From vxwexplo-errs@csg.lbl.gov Thu Feb 1 04:03:09 2001 From: Vxworks Exploder Date: Thu Feb 1 04:03:12 PST 2001 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Thu Feb 1 04:03:05 PST 2001 Subject: Re: Mot8260 with Galileo Ethernet Switch Subject: Re: BSPs Subject: c/c++ constructor/destructor problems Subject: Send signals to task from shell Subject: RE: comp.os.vxworks newsdigest Subject: performance of RT8139 based PCI cards Subject: Re: Please help - Undefined Symbol ! Subject: What is ROUTE HOST TABLE Subject: Re: Send signals to task from shell Subject: Re: Speeding Up VxWorks NFS transfers. Subject: Re: Determining free disk space in VxWorks Subject: VxWorks drivers for idt77222 Subject: Re: c/c++ constructor/destructor problems Subject: Re: c/c++ constructor/destructor problems Subject: Spawing C++ task Subject: malloc options Subject: [Q] default include directory..? Subject: Re: memShow fields Subject: Compiler version skew Subject: Memory leak tools? Subject: How do I build a compressed VxWorks? Subject: re: What is ROUTE HOST TABLE Subject: malloc and free in vxworks Subject: Re: tt with c++ that shows passed parameters? Subject: Long filename support on dosFs? Subject: [Q] Page Fault in case of PCI Device Subject: [Q] Page Fault in case of PCI Device Subject: VxWorks Network Facilities Subject: RE: malloc and free in vxworks Subject: Re: Mot8260 with Galileo Ethernet Switch Subject: Re: vxWorks for IPX1200 Subject: Re: [Q] default include directory..? Subject: Re: Memory leak tools? Subject: Re: How do I build a compressed VxWorks? Subject: Re: Speeding Up VxWorks NFS transfers. Subject: Re: Spawing C++ task Subject: Re: Send signals to task from shell Subject: Re: OS structure Subject: TaskDelay( ) bug? Subject: managment agent Subject: Re: TaskDelay( ) bug? Subject: AlphaVME and VxWorks Subject: resource name Subject: Re: Tornado/Project location independent Makefile Subject: Web Hosting only cost HK$50 ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: Mot8260 with Galileo Ethernet Switch Date: 31 Jan 2001 07:49:07 -0500 From: vanbaren@falcon.si.com (vanbaren_gerald) Organization: Smiths Industries Message-ID: <3a7809c3$1@news.si.com> References: <9585a1$h0s$1@nnrp1.deja.com> nickp423n@my-deja.com writes: >Hi >I am trying to get the VxWorks network stack up and running on a >Motorola MPC8260. The 8260 is connected directly to a Galileo GT-48315 >Ethernet switch and is not connected to a standard phy device. Does >anyone have any experience with this? How do you configure motFcc >ethernet driver to operate without a Phy device. I've set the >MOT_FCC_USR_PHY_ISO user flag. Is this sufficient? >thanks in advance >-Nick Sorry, no. I think you are going to have to work with WindRiver to figure this one out. The PHY isolation sends a "PHY isolate" command via the MII port to the PHY, which is not what you want. You want to ditch the MII entirely. Your best bet is probably to write a "null" MII driver which is sufficient to keep the WR ethernet driver happy but doesn't really do anything. As I said, you will probably have to ask WR for at least some basic information to accomplish this. If you ask nice enough, they may provide you with the source to the MotFccEnd and MiiLib drivers. gvb - -- +---------------------------------------------------------------------------+ | Jerry Van Baren / vanbaren_gerald@si.com / Grand Rapids Mi / 616-241-7973 | | My employer is a company. Companies are artifacts of a legal system. | |________________Artifacts are incapable of having opinions.