m8260IntrDeMux(), right > after > it enables only higher level priorities in the interrupt mask registers > it calls the interrupt unlock which causes an unknown interrupt to be > generated because the values in the registers haven't settled yet. > By adding a call to CACHE_PIPE_FLUSH() the error message disappears and > everything seems to work OK. Here's a snippet of the code: > > ... > /* enable only the higher priority interrupts */ > > * M8260_SIMR_L(immrVal)&= > enableHighPriInts[IVEC_TO_INUM((VOIDFUNCPTR *)intVec)].simr_l; > * M8260_SIMR_H(immrVal)&= > enableHighPriInts[IVEC_TO_INUM((VOIDFUNCPTR *)intVec)].simr_h; > > CACHE_PIPE_FLUSH (); > > /* unlock the interrupt */ > > intUnlock (_PPC_MSR_EE); > > /* call the Interrupt Handler */ --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Did anyone implement 4Ms HDLC channels ON MP860? Date: 28 Jan 2002 23:58:05 -0800 From: houcheng@ms17.hinet.net (houcheng) Organization: http://groups.google.com/ Message-ID: References: <425faaf3.0201281855.5945e6d0@posting.google.com> lionjackyfish@sina.com.cn (JackyLiu) wrote in message news:<425faaf3.0201281855.5945e6d0@posting.google.com>... > Did anyone implement 4Ms HDLC channels ON MP860? PLEASE CONTACT WITH ME . > THANKS. Hmm.. I've implement 860SAR IP Over ATM driver on FADS860 demoboard. what's up ? my email is houcheng@NoSpam.turbonetworks.comt.w --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Uninitialized Interrupt! Vector number 47 Date: 29 Jan 2002 00:04:44 -0800 From: houcheng@ms17.hinet.net (houcheng) Organization: http://groups.google.com/ Message-ID: References: nukala_s@yahoo.com (Srinivas) wrote in message news:... > I am using VxWorks for PentiumPro with Intel PRO/100B. When I am doing > continous read/write over the network I get the following message: > > Uninitialized Interrupt! > Vector number 47 (0-255) > .... > .... > > If I am not mistaken this vector number corresponds to IRQ 15. My > Intel > PRO/100B NIC card is assigned IRQ 11. But I suspect that the occurance > of this spurious interrupt has something to do with the NIC card. > > I would like to know if anyone had similar problems and > already have a solution for it. I would greatly appreciate any > pointers. Yes, I has met the same problem with you. but the "Uninitialized Interrupt!" message is not harmful to your system. In previous discussions, someone has post the solution to it. Try to find that post. Lin Houcheng houcheng@NoSpam.turbonetworks.com.tw --------------------------- Newsgroups: comp.os.vxworks Subject: Re: WWW.GIWERSWORLD.ORG Date: Tue, 29 Jan 2002 10:10:42 +0200 From: "Ofer Goren" Organization: Verio Message-ID: References: <27238F021CD1@62.42.230.225> "The goal of the establishment of israel is not to fight antisemisizim, but to ignore it." Levi Eshkol, Isreal Prime Minister, 1967. - -- Ofer Goren "Matt Giwer" wrote in message news:27238F021CD1@62.42.230.225... > FOR ALL YOUR ANTISEMITIC NEEDS. > > > REMEMBER: TODAY'S KIKE IS TOMORROW'S LAMPSHADE. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: what is af0?/arp requests missing in tornado 2.1.1 Date: Tue, 29 Jan 2002 16:47:18 +0800 From: "kola" Organization: Changzhou Telecom Message-ID: References: "can't handle af0" is printf at network device driver ouput function, it means that the driver can not find the "Address families",and af0 means "AF_UNSPEC". So there must some wrong of your network device driver or your configuration of the network device driver has some wrong. And which network device did you use? If your network device driver is end driver, you must sure your pEnd->flags has not "IFF_NOARP" flags,and you can change your flags by xxxioctl() function. Best Regards! Kola "Markus Pietrek" wrote in message news:a33unl$6pc$03$1@news.t-online.com... > Hi folks, > > I'm currently trying to port a BSP from Tornado 2.0 to Tornado 2.1.1 > (NetARM 50), based on pid7t. > > My main problem is that now the network interface is not working any more. > Circumstances are as following: > After booting, bootrom tries to download the kernel with FTP and > prints this message > > "Attached TCP/IP interface to ln0. > 0x8fdac50 (tBoot): ln0: can't handle af0 > 0x8fdac50 (tBoot): ln0: can't handle af0 > 0x8fdac50 (tBoot): ln0: can't handle af0 > 0x8fdac50 (tBoot): ln0: can't handle af0 > 0x8fdac50 (tBoot): ln0: can't handle af0 > Attaching network interface lo0... done. > Loading... > " > > Then it hangs in "Loading". With a lan analyzer I was able see that it > tries to establish a FTP connection with correct IP addresses. So the > ethernet drive should work fine. But the preliminary ARP request was > missing, therefore the wrong MAC address was used. > > Manually trying to add the MAC address with arpAdd() didn't worked either > as it wasn't stored in the table. > > So I tried to ping the target from my host, but this produced a lot of more > "can't handle af0" messages, this time from the tNetTask and with no > response on the network by the target. > > The message "can't handle af0" is being produced in > /tornado/target/lib/obj*/if_loop.o by the command loioctl. That was all I > could figure out. > > So my questions. > > What does that message "can't handle af0" means and how can I get rid of it? > How am I'm able to enable ARP requests? > > Thanks in advance, > Markus Pietrek --------------------------- Newsgroups: comp.os.vxworks Subject: Re: About STLport Date: Tue, 29 Jan 2002 11:04:07 +0200 From: "Ofer Goren" Organization: Verio Message-ID: References: r u sure it is ../include/string.h? Shouldn't it be ../h/string.h? - -- Ofer Goren Iamba Technologies "Yasuyuki Kino" wrote in message news:ur8o9zs3j.fsf@mx.omronsoft.co.jp... > > I'm trying to make vxWorks module with STLport 4.5.1 . > > However I got the error message. > > ----------------------------------------------- > d:/home/kino/vss/pg-x/develop/src/STLport-4.5.1/stlport/cstring:27: > ../include/string.h: No such file or directory > In file included from > d:/home/kino/vss/pg-x/develop/src/STLport-4.5.1/stlport/stl/_algobase.h:47 > : > : > ----------------------------------------------- > > I put STLport files on my project directory, and added the > compile option "-I./STLport-4.5.1/stlport". > > I'm using Tornado 2.1 , vxWorks 5.4, gcc 2.96 . > > Please help. > > Thanks. > > -- > kino --------------------------- Newsgroups: comp.os.vxworks Subject: Curses lib under vxworks Date: Tue, 29 Jan 2002 16:56:32 +0800 From: "Nison.liu" Organization: Changzhou Telecom Message-ID: Hi: Do you have ever use Linux/Unix curses lib under vxWorks? I want to port a Linux/Unit text editor to vxworks, then under vxworks target without host,i can edit some simple text files.But almost all these text editors need curses lib support,under vxworks how can i use it and do it? Thanks! Dongliang Liu 01/29/2002 Nison.liu@utstar.com --------------------------- Newsgroups: comp.os.vxworks Subject: Re: New WindSerf, TSR's and SPR's. Date: 29 Jan 2002 10:18:12 GMT From: Johan Borkhuis Organization: Agere Systems Message-ID: References: <488e459a.0201221531.41030518@posting.google.com> Bill Pringlemeir wrote: > I couldn't find it either. I sent an email about this. They told me > to call tech support. I imagine tech support will love 100's of > developers calling. Has anybody found TSR/SPRs using Netscape? I > thought that maybe my contract had run out, but the email confirmed > that it is active (as of today...). I can not access the TSR or SPRs > as John said ;-( You could try the new message boards, there is one for WindSurf, but I don't know if that will help much. Groeten, Johan - -- o o o o o o o . . . _____________________________ o _____ || Johan Borkhuis | .][__n_n_|DD[ ====_____ | borkhuis@agere.com | >(________|__|_[_________]_|__________________________| _/oo OOOOO oo` ooo ooo 'o!o!o o!o!o` === VxWorks FAQ: http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html === --------------------------- Newsgroups: comp.os.vxworks Subject: problem with -mlongcall Date: 29 Jan 2002 03:17:52 -0800 From: kumarpavan@lycos.com (pavan) Organization: http://groups.google.com/ Message-ID: <307af50b.0201290317.689ad29e@posting.google.com> hello all, I am just facing some strange behaviour with -mlongcall option. I am using t2cp4 , vxworks5.4, PPC604 target with ccppc compiler (cygnus-2.7.2-960126 egcs-971225 tornado 2.0 ). The following piece of code when compiled with -mlongcall option gives a machine check exception but works fine if you remove -mlongcall option. #include class A { int i; public: A():i(0){} ~A(){} }; extern "C" int checkMine() { while(true) { A a; } return 0; } output: - ->sp checkMine task spawned: id = 0x143a5d0, name = t16 value = 21210576 = 0x143a5d0 - -> machine check : Master Abort Received MPC107, Error Detection Register 1 (0xc1) = 0x00 MPC107, Error Detection Register 2 (0xc5) = 0x00 MPC107, 60x Bus Error Status Register (0xc3) = 0x50 MPC107, PCI Bus Error Status Register (0xc7) = 0x00 MPC107, Bus Error Address Register (0xc8) = 0x20000000 machine check Exception next instruction address: 0x0143a858 Machine Status Register: 0x0004b030 Condition Register: 0x22000000 Task: 0x143a5d0 "t16" - ->0x0143a858 value = 21211224 = 0x143a858 = checkMine + 0x70 any help will be appreciated. TIA, pavan. --------------------------- Newsgroups: comp.os.vxworks,comp.realtime,comp.embedded,comp.answers,news.answers Subject: [comp.os.vxworks] FAQ - index Date: 29 Jan 2002 11:25:45 GMT From: johan@borksoft.xs4all.nl (Johan Borkhuis) Organization: none Message-ID: Followup-To: comp.os.vxworks Archive-name: vxworks-faq/index Posting-Frequency: monthly Last-modified: 2001/12/31 Version: 1.41 URL: http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html Copyright: (c) 1999 - 2002 Johan Borkhuis Maintainer: Johan Borkhuis This is the index of the FAQ for the newsgroup comp.os.vxworks. The FAQ itself can be found at http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html Contents: 1. Tool problems 1.1 Compiler and Linker 1.2 Debugger 1.3 FTP 1.4 Host tools 1.5 Installation 1.6 Make 1.7 Project facility 1.7.1 Hierarchical projects 1.7.2 Super projects 1.8 Target Server 1.9 Target Shell 1.10 Telnet 1.11 Tornado 1.11.1 Tornado (General) 1.11.2 Tornado (Windows) 1.12 Version Control 1.13 Visual Studio integration 1.14 Windsh 1.15 WindView 2. Changes/additions to be made to BSP's 2.1 Different BSP's 2.2 Processor specific issues 2.2.1 Power PC 2.2.2 I960 2.2.3 MIPS 2.2.4 ARM 2.3 Bootstrap 2.4 Adding component to your VxWorks configuration 2.5 VME problems 2.6 VxWorks environment 2.6.1 usrSerial.c 2.6.2 prj_vxworks.tcl 2.7 PCI problems 3. File System problems 3.1 Dos file system 3.2 Flash File system 3.3 Floppy-disk File system 3.4 RAM-disk File system 3.5 General file system questions 4. Network 4.1 Configuration problems 4.2 ARP 4.3 DHCP 4.4 FTP and TFTP 4.5 PPP 4.5.1 PPP on Windows 95 4.5.2 PPP on Windows NT 4.5.3 PPP on Solaris 4.6 Sockets 4.7 Telnet 4.8 Other network related questions 5. VxWorks Questions 5.1 C++ issues 5.2 Communication problems 5.3 Interrupts 5.4 LoadModule problems 5.5 pthreads 5.6 Reboot 5.7 Semaphores 5.8 Simulator (VxSim) 5.9 Task related items 5.10 Time/timer related items 5.11 Wind Web Server related items 5.12 Zinc/windML related items 5.13 Other items 6. Extra information 6.1 Hard delay 6.2 Memory leaks 6.3 Corba engines 6.4 Web servers 6.5 NTP usage 6.6 Performace / Benchmarks 6.7 SNMP 6.8 Lint 6.9 Encryption Other Links The News Department Overall question index Overall Sources index Change history - -- o o o o o o o . . . _____J_o_h_a_n___B_o_r_k_h_u_i_s___ o _____ || http://www.xs4all.nl/~borkhuis | .][__n_n_|DD[ ====_____ | johan@borksoft.xs4all.nl | >(________|__|_[_________]_|________________________________| _/oo OOOOO oo` ooo ooo 'o!o!o o!o!o` ==== VxWorks-FAQ: http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html ==== --------------------------- Newsgroups: comp.os.vxworks Subject: Why "dosFsDevInit failed"? Date: 29 Jan 2002 11:00:21 GMT From: Charly Message-ID: <2002129-12021-37298@foorum.com> Hello, When I start a big application on my PCPentium target, the shell display "dosFsDevInit failed" And the target seems dead, no Shell, no targer server connection. Do you have an idea where I could look? VxWorks configuration? - -- Use our news server 'news.foorum.com' from anywhere. More details at: http://nnrpinfo.go.foorum.com/ --------------------------- Newsgroups: comp.os.vxworks Subject: Death of VxWorks when "sysClkConnect" Date: 29 Jan 2002 11:09:02 GMT From: Charly Message-ID: <2002129-1292-461828@foorum.com> Death of VxWorks when "sysClkConnect" Hello, when I debug, The following lines make a stop of VxWorks debug. The Task "Wdb.." is suspended and the target server connection is in default. sysClkDisable(); sysClkConnect((FUNCPTR)vxw_mon_it, 0); sysClkEnable(); How could I avoid this problem? Thanks charly - -- Use our news server 'news.foorum.com' from anywhere. More details at: http://nnrpinfo.go.foorum.com/ --------------------------- Newsgroups: comp.os.vxworks Subject: 860SAR UTOPIA.... What is syncronization signals? Date: 29 Jan 2002 03:59:49 -0800 From: 10month@hanmail.net (Changwon Lee) Organization: http://groups.google.com/ Message-ID: Hi... I'm reading an article, "MPC8xx ATM Supplement", to activate UTOPIA interface in 860SAR. But I can not understand the meaning of syncronization signal. Context in "MPC8xx ATM Supplement"... 5.1.6 APC Initialization and Operating Considerations As long as the APC timer(timer4) is not active, the APC parameters may be initialized in any order. Timer 4 must be initialized last, following the initialization of the APC tables of all the ATM ports in the system. Failure to initialize timer 4 last will allow the scheduling algorithm to start prematurely, resulting in unpredictable behavior. Furthermore, the APC table parameters must be initialized before any TRANSMIT_CHANNEL_ACTIVATE commands are issued. However, these commands may be issued at any time, wheter the APC timer is active or inactive. For more information, see the description of the TRANSMIT_ACTIVATE_CAHNNEL command. Also, note that the physical interface (serial or UTOPIA) must be enabled and its associated clocks and SYNCHRONIZATION SIGNALS must be active before the APC timer is activated. Otherwise, the transmit queue will simply overflow (causing APCO interrupt). - ---------------------------------- What is the syncronization signal? What should I do for this syncronization signal in H/W and S/W. Thank you for further comments! ^_^ Have a nice day or night! --------------------------- End of New-News digest ********************** From vxwexplo-errs@csg.lbl.gov Wed Jan 30 04:03:23 2002 From: Vxworks Exploder Date: Wed Jan 30 04:03:25 PST 2002 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Wed Jan 30 04:03:18 PST 2002 Subject: Re: what is af0?/arp requests missing in tornado 2.1.1 Subject: A shell command to reboot Vxworks? Subject: IPCP ippool? Subject: Re: A shell command to reboot Vxworks? Subject: Re: problems and bursts in using pci+cache Subject: Re: Curses lib under vxworks Subject: WINDWEB_AUTO and javascript Subject: Re: what is af0?/arp requests missing in tornado 2.1.1 Subject: Tornado doesn't find stdc++ headers Subject: Tornado VxSim & PEG GUI Subject: Ton of errors when creating a project. Subject: Re: A shell command to reboot Vxworks? Subject: Re: About STLport Subject: Re: Curses lib under vxworks Subject: Re: Death of VxWorks when "sysClkConnect" Subject: Re: curses Subject: Re: Problem in memory which allocated by tRootTask and using it by application Subject: Re: problem with -mlongcall Subject: How to get actual addresses for boot codes (romInit.s) Subject: Re: Death of VxWorks when "sysClkConnect" Subject: Re: VxWorks Pipes Subject: Here is a short example which crash Subject: Re: Here is a short example which crash Subject: Re: How to get actual addresses for boot codes (romInit.s) Subject: Re: New WindSerf, TSR's and SPR's. Subject: changing the target name Subject: Re: WINDWEB_AUTO and javascript Subject: Re: changing the target name Subject: Re: How to get actual addresses for boot codes (romInit.s) Subject: Anyone tried to modify 8260 interrupt priority? Subject: Re: VxWorks AE C++ Static constructor Subject: [Help Me] Socket binding error Subject: Modify MAC addr For Enet Driver Subject: Re: missing ftime and alarm?? Subject: Re: New WindSerf, TSR's and SPR's. Subject: Re: Ton of errors when creating a project. Subject: Re: Cache problem Subject: Re: Cache problem Subject: Binding a socket to a interface in VxWorks Subject: login issue Subject: Re: VxWorks AE C++ Static constructor Subject: Configuration of a serial device to run full duplex Subject: Easily Make $1000's Per Month Selling On Ebay! Subject: Re: Death of VxWorks when Subject: IOP IQ80310 - What addr do i download the VxWorks Image ? Subject: Re: Suggestions? ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: what is af0?/arp requests missing in tornado 2.1.1 Date: Tue, 29 Jan 2002 13:24:42 +0100 From: Markus Pietrek Organization: FS Forth Systeme GmbH Message-ID: References: kola wrote: > "can't handle af0" is printf at network device driver ouput function, > it > means that the driver can not > find the "Address families",and af0 means "AF_UNSPEC". > So there must some wrong of your network device driver or your > configuration of the network device driver > has some wrong. And which network device did you use? I'm using the builtin ethernet device of the NetARM with an end driver. > If your network device driver is end driver, you must sure your > pEnd->flags has not "IFF_NOARP" flags,and > you can change your flags by xxxioctl() function. Thank you. There are some ioctl features missing (like EIOCGHDRLEN and EIOCGNPT and some more). Now I have to find out what they should do :-/ Bye, Markus Pietrek --------------------------- Newsgroups: comp.os.vxworks Subject: A shell command to reboot Vxworks? Date: 29 Jan 2002 12:37:43 GMT From: Charly Message-ID: <2002129-133743-405035@foorum.com> Hello again, Is there a possibility from Shell to restart VxWorks? Thanks. charly - -- Use our news server 'news.foorum.com' from anywhere. More details at: http://nnrpinfo.go.foorum.com/ --------------------------- Newsgroups: comp.os.vxworks Subject: IPCP ippool? Date: 29 Jan 2002 04:48:42 -0800 From: sardine_sz@163.net (sardine) Organization: http://groups.google.com/ Message-ID: hi gurus, now i use windNet PPP&PPPOE1.2+patch1.0? after config usrBasicPPPCfg.c and usrBasicPPPOECfg.c ,i can dial to the PPPOE Server successful, but there have a problem: when two pppoe client dial to the PPPOE AC,the all client get a same ip address? how i config ip pool on IPCP of PPP ??? any advice will be apprieciate !!! thx first!!! --------------------------- Newsgroups: comp.os.vxworks Subject: Re: A shell command to reboot Vxworks? Date: 29 Jan 2002 12:42:15 GMT From: Charly Message-ID: <2002129-134215-365133@foorum.com> References: <2002129-133743-405035@foorum.com> I found it: "reboot" Thanks - -- Use our news server 'news.foorum.com' from anywhere. More details at: http://nnrpinfo.go.foorum.com/ --------------------------- Newsgroups: comp.os.vxworks Subject: Re: problems and bursts in using pci+cache Date: 29 Jan 2002 05:43:09 -0800 From: ruchama@septier.com (Ruchama Goldshmidt) Organization: http://groups.google.com/ Message-ID: <5df340e0.0201290543.3604a3d8@posting.google.com> References: <5df340e0.0201282214.10605ee1@posting.google.com> ruchama@septier.com (Ruchama Goldshmidt) wrote in message news:<5df340e0.0201282214.10605ee1@posting.google.com>... Hello!! just want to add that I used a caopyback mode of the cache!! Thanks!! > hello vxWorkers! > > I develop a ppc8260 board, using a shared memory when the master on > the bus is a MPC750 board. > > When enable the cache and the MMU I see that even though I marked the > PCI region as cache inhibited the access to the bus is in bursts that > means that the processor thinks that this area is cached. > > When I run a simple application which only used an UDP protocol to > send packets from the 750 to the 8260 boards via the PCI bridge, this > (or maybe another reason...)caused to many many (almost 80%!!) missed > packets . > If disable the cache there are no missed packets!!! > > my definitions in the BAT are: > > /* D BAT 2 - PCI */ > ((PCI_MEM_ADRS & _MMU_UBAT_BEPI_MASK) | _MMU_UBAT_BL_256M | > _MMU_UBAT_VS | _MMU_UBAT_VP), > ((PCI_MEM_ADRS & _MMU_LBAT_BRPN_MASK) | _MMU_LBAT_PP_RW | > _MMU_LBAT_CACHE_INHIBIT | _MMU_LBAT_GUARDED), > > and in the IBAT is the same and just the guarded bit is off. > > A. Is anyone know why the access is in bursts , (maybe my definitions > are wrong?) > > B. why there are so many missed packets?? > > TIA!! > > regards, > Ruchama --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Curses lib under vxworks Date: 29 Jan 2002 13:57:20 GMT From: Johan Borkhuis Organization: Agere Systems Message-ID: References: "Nison.liu" wrote: > Hi: > > Do you have ever use Linux/Unix curses lib under vxWorks? > I want to port a Linux/Unit text editor to vxworks, then under > vxworks > target without host,i can edit some simple text files.But almost all > these text editors need curses lib support,under vxworks how can i use > it and do it? > Thanks! ftp://ftp.atd.ucar.edu/pub/archive/vxworks/vx/ There you can find the files vwcurses01..08. Groeten, Johan - -- o o o o o o o . . . _____________________________ o _____ || Johan Borkhuis | .][__n_n_|DD[ ====_____ | borkhuis@agere.com | >(________|__|_[_________]_|__________________________| _/oo OOOOO oo` ooo ooo 'o!o!o o!o!o` === VxWorks FAQ: http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html === --------------------------- Newsgroups: comp.os.vxworks Subject: WINDWEB_AUTO and javascript Date: Tue, 29 Jan 2002 09:25:51 -0500 From: david lindauer Message-ID: <3C56B0EF.9070502@notifier-is.net> hi, we want to use some javascript in conjunction with form input. For example declare an 'onchange' event for combo boxes. We put the onchange event in the tag, but vxworks will not propagate that when creating the web page with the SSI functions. Viewing the page source indicates that it is not uploaded to the browser. Is there any work around to get this to work? Thanks, David --------------------------- Newsgroups: comp.os.vxworks Subject: Re: what is af0?