________________| --------------------------- Newsgroups: comp.os.vxworks Subject: Re: BSPs Date: Wed, 31 Jan 2001 14:11:09 +0000 From: j.news.0101@csd.co.uk Message-ID: <3A781CFD.DE8289BB@csd.co.uk> References: <9563bd$lpt$1@nnrp1.deja.com> Stuart > Can anyone point me in the direction of people who do VxWorks BSPs for > ARM7 kits. I have tried WindRiver but they don't do quite what I'm > after. Our company (CSD) carried out the port of VxWorks to ARM and have worked on it ever since. We have written a large number of ARM BSPs and are very familiar with the likely problems and solutions. If you would like to take this further, please email me or call me on +44 (0)1223 518815. - -- Jem Davies --------------------------- Newsgroups: comp.os.vxworks Subject: c/c++ constructor/destructor problems Date: Wed, 31 Jan 2001 14:32:55 GMT From: kimp@avtecinc.com (Kim Putnam) Organization: Info Avenue Internet Services Message-ID: <3a782073.6222625@192.168.1.214> Hi all, I am reposting this under a new subject, in hopes that I will get a response. We are trying to create a vxWorks image that boots via NDDS. We are using thier multicast features, and have a fileserver. This code is in cpp. It connects to the fileserver, creates a memdrv, and receives the vxworks image in blocks. I write these blocks out the the memdrv, and use the fd to call bootLoadModule. I created a test function, to insure that my ndds calls work, and commented out bootLoadModule, and downloaded this to my target. The ndds calls work fine. I then uncommented bootLoadModule,and had to include libPENTIUMgnuvx.a. I am able to download this image, but when i try to execute the function via the debugger. The box crashes, and I get the errors like "can't call constructors on target", and "can' read constructor address in target memory". Any ideas? Thanks Kim --------------------------- Newsgroups: comp.os.vxworks Subject: Send signals to task from shell Date: Wed, 31 Jan 2001 15:29:57 GMT From: daune@iba.be (Jean-François Daune) Organization: KPNQwest customer news service Message-ID: <3a782e48.27615018@news.belgium.eu.net> Hi all, is there a way to send signals to a task (SIGINT, SIGTERM,...) from the Tornado II shell ?? J-F Daune --------------------------- Newsgroups: comp.os.vxworks Subject: RE: comp.os.vxworks newsdigest Date: Sun, 28 Jan 2001 21:18:36 +0530 From: "LakshmiNarayan" Organization: Lawrence Berkeley National Laboratory Message-ID: <959do1$e4c$1@overload.lbl.gov> This is a multi-part message in MIME format. - ------=_NextPart_000_001A_01C0896F.E0828840 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: 7bit Hi I am on leave , I will be back in office on 26th Feb . For NIT192 project details contact shashank.kulkarni@wipro.com rgds Lakshminarayan S Wipro Technologies .. # 30 ,I Mn Road ,S.R Nagar ,Mission Road Bangalore,India. @: 91-80-2275590/2241730 Extn 1131 - ------=_NextPart_000_001A_01C0896F.E0828840 Content-Type: application/ms-tnef; name="winmail.dat" Content-Transfer-Encoding: base64 Content-Disposition: attachment; filename="winmail.dat" eJ8+IiQPAQaQCAAEAAAAAAABAAEAAQeQBgAIAAAA5AQAAAAAAADoAAEIgAcAGAAAAElQTS5NaWNy b3NvZnQgTWFpbC5Ob3RlADEIAQ2ABAACAAAAAgACAAEGgAMADgAAANEHAQAcABUAEgAAAAAAHAEB A5AGADAGAAAoAAAACwACAAEAAAALACMAAAAAAAMAJgAAAAAACwApAAAAAAADADYAAAAAAB4AcAAB