/arp requests missing in tornado 2.1.1 Date: Tue, 29 Jan 2002 17:31:14 +0100 From: Markus Pietrek Organization: FS Forth Systeme GmbH Message-ID: References: Ok, got it working with using this line: if (END_MIB_INIT (pEndObj, M2_ifType_ethernet_csmacd , ieAdrs, 6, ETHERMTU, EI_SPEED) == ERROR) instead of if (END_MIB_INIT (pEndObj, M2_ifType_iso88023_csmacd , ieAdrs, 6, ETHERMTU, EI_SPEED) == ERROR) Bye, Markus Pietrek --------------------------- Newsgroups: comp.os.vxworks Subject: Tornado doesn't find stdc++ headers Date: 29 Jan 2002 16:46:17 GMT From: charles@pentek.com (Charles Krug) Organization: Concentric Internet Services Message-ID: Reply-To: charles@pentek.com List: I'm having trouble building a project from c++ sources under Tornado (WinNT). It's complaining "no rule to make iostream." iostream, along with the rest of the stdc++ headers, is present in the Tornado heirarchy, but the compiler isn't finding them. Where do I set the include search paths? Most environments I use correctly resolve include paths based on the source file type, in this case cxx. Is .cxx the correct extension for c++ files under Tornado 2.0-WinNT? Thank you. Charles - -- Charles Krug, Jr. Systems Engineer Pentek Corp 1 Park Way Upper Saddle River, NJ 07458 --------------------------- Newsgroups: comp.os.vxworks Subject: Tornado VxSim & PEG GUI Date: Tue, 29 Jan 2002 18:10:34 +0100 From: "Mehmet Demirel" Organization: Swisscom IP+ (post doesn't reflect views of Swisscom) Message-ID: Reply-To: "Mehmet Demirel" Hi, looking for people who ported PEG GUI from Swellsoftware to VxWorks ? I would like to discuss somethings concirning output in VxSim and so on. Cheers Mehmet --------------------------- Newsgroups: comp.os.vxworks Subject: Ton of errors when creating a project. Date: Tue, 29 Jan 2002 12:29:23 -0500 From: Doug Dotson Organization: Verio Message-ID: <3C56DBF3.84AFD130@digidata.com> I am trying to create a bootable project based upon a new BSP. The BSP is properly located in the \tornado\target\config, the Makefile has been modified to have proper paths to the commDrv and CoreDrv directories. When I create the project a window "Project Creation Error" pops up with hunderds of #define statements and further dow it appears that some include files could not be located stating "Permission Denied". I am working on an NT box with Tornado 2.0.2 for MIPS. I have already successfully built the same type of project using a different BSP. Thanks, Doug --------------------------- Newsgroups: comp.os.vxworks Subject: Re: A shell command to reboot Vxworks? Date: 29 Jan 2002 09:59:12 -0800 From: saint_p@yahoo.com (Brian St. Pierre) Organization: http://groups.google.com/ Message-ID: <7363e2a5.0201290959.22a6b8f8@posting.google.com> References: <2002129-133743-405035@foorum.com> <2002129-134215-365133@foorum.com> Charly wrote in message news:<2002129-134215-365133@foorum.com>... > I found it: "reboot" > Thanks Ctrl-X will also cause a reboot. - -Brian --------------------------- Newsgroups: comp.os.vxworks Subject: Re: About STLport Date: Tue, 29 Jan 2002 12:47:01 -0500 From: Fran Horan Organization: Johns Hopkins University Applied Physics Lab, Laurel, MD, USA Message-ID: <3C56E012.91EC81FC@jhuapl.edu> References: You should have a STL supplied as part of VxWorks 5.4 and could try using that instead of STLport. Take a look at target/h/*stl* to see if you have it. Yasuyuki Kino wrote: > I'm trying to make vxWorks module with STLport 4.5.1 . > > However I got the error message. > > ----------------------------------------------- > d:/home/kino/vss/pg-x/develop/src/STLport-4.5.1/stlport/cstring:27: > ../include/string.h: No such file or directory > In file included from > d:/home/kino/vss/pg-x/develop/src/STLport-4.5.1/stlport/stl/_algobase.h:47 > : > : > ----------------------------------------------- > > I put STLport files on my project directory, and added the > compile option "-I./STLport-4.5.1/stlport". > > I'm using Tornado 2.1 , vxWorks 5.4, gcc 2.96 . > > Please help. > > Thanks. > > -- > kino --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Curses lib under vxworks Date: 29 Jan 2002 10:02:02 -0800 From: saint_p@yahoo.com (Brian St. Pierre) Organization: http://groups.google.com/ Message-ID: <7363e2a5.0201291002.56d11e59@posting.google.com> References: <7363e2a5.0201290733.52dc8f9b@posting.google.com> saint_p@yahoo.com (Brian St. Pierre) wrote in message news:<7363e2a5.0201290733.52dc8f9b@posting.google.com>... > "Nison.liu" wrote in message news:... > > > > I want to port a Linux/Unit text editor to vxworks, then under vxworks > > There is a curses lib at ftp://ftp.atd.ucar.edu/pub/archive/vxworks. I > saw a reference in this NG for a vi clone in that archive as well, but Found it: get 'stevie' from this archive. It is a vi clone, which will probably be easier than going to the trouble of doing your own port... - -Brian --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Death of VxWorks when "sysClkConnect" Date: Tue, 29 Jan 2002 14:17:55 -0000 From: "Dave Korn" Organization: Lumber Cartel (tinlc) Members #2234-2237 (owing to browser refresh) Message-ID: <9JB58.37$5o.67861@newsr2.u-net.net> References: <2002129-1292-461828@foorum.com> "Charly" wrote in message news:2002129-1292-461828@foorum.com... > > Death of VxWorks when "sysClkConnect" > Hello, > when I debug, The following lines make a stop of VxWorks debug. The Task "Wdb.." > is suspended and the target server connection is in default. > > sysClkDisable(); > sysClkConnect((FUNCPTR)vxw_mon_it, 0); > sysClkEnable(); > > How could I avoid this problem? Fix the bug in vxw_mon_it that is causing a crash when it is called from the clock interrupt. Or, if you don't want vxw_mon_it to be called by the clock interrupt, don't connect it to the clock interrupt. DaveK - -- moderator of alt.talk.rec.soc.biz.news.comp.humanities.meow.misc.moderated.meow Burn your ID card! http://www.optional-identity.org.uk/ Help support the campaign, copy this into your .sig! Proud Member of the Exclusive "I have been plonked by Davee because he thinks I'm interesting" List Member # Master of Many Meowing Minions --------------------------- Newsgroups: comp.os.vxworks Subject: Re: curses Date: Tue, 29 Jan 2002 15:49:06 -0000 From: "Dave Korn" Organization: Lumber Cartel (tinlc) Members #2234-2237 (owing to browser refresh) Message-ID: References: "Weddick" wrote in message news:KgqX7.12617$kK1.91125@rwcrnsc52... > Has anyone done a port or know of a port for using curses on a X86 Vxworks > System? > > Thanks, Did you get an answer to this one yet? If not, do a google search for "vxworks ncurses ftp.ucar.edu" and you should find it. DaveK - -- moderator of alt.talk.rec.soc.biz.news.comp.humanities.meow.misc.moderated.meow Burn your ID card! http://www.optional-identity.org.uk/ Help support the campaign, copy this into your .sig! Proud Member of the Exclusive "I have been plonked by Davee because he thinks I'm interesting" List Member # Master of Many Meowing Minions --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Problem in memory which allocated by tRootTask and using it by application Date: Tue, 29 Jan 2002 15:59:41 -0000 From: "Dave Korn" Organization: Lumber Cartel (tinlc) Members #2234-2237 (owing to browser refresh) Message-ID: References: <799f3d42.0201252300.a1cbcc@posting.google.com> <3c5575dd$1@brateggebdc5.br-automation.co.at> "Werner Schiendl" wrote in message news:3c5575dd$1@brateggebdc5.br-automation.co.at... > [SNIP] or you overwrite some bytes _before_ the allocated area. > (The information for memory manager in VxWorks is stored here). Also *after* the allocated area, where there is often the memory block header for the next free/used block. Pedantic, I know, but worth pointing out that it's just as deadly to run off the end of an array as to negative-index your way off the start. DaveK - -- moderator of alt.talk.rec.soc.biz.news.comp.humanities.meow.misc.moderated.meow Burn your ID card! http://www.optional-identity.org.uk/ Help support the campaign, copy this into your .sig! Proud Member of the Exclusive "I have been plonked by Davee because he thinks I'm interesting" List Member # Master of Many Meowing Minions --------------------------- Newsgroups: comp.os.vxworks Subject: Re: problem with -mlongcall Date: Tue, 29 Jan 2002 15:12:09 -0000 From: "Dave Korn" Organization: Lumber Cartel (tinlc) Members #2234-2237 (owing to browser refresh) Message-ID: <9JB58.38$5o.67861@newsr2.u-net.net> References: <307af50b.0201290317.689ad29e@posting.google.com> "pavan" wrote in message news:307af50b.0201290317.689ad29e@posting.google.com... > hello all, > > I am just facing some strange behaviour with -mlongcall option. > I am using t2cp4 , vxworks5.4, PPC604 target with ccppc compiler > (cygnus-2.7.2-960126 egcs-971225 tornado 2.0 ). > The following piece of code when compiled with -mlongcall option gives > a machine check exception but works fine if you remove -mlongcall > option. Couldn't reproduce that here using either 2.7.2 or 2.95.4-vec. Strange. > ->sp checkMine > task spawned: id = 0x143a5d0, name = t16 > value = 21210576 = 0x143a5d0 > -> > machine check : Master Abort Received > MPC107, Error Detection Register 1 (0xc1) = 0x00 > MPC107, Error Detection Register 2 (0xc5) = 0x00 > MPC107, 60x Bus Error Status Register (0xc3) = 0x50 > MPC107, PCI Bus Error Status Register (0xc7) = 0x00 > MPC107, Bus Error Address Register (0xc8) = 0x20000000 OK, so something in the code has tried to read/write address 0x20000000, and you've got nothing mapped there, so the bus controller generates a fault when no device answers on the pci bus after a timeout. > machine check > Exception next instruction address: 0x0143a858 > Machine Status Register: 0x0004b030 > Condition Register: 0x22000000 > Task: 0x143a5d0 "t16" > ->0x0143a858 > value = 21211224 = 0x143a858 = checkMine + 0x70 Right, at that stage it would have been useful to also do a "ti 0x143a5d0" to see what was in all the registers, and a "l checkMine, 50" to show us the actual assembler code that was being executed. It might also be informative to add a line "printf ("Address of A is $%08x\n", &a) during the while (true) loop. And using the -save-temps flag when compiling in order to have a look at the .s file generated by this code > while(true) I couldn't find where in the header files this lower-case "true" gets defined, usually people use the upper case version. The only place I *could* find it defined was in the stl_config.h header file. You should perhaps verify that this is defined the way you expect it is. DaveK - -- moderator of alt.talk.rec.soc.biz.news.comp.humanities.meow.misc.moderated.meow Burn your ID card! http://www.optional-identity.org.uk/ Help support the campaign, copy this into your .sig! Proud Member of the Exclusive "I have been plonked by Davee because he thinks I'm interesting" List Member # Master of Many Meowing Minions --------------------------- Newsgroups: comp.os.vxworks Subject: How to get actual addresses for boot codes (romInit.s) Date: 29 Jan 2002 10:39:50 -0800 From: ryeung@earthlink.net (Raymond Yeung) Organization: http://groups.google.com/ Message-ID: I need to troubleshoot a problem that occurs during system bootup. The board uses MPC8260. Logic analyzer indicates the board branches abnormally to an address. One way to track this problem is with emulator. Other than this approach, is there any way that I can generate a file that maps out actual addresses of all instructions in boot code (esp. those assembler code in romInit.s)? Since an emulator knows where to load the text,bss,data segments, the information must have been built somewhere. Which utilities should I use? What's the procedure? --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Death of VxWorks when "sysClkConnect" Date: Tue, 29 Jan 2002 19:49:23 +0100 From: "Werner Schiendl" Message-ID: <3c56eeb5@brateggebdc5.br-automation.co.at> References: <2002129-1292-461828@foorum.com> Hi, what are you trying exactly? Debugging the code in your post? I don't think this can work if you are debugging with the debugger integrated with Tornado II in task mode. You need to debug in system mode, because while you change the system clock there will probably be a problem with anything that relies on time (like taskDelay() calls). The normal network driver/stack probably will depend on a working system clock. System mode debugging uses polled mode drivers which should work without system clock. hth Werner "Charly" wrote in message news:2002129-1292-461828@foorum.com... > > Death of VxWorks when "sysClkConnect" > Hello, > when I debug, The following lines make a stop of VxWorks debug. The Task "Wdb.." > is suspended and the target server connection is in default. > > sysClkDisable(); > sysClkConnect((FUNCPTR)vxw_mon_it, 0); > sysClkEnable(); > > How could I avoid this problem? > Thanks > charly > > -- > Use our news server 'news.foorum.com' from anywhere. > More details at: http://nnrpinfo.go.foorum.com/ --------------------------- Newsgroups: comp.os.vxworks Subject: Re: VxWorks Pipes Date: Tue, 29 Jan 2002 20:00:16 +0100 From: "Werner Schiendl" Message-ID: <3c56f141$1@brateggebdc5.br-automation.co.at> References: <38c6d574.0201290739.36ed5aa6@posting.google.com> Hi, "Keith Jarvis" wrote in message news:38c6d574.0201290739.36ed5aa6@posting.google.com... > Hi, > > I am somewhat new to the VxWorks world and would just like some > guidance on a couple of things in regards to using pipes on VxWorks. > Please excuse my ignorance but: Which version of VxWorks are you using? I'm in VxWorks 5.4 at this time, so my comments apply to this version. VxWorks AE may be different. > > First, does VxWorks support stream pipes and/or named pipes? AFAIK, no. VxWorks' pipes are message oriented - not stream oriented. > > Second, is there away to handle connections to a pipe in VxWorks. You could probably use a socket for that, depending on your needs this may or may not be adequat. Otherwise, you can write your own pipe driver. The VxWorks Programmer's Guide explains how to write a character device driver. I don't know for sure if the pipe of VxWorks has such thing - I doubt it. Further, the VxWorks pipe as I understand it has no concept of more than one client (i. e. it does not multiply the stuff you put in for every client). > Meaning, my server want to be able to account for each tasks that > connects to it. I understand that pipes in VxWorks are just an I/O > device on top of the Message Queue (If I misunderstood this concept, > please forgive me). I understand about using file descriptors and > waiting for some action to occur on them when using sockets. Does > this work in the same way? Client side yes (but only one client), at the server side, sockets are special. > > If there are any sites that anyone knows of with any more of this > information, can you list them for me. Any help would be greatly > appreciated. > If you want try the socket approach, start of with the TCP Server/Client example from the Network Programmers Guide. It'll get you started quickly. hth Werner --------------------------- Newsgroups: comp.os.vxworks Subject: Here is a short example which crash Date: 29 Jan 2002 17:26:17 GMT From: Charly Message-ID: <2002129-182617-871771@foorum.com> Hello all, On the following short example, if type in Shell: - ->es6() VxWorks crashs (boot or exception or ...) If I comment the line with fopen(line (1)), All is ok. "fopen" is connected to a FTP server. If I comment the line(2)), It works fine also. Any idea to help me? Thanks - ------------------------------ void vide(char* st) { int v,a=0; v=0; a=fopen(st,"r"); /* line (1)*/ taskDelay(100); if (a==0) { printf("es2 err"); } else { fclose(a);}; printf("end vide"); } void es6() { int t1,t2; t1 = taskSpawn ("SpawnT1", 200, 0, 2000, (FUNCPTR)vide,"ctdt.c",0,0,0,0,0,0,0,0, 0); printf("Spawn vide1"); taskDelay(100); t2 = taskSpawn ("SpawnT2", 201, 0, 2000, (FUNCPTR)vide,"ctdt.o",0,0,0,0,0,0,0,0, 0); /*line (2)*/ printf("Spawn vide2"); } - -- Use our news server 'news.foorum.com' from anywhere. More details at: http://nnrpinfo.go.foorum.com/ --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Here is a short example which crash Date: 29 Jan 2002 11:58:51 -0800 From: Joe Chung Message-ID: References: <2002129-182617-871771@foorum.com> 1. What happens if you do - -> copy "ctdt.c" from your windsh instead of executing es6()? 2. What happens if you call the vide() function directly from es6() instead of taskSpawn'ing it? 3. What happens if you increase the stack size of the spawned tasks from 2000 to 65536? - -jc - -- (apply 'concat (reverse (list "com" (char-to-string 46) "yahoo" (char-to-string 64) "joechung"))) --------------------------- Newsgroups: comp.os.vxworks Subject: Re: How to get actual addresses for boot codes (romInit.s) Date: Tue, 29 Jan 2002 20:00:26 GMT From: "Jim" Organization: Magma Communications Ltd. Message-ID: References: What is the address your jumping too ? If your hardware reset word is messed up you may not boot properly ... specifically there is a field in your HW reset word the field IIP causes your board to boot at either FFF00100 or 00000100 ... depending on how big you boot device if you use FFF00100 you could have some problems this ... your HW guys should be able to check this ... Also you can dump your boot image cor file with the function objdumpppc in the tonado directory ... if you use -x option you will get location of everything in image ... Good luck, Jim "Raymond Yeung" wrote in message news:d83bb8e3.0201291039.33d4568a@posting.google.com... > I need to troubleshoot a problem that occurs during system bootup. > The board uses MPC8260. Logic analyzer indicates the board > branches abnormally to an address. One way to track this problem > is with emulator. Other than this approach, is there any way > that I can generate a file that maps out actual addresses of all > instructions in boot code (esp. those assembler code in romInit.s)? > > Since an emulator knows where to load the text,bss,data segments, > the information must have been built somewhere. Which utilities > should I use? What's the procedure? --------------------------- Newsgroups: comp.os.vxworks Subject: Re: New WindSerf, TSR's and SPR's. Date: Tue, 29 Jan 2002 15:32:04 -0600 From: Dinker Charak Organization: Fermi National Accelerator Lab Message-ID: References: <488e459a.0201221531.41030518@posting.google.com> Reply-To: dinker@fnal.gov Hi, I am using Netscape and I have been looking at SPRs etc easily for quite sometime. But maybe because I have FutureSplash Player and Shockwave Flash plugins. On going to Products Tab I see the SPR link and search on the page. You don't see even that? http://web1.windriver.com/windsurf/kpl/ While you get it working maybe this link is of some help. Of course it will ask for a login name/password. This is the SPR search page. HTH DInker Bill Pringlemeir wrote: >>>> Bill Pringlemeir wrote: > > >>> How does one access SPRs under the new WindSerf? It is hard to > >>> install shockwave plug-ins on lynx [in case the info is in the > >>> tutorial] ... > >>>> john_94501@yahoo.com (John) wrote, > > >> You need to go through the Java applet thing on the left side of > >> the main page - it is labelled Knowledge Base. That gives you an > >> interactive folder tree to open and lets you get at the SPRs. > >> > >> For the TSRs, there is a link in the problem solving panel that > >> will let you get the status of one of your TSRs. > >>>> "Charles" == Charles Johnson writes: > > Charles> They don't appear to have much of a search mechanism though. > > I couldn't find it either. I sent an email about this. They told me > to call tech support. I imagine tech support will love 100's of > developers calling. Has anybody found TSR/SPRs using Netscape? I > thought that maybe my contract had run out, but the email confirmed > that it is active (as of today...). I can not access the TSR or SPRs as > John said ;-( > > tia, > Bill Pringlemeir. > --------------------------- Newsgroups: comp.os.vxworks Subject: changing the target name Date: 29 Jan 2002 13:54:09 -0800 From: trish.bennett@caenewnes.com (Trish Bennett) Organization: http://groups.google.com/ Message-ID: We simply (?) want to change the name of the target from xx.out to xx.o. How can this be done? I tried replacing all occurences of xx.out to xx.o in the .wpj file (outside of the Tornado IDE), but that doesn't work. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: WINDWEB_AUTO and javascript Date: Tue, 29 Jan 2002 17:19:57 -0500 From: david lindauer Message-ID: <3C57200D.2000208@notifier-is.net> References: <3C56B0EF.9070502@notifier-is.net> I solved it by writing copycat versions of the appropriate WindWeb SSI functions that did what I wanted. David david lindauer wrote: > hi, > we want to use some javascript in conjunction with form input. For > example declare an 'onchange' event for combo boxes. We put the > onchange event in the tag, but vxworks will not > propagate that when creating the web page with the SSI functions. > Viewing the page source indicates that it is not uploaded to the > browser. Is there any work around to get this to work? > > Thanks, > > David > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: changing the target name Date: Wed, 30 Jan 2002 09:18:58 +1030 From: Tim Shaw Organization: Defence Science and Technology Organisation Message-ID: <3C5726DA.A30B2B06@dsto.defence.gov.au> References: You can either create a new rule in the project properties menu, or add mv xx.out xx.o to the post build rule in the properties/macros tab. Trish Bennett wrote: > We simply (?) want to change the name of the target from xx.out to > xx.o. How can this be done? I tried replacing all occurences of > xx.out to xx.o in the .wpj file (outside of the Tornado IDE), but that > doesn't work. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: How to get actual addresses for boot codes (romInit.s) Date: 29 Jan 2002 15:29:11 -0800 From: ryeung@earthlink.net (Raymond Yeung) Organization: http://groups.google.com/ Message-ID: References: Thanks a lot Jim. objdumpppc is exactly what I'm looking for. Also, for my own record, command "l" would also disassemable the boot code for me, provided I have a good board running the same boot image. As an aside, how do people usually use the .map file? Where can I find information on how to interpret its content, anybody? "Jim" wrote in message news:... > What is the address your jumping too ? > > If your hardware reset word is messed up you may not boot properly ... > specifically there is a field in your HW reset word the field IIP causes > your board to boot at either FFF00100 or 00000100 ... depending on how big > you boot device if you use FFF00100 you could have some problems this ... > your HW guys should be able to check this ... > > Also you can dump your boot image cor file with the function objdumpppc in > the tonado directory ... if you use -x option you will get location of > everything in image ... > > Good luck, > Jim --------------------------- Newsgroups: comp.os.vxworks Subject: Anyone tried to modify 8260 interrupt priority? Date: 29 Jan 2002 15:59:34 -0800 From: ryeung@earthlink.net (Raymond Yeung) Organization: http://groups.google.com/ Message-ID: I need to modify Timer2's interrupt priority such that auxiliary clock used by SPY has the highest priority on the processor. This would allow me to measure ISR CPU utilization as well. I first thought all I need to do is to program SICR with the interrupt number of Timer2. However this doesn't work. It seems VxWorks interrupt controller masks out Timer2 interrupts in other high speed device (e.g. FCC) ISRs. I'd need to modify the hardcoded interrupt masks, such that Timer2 interrups are no longer masked by interrupts I want to measure. Then regenerate the library archive and rebuild my vxWorks image. Does this sound right? --------------------------- Newsgroups: comp.os.vxworks Subject: Re: VxWorks AE C++ Static constructor Date: Tue, 29 Jan 2002 23:28:26 +0000 From: molochai Message-ID: References: <3C55A2AB.C6A504D0@ele.etsmtl.ca.NOSPAM> David Roberge wrote: > Hi, > > I have a question about C++ static constructors in VxWorks AE. > > Here is an example of the situation. I have two components, one user > shared lib protection domain and one application domain. > > 1. component A downloaded in user shared lib protection domain > 2. component B downloaded in user shared lib protection domain > 3. application domain attached to user shared lib protection domain > > The static constructors of component A & B are called when the > application domain attach to the shared library. However I need the > constructors of component A to be executed before the constructors of > component B. > > Is there a way to tell VxWorks the order of execution for static C++ > constructors in this dynamic loading context ? > Can't comment on the VxWorks AE apsects of this, but in general C++ provides no guarantees about the order of calling of static constructors. I can't really see how a system to specify the ordering of calls might work. If it matters, it's up to you to enforce it yourself using the singleton pattern and construct-on-first-use. Count yourself lucky that AE uses a later version of GCC than that shipped with T2/VxWorks5.4 - on PPC architectures at least, GCC produces code that can call static constructors (and even worse destructors) more than once ... --------------------------- Newsgroups: comp.os.vxworks Subject: [Help Me] Socket binding error Date: Wed, 30 Jan 2002 03:20:04 GMT From: "Kim, Jeong-Hwan" Organization: Dacom Message-ID: Hi, everyone When socket server task is loaded again after it is unloaded ( not board reset ), socket binding routine occurs error indicating "address in use". So I set socket option value "SO_REUSEADDR" but it also has same error. Please let me know how I can solve this problem ?? My socket server code is as follows : ................ m_hSocket = socket(AF_INET, SOCK_STREAM, 0); if ( m_hSocket == INVALID_SOCKET ) { error_print("(%s, %d) DM Server socket initialization failure", file_name, __LINE__); perror("[DM SERVER : socket]"); return(1); } else printf("[DM SERVER] SOCKET created\n"); optval = 1; setsockopt(m_hSocket, SOL_SOCKET, SO_REUSEADDR, (char *)&optval, sizeof(int)); server_address.sin_family = AF_INET; server_address.sin_addr.s_addr = htonl(INADDR_ANY); /* inet_addr("127.0.0.1"); */ server_address.sin_port = htons(tcp_port_number); err = bind(m_hSocket, (struct sockaddr *)&server_address, sizeof(server_address)); if ( err ) { error_print("(%s, %d) DM Server socket bind failure", file_name, __LINE__); perror("[DM SERVER : bind]"); close(m_hSocket); return(1); } Thanks in advance Kim --------------------------- Newsgroups: comp.os.vxworks Subject: Modify MAC addr For Enet Driver Date: 29 Jan 2002 17:33:34 -0800 From: asherg@hotmail.com (phil) Organization: http://groups.google.com/ Message-ID: Does anybody know if there is a clean way (or not) to modify the MAC address for an Ethernet chip after the stack is up? Something similar to ifAddrAdd (IP) but for MAC? I am using Tornado 2.0 with the FEI END driver. Thanks. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: missing ftime and alarm?? Date: 29 Jan 2002 19:55:46 -0800 From: vdalal@hotmail.com (Vasu Dalal) Organization: http://groups.google.com/ Message-ID: <363d071b.0201291955.44908dca@posting.google.com> References: <4e22fcd6.0201251941.4f274089@posting.google.com> Hello Ganesh : Take a look at this link : http://www.xs4all.nl/~borkhuis/vxworks/vxw_pt6.html#6.2 HTH, vasu ref: ganesh_kumarg@rediffmail.com (ganesh godavari) wrote in message news:<4e22fcd6.0201251941.4f274089@posting.google.com>... > hello everyone, > i am trying to port openssl-0.96b onto VxWorks. i found the ftime and > alarm functions missing in VxWorks. > > does VxWorks have related function for ftime and alarm?. if so can u > tell me. > > has anyone ported openssl onto VxWorks? if so can u share the make > files? > > Thanks, > gkgodava --------------------------- Newsgroups: comp.os.vxworks Subject: Re: New WindSerf, TSR's and SPR's. Date: 29 Jan 2002 20:18:26 -0500 From: Bill Pringlemeir Organization: Bell Sympatico Message-ID: References: <488e459a.0201221531.41030518@posting.google.com> Sender: bpringlemeir@DeadDuck >>>>> "Dinker" == Dinker Charak writes: Dinker> Hi, I am using Netscape and I have been looking at SPRs etc Dinker> easily for quite sometime. But maybe because I have Dinker> FutureSplash Player and Shockwave Flash plugins. Dinker> On going to Products Tab I see the SPR link and search on the Dinker> page. You don't see even that? Dinker> http://web1.windriver.com/windsurf/kpl/ Thanks that does it for me! I have Navigator too (4.75), and it uses Java for the applette at the left. The SPR's aren't under `manual' are they? The URL you listed queries for my credentials and then it lets me search for SPRs... regards, Bill Pringlemeir. - -- I'm not just a gardener, I'm a plant manager. vxWorks FAQ, "http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html" --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Ton of errors when creating a project. Date: Wed, 30 Jan 2002 06:51:56 GMT From: "Gary M" Organization: AT&T Broadband Message-ID: References: <3C56DBF3.84AFD130@digidata.com> Could you elaborate what you mean by "proper paths"? Are they relative to WIND_BASE, or what? Have you tried to build a boot ROM in the BSP directory? "Doug Dotson" wrote in message news:3C56DBF3.84AFD130@digidata.com... > I am trying to create a bootable project based upon a new > BSP. The BSP is properly located in the \tornado\target\config, > the Makefile has been modified to have proper paths to the > commDrv and CoreDrv directories. When I create the project > a window "Project Creation Error" pops up with hunderds of > #define statements and further dow it appears that some > include files could not be located stating "Permission Denied". > I am working on an NT box with Tornado 2.0.2 for MIPS. > I have already successfully built the same type of project > using a different BSP. > > Thanks, > Doug > > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Cache problem Date: Wed, 30 Jan 2002 06:52:49 GMT From: "Gary M" Organization: AT&T Broadband Message-ID: References: Actually, chances are that the cache is properly enabled. The problem is likely the speed of the local bus, which is surely running at 40 Mhz with the 80 Mhz processor. The slower speed of the bus and the longer access time for RAM offsets the gains made by the processor speed. The 860T also has a small cache (4K+4K), which a complex application will quickly gobble up. "Boris" wrote in message news:a2vlv5$2i8$1@mail.cn99.com... > Hi,all, > Our test found that our program can't run faster on > a 860T 80MHz chip than on a 50MHz chips,the OS on our > target board is vxWorks 5.4,We doubt if we properly set > the cache.Can anyone provide useful information on > how to enable cache in vxWorks? > Thank you in advance. > > Sincerely yours, > Xiao > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Cache problem Date: Wed, 30 Jan 2002 06:53:00 GMT From: "Gary M" Organization: AT&T Broadband Message-ID: References: Actually, chances are that the cache is properly enabled. The problem is likely the speed of the local bus, which is surely running at 40 Mhz with the 80 Mhz processor. The slower speed of the bus and the longer access time for RAM offsets the gains made by the processor speed. The 860T also has a small cache (4K+4K), which a complex application will quickly gobble up. "Boris" wrote in message news:a2vlv5$2i8$1@mail.cn99.com... > Hi,all, > Our test found that our program can't run faster on > a 860T 80MHz chip than on a 50MHz chips,the OS on our > target board is vxWorks 5.4,We doubt if we properly set > the cache.Can anyone provide useful information on > how to enable cache in vxWorks? > Thank you in advance. > > Sincerely yours, > Xiao > --------------------------- Newsgroups: comp.dcom.sys.cisco,comp.os.vxworks Subject: Binding a socket to a interface in VxWorks Date: 29 Jan 2002 23:16:01 -0800 From: naveens@ishoni.com (Navaneetham) Organization: http://groups.google.com/ Message-ID: <6c5f1af7.0201292316.58ff333d@posting.google.com> hi, how socket library will behave when two interfaces are having same ip address? (ppp & eth both interfaces are having same ip address. ppp is unnumbered interface.) problem I am facing is creating socket for INADDR_ANY and bind to any one port no. I am not receving packets from ppp interface. do any one know what is the problem? Is any solution for this??? working with VxWorks stack. ------------------------ | | | | | eth ppp | ---|--------------|----- | | | | 192.168.1.1 192.168.1.1 (x port) (x port) Regards Navaneetham --------------------------- Newsgroups: comp.os.vxworks Subject: login issue Date: 30 Jan 2002 00:30:21 -0800 From: senthilb@comneti.com (senthil) Organization: http://groups.google.com/ Message-ID: <26ca8a8f.0201300030.47a0fa90@posting.google.com> We developed a login table. Where the admin user comes first . The admin user is "target" as found in usrConfig.c . I could login using target and add new users into the login table. But once i logout , i could never login using the new added user names. I dont have any problem with the "target" username which is the first entry. The funtion loginUserVerify( ) always returns -1 when i try to login with my new user name and password. We even installed our encryption routine. Do any one have the code for loginUserVerify( ) or can any one suggest how to overcome the problem with loginUserVerify() senthil --------------------------- Newsgroups: comp.os.vxworks Subject: Re: VxWorks AE C++ Static constructor Date: Wed, 30 Jan 2002 08:27:09 +0000 From: Danny Chapman Organization: Lucent Technologies, Columbus, Ohio Message-ID: References: <3C55A2AB.C6A504D0@ele.etsmtl.ca.NOSPAM> On Tue, 29 Jan 2002, molochai wrote: > Can't comment on the VxWorks AE apsects of this, but in general C++ > provides no guarantees about the order of calling of static constructors. That's only true of the creation of non-local static objects. Within a single translation unit, the order is guaranteed. - - Danny --------------------------- Newsgroups: comp.os.vxworks Subject: Configuration of a serial device to run full duplex Date: Wed, 30 Jan 2002 09:35:43 +0100 From: "Jakob Holm-Petersen" Organization: TDC Internet Message-ID: <3c57b060$0$89074$edfadb0f@dspool01.news.tele.dk> Hi, I'm right now working on a project where we need to make a physical loop back on a serial device. The loop back function is used to verify that all transmitted chars on the device are properly transmitted. My problem is that it seems like that the serial device is running half duplex. My source code looks like this: fd = open ("/tyCo/1", O_RDWR, 1); write (fd, "abcde", 5); i=0; while (i < 5) { i += read (fd, &rxBuffer[i], 1); } The result is that the while loop is newer completed because the read function only read the last 2 chars "de" in the string "abcde". Do anyone have a solution of an explanation of this problem? Best regards Jakob Holm-Petersen --------------------------- Newsgroups: comp.os.vxworks Subject: Easily Make $1000's Per Month Selling On Ebay! Date: 30 Jan 2002 08:59:45 GMT From: nathan Organization: me Message-ID: <3c57b601$1$12608$afc38c87@news.optusnet.com.au> Recieving over 1.5 billion page views per month, Ebay is the ULTIMATE venue for selling virtually any goods and making huge profits with almost no effort. But you have to know what to sell and how to sell. That's where Ebay Marketing 2001 comes in. This manual provides you with easy to understand and detailed instructions for maximizing your profits with selling strategies that are PROVEN WINNERS. This manual provides you with easy to understand and detailed instructions for maximizing your profits with selling strategies that are PROVEN WINNERS. This information will make you $1000's on Ebay. Although this book was based on eBay auctions, the overall information will work for ALL online auctions. Ebay Marketing 2001 teaches you effective SELLING STRATEGIES and you don't need any specialized computer knowledge. I'm going to pass on the SECRET SELLING TECHNIQUES that I use each and every day to bring in hundreds of thousands of dollars selling my products on internet auctions. THE INTERNET CAN BE YOUR GOLDMINE! - ---------------------------------- Why aren't you digging out the gold? Because you don't have the right tools. For more information e-mail: joshabrahams@dr.com/ put "info" inthe subject header --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Death of VxWorks when Date: 30 Jan 2002 10:53:27 GMT From: Charly Message-ID: <2002130-115327-540197@foorum.com> References: <2002129-1292-461828@foorum.com> <9JB58.37$5o.67861@newsr2.u-net.net> Here is the code for "vxw_mon_it" void vxw_mon_it() { tickAnnounce(); } - -- Use our news server 'news.foorum.com' from anywhere. More details at: http://nnrpinfo.go.foorum.com/ --------------------------- Newsgroups: comp.os.vxworks Subject: IOP IQ80310 - What addr do i download the VxWorks Image ? Date: 30 Jan 2002 03:40:36 -0800 From: deepaks@sasken.com (Deepak Shetty) Organization: http://groups.google.com/ Message-ID: <421ece65.0201300340.8019fe1@posting.google.com> Hi, folks I am currently working on an Intel IQ80310 IOP board and need to downlaod the VxWorks image bootrom.bin provided on to the on-board flash unfortunately the board manual doesnt describe the memory location / address where i need to put in the image for proper execution of the same. the board has redboot as the boot monitor, and i need to use the FRU ( flash recovery utility ) provided by the intel people to download the image ( redboot as well as any other, for eg: vxwrks ) onto the on board flash via the PCI interface of the PC to which the board is connected. but i guess i need to downlaod the image to a specified address of the flash as there is a rotary switch setting provided on the board if the image downlaoded is of vxworks, but the manual doesnt give the address of the same any idea where i need to download the image of vxworks ? in case any one has dealt with such similar prob before ? I am unable to get any support from wrs / intel on the same :( also usualy how is the memory mapped for the flash ...? redboot shows some info when i enter fis list & fis free, but i am confused ... can someone tell me how is the memory mapping present betn the flash and the onboard 32MB RAM ? Thanx, deepak --------------------------- Newsgroups: comp.realtime,comp.realtime.jobs,comp.os.vxworks,comp.os.qnx Subject: Re: Suggestions? Date: 30 Jan 2002 01:17:26 -0800 From: amcheeseus@yahoo.com (amit choudhary) Organization: http://groups.google.com/ Message-ID: <4cd34f34.0201300117.5e7b9fc@posting.google.com> References: <4cd34f34.0201261035.4561ad4@posting.google.com> bsd_mike@hotmail.com (Mike) wrote in message news:... > I owuld sure love to see it! > -Mike > In few days time I will be putting up the kernel with documentation for free download, amit. --------------------------- End of New-News digest ********************** From vxwexplo-errs@csg.lbl.gov Thu Jan 31 04:03:28 2002 From: Vxworks Exploder Date: Thu Jan 31 04:03:32 PST 2002 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Thu Jan 31 04:03:23 PST 2002 Subject: Re: problem with -mlongcall Subject: Re: 860SAR UTOPIA.... What is syncronization signals? Subject: Default scheduler in VxSim ? Subject: Connecting tornado tools to target over shared mem? Subject: Tornado2 and PVCS Subject: Re: login issue Subject: Re: Here is a short example which crash Subject: "dosFsDevInit failed". Where could I look in VxW configuration? Subject: Re: Ton of errors when creating a project. Subject: Timers in VxSIM Subject: 16550 UART driver Subject: Re: [Help Me] Socket binding error Subject: Re: "dosFsDevInit failed". Where could I look in VxW configuration? Subject: Re: Connecting tornado tools to target over shared mem? Subject: Re: Default scheduler in VxSim ? Subject: Re: Tornado2 and PVCS Subject: How do I use intConnect Subject: PPPoE and vxWorks Subject: rapidcontrol for CLI Subject: Re: login issue Subject: "__fixdfdi" reference Subject: Source for WRS build utilities? Subject: Re: Modify MAC addr For Enet Driver Subject: Re: 16550 UART driver Subject: Re: Curses lib under vxworks Subject: Re: Modify MAC addr For Enet Driver Subject: Re: Binding a socket to a interface in VxWorks Subject: VxWorks Pipes Subject: Gotcha with C++ static constructors and GCC 2.95.x (PPC) Subject: Re: Subject: Re: "__fixdfdi" reference Subject: MPC8260 Ethernet Driver Subject: Re: login issue Subject: Re: Subject: tffsDevFormat fails.. Subject: Re: Subject: Re: How do I use intConnect Subject: Re: Tornado2 and PVCS Subject: Re: Source for WRS build utilities? Subject: Re: Connecting tornado tools to target over shared mem? Subject: Re: Source for WRS build utilities? Subject: Does nucleus plus support MMU? Subject: Re: tffsDevFormat fails.. Subject: semTake and ISR Subject: Re: Source for WRS build utilities? Subject: Re: Modify MAC addr For Enet Driver Subject: Where is my image? Subject: Re: semTake and ISR Subject: Re: Default scheduler in VxSim ? Subject: Re: login issue Subject: Connect more than one ISR to single IV?? Subject: what are these Assembler messages... Subject: Re: Tornado2 and PVCS Subject: Re: 16550 UART driver Subject: Re: How do I use intConnect Subject: Re: How do I use intConnect Subject: compilation> Subject: Re: Default scheduler in VxSim ? Subject: Re: Tornado2 and PVCS Subject: Re: Connect more than one ISR to single IV?? Subject: Re: what are these Assembler messages... Subject: Re: login issue Subject: Re: 16550 UART driver Subject: Re: About STLport Subject: Re: semTake and ISR Subject: Re: what are these Assembler messages... Subject: Re: Tornado2 and PVCS Subject: Re: problem with -mlongcall ( with one more interesting point) Subject: How do I access both serial ports? ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: problem with -mlongcall Date: 30 Jan 2002 04:35:28 -0800 From: kumarpavan@lycos.com (pavan) Organization: http://groups.google.com/ Message-ID: <307af50b.0201300435.637b546@posting.google.com> References: <307af50b.0201290317.689ad29e@posting.google.com> <9JB58.38$5o.67861@newsr2.u-net.net> Dave, Thanks for the response.i am not sure if i have missed out any compiler flag in my compilation.The make output is shown below: + /home/tornado/host/sun4-solaris2/bin/ccppc -mlongcall -DCPU=PPC604 - -Wall - -I.. -I/home/tornado/target/h/arch/ppc - -I/home/tornado/target/h - -I/home/tornado/host/sun4-solaris2/lib/gcc-lib/powerpc-wrs-vxworks/cygnus-2.7.2-960126/include - -c ../test.cpp could you please let me know if you had any more flags in your compilation. BTW i have a target with 256MB RAM. TIA, pavan. "Dave Korn" wrote in message news:<9JB58.38$5o.67861@newsr2.u-net.net>... > "pavan" wrote in message > news:307af50b.0201290317.689ad29e@posting.google.com... > > hello all, > > > > I am just facing some strange behaviour with -mlongcall option. > > I am using t2cp4 , vxworks5.4, PPC604 target with ccppc compiler > > (cygnus-2.7.2-960126 egcs-971225 tornado 2.0 ). > > The following piece of code when compiled with -mlongcall option gives > > a machine check exception but works fine if you remove -mlongcall > > option. > > Couldn't reproduce that here using either 2.7.2 or 2.95.4-vec. Strange. > > > ->sp checkMine > > task spawned: id = 0x143a5d0, name = t16 > > value = 21210576 = 0x143a5d0 > > -> > > machine check : Master Abort Received > > MPC107, Error Detection Register 1 (0xc1) = 0x00 > > MPC107, Error Detection Register 2 (0xc5) = 0x00 > > MPC107, 60x Bus Error Status Register (0xc3) = 0x50 > > MPC107, PCI Bus Error Status Register (0xc7) = 0x00 > > MPC107, Bus Error Address Register (0xc8) = 0x20000000 > > OK, so something in the code has tried to read/write address 0x20000000, > and you've got nothing mapped there, so the bus controller generates a fault > when no device answers on the pci bus after a timeout. > > > machine check > > Exception next instruction address: 0x0143a858 > > Machine Status Register: 0x0004b030 > > Condition Register: 0x22000000 > > Task: 0x143a5d0 "t16" > > ->0x0143a858 > > value = 21211224 = 0x143a858 = checkMine + 0x70 > > Right, at that stage it would have been useful to also do a "ti 0x143a5d0" > to see what was in all the registers, and a "l checkMine, 50" to show us the > actual assembler code that was being executed. It might also be informative > to add a line "printf ("Address of A is $%08x\n", &a) during the while > (true) loop. And using the -save-temps flag when compiling in order to have > a look at the .s file generated by this code > > > while(true) > > I couldn't find where in the header files this lower-case "true" gets > defined, usually people use the upper case version. The only place I > *could* find it defined was in the stl_config.h header file. You should > perhaps verify that this is defined the way you expect it is. > > DaveK > -- > moderator of > alt.talk.rec.soc.biz.news.comp.humanities.meow.misc.moderated.meow > Burn your ID card! http://www.optional-identity.org.uk/ > Help support the campaign, copy this into your .sig! > Proud Member of the Exclusive "I have been plonked by Davee because he > thinks I'm interesting" List > Member # > Master of Many Meowing Minions --------------------------- Newsgroups: comp.os.vxworks Subject: Re: 860SAR UTOPIA.... What is syncronization signals? Date: 30 Jan 2002 04:50:44 -0800 From: houcheng@ms17.hinet.net (houcheng) Organization: http://groups.google.com/ Message-ID: References: Hi Lee: I am also writing 860SAR driver and happy to see people writing driver on it. The last section means that before the APC timer ( or transmiter ) operation, the physical interface ( or physical layer) must be activated! Otherwise the transmiter would be locked since it would try to send ATM cells to a in-activated phyical interface! houcheng@no_spam.turbonetworks.com.tw Lin Houcheng 10month@hanmail.net (Changwon Lee) wrote in message news:... > Hi... > > I'm reading an article, "MPC8xx ATM Supplement", to activate UTOPIA > interface in 860SAR. But I can not understand the meaning of > syncronization signal. > > Context in "MPC8xx ATM Supplement"... > > 5.1.6 APC Initialization and Operating Considerations > > As long as the APC timer(timer4) is not active, the APC parameters may > be initialized in any order. Timer 4 must be initialized last, > following the initialization of the APC tables of all the ATM ports in > the system. Failure to initialize timer 4 last will allow the > scheduling algorithm to start prematurely, resulting in unpredictable > behavior. > > Furthermore, the APC table parameters must be initialized before any > TRANSMIT_CHANNEL_ACTIVATE commands are issued. However, these commands > may be issued at any time, wheter the APC timer is active or inactive. > For more information, see the description of the > TRANSMIT_ACTIVATE_CAHNNEL command. > > Also, note that the physical interface (serial or UTOPIA) must be > enabled and its associated clocks and SYNCHRONIZATION SIGNALS must be > active before the APC timer is activated. Otherwise, the transmit > queue will simply overflow (causing APCO interrupt). > > ---------------------------------- > > What is the syncronization signal? > What should I do for this syncronization signal in H/W and S/W. > > Thank you for further comments! ^_^ > Have a nice day or night! --------------------------- Newsgroups: comp.os.vxworks Subject: Default scheduler in VxSim ? Date: 30 Jan 2002 05:25:39 -0800 From: deepaks@sasken.com (Deepak Shetty) Organization: http://groups.google.com/ Message-ID: <421ece65.0201300525.73bf4466@posting.google.com> Hi, friends I have a basic doubt on the default scheduler thats active in VxSim under Win98 I read the vxworks_guide and other documentation & found out that the default scheduler is a priority based scheduler which expects the tasks to be co-operative, in the sense that if one task uses an infinite loop to hog CPU, other tasks will never get a chance to run. Why is this kind of scheduling present at all as its not good for realtime applications ? and what is the reason behind makign this as the default scheduler instead of the otehr round-robin ( pre-emptive) scheduler which can be invoked using kernelTimeSlice( ticks ) function ? I think that the round-robin is more suited for real time apps as its rtuly pre-emptive and can deal with arrogant tasks by pre-empting them in case it tried to hog the cpu. also the kernelTimeSlice( ticks ) functions expects the no of ticks as an argument, whats the typical value suited for a real time app ? Thanx, deepak --------------------------- Newsgroups: comp.os.vxworks Subject: Connecting tornado tools to target over shared mem? Date: 30 Jan 2002 14:32:59 +0100 From: Dennis Roren Organization: Newshosting.com - Highest quality at a great price! www.newshosting.com Message-ID: <86wuy0q7zo.fsf@aleph.hack.org> Hi! I was wondering if anyone knows how, or if its at all possible to connect the Tornado tools (f.ex. ProfileScope), using a shared memory network on the hardware? Our system consists of one PB (Processor Board), and several CB's (Controller Boards), and each of the CB's have one or two PMC modules on them. The communication between the PB's, the CB's and the PMC's take place over a shared memory network. What we would like to be able to do is to connect to the PMC modules using this network instead of the serial connector, as the speed is too slow to be able to give us enough information. If it is possible to get information about how the targetserver connects to the PB's over the usual network, (i.e. which ports), it should be possible to make a "proxy" task on the PB that sends all transmission on to the PMC of our choise. Is there anyone that has tried to do something like this? And is it possible? In our setup we also have a network connection to the PB's (Processor Boards), via a network adapter in PMC slot1. It's used to connect to the PB's from our network. We have a external application that uses among other things NFS mounted directorys. The problem comes when we want to monitor (with for example ProfileScope), the PMC 860 modules on one of the CB's. Today we have to do a lot of configuration to be able to run over the serial interface on these modules, and we are not really satisfied with the result. Is there anyone that have tried to do the same? (Hardware: MVME 2400 cards with 860 PMC-modules) - --------------- Sorry if this posting is a little disoriented, but I posted this on the WindRiver WindSurf, and it's two postings made into one. With kind regards // Dennis P.s. I would be happy if you could send a CC: to my emailadress too D.s. - -- Dennis Rören |Dennis.Roren@era.ericsson.se Ericsson Radio Systems AB|Tel: +46 13 32 27 58 Datalinjen 3 |Fax: +46 13 21 29 28 581 12 Linköping |(LM/ERA/SVZ/TD) --------------------------- Newsgroups: comp.os.vxworks Subject: Tornado2 and PVCS Date: Wed, 30 Jan 2002 15:58:06 +0200 From: "Ofer Goren" Organization: Verio Message-ID: Hi. Does anyone work with this combination, and if so, any recommendations about the best way to work? Tornado2.02 for win2kSp2 PVCS for windows, version 6.7.11 Best reagrds, - -- Ofer Goren Iamba Technologies --------------------------- Newsgroups: comp.os.vxworks Subject: Re: login issue Date: 30 Jan 2002 14:13:13 GMT From: Johan Borkhuis Organization: Agere Systems Message-ID: References: <26ca8a8f.0201300030.47a0fa90@posting.google.com> senthilb@comneti.com (senthil) wrote: > We developed a login table. Where the admin user comes first . The > admin user is "target" as found in usrConfig.c . I could login using > target and add new users into the login table. But once i logout , i > could never login using the new added user names. I dont have any > problem with the "target" username which is the first entry. The > funtion loginUserVerify( ) always returns -1 when i try to login with > my new user name and password. We even installed our encryption > routine. Do any one have the code for loginUserVerify( ) or can any > one suggest how to overcome the problem with loginUserVerify() Did you enter the passwords as passwords or as encrypted passwords as returned by the vxencrypt untility? Groeten, Johan - -- o o o o o o o . . . _____________________________ o _____ || Johan Borkhuis | .][__n_n_|DD[ ====_____ | borkhuis@agere.com | >(________|__|_[_________]_|__________________________| _/oo OOOOO oo` ooo ooo 'o!o!o o!o!o` === VxWorks FAQ: http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html === --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Here is a short example which crash Date: 30 Jan 2002 12:47:46 GMT From: Charly Message-ID: <2002130-134746-357809@foorum.com> References: <2002129-182617-871771@foorum.com> Thanks, you had a good nose. The problem was with the stack size. Regards charly - -- Use our news server 'news.foorum.com' from anywhere. More details at: http://nnrpinfo.go.foorum.com/ --------------------------- Newsgroups: comp.os.vxworks Subject: "dosFsDevInit failed". Where could I look in VxW configuration? Date: 30 Jan 2002 13:34:11 GMT From: Charly Message-ID: <2002130-143411-880483@foorum.com> the shell display "dosFsDevInit failed". what could be the reason. I have no idea where to look in Vxworks configuration - -- Use our news server 'news.foorum.com' from anywhere. More details at: http://nnrpinfo.go.foorum.com/ --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Ton of errors when creating a project. Date: Wed, 30 Jan 2002 09:38:09 -0500 From: Doug Dotson Organization: Verio Message-ID: <3C580551.D16E2BCF@digidata.com> References: <3C56DBF3.84AFD130@digidata.com> The Makefile has two macros that point to the "coreDrv" and "commDrv" subdirectories. They are absolute paths like "c:/tornado/target/config/ev-64240A/coreDrv" and "c:/tornado/target/config/ev-64240A/commDrv". I can't build a boot ROM because I can't create the project. I captured the error messages in a file. I will send it to you directly. thanks, doug Gary M wrote: > Could you elaborate what you mean by "proper paths"? Are they relative to > WIND_BASE, or what? > Have you tried to build a boot ROM in the BSP directory? > > "Doug Dotson" wrote in message > news:3C56DBF3.84AFD130@digidata.com... > > I am trying to create a bootable project based upon a new > > BSP. The BSP is properly located in the \tornado\target\config, > > the Makefile has been modified to have proper paths to the > > commDrv and CoreDrv directories. When I create the project > > a window "Project Creation Error" pops up with hunderds of > > #define statements and further dow it appears that some > > include files could not be located stating "Permission Denied". > > I am working on an NT box with Tornado 2.0.2 for MIPS. > > I have already successfully built the same type of project > > using a different BSP. > > > > Thanks, > > Doug > > > > --------------------------- Newsgroups: comp.os.vxworks Subject: Timers in VxSIM Date: 29 Jan 2002 07:07:21 -0800 From: bryanramesh@yahoo.com (Bryan) Organization: http://groups.google.com/ Message-ID: Hi, Is the timers provided by VxWorks will be reliable in VxSim as well? Will it provide the full timer facility. I have full VxSIM installed. Is there any restrictions with respect to timers usage in using the VxSIM. Kindly clear me. Thanks, Bryan --------------------------- Newsgroups: comp.os.vxworks Subject: 16550 UART driver Date: Wed, 30 Jan 2002 16:22:48 +0100 From: Emmanuel Herbreteau Organization: SEPRO Robotique (BEP) Message-ID: <3C580FC8.8DAA7C55@sepro-robotique.com> Hi, I'm looking for a driver for the NS16550 UART with FIFO enabled (usefull for speed over 19200 bauds). The default driver provides with Tornado/VxWorks 5.4 doesn't use the integrated 16 bytes FIFO... (it reads only one byte for each interrupts) Does anyone know where to find such driver ?? Thanks a lot. Regards Emmanuel --------------------------- Newsgroups: comp.os.vxworks Subject: Re: [Help Me] Socket binding error Date: Wed, 30 Jan 2002 16:28:32 +0100 From: "Werner Schiendl" Message-ID: <3c581122$1@brateggebdc5.br-automation.co.at> References: Hi, Do you close the socket before you unload your code? You cannot have to sockets bind to the same port hth Werner "Kim, Jeong-Hwan" wrote in message news:EDJ58.9680$Of.192302@news.bora.net... > Hi, everyone > > When socket server task is loaded again after it is unloaded ( not board > reset ), > socket binding routine occurs error indicating "address in use". > So I set socket option value "SO_REUSEADDR" but it also has same error. > Please let me know how I can solve this problem ?? > > > My socket server code is as follows : > > > > ................ > > m_hSocket = socket(AF_INET, SOCK_STREAM, 0); > > if ( m_hSocket == INVALID_SOCKET ) > { > error_print("(%s, %d) DM Server socket initialization failure", > file_name, __LINE__); > perror("[DM SERVER : socket]"); > return(1); > } > else > printf("[DM SERVER] SOCKET created\n"); > > optval = 1; > setsockopt(m_hSocket, SOL_SOCKET, SO_REUSEADDR, (char *)&optval, > sizeof(int)); > > server_address.sin_family = AF_INET; > server_address.sin_addr.s_addr = htonl(INADDR_ANY); /* > inet_addr("127.0.0.1"); */ > server_address.sin_port = htons(tcp_port_number); > > > err = bind(m_hSocket, (struct sockaddr *)&server_address, > sizeof(server_address)); > > if ( err ) > { > error_print("(%s, %d) DM Server socket bind failure", file_name, > __LINE__); > perror("[DM SERVER : bind]"); > close(m_hSocket); > return(1); > } > > > > > Thanks in advance > > > Kim > > > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: "dosFsDevInit failed". Where could I look in VxW configuration? Date: Wed, 30 Jan 2002 16:45:34 +0100 From: "Werner Schiendl" Message-ID: <3c581520$1@brateggebdc5.br-automation.co.at> References: <2002130-143411-880483@foorum.com> Does it display an error code as well? Try to initialize the disk manually, if not. hth Werner "Charly" wrote in message news:2002130-143411-880483@foorum.com... > > the shell display "dosFsDevInit failed". > > what could be the reason. I have no idea where to look in Vxworks configuration > > > -- > Use our news server 'news.foorum.com' from anywhere. > More details at: http://nnrpinfo.go.foorum.com/ --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Connecting tornado tools to target over shared mem? Date: Wed, 30 Jan 2002 16:43:10 +0100 From: "Werner Schiendl" Message-ID: <3c58148f$1@brateggebdc5.br-automation.co.at> References: <86wuy0q7zo.fsf@aleph.hack.org> Hi, is your shared memory network a normal IP network? You should be done setting up proper routing information. Be sure to set the routing information on the VxWorks target(s) as well as one the host. This works at least with WDB and the normal Tornado Tools (Shell, debugger,...) Does ProfileScope use the target server communication? Then this will work. hth Werner "Dennis Roren" wrote in message news:86wuy0q7zo.fsf@aleph.hack.org... > > Hi! > I was wondering if anyone knows how, or if its at all possible to > connect the Tornado tools (f.ex. ProfileScope), using a shared > memory network on the hardware? > > Our system consists of one PB (Processor Board), and several > CB's (Controller Boards), and each of the CB's have one or two > PMC modules on them. The communication between the PB's, the > CB's and the PMC's take place over a shared memory network. > > What we would like to be able to do is to connect to the PMC > modules using this network instead of the serial connector, as > the speed is too slow to be able to give us enough information. > > If it is possible to get information about how the targetserver > connects to the PB's over the usual network, (i.e. which ports), > it should be possible to make a "proxy" task on the PB that sends > all transmission on to the PMC of our choise. > > Is there anyone that has tried to do something like this? And > is it possible? > > In our setup we also have a network connection to the PB's > (Processor Boards), via a network adapter in PMC slot1. It's > used to connect to the PB's from our network. We have a external > application that uses among other things NFS mounted directorys. > > The problem comes when we want to monitor (with for example > ProfileScope), the PMC 860 modules on one of the CB's. > Today we have to do a lot of configuration to be able to run > over the serial interface on these modules, and we are not > really satisfied with the result. > > Is there anyone that have tried to do the same? > > (Hardware: MVME 2400 cards with 860 PMC-modules) > > --------------- > > Sorry if this posting is a little disoriented, but I posted > this on the WindRiver WindSurf, and it's two postings made into > one. > > With kind regards > > // Dennis > P.s. I would be happy if you could send a CC: to my emailadress > too D.s. > > -- > Dennis Rören |Dennis.Roren@era.ericsson.se > Ericsson Radio Systems AB|Tel: +46 13 32 27 58 > Datalinjen 3 |Fax: +46 13 21 29 28 > 581 12 Linköping |(LM/ERA/SVZ/TD) --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Default scheduler in VxSim ? Date: Wed, 30 Jan 2002 16:39:03 +0100 From: "Werner Schiendl" Message-ID: <3c581398$1@brateggebdc5.br-automation.co.at> References: <421ece65.0201300525.73bf4466@posting.google.com> "Deepak Shetty" wrote in message news:421ece65.0201300525.73bf4466@posting.google.com... > Hi, friends > I have a basic doubt on the default scheduler thats active in VxSim > under Win98 > I read the vxworks_guide and other documentation & found out that the > default scheduler is a priority based scheduler which expects the > tasks to be co-operative, in the sense that if one task uses an > infinite loop to hog CPU, other tasks will never get a chance to run. This makes absolutely sense, because there are usually more and less time critical parts in a real-time application. It is an important part of a real-time design, to decide the correct priority and timing schemes. You cannot accept some scheduler that donates some time from a life critical task to the low priority parts. > > Why is this kind of scheduling present at all as its not good for > realtime applications ? and what is the reason behind makign this as > the default scheduler instead of the otehr round-robin ( pre-emptive) > scheduler which can be invoked using kernelTimeSlice( ticks ) function > ? > Round robin scheduling is only working with tasks of the same priority. It's probably not used much in typical VxWorks apps. > I think that the round-robin is more suited for real time apps as its > rtuly pre-emptive and can deal with arrogant tasks by pre-empting them > in case it tried to hog the cpu. also the kernelTimeSlice( ticks ) > functions expects the no of ticks as an argument, whats the typical > value suited for a real time app ? > If you have an arrogant task on your system you have a BIG problem. If this task has high priority, than you probably have a HUGE problem. Think for it for a while: this is an realtime system - not a windoze box where some fool installs useless tools. To make sure everything works as expected, you usually have watchdogs guarding the software. There are software watchdogs (like wdLib in VxWorks) and hardware watchdogs. hth Werner --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Tornado2 and PVCS Date: Wed, 30 Jan 2002 16:49:31 +0100 From: "Martin Raabe" Organization: T-Online Message-ID: References: Hello Ofer, I recommend to use SNiFF+ PRO as a WindPowerTool on top of Tornado. For more derails look at: www.windriver.com for a free eval download. Ciao Martin Raabe "Ofer Goren" schrieb im Newsbeitrag news:newscache$ku9rqg$lck$1@lnews.actcom.co.il... > Hi. > Does anyone work with this combination, and if so, any recommendations about > the best way to work? > > Tornado2.02 for win2kSp2 > PVCS for windows, version 6.7.11 > > Best reagrds, > > -- > Ofer Goren > Iamba Technologies > > --------------------------- Newsgroups: comp.os.vxworks Subject: How do I use intConnect Date: 30 Jan 2002 07:58:33 -0800 From: danielmartinfuchs@yahoo.co.uk (Daniel Fuchs) Organization: http://groups.google.com/ Message-ID: <78cc994f.0201300758.3d442448@posting.google.com> Hello, I am new to VxWorks/Real-time programming. I have been given the task to connect my HDLC driver into a VxWorks BSP. I obviously need to use the intConnect command. I know that this is probably a simple question/solution but... the problem I am having is that I do not understand the 3 inputs needed for intConnect. I know that they are: VOIDFUNCPTR *vector VOIDFUNCPTR routine int parameter But do not understand what they mean. Would someone be able to help? Regards, Daniel --------------------------- Newsgroups: comp.os.vxworks Subject: PPPoE and vxWorks Date: Wed, 30 Jan 2002 11:34:23 -0500 From: "Stephane Ross" Message-ID: Hi all, Anybody knows if there is a vxWorks PPPoE implementation out there beside WindNET PPP/WindNetPPPoE? Thanks Stephane --------------------------- Newsgroups: comp.os.vxworks Subject: rapidcontrol for CLI Date: 30 Jan 2002 08:43:51 -0800 From: ilyasg@yahoo.com (Ilyas G) Organization: http://groups.google.com/ Message-ID: hi i am using rapid control for CLI We know that : The handlers are executed one by one through the command tree ie : For a command "create ip route", the handlers of the "create" ,"ip" , "route" will be executed one by one. the problem is with the no form. if i want to have "no" form for this command, i have to declare the handler within the last node "route". and the command will be # create ip no route wich is not good, i prefer to have # no create ip route Of course i can test on the rest of the command in the first handler, but if many nodes commands exists it will be as hard as developing a new "no" branch. i think. any hints ???? --------------------------- Newsgroups: comp.os.vxworks Subject: Re: login issue Date: Wed, 30 Jan 2002 17:39:16 +0000 (UTC) From: David Laight Organization: BT Openworld Message-ID: <3C583019.77D3F8D0@l8s.co.uk> References: <26ca8a8f.0201300030.47a0fa90@posting.google.com> > Did you enter the passwords as passwords or as encrypted passwords as > returned by the vxencrypt untility? FWIW don't assume any security from the 'vxencrypt' function. I think it is reversable by hand! Certainly not hard in sw. The generated password is: sum( p[i] * i ^ i )) * 0x1e3a1d5 converted to a string using different parts of the alphabet for different nibbles! David --------------------------- Newsgroups: comp.os.vxworks Subject: "__fixdfdi" reference Date: 30 Jan 2002 10:13:05 -0800 From: Jeffrey.Steinsnyder@L-3Com.com (Jeff Steinsnyder) Organization: http://groups.google.com/ Message-ID: We are using VxWorks v5.4. When the 64-bit floating point number converted to 64-bit integer the compiler generates a reference to "__fixdfdi" which can not be resolved at link time. What do I need to do to resolve this reference? --------------------------- Newsgroups: comp.os.vxworks Subject: Source for WRS build utilities? Date: 30 Jan 2002 12:59:07 -0500 From: "Paul D. Smith" Organization: Nortel Networks, Inc. / Billerica, MA Message-ID: Reply-To: pausmith@nortelnetworks.com Hi folks; I'm looking for the source code to, or replacements for, the following parts of the standard WRS build tools (these are from 5.3.1...): 1) elfToBin. I read somewhere that this could be replaced by objcopy -O binary, but the output file produced by elfToBin is about 2M larger than the one produced by objcopy; that doesn't seem right to me :). (This is out of a 16M image, before compression). 2) deflate. I read somewhere that this tool used normal zlib compression, but I haven't been able to get any zlib-based tool I have to uncompress a file created by deflate, nor to compress a file to be the same as the one deflate creates. Ideas or pointers, anyone? - -- - ------------------------------------------------------------------------------- Paul D. Smith HASMAT--HA Software Mthds & Tools "Please remain calm...I may be mad, but I am a professional." --Mad Scientist - ------------------------------------------------------------------------------- These are my opinions---Nortel Networks takes no responsibility for them. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Modify MAC addr For Enet Driver Date: Wed, 30 Jan 2002 10:25:27 -0800 From: "Dan Gold" Organization: Mailgate.ORG Server - http://www.Mailgate.ORG Message-ID: References: Sender: gold@ensemble.com "phil" wrote in message news:dd61e8fb.0201291733.6377d63c@posting.google.com... > Does anybody know if there is a clean way (or not) to modify the MAC address > for an Ethernet chip after the stack is up? Something similar to ifAddrAdd > (IP) but for MAC? > As a purist, there is no clean way - you shouldn't be doing it :-) And if you must, the way to do it is somewhat specific to your Ethernet chip and driver. It's possible your driver ioctl may support EIOCSADDR, but you'd need to look at it to make sure it does everything it's supposed to. I think if you do a Google search on this topic, there's a long thread about why not to do it... hth, Dan Gold gold@ensemble.com --------------------------- Newsgroups: comp.os.vxworks Subject: Re: 16550 UART driver Date: 30 Jan 2002 13:32:43 -0500 From: Bill Pringlemeir Organization: Bell Sympatico Message-ID: References: <3C580FC8.8DAA7C55@sepro-robotique.com> Sender: bpringlemeir@DeadDuck >>>>> "EH" == Emmanuel Herbreteau writes: EH> Hi, I'm looking for a driver for the NS16550 UART with FIFO EH> enabled (usefull for speed over 19200 bauds). EH> The default driver provides with Tornado/VxWorks 5.4 EH> doesn't use the integrated 16 bytes FIFO... (it reads only EH> one byte for each interrupts) EH> Does anyone know where to find such driver ?? Yes, you can find one in your imagination! The source code is in $(WIND_BASE)/target/src/drv/sio/ns16550Sio.c. I am guessing that you didn't know this? *** 258,264 **** REG(MCR, pChan) = pChan->mcr; /* now reset the channel mode registers */ ! REG(FCR, pChan) = (RxCLEAR | TxCLEAR | FIFO_ENABLE | FCR_RXTRIG_L); if (options & CREAD) pChan->ier |= RxFIFO_BIT; - --- 256,263 ---- REG(MCR, pChan) = pChan->mcr; /* now reset the channel mode registers */ ! ! REG(FCR, pChan) = (RxCLEAR | TxCLEAR | FIFO_ENABLE); if (options & CREAD) pChan->ier |= RxFIFO_BIT; *************** *** 335,342 **** pChan->mcr |= (MCR_DTR | MCR_RTS); REG(MCR, pChan) = pChan->mcr; ! /* clear Tx and receive and enable FIFO of 8 bytes */ ! REG(FCR, pChan) = (RxCLEAR | TxCLEAR | FIFO_ENABLE | FCR_RXTRIG_L); intUnlock (oldlevel); } - --- 334,342 ---- pChan->mcr |= (MCR_DTR | MCR_RTS); REG(MCR, pChan) = pChan->mcr; ! /* clear Tx and receive and enable FIFO */ ! ! REG(FCR, pChan) = (RxCLEAR | TxCLEAR | FIFO_ENABLE); intUnlock (oldlevel); } *************** Regards, Bill Pringlemeir. - -- Have you ever tasted brussel sprouts and thought it was cauliflower? Or been called by your sister while having sex with broccoli? You will. And the company that will bring it to you: AT&T. vxWorks FAQ, "http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html" --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Curses lib under vxworks Date: 29 Jan 2002 07:33:01 -0800 From: saint_p@yahoo.com (Brian St. Pierre) Organization: http://groups.google.com/ Message-ID: <7363e2a5.0201290733.52dc8f9b@posting.google.com> References: "Nison.liu" wrote in message news:... > Hi: > > Do you have ever use Linux/Unix curses lib under vxWorks? > I want to port a Linux/Unit text editor to vxworks, then under vxworks > target without host,i can edit some simple text files.But almost all these > text editors need curses lib support,under vxworks how can i use it and do > it? Hi - There is a curses lib at ftp://ftp.atd.ucar.edu/pub/archive/vxworks. I saw a reference in this NG for a vi clone in that archive as well, but I don't see it there now. I have used a vi clone under vxworks before, but I can't place my hands on it right now... - -Brian [don't bother replying to the from: address, it's a throwaway] reply to vxworks at bstpierre (dot) org --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Modify MAC addr For Enet Driver Date: Wed, 30 Jan 2002 18:45:02 +0000 (UTC) From: David Laight Organization: BT Openworld Message-ID: <3C583F82.62213CB6@l8s.co.uk> References: > > Does anybody know if there is a clean way (or not) to modify the MAC > > address for an Ethernet chip after the stack is up? Something similar > > to ifAddrAdd (IP) but for MAC? > > > As a purist, there is no clean way - you shouldn't be doing it :-) > And if you must, the way to do it is somewhat specific to your Ethernet > chip and driver. It's possible your driver ioctl may support EIOCSADDR, > but you'd need to look at it to make sure it does everything it's > supposed to. However the ioctl will not change the address that the vxWorks IP stack reports as its MAC address in arp responses. This might be at ipDrvCtrl[unit].idr.ac_enaddr. You also need to 'persuade' any remote systems to lose their arp table entry. > I think if you do a Google search on this topic, there's a long > thread about why not to do it... And some reasons why you might want to: 1) you are running DECnet 2) You have just 'hot swapped' the network card, and wish to use the new cards real address (somewhat safer than trying to un-plumb a vxworks network interface). David --------------------------- Newsgroups: comp.dcom.sys.cisco,comp.os.vxworks Subject: Re: Binding a socket to a interface in VxWorks Date: 30 Jan 2002 13:44:06 -0500 From: Bill Pringlemeir Organization: Bell Sympatico Message-ID: References: <6c5f1af7.0201292316.58ff333d@posting.google.com> Followup-To: comp.os.vxworks Sender: bpringlemeir@DeadDuck >>>>> "Nav" == Navaneetham writes: Nav> hi, how socket library will behave when two interfaces are Nav> having same ip address? (ppp & eth both interfaces are having Nav> same ip address. ppp is unnumbered interface.) Nav> problem I am facing is creating socket for INADDR_ANY and bind Nav> to any one port no. I am not receving packets from ppp Nav> interface. do any one know what is the problem? Is any solution Nav> for this??? working with VxWorks stack. You can not do this. The IP software layer doesn't know anything about interfaces. This is done at at different layer. I have never typed in "www.yahoo.com:atm_if3" or "64.58.76.222:tr4". How will the stack and everything else route this? It does make some sense and I see how you would fall into the trap of thinking that you can do this. However, you can't! The IP layer and routing is done base on IP address alone. There is no attached interface information. In my opinion this is good software and protocol design. You wish to clearly separate layers and IP (UDP/TCP) does this very well. It is a quite popular; perhaps for this reason. regards, Bill Pringlemeir. - -- Enjoy vril and odyle! vxWorks FAQ, "http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html" --------------------------- Newsgroups: comp.os.vxworks Subject: VxWorks Pipes Date: 29 Jan 2002 07:39:40 -0800 From: keith_p_jarvis@md.northgrum.com (Keith Jarvis) Organization: http://groups.google.com/ Message-ID: <38c6d574.0201290739.36ed5aa6@posting.google.com> Hi, I am somewhat new to the VxWorks world and would just like some guidance on a couple of things in regards to using pipes on VxWorks. Please excuse my ignorance but: First, does VxWorks support stream pipes and/or named pipes? Second, is there away to handle connections to a pipe in VxWorks. Meaning, my server want to be able to account for each tasks that connects to it. I understand that pipes in VxWorks are just an I/O device on top of the Message Queue (If I misunderstood this concept, please forgive me). I understand about using file descriptors and waiting for some action to occur on them when using sockets. Does this work in the same way? If there are any sites that anyone knows of with any more of this information, can you list them for me. Any help would be greatly appreciated. Thanks in advance Keith --------------------------- Newsgroups: comp.os.vxworks Subject: Gotcha with C++ static constructors and GCC 2.95.x (PPC) Date: Wed, 30 Jan 2002 19:19:58 GMT From: sj@shell1.aracnet.com (Scott Johnson) Organization: aracnet.com -- Portland's loudest electrons (info@aracnet.com) Message-ID: After several hours of debugging and pulling out my hair :), I've just found a (afaik undocumented) bug with GCC 2.95.x which may cause static constructors/destructors to NOT work when loaded onto the target using the target-based loader (the ld/loadModule) commands. Don't know if this effects the host-based shell or not. Essentially, the problem is as follows: The munch script can be told to produce assembly language output (actually a bunch of __asm__ directives for the C compiler to swallow) or C output. When the latter is used, additional flags must be passed to the compiler (-fdollars-in-identifiers) to get the output of munch to compile; otherwise the presence of $ in the symbols will cause the compile to fail. For this reason, the Tornado makefiles default to using the -asm option; the assembler has no problem with dollar signs. The munched file, when compiled (by either method) containts two arrays of function pointers, _ctors and _dtors; these are a NULL-terminated list of the "constructor functions" and "destructor functions" (functions, one per file that has static ctors/dtors, which call all the static ctors/dtors in the file). This seems to work fine with the 2.7.2 toolchain. But with the 2.95.x toolchain; using the assembler rather than the compiler causes a subtle change in the symbol table entry for the _ctors and _dtors symbols (one that will NOT show up with nmppc; only objdumpppc -t reveals this). The "object" bit, which I suppose didn't exist with the 2.7 binutils; is NOT set when the input to the compiler is assembly code. When the output of munch is linked back in with the module; everything is fine until you try loading it on the target. For some reason; loadLib DOES NOT RECOGNIZE _ctors and _dtors as valid symbols and thus does NOT load them into the symbol table. When cplusCtors is run (either manually, or automatically if cplusXtorSet is used); it looks in the symbol table to find _ctors; cannot find it; assumes that the module has no constructors, and silently exits. (The same applies for cplusDtors and _dtors). The workaround is to add the line MUNCH_FLAGS= to your makefiles. This will force munch to output C code (which must be compiled with -fdollars-in-identifiers); which will have the "object" bit set in the generated object file. loadLib will then properly generate symbol table entries for _ctors and _dtors, and all will be well. Again, this affects GCC 2.95.x (using DaveK's stuff) on PowerPC. Don't know about other architectures. The 2.7.2 toolchain that ships with Tornado also seems to be unaffected. Hope this helps someone out there... e_s - -- - ------------------------------------------------------------------------ engineer_scotty (no, not that one) -- consumer of fine ales and lagers some days you're the fire hydrant, some days you're the dog | go blazers no small furry creatures were harmed in the creation of this .signature --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Date: 30 Jan 2002 18:16:51 GMT From: Charly Message-ID: <2002130-191651-67831@foorum.com> References: <2002130-143411-880483@foorum.com> <3c581520$1@brateggebdc5.br-automation.co.at> There is no hard disk. The files are only on host. The application "fopen" the files throught FTP. - -- Use our news server 'news.foorum.com' from anywhere. More details at: http://nnrpinfo.go.foorum.com/ --------------------------- Newsgroups: comp.os.vxworks Subject: Re: "__fixdfdi" reference Date: Wed, 30 Jan 2002 19:44:25 GMT From: sj@shell1.aracnet.com (Scott Johnson) Organization: aracnet.com -- Portland's loudest electrons (info@aracnet.com) Message-ID: References: In article , Jeff Steinsnyder wrote: >We are using VxWorks v5.4. When the 64-bit floating point number >converted to 64-bit integer the compiler generates a reference to >"__fixdfdi" which can not be resolved at link time. What do I need to >do to resolve this reference? I'm asuming you are using gcc. If you are using Diab or another compiler; I do not know the answer. Try adding -lgcc to your final link (either of your kernel/app, or of your loaded module.) You might also have to kill the -nostdlib option that is getting passed to the compiler/linker. The necessary function is in libgcc.a (the compiler's support library) which Wind River likes to pretend doesn't exist. Look in $WIND_BASE/host/hosttype/lib/gcc-lib/powerpc-wrs-vxworks/cygnus-whatever/libgcc.a e_s - -- - ------------------------------------------------------------------------ engineer_scotty (no, not that one) -- consumer of fine ales and lagers some days you're the fire hydrant, some days you're the dog | go blazers no small furry creatures were harmed in the creation of this .signature --------------------------- Newsgroups: comp.os.vxworks Subject: MPC8260 Ethernet Driver Date: Wed, 30 Jan 2002 15:13:02 -0500 From: "Grant" Organization: Bell Sympatico Message-ID: Hi I am having a problem with the ethernet driver WRS provides for MPC8260. The default ethernet driver coming with Vxworks 5.4 and MPC 8260 BSP only supports copy on transmission. Although inside motFccEnd.c there is a function called motFccPktTransmit() which is supposed to transmit packets with zero copy, the reality is if I use this API instead of motFccPktCopyTransmit(), the vxworks cannot boot up. Actually it hangs when SYS_FCC_ENET_ENABLE is invoked. I am wondering if anybody tried the zero-copy transmission or this is a hardware problem that MPC8260 only supports copy and transmit? Thanks a lot for any comments - -- Grant --------------------------- Newsgroups: comp.os.vxworks Subject: Re: login issue Date: Wed, 30 Jan 2002 16:15:55 -0500 From: david lindauer Message-ID: <3C58628B.50702@notifier-is.net> References: <26ca8a8f.0201300030.47a0fa90@posting.google.com> - --------------020504000102020508080108 Content-Type: text/plain; charset=us-ascii; format=flowed Content-Transfer-Encoding: 7bit we're having the same problem with the FTP server. We do encrypt the passwords by vxencrypt. but new passwords/users won't log in, while the default works fine. Another problem is that if we delete the default and re-add it that quits working too. Bear in mind you have to know the password to delete it and we do reencrypt before adding it back in. Anyone got any ideas on this? David Johan Borkhuis wrote: >senthilb@comneti.com (senthil) wrote: > >>We developed a login table. Where the admin user comes first . The >>admin user is "target" as found in usrConfig.c . I could login using >>target and add new users into the login table. But once i logout , i >>could never login using the new added user names. I dont have any >>problem with the "target" username which is the first entry. The >>funtion loginUserVerify( ) always returns -1 when i try to login with >>my new user name and password. We even installed our encryption >>routine. Do any one have the code for loginUserVerify( ) or can any >>one suggest how to overcome the problem with loginUserVerify() >> > >Did you enter the passwords as passwords or as encrypted passwords as >returned by the vxencrypt untility? > >Groeten, > Johan > - --------------020504000102020508080108 Content-Type: text/html; charset=us-ascii Content-Transfer-Encoding: 7bit we're having the same problem with the FTP server.  We do encrypt the passwords by vxencrypt.
but new passwords/users won't log in, while the default works fine.  Another problem is that if we delete the default and re-add it that quits working too.  Bear in mind you have to know the password to delete it and we do reencrypt before adding it back in.

Anyone got any ideas on this?

David

Johan Borkhuis wrote:
senthilb@comneti.com (senthil) wrote:

We developed a login table. Where the admin user comes first . The
admin user is "target" as found in usrConfig.c . I could login using
target and add new users into the login table. But once i logout , i
could never login using the new added user names. I dont have any
problem with the "target" username which is the first entry. The
funtion loginUserVerify( ) always returns -1 when i try to login with
my new user name and password. We even installed our encryption
rou
tine. Do any one have the code for loginUserVerify( ) or can any
one suggest how to overcome the problem with loginUserVerify()

Did you enter the passwords as passwords or as encrypted passwords as 
returned by the vxencrypt untility?

Groeten,
        Johan


- --------------020504000102020508080108-- --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Date: Wed, 30 Jan 2002 16:20:30 -0500 From: david lindauer Message-ID: <3C58639E.1080308@notifier-is.net> References: <2002130-143411-880483@foorum.com> <3c581520$1@brateggebdc5.br-automation.co.at> <2002130-191651-67831@foorum.com> you don't do a dosfsdevinit if you don't have a hard drive; the host has the file system. I don't think there is a configuration that will automatically install a file system like this; I think you are calling dosFsDevInit somewhere in your own code. You should probably take it out, although it is just a nuisance message as long as you aren't trying to use it... David Charly wrote: >There is no hard disk. >The files are only on host. The application "fopen" the files throught FTP. > > --------------------------- Newsgroups: comp.os.vxworks Subject: tffsDevFormat fails.. Date: 30 Jan 2002 13:52:45 -0800 From: roswel_ajf@hotmail.com (roswell) Organization: http://groups.google.com/ Message-ID: <6dcae13d.0201301352.383bde3d@posting.google.com> Hi vxfolks.. well, tffsDevFormat 0, 0 fails. (value = -1 = 0xffffffff = udbinfo + 0xffed84f3). I guess, it cannot find my write/erase routines. yes, i never got the printout in my cfiscsIdentify() routine. But, i saw my xxxRegiset() is getting called. i can see noOfDrives is 1. BTW, i erased all erseunit before doing this tffsDevformat thing. now, my entire flash filled with 0xfffff...till theend. i think, the connection between FlFlash and FlSocket is not established, may. I thought the cfiscsIdentify() makes the connection. then, tffsDevFormat needs to write/erase etc.. please please help. thanks in advance (THe most funny thing is one time tffsDevFormat was actually invoking my MTD's write routine, but half was failed with the same error as above. then, via other project i got eraseunit() to erase the entire flash, then trying tffsDevFormat() fails. Does the tffsDevFormat() expect any manufacture data to be foudn in flash? ). my core is ppc750/vx5.2 --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Date: Wed, 30 Jan 2002 23:32:31 +0100 From: "Werner Schiendl" Message-ID: <3c587481$1@brateggebdc5.br-automation.co.at> References: <2002130-143411-880483@foorum.com> <3c581520$1@brateggebdc5.br-automation.co.at> <2002130-191651-67831@foorum.com> Hi, check your current working directory (on the target!) if you do not have the pwd utility compiled in your image, at least set the host as current working directory. See the log of your FTP server (or use some network monitor, e. g. the free EtherReal software) to see if your target tries to connect to the FTP server. I guess you are accidentially trying to read the file from a local drive - which does not exist. There is some mechanism to 'mount' the disk on demand - or at least delay some (file system) checks - until the first access. hth Werner "Charly" wrote in message news:2002130-191651-67831@foorum.com... > > There is no hard disk. > The files are only on host. The application "fopen" the files throught FTP. > > > -- > Use our news server 'news.foorum.com' from anywhere. > More details at: http://nnrpinfo.go.foorum.com/ --------------------------- Newsgroups: comp.os.vxworks Subject: Re: How do I use intConnect Date: Wed, 30 Jan 2002 23:45:07 +0100 From: "Werner Schiendl" Message-ID: <3c587774$1@brateggebdc5.br-automation.co.at> References: <78cc994f.0201300758.3d442448@posting.google.com> Hi, vector: the interrupt vector you wish to set - make it with the macro INUM_TO_IVEC(interrupt number) routine: your ISR parameter: This data is passed to your ISR as parameter, you can pass whatever info you consider helpful. If your device is a PCI device, you should use pciIntConnect instead, otherwise interrupt sharing will break your stuff. hth Werner "Daniel Fuchs" wrote in message news:78cc994f.0201300758.3d442448@posting.google.com... > Hello, > > I am new to VxWorks/Real-time programming. I have been given the task > to connect my HDLC driver into a VxWorks BSP. I obviously need to use > the intConnect command. > > I know that this is probably a simple question/solution but... the > problem I am having is that I do not understand the 3 inputs needed > for intConnect. I know that they are: > > VOIDFUNCPTR *vector > VOIDFUNCPTR routine > int parameter > > But do not understand what they mean. > > > > Would someone be able to help? > > > Regards, > Daniel --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Tornado2 and PVCS Date: Wed, 30 Jan 2002 23:40:04 +0100 From: "Werner Schiendl" Message-ID: <3c587646$1@brateggebdc5.br-automation.co.at> References: Hi, I don't know PVCS, we use MS SourceSafe for some VxWorks projects. As far as I was told, PVCS and SourceSafe are pretty much the same stuff (e. g. exclusive check-out paradigm). There is no need for any third party tools (especially if they are not for free - or you own them anyway) Just use the Tool that comes with your sourcecode management system. (for SourceSafe, its named "Visual SourceSafe Explorer" - and allows browsing the repository, checking-in and out, getting the latest version....) + you can configure Tornado to allow check-in and check-out directly using command-line (we do not use this) One thing to be careful are the project and workspace file. They contain Unix line endings by default, and if they are changed to DOS Line-Endings, Tornado will cease to work. In SourceSafe, we checked them in as binary files - this solves the problem for us. Alternatively, you can patch the TCL files of your Tornado install. Search the group, there was a description on how to do it some time ago (maybe its in the FAQ too). hth Werner "Martin Raabe" wrote in message news:a394lp$dbu$03$1@news.t-online.com... > Hello Ofer, > I recommend to use SNiFF+ PRO as a WindPowerTool on top of Tornado. > For more derails look at: > www.windriver.com for a free eval download. > > Ciao > > Martin Raabe > > "Ofer Goren" schrieb im Newsbeitrag > news:newscache$ku9rqg$lck$1@lnews.actcom.co.il... > > Hi. > > Does anyone work with this combination, and if so, any recommendations > about > > the best way to work? > > > > Tornado2.02 for win2kSp2 > > PVCS for windows, version 6.7.11 > > > > Best reagrds, > > > > -- > > Ofer Goren > > Iamba Technologies > > > > > > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Source for WRS build utilities? Date: Wed, 30 Jan 2002 23:50:23 +0100 From: "Werner Schiendl" Message-ID: <3c5878b0$1@brateggebdc5.br-automation.co.at> References: Hi, http://www.gzip.org/zlib/ maybe a start for point 2 hth Werner "Paul D. Smith" wrote in message news:p5lmefsot0.fsf@lemming.engeast.baynetworks.com... > Hi folks; I'm looking for the source code to, or replacements for, the > following parts of the standard WRS build tools (these are from > 5.3.1...): > > 1) elfToBin. I read somewhere that this could be replaced by objcopy > -O binary, but the output file produced by elfToBin is about 2M > larger than the one produced by objcopy; that doesn't seem right to > me :). (This is out of a 16M image, before compression). > > 2) deflate. I read somewhere that this tool used normal zlib > compression, but I haven't been able to get any zlib-based tool I > have to uncompress a file created by deflate, nor to compress a > file to be the same as the one deflate creates. > > > Ideas or pointers, anyone? > > -- > -------------------------------------------------------------------------- - ----- > Paul D. Smith HASMAT--HA Software Mthds & Tools > "Please remain calm...I may be mad, but I am a professional." --Mad Scientist > -------------------------------------------------------------------------- - ----- > These are my opinions---Nortel Networks takes no responsibility for them. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Connecting tornado tools to target over shared mem? Date: Thu, 31 Jan 2002 10:03:50 +1030 From: Tim Shaw Organization: Defence Science and Technology Organisation Message-ID: <3C5882DE.6FFE9235@dsto.defence.gov.au> References: <86wuy0q7zo.fsf@aleph.hack.org> If you've connected the PMCs to the shared memory network and configured the route tables on the host so that the host can talk to the PMCs, then the Tornado tools will work fine. We've not done quite that, but we've used Tornado tools on targets on the VME bus using the shared memory network. Dennis Roren wrote: > Hi! > I was wondering if anyone knows how, or if its at all possible to > connect the Tornado tools (f.ex. ProfileScope), using a shared > memory network on the hardware? > > Our system consists of one PB (Processor Board), and several > CB's (Controller Boards), and each of the CB's have one or two > PMC modules on them. The communication between the PB's, the > CB's and the PMC's take place over a shared memory network. > > What we would like to be able to do is to connect to the PMC > modules using this network instead of the serial connector, as > the speed is too slow to be able to give us enough information. > > If it is possible to get information about how the targetserver > connects to the PB's over the usual network, (i.e. which ports), > it should be possible to make a "proxy" task on the PB that sends > all transmission on to the PMC of our choise. > > Is there anyone that has tried to do something like this? And > is it possible? > > In our setup we also have a network connection to the PB's > (Processor Boards), via a network adapter in PMC slot1. It's > used to connect to the PB's from our network. We have a external > application that uses among other things NFS mounted directorys. > > The problem comes when we want to monitor (with for example > ProfileScope), the PMC 860 modules on one of the CB's. > Today we have to do a lot of configuration to be able to run > over the serial interface on these modules, and we are not > really satisfied with the result. > > Is there anyone that have tried to do the same? > > (Hardware: MVME 2400 cards with 860 PMC-modules) > > --------------- > > Sorry if this posting is a little disoriented, but I posted > this on the WindRiver WindSurf, and it's two postings made into > one. > > With kind regards > > // Dennis > P.s. I would be happy if you could send a CC: to my emailadress > too D.s. > > -- > Dennis Rören |Dennis.Roren@era.ericsson.se > Ericsson Radio Systems AB|Tel: +46 13 32 27 58 > Datalinjen 3 |Fax: +46 13 21 29 28 > 581 12 Linköping |(LM/ERA/SVZ/TD) --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Source for WRS build utilities? Date: 30 Jan 2002 19:09:52 -0500 From: Bill Pringlemeir Organization: Bell Sympatico Message-ID: References: Sender: bpringlemeir@DeadDuck >>>>> "Paul" == Paul D Smith writes: Paul> Hi folks; I'm looking for the source code to, or replacements Paul> for, the following parts of the standard WRS build tools (these Paul> are from 5.3.1...): Paul> 1) elfToBin. I read somewhere that this could be replaced by Paul> objcopy -O binary, but the output file produced by elfToBin is Paul> about 2M larger than the one produced by objcopy; that doesn't Paul> seem right to me :). (This is out of a 16M image, before Paul> compression). I am not quite sure what is going on. However, I do think that elfToBin is for making `ROM' images. For this reason, it may pad from zero to the start of your image with zeros. The best way to handle this is to use a linker file to insert the appropriate blank sections. Of course, you may not wish to have this padding depending on your circumstances. A way to divine the difference is to use a hex editor (and difference the results). For example, od -X foo1.bin > foo1.hex od -X foo2.bin > foo2.hex diff foo1.hex foo2.hex Or use the almighty Emacs' hexl and ediff. Paul> 2) deflate. I read somewhere that this tool used normal zlib Paul> compression, but I haven't been able to get any zlib-based tool Paul> I have to uncompress a file created by deflate, nor to compress Paul> a file to be the same as the one deflate creates. The vxWorks deflate is modified from the zlib routines. It may have a pre-configured table or something like that. The worst thing about it is that it doesn't supply an output size or a callback for kicking a watchdog [the kind that resets a processor]. I would ditch the delate and use the minizip or whatever comes with the zlib tools. I wrote my own compression user routines along with a `objcopy' for the ARM COFF format that removed a lot of cruft from the image. For instance, a typical ARM `.out' file has numerous ".LC0", "__gnu_compiled__", "__YOUR-mom.WEAR$ARMY___BOOTS_", etc. These labels/symbols are not necessary to duplicate the `loadModule' functionality. Removing the symbols and compressing can result in much smaller (70%+ reduction). I have also ported and used bzip2. It does about 10% better on ARM code, but it takes about 5x as long! Don't do this unless you have a blazingly fast processor (200Mhz+), a small image or patient users... yah right! hth, Bill Pringlemeir. - -- Keep things as simple as possible, but no simpler. - A. Einstein vxWorks FAQ, "http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html" --------------------------- Newsgroups: comp.arch.embedded,comp.os.vxworks Subject: Does nucleus plus support MMU? Date: Wed, 30 Jan 2002 16:58:00 -0800 From: "WJ" Organization: Posted via Supernews, http://www.supernews.com Message-ID: Hi, I'd like to know if Nucleus Plus supports MMU for MIPS4K? Thanks WJ --------------------------- Newsgroups: comp.os.vxworks Subject: Re: tffsDevFormat fails.. Date: Thu, 31 Jan 2002 01:33:11 GMT From: Andray Kaganovsky Organization: Primus Canada Message-ID: References: <6dcae13d.0201301352.383bde3d@posting.google.com> [AndrayK]. What is the value of int variable 'noOfMTDs' before and after call to your MTD's registration routine ? If TFFS runs out of available slots in MTD table, it will silently abort MTD installation, and won't increment 'noOfMTDs'. Order in which you install MTDs in BSP's sysTffs.c is also important. Make sure your MTD is the first one in the MTD table. I also suggest to set breakpoints on all socket's routines, and let us know the order in which they are called, and their arguments. That could help us to figure out where inside TFFS things go wrong. Andray roswel_ajf@hotmail.com (roswell) wrote in news:6dcae13d.0201301352.383bde3d@posting.google.com: > Hi vxfolks.. > > well, tffsDevFormat 0, 0 fails. (value = -1 = 0xffffffff = udbinfo + > 0xffed84f3). I guess, it cannot find my write/erase routines. yes, i > never got the printout in my cfiscsIdentify() routine. But, i saw my > xxxRegiset() is getting called. i can see noOfDrives is 1. BTW, i > erased all erseunit before doing this tffsDevformat thing. now, my > entire flash filled with 0xfffff...till theend. > > i think, the connection between FlFlash and FlSocket is not > established, may. I thought the cfiscsIdentify() makes the connection. > then, tffsDevFormat needs to write/erase etc.. > > please please help. > > thanks in advance > > (THe most funny thing is one time tffsDevFormat was actually invoking > my MTD's write routine, but half was failed with the same error as > above. then, via other project i got eraseunit() to erase the entire > flash, then trying tffsDevFormat() fails. Does the tffsDevFormat() > expect any manufacture data to be foudn in flash? ). my core is > ppc750/vx5.2 > --------------------------- Newsgroups: comp.os.vxworks Subject: semTake and ISR Date: 30 Jan 2002 18:13:24 -0800 From: mark_ck_chang@yahoo.com (Mark Chang) Organization: http://groups.google.com/ Message-ID: <9b6f8883.0201301813.e5ac79c@posting.google.com> I know that The semTake( ) routine is not callable from interrupt service routines. So what happens when I have a semTake in an IRS? Does it just return? Any damage? Since semTake basically does an IntLock. And to answer your question as to why I would do this. I don't, but some code I was debugging did and I'm wondering about the effects, if any. TIA Mark --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Source for WRS build utilities? Date: 30 Jan 2002 21:09:58 -0500 From: "Paul D. Smith" Organization: Nortel Networks, Inc. / Billerica, MA Message-ID: References: Reply-To: pausmith@nortelnetworks.com %% Bill Pringlemeir writes: Paul> 1) elfToBin. I read somewhere that this could be replaced by Paul> objcopy -O binary, but the output file produced by elfToBin is Paul> about 2M larger than the one produced by objcopy; that doesn't Paul> seem right to me :). (This is out of a 16M image, before Paul> compression). bp> I am not quite sure what is going on. However, I do think that bp> elfToBin is for making `ROM' images. For this reason, it may pad bp> from zero to the start of your image with zeros. Well, I did an "objdump -b binary --full" on both, and the elfToBin didn't appear to have lots of empty space... maybe objdump wouldn't show it? On the other hand deflate does give me a 73% compression, which implies there might be a lot of empty space there. One thing I did realize by doing this was that the objcopy output contained the ELF .comment sections, while the elfToBin didn't... but that would make the objcopy results even _smaller_, and it was already "too small" to begin with :-/. bp> A way to divine the difference is to use a hex editor (and bp> difference the results). For example, bp> od -X foo1.bin > foo1.hex bp> od -X foo2.bin > foo2.hex bp> diff foo1.hex foo2.hex bp> Or use the almighty Emacs' hexl and ediff. OK, I'll look at it this way. Maybe objdump is faking me out. Thanks. Paul> 2) deflate. I read somewhere that this tool used normal zlib Paul> compression, but I haven't been able to get any zlib-based tool Paul> I have to uncompress a file created by deflate, nor to compress Paul> a file to be the same as the one deflate creates. bp> The vxWorks deflate is modified from the zlib routines. It may bp> have a pre-configured table or something like that. The worst bp> thing about it is that it doesn't supply an output size or a bp> callback for kicking a watchdog [the kind that resets a bp> processor]. I would ditch the delate and use the minizip or bp> whatever comes with the zlib tools. bp> I wrote my own compression user routines along with a `objcopy' bp> for the ARM COFF format that removed a lot of cruft from the bp> image. For instance, a typical ARM `.out' file has numerous bp> ".LC0", "__gnu_compiled__", "__YOUR-mom.WEAR$ARMY___BOOTS_", etc. bp> These labels/symbols are not necessary to duplicate the bp> `loadModule' functionality. Removing the symbols and compressing bp> can result in much smaller (70%+ reduction). I think the ELF format collects most of that stuff into a separate section anyway, which can be removed. ELF is a lot nicer about that kind of thing than COFF. Here's the problem, though: what I need to do is reproduce _exactly_ the current image using the new tools. The goal is to "prove" that we can build the image using our own toolchain. So, while it's possible that the new image is smaller than the old but still gives the same functionality, that's not what I need, at least for now... :( Thanks for your help so far! - -- - ------------------------------------------------------------------------------- Paul D. Smith HASMAT--HA Software Mthds & Tools "Please remain calm...I may be mad, but I am a professional." --Mad Scientist - ------------------------------------------------------------------------------- These are my opinions---Nortel Networks takes no responsibility for them. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Modify MAC addr For Enet Driver Date: 30 Jan 2002 18:31:35 -0800 From: asherg@hotmail.com (phil) Organization: http://groups.google.com/ Message-ID: References: asherg@hotmail.com (phil) wrote in message news:... > Does anybody know if there is a clean way (or not) to modify the MAC address > for an Ethernet chip after the stack is up? Something similar to ifAddrAdd > (IP) but for MAC? > > I am using Tornado 2.0 with the FEI END driver. > > Thanks. Thank you for your replies. As it turns out there isn't a "clean way" to do it like I thought. I had to modify the MAC address on the chip and update the IP code by modifying the ipDrvCtrl[unit].idr.ac_enaddr entry. - -Phil. --------------------------- Newsgroups: comp.os.vxworks Subject: Where is my image? Date: 29 Jan 2002 11:07:39 -0800 From: igtorque@eliop.es (=?ISO-8859-1?Q?Ignacio_Gonz=E1lez?=) Organization: http://groups.google.com/ Message-ID: <9ec53d11.0201291107.344fc880@posting.google.com> I have a PC Pentium with a little modified pcPentium BSP, and I'm trying to add a driver for a PCI BT848 based frame-grabber. The frame grabber has an on-chip DMA controller that transfers the digitized pixels to system memory, according to a RISC program (WRITE nbytes address, SYNC, JUMP, and something else) that the driver puts in memory. I have checked the RISC program and it seems to be ok. I have checked the RISC program counter (on the bt848 registers) and it seems to be running the program. But I get only two video lines. Two lines always, even for different image sizes. Buffers are allocated with malloc and addresses are put in the bt848 registers for DMA controller's use. I have tried the cache coherency issues, using cacheDmaMalloc for allocating memory, CACHE_DMA_FLUSH for the RISC program I write, and CACHE_DMA_INVALIDATE for the image buffer the DMA controller writes. That way I loose the two video lines. Is there any issue on dma (reserve channel, ..) I'm not addressing? Thanks in advance Ignacio --------------------------- Newsgroups: comp.os.vxworks Subject: Re: semTake and ISR Date: Wed, 30 Jan 2002 21:39:41 -0800 From: "GaryM" Message-ID: References: <9b6f8883.0201301813.e5ac79c@posting.google.com> You *might* get away with it if NO_WAIT is specified; otherwise, the outcome would be unpredictable and can cause system-wide side effects. One could contrive a reason to use semTake() in an ISR, perhaps to preemptively attempt to reclaim the lock, but a better design is called for. "Mark Chang" wrote in message news:9b6f8883.0201301813.e5ac79c@posting.google.com... > I know that > The semTake( ) routine is not callable from interrupt service routines. > > So what happens when I have a semTake in an IRS? > Does it just return? > > Any damage? Since semTake basically does an IntLock. > > And to answer your question as to why I would > do this. I don't, but some code I was debugging > did and I'm wondering about the effects, if any. > > TIA > Mark --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Default scheduler in VxSim ? Date: 30 Jan 2002 21:42:58 -0800 From: deepaks@sasken.com (Deepak Shetty) Organization: http://groups.google.com/ Message-ID: <421ece65.0201302142.39ce2dba@posting.google.com> References: <421ece65.0201300525.73bf4466@posting.google.com> <3c581398$1@brateggebdc5.br-automation.co.at> HI, werner, Thanx for the explanation, i understood that real tiem apps needs to understand the imp of time and be co-operative & using proper priorities to prioritize the diff tasks is a pretty good design challenge faced by real time developers Sone of my doubts inline ... "Werner Schiendl" wrote in message news:<3c581398$1@brateggebdc5.br-automation.co.at>... > "Deepak Shetty" wrote in message > news:421ece65.0201300525.73bf4466@posting.google.com... > > Hi, friends > > I have a basic doubt on the default scheduler thats active in VxSim > > under Win98 > > I read the vxworks_guide and other documentation & found out that the > > default scheduler is a priority based scheduler which expects the > > tasks to be co-operative, in the sense that if one task uses an > > infinite loop to hog CPU, other tasks will never get a chance to run. > > This makes absolutely sense, because there are usually more and less time > critical parts in a real-time application. agreed ! > > It is an important part of a real-time design, to decide the correct > priority and timing schemes. > You cannot accept some scheduler that donates some time from a life critical > task to the low priority parts. I have a basic doubt here, since i am a newbie to real time design . Say i have 2 tasks, the task1 spawning the task 2 , and inside task2 i have started a timer whose timer handler function is also defined & connected to the timer. Now since the task2 in which the timer was invoked needs to be present always until the timer handler executes, bcos as per the doc , the timer handler executes in the context of the task in which it was created which in this case in task2. How do i achieve this, i mean how do i keep the task2 active ( w/o lettign it die of ) and also not hogging the cpu ?, so that any work in task1 can also proceed and when the timer expires my timer handler is called in the context of tas2. if i just use a big whiel loop to keep the task2 active, it will hog the cpu, not allowing the task1 to run and probably the timer handler also. do i need to use taskdelay(), pls clarify. Assume the default schedluer ( priority based and not round robin ) > > > > > Why is this kind of scheduling present at all as its not good for > > realtime applications ? and what is the reason behind makign this as > > the default scheduler instead of the otehr round-robin ( pre-emptive) > > scheduler which can be invoked using kernelTimeSlice( ticks ) function > > ? > > > > Round robin scheduling is only working with tasks of the same priority. > It's probably not used much in typical VxWorks apps. Any idea , in what cases is round robin preferred ? since its provided in vxworks, there shud be some motive behind it ? > > in case it tried to hog the cpu. also the kernelTimeSlice( ticks ) > > functions expects the no of ticks as an argument, whats the typical > > value suited for a real time app ? > > whats the typical value used for the ticks parameter ? > To make sure everything works as expected, you usually have watchdogs > guarding the software. > There are software watchdogs (like wdLib in VxWorks) and hardware watchdogs. How is this guarding achieved ? Thanx for ur time deepak --------------------------- Newsgroups: comp.os.vxworks Subject: Re: login issue Date: 30 Jan 2002 22:03:30 -0800 From: senthilb@comneti.com (senthil) Organization: http://groups.google.com/ Message-ID: <26ca8a8f.0201302203.893596c@posting.google.com> References: <26ca8a8f.0201300030.47a0fa90@posting.google.com> <3C58628B.50702@notifier-is.net> Initally, i got the encrypted password using the exe vxencrpt in bin folder. Even after adding a new user with the encrypted password i was not able to login. Inorder to override the encrypt call back we installed our routine at the start of the application. As below /**** insall user defined encrption routine ***/ loginEncryptInstall((FUNCPTR)encryptRoutine,0); /* our encryption test routine just have the same passoword added */ STATUS encryptRoutine(char *password, char *encryptedPassword) { strcpy(encryptedPassword, password); return OK; } Using this encryption routine the default user : target pswrd: password got added as we expected. Then we added another user . We were able to login using target user but not with any other user. We even had some debug mechanisms to prove that the user defined encryptRouting( ) is called when the loginUserVerify( ) function is called. The loginUserVerify( ) entered into our encrypt routine but it always returned -1 for new users added and was sucessful for target user. Please let me know how to overcome this issue. regards senthil --------------------------- Newsgroups: comp.os.vxworks Subject: Connect more than one ISR to single IV?? Date: 30 Jan 2002 22:11:00 -0800 From: mistry_amit@yahoo.com (Amit) Organization: http://groups.google.com/ Message-ID: <3ca555c5.0201302211.4f9744ef@posting.google.com> Hi friends, Do I connect more than one Interrupt Service routine to same Interrupt Vector means same Interrupt source?? I want to connect another ISR to Keyboard interrupt source. Is there any other way to achieve this?? Expecting replies from VxWorks experts!! Thanks, Amit --------------------------- Newsgroups: comp.os.vxworks Subject: what are these Assembler messages... Date: 30 Jan 2002 23:44:37 -0800 From: alimohd@logiceastern.com (Ali) Organization: http://groups.google.com/ Message-ID: Hi All I am currently working on BSP for our custom board having MPC8260 processor. When i am compiling romInit.s file it shows me the assembler messages like shown below.. as a result my file does not get build can any body tell me why this is so and what actually are these messages .. How can i get rid of these messages.I am using tornado2.0 tool and my operating system is vxworks. The gnu tool im using is PPCEC603 . What have to do so that i can build my file without this all going on. thanx in advance ali here is snapshot of these messages cd /home/alimohd/tornado/target/proj/eps8260Asm/PPCEC603gnu make -f ../Makefile BUILD_SPEC=PPCEC603gnu rominit.o ccppc -g -mstrict-align -ansi -nostdinc -fvolatile -fno-builtin - -fno-for-scope -P -x assembler-with-cpp -I. - -I/home/alimohd/tornado/target/h -DCPU=PPCEC603 -c /home/alimohd/tornado/target/config/eps8260/rominit.s -o rominit.o /var/tmp/ccpuaWfd.s: Assembler messages: /var/tmp/ccpuaWfd.s:1: Error: Rest of line ignored. First ignored character valued 0xd. /var/tmp/ccpuaWfd.s:2: Error: Rest of line ignored. First ignored character valued 0xd. /var/tmp/ccpuaWfd.s:3: Error: Rest of line ignored. First ignored character valued 0xd. /var/tmp/ccpuaWfd.s:4: Error: Bad expression /var/tmp/ccpuaWfd.s:5: Error: Rest of line ignored. First ignored character valued 0xd. /var/tmp/ccpuaWfd.s:6: Error: Rest of line ignored. First ignored character valued 0xd. /var/tmp/ccpuaWfd.s:7: Error: Rest of line ignored. First ignored character valued 0xd. /var/tmp/ccpuaWfd.s:14: Error: Rest of line ignored. First ignored character valued 0xd. /var/tmp/ccpuaWfd.s:15: Error: Rest of line ignored. First ignored character valued 0xd. /var/tmp/ccpuaWfd.s:27: Error: Rest of line ignored. First ignored character valued 0xd. /var/tmp/ccpuaWfd.s:28: Error: Rest of line ignored. First ignored character valued 0xd. /var/tmp/ccpuaWfd.s:4135: Error: Rest of line ignored. First ignored character valued 0xd. /var/tmp/ccpuaWfd.s:4138: Error: Rest of line ignored. First ignored character valued 0xd. /var/tmp/ccpuaWfd.s:4139: Error: Rest of line ignored. First ignored character valued 0xd. /var/tmp/ccpuaWfd.s:4140: Error: Rest of line ignored. First ignored character valued 0xd. /var/tmp/ccpuaWfd.s:4141: Error: Rest of line ignored. First ignored character valued 0xd. /var/tmp/ccpuaWfd.s:4142: Error: Rest of line ignored. First ignored character valued 0xd. /var/tmp/ccpuaWfd.s:4143: Error: Rest of line ignored. First ignored character valued 0xd. /var/tmp/ccpuaWfd.s:4144: Error: Rest of line ignored. First ignored character valued 0xd. /var/tmp/ccpuaWfd.s:4145: Error: Rest of line ignored. First ignored character valued 0xd. /var/tmp/ccpuaWfd.s:4147: Error: Rest of line ignored. First ignored character valued 0xd. /var/tmp/ccpuaWfd.s:4196: Error: Rest of line ignored. First ignored character valued 0xd. /var/tmp/ccpuaWfd.s:4197: Error: Rest of line ignored. First ignored character valued 0xd. /var/tmp/ccpuaWfd.s:4199: Error: Rest of line ignored. First ignored character valued 0xd. /var/tmp/ccpuaWfd.s:4200: Error: Bad expression /var/tmp/ccpuaWfd.s:4201: Error: Rest of line ignored. First ignored character valued 0xd. /var/tmp/ccpuaWfd.s:4202: Error: Rest of line ignored. First ignored character valued 0xd. /var/tmp/ccpuaWfd.s:4203: Error: Rest of line ignored. First ignored character valued 0xd. /var/tmp/ccpuaWfd.s:4204: Error: Rest of line ignored. First ignored character valued 0xd. /var/tmp/ccpuaWfd.s:4205: Error: Rest of line ignored. First ignored character valued 0xd. /var/tmp/ccpuaWfd.s:4206: Error: Rest of line ignored. First ignored character valued 0xd. /var/tmp/ccpuaWfd.s:4852: Error: Rest of line ignored. First ignored character valued 0xd. ' but expected `,'s:4853: Error: syntax error; found ` /var/tmp/ccpuaWfd.s:4855: Error: Rest of line ignored. First ignored character valued 0xd. ' but expected `,'s:4856: Error: syntax error; found ` ' but expected `,'s:4857: Error: syntax error; found ` ' but expected `,'s:4858: Error: syntax error; found ` ' but expected `,'s:4859: Error: syntax error; found ` /var/tmp/ccpuaWfd.s:4861: Error: Rest of line ignored. First ignored character valued 0xd. /var/tmp/ccpuaWfd.s:4862: Error: Rest of line ignored. First ignored character valued 0xd. /var/tmp/ccpuaWfd.s:4863: Error: Rest of line ignored. First ignored character valued 0xd. ' but expected `,'s:4864: Error: syntax error; found ` ' but expected `,'s:4866: Error: syntax error; found ` ' but expected `,'s:4868: Error: syntax error; found ` ' but expected `,'s:4870: Error: syntax error; found ` ' but expected `,'s:4872: Error: syntax error; found ` ' but expected `,'s:4874: Error: syntax error; found ` ' but expected `,'s:4876: Error: syntax error; found ` ' but expected `,'s:4878: Error: syntax error; found ` ' but expected `,'s:4880: Error: syntax error; found ` ' but expected `,'s:4882: Error: syntax error; found ` ' but expected `,'s:4884: Error: syntax error; found ` ' but expected `,'s:4886: Error: syntax error; found ` ' but expected `,'s:4888: Error: syntax error; found ` ' but expected `,'s:4890: Error: syntax error; found ` ' but expected `,'s:4892: Error: syntax error; found ` ' but expected `,'s:4894: Error: syntax error; found ` /var/tmp/ccpuaWfd.s:4896: Error: Rest of line ignored. First ignored character valued 0xd. /var/tmp/ccpuaWfd.s:4897: Error: Rest of line ignored. First ignored character valued 0xd. /var/tmp/ccpuaWfd.s:4898: Error: Rest of line ignored. First ignored character valued 0xd. ' but expected `,'s:4899: Error: syntax error; found ` ' but expected `,'s:4901: Error: syntax error; found ` ' but expected `,'s:4903: Error: syntax error; found ` ' but expected `,'s:4905: Error: syntax error; found ` /var/tmp/ccpuaWfd.s:4907: Error: Rest of line ignored. First ignored character valued 0xd. /var/tmp/ccpuaWfd.s:4908: Error: Rest of line ignored. First ignored character valued 0xd. /var/tmp/ccpuaWfd.s:4909: Error: Rest of line ignored. First ignored character valued 0xd. ' but expected `,'s:4910: Error: syntax error; found ` ' but expected `,'s:4912: Error: syntax error; found ` ' but expected `,'s:4914: Error: syntax error; found ` ' but expected `,'s:4916: Error: syntax error; found ` /var/tmp/ccpuaWfd.s:4918: Error: Rest of line ignored. First ignored character valued 0xd. /var/tmp/ccpuaWfd.s:4919: Error: Rest of line ignored. First ignored character valued 0xd. /var/tmp/ccpuaWfd.s:4920: Error: Rest of line ignored. First ignored character valued 0xd. ' but expected `,'s:4921: Error: syntax error; found ` ' but expected `,'s:4922: Error: syntax error; found ` ' but expected `,'s:4923: Error: syntax error; found ` /var/tmp/ccpuaWfd.s:4924: Error: Rest of line ignored. First ignored character valued 0xd. /var/tmp/ccpuaWfd.s:4925: Error: Rest of line ignored. First ignored character valued 0xd. /var/tmp/ccpuaWfd.s:4926: Error: Rest of line ignored. First ignored character valued 0xd. /var/tmp/ccpuaWfd.s:4996: Error: Rest of line ignored. First ignored character valued 0xd. ' but expected `,'s:4997: Error: syntax error; found ` ' but expected `,'s:4998: Error: syntax error; found ` /var/tmp/ccpuaWfd.s:4999: Error: Rest of line ignored. First ignored character valued 0xd. ' but expected `,'s:5000: Error: syntax error; found ` /var/tmp/ccpuaWfd.s:5001: Error: Rest of line ignored. First ignored character valued 0xd. make: *** [rominit.o] Error 1 hit ENTER to exit --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Tornado2 and PVCS Date: Thu, 31 Jan 2002 08:47:16 +0100 From: "Martin Raabe" Organization: T-Online Message-ID: References: <3c587646$1@brateggebdc5.br-automation.co.at> Hello Werner, I agree with you, when you are talking about pure Versioning. But then I think MS Source Safe also isn't for free! For pure versioning RCS or cvs is a good approach without any payment in Version Control Software to a vendor. I experienced that many projects need more than that, which is COnfiguration management. MS Source Safe lacks in offering sub-version numbering, which means, the version number is only one number like 1 or 534 or 3456. RCS, PVCS, CVS or others provide a better numbering scheme to reflect what happens in your project. like: 1.2.1 or 1.4.2 or 3.6.5.9.3 Also MS Source Safe doe not support branches, which might be changes being - - customer speciffic - - bug fixing - - new feature testing - - etc. I mentioned SNiFF+ here, because I don't do any development without it, is it doen with Tornado or pRISM+ or Visual Studio. Just my two new Euro Cents. Ciao Martin Raabe "Werner Schiendl" schrieb im Newsbeitrag news:3c587646$1@brateggebdc5.br-automation.co.at... > Hi, > > I don't know PVCS, we use MS SourceSafe for some VxWorks projects. > > As far as I was told, PVCS and SourceSafe are pretty much the same stuff (e. > g. exclusive check-out paradigm). > > There is no need for any third party tools (especially if they are not for > free - or you own them anyway) > Just use the Tool that comes with your sourcecode management system. > (for SourceSafe, its named "Visual SourceSafe Explorer" - and allows > browsing the repository, checking-in and out, getting the latest > version....) > > + you can configure Tornado to allow check-in and check-out directly using > command-line (we do not use this) > > One thing to be careful are the project and workspace file. > They contain Unix line endings by default, and if they are changed to DOS > Line-Endings, Tornado will cease to work. > In SourceSafe, we checked them in as binary files - this solves the problem > for us. > > Alternatively, you can patch the TCL files of your Tornado install. > Search the group, there was a description on how to do it some time ago > (maybe its in the FAQ too). > > > hth > Werner > > > "Martin Raabe" wrote in message > news:a394lp$dbu$03$1@news.t-online.com... > > Hello Ofer, > > I recommend to use SNiFF+ PRO as a WindPowerTool on top of Tornado. > > For more derails look at: > > www.windriver.com for a free eval download. > > > > Ciao > > > > Martin Raabe > > > > "Ofer Goren" schrieb im Newsbeitrag > > news:newscache$ku9rqg$lck$1@lnews.actcom.co.il... > > > Hi. > > > Does anyone work with this combination, and if so, any recommendations > > about > > > the best way to work? > > > > > > Tornado2.02 for win2kSp2 > > > PVCS for windows, version 6.7.11 > > > > > > Best reagrds, > > > > > > -- > > > Ofer Goren > > > Iamba Technologies > > > > > > > > > > > > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: 16550 UART driver Date: Thu, 31 Jan 2002 09:35:49 +0100 From: Emmanuel Herbreteau Organization: SEPRO Robotique (BEP) Message-ID: <3C5901E5.B7FC9082@sepro-robotique.com> References: <3C580FC8.8DAA7C55@sepro-robotique.com> Hi, I'm very surprised to find this driver in /src/drv/serial and (also) in /src/drv/sio. And it's not the same driver ! But, unfortunately, despite the FIFO_ENABLE bit, i'm quite sure it's still an old driver for 16450 (without FIFO) because the interrupts routines looks like : void ns16550IntRd (NS16550_CHAN * pChan ) { char inchar; /* read character from Receive Holding Reg. */ inchar = REG(RBR, pChan); (*pChan->putRcvChar) (pChan->putRcvArg, inchar); } ..and it reads only one byte per interrupt. The interrupt frequency is more than 3200 Hz for 38400 bauds (measured on a oscilloscope). It's too much to do PPP/IP during my real-time process. Does anyone know how to enable the FIFO mode on 16550 ? - -- Regards Emmanuel Bill Pringlemeir wrote: > >>>>> "EH" == Emmanuel Herbreteau writes: > EH> Hi, I'm looking for a driver for the NS16550 UART with FIFO > EH> enabled (usefull for speed over 19200 bauds). > EH> Does anyone know where to find such driver ?? > Yes, you can find one in your imagination! The source code is in > $(WIND_BASE)/target/src/drv/sio/ns16550Sio.c. I am guessing that you > didn't know this? --------------------------- Newsgroups: comp.os.vxworks Subject: Re: How do I use intConnect Date: Thu, 31 Jan 2002 11:00:12 +0100 From: Bertram Metz Organization: SBS Technologies Message-ID: References: <78cc994f.0201300758.3d442448@posting.google.com> <3c587774$1@brateggebdc5.br-automation.co.at> Hi, >If your device is a PCI device, you should use pciIntConnect instead, >otherwise interrupt sharing will break your stuff. Some BSP's (for instance some PowerPC BSP's) have central interrupt handlers, which maintain linked lists for all connected interrupt handlers. In those BSP's it's not necessary to use pciIntConnect(). intConnect() does quite the same in this case and it would cause some overhead when pciIntConnect() is used. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: How do I use intConnect Date: Thu, 31 Jan 2002 11:05:42 +0100 From: Bertram Metz Organization: SBS Technologies Message-ID: <7f5i5ukjchuibfa52uobmhf0clcthlcqsd@4ax.com> References: <78cc994f.0201300758.3d442448@posting.google.com> <3c587774$1@brateggebdc5.br-automation.co.at> Sorry, there was a wrong email address in my last posting. Bertram --------------------------- Newsgroups: comp.os.vxworks Subject: compilation> Date: Thu, 31 Jan 2002 19:26:07 +0900 From: "Girish Gulawani" Message-ID: <3c591ba1$0$27606$44c9b20d@news3.asahi-net.or.jp> hello, all. i have a LSI MIPS & a huge source code. our cpu in cached, both write through/back modes, supports page size of 16K & above. as you might know already the default compile builds them with 4K page size. what i'd like to know is, how to build them with page size 16K?? since its a static linking, the linker is not used hence i dont have choice of using ALIGN from linker script nor i can add #pragma align to the code & for this reason being its a huge collection of c & header files. my wildest guess is i should be able to do this via "specs" file by adding some directive, & build all the programs. i've already tried the option --n16384 & -Wl,n16384, no surprise it did not help. could you please tell me what is correct compiler option to build with 16K page size? [i know atleast one guy who can help me, DaveK. since i dont want to spam hence this question is posted to newsgroup & any help or idea from anybody is welcome.] please please help. many thanks in advance & best regards, girish. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Default scheduler in VxSim ? Date: Thu, 31 Jan 2002 11:17:46 +0100 From: "Werner Schiendl" Message-ID: <3c5919cb@brateggebdc5.br-automation.co.at> References: <421ece65.0201300525.73bf4466@posting.google.com> <3c581398$1@brateggebdc5.br-automation.co.at> <421ece65.0201302142.39ce2dba@posting.google.com> Hi, "Deepak Shetty" wrote in message news:421ece65.0201302142.39ce2dba@posting.google.com... > HI, werner, > Thanx for the explanation, i understood that real tiem apps needs to > understand the imp of time and be co-operative & using proper > priorities to prioritize the diff tasks is a pretty good design > challenge faced by real time developers > > Sone of my doubts inline ... > > "Werner Schiendl" wrote in message news:<3c581398$1@brateggebdc5.br-automation.co.at>... > > "Deepak Shetty" wrote in message > > news:421ece65.0201300525.73bf4466@posting.google.com... > > > Hi, friends > > > I have a basic doubt on the default scheduler thats active in VxSim > > > under Win98 > > > I read the vxworks_guide and other documentation & found out that the > > > default scheduler is a priority based scheduler which expects the > > > tasks to be co-operative, in the sense that if one task uses an > > > infinite loop to hog CPU, other tasks will never get a chance to run. > > > > This makes absolutely sense, because there are usually more and less time > > critical parts in a real-time application. > > agreed ! > > > > > It is an important part of a real-time design, to decide the correct > > priority and timing schemes. > > You cannot accept some scheduler that donates some time from a life critical > > task to the low priority parts. > > I have a basic doubt here, since i am a newbie to real time design . > Say i have 2 tasks, the task1 spawning the task 2 , and inside task2 i > have started a timer whose timer handler function is also defined & > connected to the timer. Now since the task2 in which the timer was > invoked needs to be present always until the timer handler executes, > bcos as per the doc , the timer handler executes in the context of the > task in which it was created which in this case in task2. How do i > achieve this, i mean how do i keep the task2 active ( w/o lettign it > die of ) and also not hogging the cpu ?, so that any work in task1 can > also proceed and when the timer expires my timer handler is called in > the context of tas2. if i just use a big whiel loop to keep the task2 > active, it will hog the cpu, not allowing the task1 to run and > probably the timer handler also. do i need to use taskdelay(), pls > clarify. Assume the default schedluer ( priority based and not round > robin ) > I guess you are using the timerLib functions - which are (or at least try to be) POSIX compatible. POSIX by itself is not particularly designed for real-time apps. Anyway the timerLib builds on signals, which I did not use much 'til now. If you want some kind of alarm or trigger after a certain amount of time, I think the easiest approach in VxWorks is using wdLib (the VxWorks watchdog facility). The routine you've connected in to the watchdog will be called from interrupt context (basically the system clock interrupt). Depending on what you are using the timer for, you can e. g. give a binary semaphore to trigger a task which should perform the real action. If you are using a watchdog as a guard to make sure something does not take longer than acceptable, you may do whatever brings your device in a safe fatal-error-state (shut down, reboot, whatever is appropriate). A note to the timers: You need not keep the task running in order to receive the signal. The signal handler interrupts the current code anyway - so you should of course receive the signal, if your task e. g. pends on a semaphore - which you may then give in the signal handler. > > > > > > > > > Why is this kind of scheduling present at all as its not good for > > > realtime applications ? and what is the reason behind makign this as > > > the default scheduler instead of the otehr round-robin ( pre-emptive) > > > scheduler which can be invoked using kernelTimeSlice( ticks ) function > > > ? > > > > > > > Round robin scheduling is only working with tasks of the same priority. > > It's probably not used much in typical VxWorks apps. > > Any idea , in what cases is round robin preferred ? since its provided > in vxworks, there shud be some motive behind it ? If you have some independent tasks of the same priority that do not pend by design, this may be useful. E. g. if you have to background process data and drive a user interface at the same time? There are not very much uses for round robin scheduling in a realtime system, normally. This is an idiom suitable for desktop OS, where multiple application run concurrently as fast as possible. > > > > in case it tried to hog the cpu. also the kernelTimeSlice( ticks ) > > > functions expects the no of ticks as an argument, whats the typical > > > value suited for a real time app ? > > > > > whats the typical value used for the ticks parameter ? > It depends on your system clock and how long the individual actions may be waiting (think of a user interface, if it isn't processed every say 100ms, it is not really usable well) and the number of concurrent tasks. Normally you wish a low tick number to have the illusion of truly parallel action. On the other hand, every context switch consumes time by itself. I would think 10 ms to be a good starting point on an average, up to date CPU. This means, for the default system clock rate of 60 Hz, I'd probably start with a tick value of 1 and increase it in case to much time is wasted with the context switches. > > > To make sure everything works as expected, you usually have watchdogs > > guarding the software. > > There are software watchdogs (like wdLib in VxWorks) and hardware watchdogs. > > How is this guarding achieved ? For the software watchdog see above, and see the VxWorks documenation on watchdogs (wdLib). The action you take is completely dependent on your device. For some device, shut down may be the safest - for other devices this would not be acceptable. A hardware watchdog is a timer which must be reset from the software before it expires. If it is not, then (usually) the board is reset by hardware. This makes sure, that the system is running - if it locks up, e. g. in an crazy ISR function, the hardware watchdog will reset the system to a safe state. hth Werner --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Tornado2 and PVCS Date: Thu, 31 Jan 2002 11:30:51 +0100 From: "Werner Schiendl" Message-ID: <3c591cdc$1@brateggebdc5.br-automation.co.at> References: <3c587646$1@brateggebdc5.br-automation.co.at> Hi, "Martin Raabe" wrote in message news:a3asq2$j3r$07$1@news.t-online.com... > Hello Werner, > I agree with you, > when you are talking about pure Versioning. > But then I think MS Source Safe also isn't for free! Yes, of course. PVCS is not free, too (IIRC). The thing is not if it is free, but if one has to spend money for it. If you have the tool already, you can of course use it. > > For pure versioning RCS or cvs is a good approach without any payment > in Version Control Software to a vendor. > rcs is probably a bit out-dated (if you mean that old unix command line thing I have in mind). cvs is probably a very good choice, given the huge number of open source projects which successfully use it. However, cvs has a completely different approach to versioning (by default) than VSS (and I think PVCS too): You do not need to check-out a file in order to make changes, and other cannot edit the same file while you make your changes, like in VSS. Instead, everyone can edit the files at will and cvs helps merging the files (by preventing overwriting changes, merging changed,...) The philosophy may - or may not - be acceptable or even superior in any given shop. > I experienced that many projects need more than that, which is COnfiguration > management. > MS Source Safe lacks in offering sub-version numbering, which means, the > version number is > only one number like 1 or 534 or 3456. > RCS, PVCS, CVS or others provide a better numbering scheme to reflect what > happens in your project. > like: 1.2.1 or 1.4.2 or 3.6.5.9.3 Probably a philosophic question, VSS supports labels, which you will use for releases, patches, etc. > > Also MS Source Safe doe not support branches, which might be changes being > - customer speciffic > - bug fixing > - new feature testing > - etc. > This is not true vor VSS 6.0, it supports branching, sharing and merging of files. The idioms are different than in e. g. cvs, but not necessarily more limited. > I mentioned SNiFF+ here, because I don't do any development without it, > is it doen with Tornado or pRISM+ or Visual Studio. > > Just my two new Euro Cents. > > Ciao > I'm not a SourceSafe advocate for reasons (poor performance in WAN w/ limited bandwith, exclusive checkout only, less-than-optimal command line support). But it has good features as well (nicely integrated with MS dev tools, nice project explorer, easy to teach and use) hth Werner --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Connect more than one ISR to single IV?? Date: Thu, 31 Jan 2002 11:41:15 +0100 From: "Werner Schiendl" Message-ID: <3c591f4b$1@brateggebdc5.br-automation.co.at> References: <3ca555c5.0201302211.4f9744ef@posting.google.com> Hi, you can get the current interrupt service routine with intVecGet() Depending on your target you *may* be able to call this vector in your own handler and thus daisy chain the two. If you have the source for the first ISR (and the code that attaches it, in particular) you can probably just use pciIntConnect instead of intConnect. The former manages a list as it is used for shared interrupts on PCI bus. If you just want to do additional processing to the same interrupt (i. e. from the same device), the best way would be modifying the original ISR. Appart from getting the source code, this will cause the least headaches. hth Werner "Amit" wrote in message news:3ca555c5.0201302211.4f9744ef@posting.google.com... > Hi friends, > > Do I connect more than one Interrupt Service routine to same Interrupt > Vector means same Interrupt source?? > > I want to connect another ISR to Keyboard interrupt source. > > Is there any other way to achieve this?? > > Expecting replies from VxWorks experts!! > > Thanks, > Amit --------------------------- Newsgroups: comp.os.vxworks Subject: Re: what are these Assembler messages... Date: Thu, 31 Jan 2002 08:48:39 GMT From: "J. J. Farrell" Organization: EarthLink Inc. -- http://www.EarthLink.net Message-ID: References: "Ali" wrote in message news:b305b5bd.0201302344.6131ff25@posting.google.com... > Hi All > I am currently working on BSP for our custom board having MPC8260 > processor. > When i am compiling romInit.s file it shows me the assembler messages > like shown below.. > > /var/tmp/ccpuaWfd.s:4852: Error: Rest of line ignored. First ignored > character valued 0xd. > ' but expected `,'s:4853: Error: syntax error; found ` It's complaining about finding Carriage Return characters (0xd), presumably at the end of each line. I guess your editor is putting extra ones in, or perhaps you copied the file from a Windows machine without converting the line endings. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: login issue Date: 31 Jan 2002 00:59:57 -0800 From: prempk@comneti.com (Premkumar) Organization: http://groups.google.com/ Message-ID: <89792565.0201310059.487be7b2@posting.google.com> References: <26ca8a8f.0201300030.47a0fa90@posting.google.com> senthilb@comneti.com (senthil) wrote in message news:<26ca8a8f.0201300030.47a0fa90@posting.google.com>... > We developed a login table. Where the admin user comes first . The > admin user is "target" as found in usrConfig.c . I could login using > target and add new users into the login table. But once i logout , i > could never login using the new added user names. I dont have any > problem with the "target" username which is the first entry. The > funtion loginUserVerify( ) always returns -1 when i try to login with > my new user name and password. We even installed our encryption > routine. Do any one have the code for loginUserVerify( ) or can any > one suggest how to overcome the problem with loginUserVerify() > > senthil We had fixed the above login issue.The variable which we used to get the new user name and password should be a global variable.This solves the problem w.r.t the loginUserVerify() for the new users added to the existing login table,u should use this global variable for any kind of reference to the new users. still it is advisable to have our own encryption algorithm rather then the one using vxencrypt. regds, Premkumar --------------------------- Newsgroups: comp.os.vxworks Subject: Re: 16550 UART driver Date: 31 Jan 2002 09:12:44 GMT From: Johan Borkhuis Organization: Agere Systems Message-ID: References: <3C580FC8.8DAA7C55@sepro-robotique.com> <3C5901E5.B7FC9082@sepro-robotique.com> Emmanuel Herbreteau wrote: > Hi, > > I'm very surprised to find this driver in /src/drv/serial and (also) > in /src/drv/sio. And it's not the same driver ! > > But, unfortunately, despite the FIFO_ENABLE bit, i'm quite sure > it's still an old driver for 16450 (without FIFO) because the > interrupts routines looks like : > > void ns16550IntRd (NS16550_CHAN * pChan ) > { > char inchar; > /* read character from Receive Holding Reg. */ > inchar = REG(RBR, pChan); > (*pChan->putRcvChar) (pChan->putRcvArg, inchar); > } > > ..and it reads only one byte per interrupt. The interrupt frequency > is more than 3200 Hz for 38400 bauds (measured on a oscilloscope). You could try to improve this driver by reading more than one byte at a time. If you have the specs of the 16550 it should not be very difficult to do that. Groeten, Johan - -- o o o o o o o . . . _____________________________ o _____ || Johan Borkhuis | .][__n_n_|DD[ ====_____ | borkhuis@agere.com | >(________|__|_[_________]_|__________________________| _/oo OOOOO oo` ooo ooo 'o!o!o o!o!o` === VxWorks FAQ: http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html === --------------------------- Newsgroups: comp.os.vxworks Subject: Re: About STLport Date: Thu, 31 Jan 2002 10:25:31 +0100 From: "Geurt Vos" Organization: Rohill Technologies BV Message-ID: References: First note that I haven't yet successfully built STLport....I'm working on it. > > I'm trying to make vxWorks module with STLport 4.5.1 . > > However I got the error message. > > ----------------------------------------------- > d:/home/kino/vss/pg-x/develop/src/STLport-4.5.1/stlport/cstring:27: > ../include/string.h: No such file or directory > In file included from > d:/home/kino/vss/pg-x/develop/src/STLport-4.5.1/stlport/stl/_algobase.h:47 > : > STLport assumes you're using 'default GCC', where at least the standard C library header files are located in "../include". This is obviously not true for VxWorks. There are several macros you can change to fix this. I think it's easiest to change in config/stl_gcc.h (near line 320): _STLP_NATIVE_C_INCLUDE_PATH into something like: #define _STLP_NATIVE_C_INCLUDE_PATH \ C:/Tornado/target/h The next problem you'll encounter is that several files in target/h use 'include_next'. For instance in C:\Tornado\target\h\stddef.h #include_next which I think in my case should refer to: C:\Tornado\host\x86-win32\lib\gcc-lib\i586-wrs-vxworks\2.96-PentiumIII-99111 2\include\stddef.h This means you'll have to have this directory in your include path _before_ the STLport directory. If you don't do this, you'll get recursive includes: target/h/stddef.h - included by stlport/stddef.h - will again include stlport/stddef.h. ...and there's of course multithreading issues. You'll want some parts of STLport to be protected by mutexes... HTH, Geurt --------------------------- Newsgroups: comp.os.vxworks Subject: Re: semTake and ISR Date: Thu, 31 Jan 2002 06:26:39 -0500 From: joe durusau Organization: Bellsouth.Net Message-ID: <3C5929EF.3B8D52E2@bellsouth.net> References: <9b6f8883.0201301813.e5ac79c@posting.google.com> I've seen such things crash the whole system. In other cases, semTake returns error. I suspect that it would depend on your BSP. In any case, just wipe it out, as well as anything that is not specifically listed as ISR-callable in the programmer's guide. Speaking only for myself, Joe Durusau Mark Chang wrote: > I know that > The semTake( ) routine is not callable from interrupt service routines. > > So what happens when I have a semTake in an IRS? > Does it just return? > > Any damage? Since semTake basically does an IntLock. > > And to answer your question as to why I would > do this. I don't, but some code I was debugging > did and I'm wondering about the effects, if any. > > TIA > Mark --------------------------- Newsgroups: comp.os.vxworks Subject: Re: what are these Assembler messages... Date: Thu, 31 Jan 2002 06:29:33 -0500 From: joe durusau Organization: Bellsouth.Net Message-ID: <3C592A9D.F5FBD591@bellsouth.net> References: I haven't seen this before, but it sounds like you somehow corrupted the file by changing the \n's at the end of the lines to \r\n. Maybe putting the file through dos2unix or something similar will help. Speaking only for myself, Joe Durusau Ali wrote: > Hi All > I am currently working on BSP for our custom board having MPC8260 > processor. > When i am compiling romInit.s file it shows me the assembler messages > like shown below.. > as a result my file does not get build can any body tell me why this > is so and what actually are these messages .. > How can i get rid of these messages.I am using tornado2.0 tool and my > operating system is vxworks. > The gnu tool im using is PPCEC603 . > What have to do so that i can build my file without this all going > on. > thanx in advance > ali > here is snapshot of these messages > > cd /home/alimohd/tornado/target/proj/eps8260Asm/PPCEC603gnu > make -f ../Makefile BUILD_SPEC=PPCEC603gnu rominit.o > ccppc -g -mstrict-align -ansi -nostdinc -fvolatile -fno-builtin > -fno-for-scope -P -x assembler-with-cpp -I. > -I/home/alimohd/tornado/target/h -DCPU=PPCEC603 -c > /home/alimohd/tornado/target/config/eps8260/rominit.s -o rominit.o > /var/tmp/ccpuaWfd.s: Assembler messages: > /var/tmp/ccpuaWfd.s:1: Error: Rest of line ignored. First ignored > character valued 0xd. > /var/tmp/ccpuaWfd.s:2: Error: Rest of line ignored. First ignored > character valued 0xd. > /var/tmp/ccpuaWfd.s:3: Error: Rest of line ignored. First ignored > character valued 0xd. > /var/tmp/ccpuaWfd.s:4: Error: Bad expression > /var/tmp/ccpuaWfd.s:5: Error: Rest of line ignored. First ignored > character valued 0xd. > /var/tmp/ccpuaWfd.s:6: Error: Rest of line ignored. First ignored > character valued 0xd. > /var/tmp/ccpuaWfd.s:7: Error: Rest of line ignored. First ignored > character valued 0xd. > /var/tmp/ccpuaWfd.s:14: Error: Rest of line ignored. First ignored > character valued 0xd. > /var/tmp/ccpuaWfd.s:15: Error: Rest of line ignored. First ignored > character valued 0xd. > /var/tmp/ccpuaWfd.s:27: Error: Rest of line ignored. First ignored > character valued 0xd. > /var/tmp/ccpuaWfd.s:28: Error: Rest of line ignored. First ignored > character valued 0xd. > /var/tmp/ccpuaWfd.s:4135: Error: Rest of line ignored. First ignored > character valued 0xd. > /var/tmp/ccpuaWfd.s:4138: Error: Rest of line ignored. First ignored > character valued 0xd. > /var/tmp/ccpuaWfd.s:4139: Error: Rest of line ignored. First ignored > character valued 0xd. > /var/tmp/ccpuaWfd.s:4140: Error: Rest of line ignored. First ignored > character valued 0xd. > /var/tmp/ccpuaWfd.s:4141: Error: Rest of line ignored. First ignored > character valued 0xd. > /var/tmp/ccpuaWfd.s:4142: Error: Rest of line ignored. First ignored > character valued 0xd. > /var/tmp/ccpuaWfd.s:4143: Error: Rest of line ignored. First ignored > character valued 0xd. > /var/tmp/ccpuaWfd.s:4144: Error: Rest of line ignored. First ignored > character valued 0xd. > /var/tmp/ccpuaWfd.s:4145: Error: Rest of line ignored. First ignored > character valued 0xd. > /var/tmp/ccpuaWfd.s:4147: Error: Rest of line ignored. First ignored > character valued 0xd. > /var/tmp/ccpuaWfd.s:4196: Error: Rest of line ignored. First ignored > character valued 0xd. > /var/tmp/ccpuaWfd.s:4197: Error: Rest of line ignored. First ignored > character valued 0xd. > /var/tmp/ccpuaWfd.s:4199: Error: Rest of line ignored. First ignored > character valued 0xd. > /var/tmp/ccpuaWfd.s:4200: Error: Bad expression > /var/tmp/ccpuaWfd.s:4201: Error: Rest of line ignored. First ignored > character valued 0xd. > /var/tmp/ccpuaWfd.s:4202: Error: Rest of line ignored. First ignored > character valued 0xd. > /var/tmp/ccpuaWfd.s:4203: Error: Rest of line ignored. First ignored > character valued 0xd. > /var/tmp/ccpuaWfd.s:4204: Error: Rest of line ignored. First ignored > character valued 0xd. > /var/tmp/ccpuaWfd.s:4205: Error: Rest of line ignored. First ignored > character valued 0xd. > /var/tmp/ccpuaWfd.s:4206: Error: Rest of line ignored. First ignored > character valued 0xd. > /var/tmp/ccpuaWfd.s:4852: Error: Rest of line ignored. First ignored > character valued 0xd. > ' but expected `,'s:4853: Error: syntax error; found ` > /var/tmp/ccpuaWfd.s:4855: Error: Rest of line ignored. First ignored > character valued 0xd. > ' but expected `,'s:4856: Error: syntax error; found ` > ' but expected `,'s:4857: Error: syntax error; found ` > ' but expected `,'s:4858: Error: syntax error; found ` > ' but expected `,'s:4859: Error: syntax error; found ` > /var/tmp/ccpuaWfd.s:4861: Error: Rest of line ignored. First ignored > character valued 0xd. > /var/tmp/ccpuaWfd.s:4862: Error: Rest of line ignored. First ignored > character valued 0xd. > /var/tmp/ccpuaWfd.s:4863: Error: Rest of line ignored. First ignored > character valued 0xd. > ' but expected `,'s:4864: Error: syntax error; found ` > ' but expected `,'s:4866: Error: syntax error; found ` > ' but expected `,'s:4868: Error: syntax error; found ` > ' but expected `,'s:4870: Error: syntax error; found ` > ' but expected `,'s:4872: Error: syntax error; found ` > ' but expected `,'s:4874: Error: syntax error; found ` > ' but expected `,'s:4876: Error: syntax error; found ` > ' but expected `,'s:4878: Error: syntax error; found ` > ' but expected `,'s:4880: Error: syntax error; found ` > ' but expected `,'s:4882: Error: syntax error; found ` > ' but expected `,'s:4884: Error: syntax error; found ` > ' but expected `,'s:4886: Error: syntax error; found ` > ' but expected `,'s:4888: Error: syntax error; found ` > ' but expected `,'s:4890: Error: syntax error; found ` > ' but expected `,'s:4892: Error: syntax error; found ` > ' but expected `,'s:4894: Error: syntax error; found ` > /var/tmp/ccpuaWfd.s:4896: Error: Rest of line ignored. First ignored > character valued 0xd. > /var/tmp/ccpuaWfd.s:4897: Error: Rest of line ignored. First ignored > character valued 0xd. > /var/tmp/ccpuaWfd.s:4898: Error: Rest of line ignored. First ignored > character valued 0xd. > ' but expected `,'s:4899: Error: syntax error; found ` > ' but expected `,'s:4901: Error: syntax error; found ` > ' but expected `,'s:4903: Error: syntax error; found ` > ' but expected `,'s:4905: Error: syntax error; found ` > /var/tmp/ccpuaWfd.s:4907: Error: Rest of line ignored. First ignored > character valued 0xd. > /var/tmp/ccpuaWfd.s:4908: Error: Rest of line ignored. First ignored > character valued 0xd. > /var/tmp/ccpuaWfd.s:4909: Error: Rest of line ignored. First ignored > character valued 0xd. > ' but expected `,'s:4910: Error: syntax error; found ` > ' but expected `,'s:4912: Error: syntax error; found ` > ' but expected `,'s:4914: Error: syntax error; found ` > ' but expected `,'s:4916: Error: syntax error; found ` > /var/tmp/ccpuaWfd.s:4918: Error: Rest of line ignored. First ignored > character valued 0xd. > /var/tmp/ccpuaWfd.s:4919: Error: Rest of line ignored. First ignored > character valued 0xd. > /var/tmp/ccpuaWfd.s:4920: Error: Rest of line ignored. First ignored > character valued 0xd. > ' but expected `,'s:4921: Error: syntax error; found ` > ' but expected `,'s:4922: Error: syntax error; found ` > ' but expected `,'s:4923: Error: syntax error; found ` > /var/tmp/ccpuaWfd.s:4924: Error: Rest of line ignored. First ignored > character valued 0xd. > /var/tmp/ccpuaWfd.s:4925: Error: Rest of line ignored. First ignored > character valued 0xd. > /var/tmp/ccpuaWfd.s:4926: Error: Rest of line ignored. First ignored > character valued 0xd. > /var/tmp/ccpuaWfd.s:4996: Error: Rest of line ignored. First ignored > character valued 0xd. > ' but expected `,'s:4997: Error: syntax error; found ` > ' but expected `,'s:4998: Error: syntax error; found ` > /var/tmp/ccpuaWfd.s:4999: Error: Rest of line ignored. First ignored > character valued 0xd. > ' but expected `,'s:5000: Error: syntax error; found ` > /var/tmp/ccpuaWfd.s:5001: Error: Rest of line ignored. First ignored > character valued 0xd. > make: *** [rominit.o] Error 1 > hit ENTER to exit --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Tornado2 and PVCS Date: 31 Jan 2002 01:48:32 -0800 From: igtorque@eliop.es (=?ISO-8859-1?Q?Ignacio_Gonz=E1lez?=) Organization: http://groups.google.com/ Message-ID: <9ec53d11.0201310148.3c625ef5@posting.google.com> References: <3c587646$1@brateggebdc5.br-automation.co.at> We use RCS (free) for DOS with our Tornado2 projects, plus a MS-Windows front-end developed by myself, and a set of equivalent batch procedures for the anti-windows people. In fact, we have been using PVCS (for DOS) for many years, and we changed to RCS when we were forced to use long names, like some of the sources distributed by Tornado (not supported by our robust, efficient, 2ky-compliant but very old PVCS 1990 version.) RCS is almost identical to bare PVCS, although has a little less options (which I had to supplement with some awk scripts called from the batch files and MS-Windows front-end.) As pointed by Werner, we also had to treat the project and workspace files as binary files (-b option in RCS) and the rest as text files (-kv option.) We learned it the hard way. We don't integrate RCS as a tool in Tornado. In fact, we use MultiEdit as the main IDE. From here we launch Tornado and RCS-Windows as separte applications. We also configure Tornado to use MultiEdit as the external editor. MultiEdit has the nice feature of updating automatically the contents and attributes of the files opened in its IDE and modified by other applications (like RCS or Tornado.) "Werner Schiendl" wrote in message news:<3c587646$1@brateggebdc5.br-automation.co.at>... > Hi, > > I don't know PVCS, we use MS SourceSafe for some VxWorks projects. > > As far as I was told, PVCS and SourceSafe are pretty much the same stuff (e. > g. exclusive check-out paradigm). > > There is no need for any third party tools (especially if they are not for > free - or you own them anyway) > Just use the Tool that comes with your sourcecode management system. > (for SourceSafe, its named "Visual SourceSafe Explorer" - and allows > browsing the repository, checking-in and out, getting the latest > version....) > > + you can configure Tornado to allow check-in and check-out directly using > command-line (we do not use this) > > One thing to be careful are the project and workspace file. > They contain Unix line endings by default, and if they are changed to DOS > Line-Endings, Tornado will cease to work. > In SourceSafe, we checked them in as binary files - this solves the problem > for us. > > Alternatively, you can patch the TCL files of your Tornado install. > Search the group, there was a description on how to do it some time ago > (maybe its in the FAQ too). > > > hth > Werner > > > "Martin Raabe" wrote in message > news:a394lp$dbu$03$1@news.t-online.com... > > Hello Ofer, > > I recommend to use SNiFF+ PRO as a WindPowerTool on top of Tornado. > > For more derails look at: > > www.windriver.com for a free eval download. > > > > Ciao > > > > Martin Raabe > > > > "Ofer Goren" schrieb im Newsbeitrag > > news:newscache$ku9rqg$lck$1@lnews.actcom.co.il... > > > Hi. > > > Does anyone work with this combination, and if so, any recommendations > about > > > the best way to work? > > > > > > Tornado2.02 for win2kSp2 > > > PVCS for windows, version 6.7.11 > > > > > > Best reagrds, > > > > > > -- > > > Ofer Goren > > > Iamba Technologies > > > > > > > > > > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: problem with -mlongcall ( with one more interesting point) Date: 31 Jan 2002 01:53:21 -0800 From: kumarpavan@lycos.com (pavan) Organization: http://groups.google.com/ Message-ID: <307af50b.0201310153.36844789@posting.google.com> References: <307af50b.0201290317.689ad29e@posting.google.com> <9JB58.38$5o.67861@newsr2.u-net.net> <307af50b.0201300435.637b546@posting.google.com> hi, i am posting again for the same problem with more inputs now. One more interesting ( / strange ?? ) problem is that, if you comment out the destructor for class A then Machine check exception doesn't come. The machine check occurs only in the second iteration of the while loop with destructor for class A. here is the detailed list of outputs obtained from the test program. - -> ld 1,0, "test.o" value = 268427016 = 0xfffdf08 = A::A(void) + 0x488 - -> d 0xfffdf08, 100 0fffdf00: 001f 737c 0000 0000 * s|....* 0fffdf10: 0000 0000 7465 7374 2e6f 00ee eeee eeee *....test.o......* 0fffdf20: eeee eeee eeee eeee eeee eeee eeee eeee *................* 0fffdf30: eeee eeee eeee eeee eeee eeee eeee eeee *................* 0fffdf40: eeee eeee eeee eeee eeee eeee eeee eeee *................* 0fffdf50: eeee eeee eeee eeee eeee eeee eeee eeee *................* 0fffdf60: eeee eeee eeee eeee eeee eeee eeee eeee *................* 0fffdf70: eeee eeee eeee ee00 eeee eeee eeee eeee *................* 0fffdf80: eeee eeee eeee eeee eeee eeee eeee eeee *................* 0fffdf90: eeee eeee eeee eeee eeee eeee eeee eeee *................* 0fffdfa0: eeee eeee eeee eeee eeee eeee eeee eeee *................* 0fffdfb0: eeee eeee eeee eeee eeee eeee eeee eeee *................* 0fffdfc0: eeee eeee eeee eeee eeee eeee eeee eeee *................* value = 21 = 0x15 - -> sp checkMine task spawned: id = 0xf43b038, name = t1 value = 256094264 = 0xf43b038 - -> Address of A is $0f43afb8 machine check : Master Abort Received MPC107, Error Detection Register 1 (0xc1) = 0x00 MPC107, Error Detection Register 2 (0xc5) = 0x00 MPC107, 60x Bus Error Status Register (0xc3) = 0x14 MPC107, PCI Bus Error Status Register (0xc7) = 0x00 MPC107, Bus Error Address Register (0xc8) = 0x20000000 machine check Exception next instruction address: 0x0fffd968 Machine Status Register: 0x0004b030 Condition Register: 0x42000080 Task: 0xf43b038 "t1" - -> ti t1 NAME ENTRY TID PRI STATUS PC SP ERRNO DELAY - ---------- ------------ -------- --- ---------- -------- -------- ------- ----- t1 checkMine f43b038 100 SUSPEND fffd968 f43afb0 0 0 stack: base 0xf43b038 end 0xf436218 size 19728 high 912 margin 18816 options: 0x1d VX_SUPERVISOR_MODE VX_DEALLOC_STACK VX_FP_TASK VX_STDIO r0 = ffffffff sp = f43afb0 r2 = 0 r3 = f43afb8 r4 = 2 r5 = 1 r6 = 1 r7 = a r8 = 2 r9 = 9421ffe4 r10 = 4 r11 = f43afb0 r12 = 0 r13 = 0 r14 = 0 r15 = 0 r16 = 0 r17 = 0 r18 = 0 r19 = 0 r20 = 0 r21 = 0 r22 = 0 r23 = 0 r24 = 0 r25 = 0 r26 = 0 r27 = 0 r28 = fffda80 r29 = fffda08 r30 = 0 r31 = f43afb0 msr = 4b030 lr = fffd958 ctr = 0 pc = fffd968 cr = 42000080 xer = 0 fpcsr = 0 fr0 = NaN fr1 = NaN fr2 = NaN fr3 = NaN fr4 = NaN fr5 = NaN fr6 = NaN fr7 = NaN fr8 = NaN fr9 = NaN fr10 = NaN fr11 = NaN fr12 = NaN fr13 = NaN fr14 = NaN fr15 = NaN fr16 = NaN fr17 = NaN fr18 = NaN fr19 = NaN fr20 = NaN fr21 = NaN fr22 = NaN fr23 = NaN fr24 = NaN fr25 = NaN fr26 = NaN fr27 = NaN fr28 = NaN fr29 = NaN fr30 = NaN fr31 = NaN machine check Exception next instruction address: 0xva0lue = fffd968 Machine Status Register: 0x00004b030 Condition Register: 0x = 0x420000800 - -> l checkMine,50 checkMine: 0xfffd8f8 9421ffd0 stwu r1,-48(r1) 0xfffd8fc 7c0802a6 mfspr r0,LR 0xfffd900 93810020 stw r28,32(r1) 0xfffd904 93a10024 stw r29,36(r1) 0xfffd908 93c10028 stw r30,40(r1) 0xfffd90c 93e1002c stw r31,44(r1) 0xfffd910 90010034 stw r0,52(r1) 0xfffd914 7c3f0b78 or r31,r1,r1 0xfffd918 3d20001a lis r9,26 0xfffd91c 3ba977b0 addi r29,r9,30640 0xfffd920 7fa803a6 mtspr LR,r29 0xfffd924 4e800021 blrl 0xfffd928 7c601b78 or r0,r3,r3 0xfffd92c 7c090378 or r9,r0,r0 0xfffd930 7d204b78 or r0,r9,r9 0xfffd934 7c1d0378 or r29,r0,r0 0xfffd938 60000000 nop 0xfffd93c 48000008 b 0x0fffd944 0xfffd940 4800009c b 0x0fffd9dc 0xfffd944 387f0008 addi r3,r31,8 0xfffd948 3d201000 lis r9,4096 0xfffd94c 3b89da80 addi r28,r9,-9600 0xfffd950 7f8803a6 mtspr LR,r28 0xfffd954 4e800021 blrl 0xfffd958 7fa9eb78 or r9,r29,r29 0xfffd95c 80090000 lwz r0,0(r9) 0xfffd960 31200004 addic r9,r0,4 0xfffd964 80090000 lwz r0,0(r9) 0xfffd968 901f0010 stw r0,16(r31) 0xfffd96c 3d601000 lis r11,4096 0xfffd970 380bda08 addi r0,r11,-9720 0xfffd974 901f0014 stw r0,20(r31) 0xfffd978 381f0008 addi r0,r31,8 0xfffd97c 901f0018 stw r0,24(r31) 0xfffd980 381f0010 addi r0,r31,16 0xfffd984 90090000 stw r0,0(r9) 0xfffd988 3d201000 lis r9,4096 0xfffd98c 3869dab0 addi r3,r9,-9552 0xfffd990 389f0008 addi r4,r31,8 0xfffd994 3d200018 lis r9,24 0xfffd998 3b89b890 addi r28,r9,-18288 0xfffd99c 7f8803a6 mtspr LR,r28 0xfffd9a0 4cc63182 crxor crb6,crb6,crb6 0xfffd9a4 4e800021 blrl 0xfffd9a8 7fa9eb78 or r9,r29,r29 0xfffd9ac 80090000 lwz r0,0(r9) 0xfffd9b0 31200004 addic r9,r0,4 0xfffd9b4 81690000 lwz r11,0(r9) 0xfffd9b8 800b0000 lwz r0,0(r11) 0xfffd9bc 90090000 stw r0,0(r9) value = 268425664 = 0xfffd9c0 = checkMine + 0xc8 > > > > > while(true) i have changed the code with while(1) > > DaveK TIA, pavan --------------------------- Newsgroups: comp.os.vxworks Subject: How do I access both serial ports? Date: Thu, 31 Jan 2002 11:02:33 +0100 From: "Ole-Johan Ellingsen" Organization: Norwegian university of science and technology Message-ID: Hi. I have a Pentium target with two serial ports (http://www.eurotech.it/main/ProductNFE.asp?SubM=Catalog&Sub2M=PC/104&Sub3M= Cpu_module&Sub4M=CPU-1231&Lang=E). I use one of the serial ports, but am unable to enable the second one for use. When I type 'devs' in the console, I only get one of them listed. The target has onboard VGA and Ethernet adapter, so I don't use any serial line for communication with Tornado or anything. I have tried to remove SLIP from the project, but it still doesn't enable the last serial port. Does anyone have any idea what I can do? Regards, Ole-Johan Ellingsen --------------------------- End of New-News digest **********************