AAAAHwAAAFJFOiBjb21wLm9zLnZ4d29ya3MgbmV3c2RpZ2VzdAAAAgFxAAEAAAAWAAAAAcCJQcYi Xo+5uPesEdS1hgCASNhh/AAAAgEdDAEAAAAiAAAAU01UUDpMQUtTSE1JTkFSQVlBTi5TSElAV0lQ Uk8uQ09NAAAACwABDgEAAABAAAYOADjksEGJwAECAQoOAQAAABgAAAAAAAAACqAlemnB1BG1hQBg ZwIeTMKAAAALAB8OAQAAAAMABhCGn1aNAwAHENEAAAAeAAgQAQAAAGUAAABISUlBTU9OTEVBVkUs SVdJTExCRUJBQ0tJTk9GRklDRU9OMjZUSEZFQkZPUk5JVDE5MlBST0pFQ1RERVRBSUxTQ09OVEFD VFNIQVNIQU5LS1VMS0FSTklAV0lQUk9DT01SR0RTAAAAAAIBCRABAAAAdwEAAHMBAADBAQAATFpG dVuFGFEDAAoAcmNwZzEyNRYyAPgLYG4OEDAzM50B9yACpAPjAgBjaArAYHNldDAgBxMCgH0ZCoF1 YwBQCwN1bG7FAiBlC6YgSGkK4wqAIEkgYW0gAiAgbIBlYXZlICwgFEBVA/BsAyBiFQBiANBrfiAL gBSAASAN4BUAFJEyADZ0aCBGZWIgFi4T5BPkRgWxTklUoDE5MiBwA2BqBZD/BUABAAGQAxAEIAWg AjAA0A0FQHMRABqRbmsua20SwGsKwAMAQAPwGREuxQWgbReqcmdkEMAT82hMYWsakG0LgArAYSp5 A5FTE+RXG8IgVG0FkGgS4AkAZwiQBCAupReVIwyCIDMRYCwUQCZNA6AIAGFkFRBTLpJSB7BhZwrB LE0EAd5pFJEhohelINJCDxEHQGUFsGUhQG5kBzAXlUAAOiA5MS04MC0AMjI3NTU5MC8RJhA0MTch EUV4dPEDoDExMxJTF+YT5BHxAgApAAADABAQAAAAAAMAERAAAAAACwABgAggBgAAAAAAwAAAAAAA AEYAAAAAA4UAAAAAAAADABCACCAGAAAAAADAAAAAAAAARgAAAABShQAA8BMAAB4AEYAIIAYAAAAA AMAAAAAAAABGAAAAAFSFAAABAAAABAAAADguNQALABWACCAGAAAAAADAAAAAAAAARgAAAAAGhQAA AAAAAAMAFoAIIAYAAAAAAMAAAAAAAABGAAAAAAGFAAAAAAAAQAAYgAggBgAAAAAAwAAAAAAAAEYA AAAAYIUAAADQduXR//8fCwAfgAggBgAAAAAAwAAAAAAAAEYAAAAADoUAAAAAAAADACCACCAGAAAA AADAAAAAAAAARgAAAAAQhQAAAAAAAAMAIYAIIAYAAAAAAMAAAAAAAABGAAAAABGFAAAAAAAAAwAj gAggBgAAAAAAwAAAAAAAAEYAAAAAGIUAAAAAAAAeADOACCAGAAAAAADAAAAAAAAARgAAAAA2hQAA AQAAAAEAAAAAAAAAHgA0gAggBgAAAAAAwAAAAAAAAEYAAAAAN4UAAAEAAAABAAAAAAAAAB4ANYAI IAYAAAAAAMAAAAAAAABGAAAAADiFAAABAAAAAQAAAAAAAAALAD2ACCAGAAAAAADAAAAAAAAARgAA AACChQAAAQAAAAsAP4ALIAYAAAAAAMAAAAAAAABGAAAAAACIAAAAAAAACwBBgAsgBgAAAAAAwAAA AAAAAEYAAAAABYgAAAAAAAACAfgPAQAAABAAAAAKoCV6acHUEbWFAGBnAh5MAgH6DwEAAAAQAAAA CqAlemnB1BG1hQBgZwIeTAIB+w8BAAAAcwAAAAAAAAA4obsQBeUQGqG7CAArKlbCAABQU1RQUlgu RExMAAAAAAAAAABOSVRB+b+4AQCqADfZbgAAAEM6XFdJTkRPV1NcQXBwbGljYXRpb24gRGF0YVxN aWNyb3NvZnRcT3V0bG9va1xvdXRsb29rLnBzdAAAAwD+DwUAAAADAA00/TcAAAIBfwABAAAAMQAA ADAwMDAwMDAwMEFBMDI1N0E2OUMxRDQxMUI1ODUwMDYwNjcwMjFFNEM0NEE1MjEwMAAAAACMKA== - ------=_NextPart_000_001A_01C0896F.E0828840-- --------------------------- Newsgroups: comp.os.vxworks Subject: performance of RT8139 based PCI cards Date: Wed, 31 Jan 2001 16:09:59 GMT From: George Varndell Organization: Deja.com Message-ID: <959dce$h3e$1@nnrp1.deja.com> Has anyone Winblaster'd a Realtek 8139 based, off-the-shelf, pci ethernet card on any target under vxworks? If so, would you be willing to share your numbers? It seems 2Mbytes/second is the best I ever see -- on any target. This is at 100Mbit, full duplex. TIA Sent via Deja.com http://www.deja.com/ --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Please help - Undefined Symbol ! Date: Wed, 31 Jan 2001 11:26:59 -0500 From: mmchugh Organization: Rancho McHugh Message-ID: <3A783CD3.4615D214@attglobal.net> References: <90s5jv$ia3$1@bcarh8ab.ca.nortel.com> <912h52$43a$1@nnrp1.deja.com> Reply-To: mchug06@attglobal.net Another alternative is to type "sysCplusEnable=1" in the target shell. That allows the shell to automatically demangle the names. You can configure vxWorks to turn sysCplusEnable on automatically or you can just externally reference it in your code and set it to 1 so you don't have to type it every time. M.McHugh ddewar@my-deja.com wrote: > > A possibility is that you ace_main is a munged C++ symbol. If it is > compiled in a .cpp file and is now given C linkage (extern "C") that > could be the case. Then it is possible that the shell doesn't know how > to decode C++ munged names. Try preceeding the definition of ace_main > with extern "C" if this sounds like it could be the problem: > > extern "C" int ace_main(unsigned int) > { > ... code here... > } > > Then recompile, relink and see if you can spawn a task using ace_main. > > Hope this helps. > > -- Don > > In article <90s5jv$ia3$1@bcarh8ab.ca.nortel.com>, > "Huynh, Hong [CAR:9N23:EXCH]" wrote: > > Tornado 2.0 > > VxWorks 5.4 > > BSP: PcPentium (466 MHz Celeron) > > > > Hi All: I have problem loading the application on to the target hard > drive. > > It is OK when I do this remotely through a Tornado Shell (everything > runs > > fine including spawning the function!). But, when do this on the > target > > itself (target has vxworks image and the application), below is what > I got. > > Can anyone tell me what is wrong? I understand that the symbol is > somehow > > missing/undefined but how to get this problem fixed? Any help or > pointers > > would be greatly appreciated? > > > > -> ld > value = 134204756 = 0x7ffcd54 = bgpNode + 0x310 > > -> kernelTimeSlice(1) > > value = 0 = 0x0 > > -> sp ace_main > > undefined symbol: ace_main > > -> lkup "ace_main" > > ace_main(unsigned int) 0x07cbd934 text (router) > > global constructors keyed to ace_main(unsigned int) 0x07cc0c80 text > (router) > > global destructors keyed to ace_main(unsigned int) 0x07cc0c48 text > (router) > > value = 0 = 0x0 > > > > Thanks > > > > Hong > > > > > > Sent via Deja.com http://www.deja.com/ > Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: What is ROUTE HOST TABLE Date: Wed, 31 Jan 2001 09:07:40 -0800 (PST) From: Kumaraswamy Mudide Organization: Lawrence Berkeley National Laboratory Message-ID: <959h8h$i65$1@overload.lbl.gov> Hi ALL, When I execute routeShow() from shell prompt, it shoes two tables one is ROUTE NET TABLE and other is ROUTE HOST TABLE. I would like to know what is the difference between them. If anyone knows about this please let me know. Thanks swamy - --- the vxWorks Users Group Exploder wrote: > Submitted-by vxwexplo-errs@csg.lbl.gov Tue Jan 30 > 01:56:11 2001 > Submitted-by: "Bosis, Ilia" > > Oops... > > Yes, unfortunately it's the known bug in the Tornado > 1.0.1. END/MUX > implementation. > It makes it impossible to create more than 1 device > of same type. > I.e. You can have devices dp0 and, say, dc0, but not > dp0 and dp1. > > Ask WRS for patch or upgrade to Tornado2. > Or (if it's an option) use the old BSD 3.4 driver. > It works. > > Best regards > Ilia. > +================================= > | Bosis Ilia > | Senior Software engineer > | Terayon Information Systems inc. > | Broadband voice business unit. > | Work phone:972-3-607-7854 > | Fax: 972-3-607-7808 > | E-mail: iliab@terayon.com > | WWW: http://www.terayon.com > +================================== > > > > -----Original Message----- > > From: vxwexplo@lbl.gov [mailto:vxwexplo@lbl.gov] > > Sent: Monday, January 29, 2001 2:32 PM > > To: vxworks_users@csg.lbl.gov > > Subject: Multiple instances of a NIC in VxWorks > > > > > > Submitted-by vxwexplo-errs@csg.lbl.gov Mon Jan 29 > 04:27:13 2001 > > Submitted-by: "Manish Kumar C" > > > > > Hi VxWorks/Tornado gurus, > > > > I have one problem. This is to do with the > multiple card > > support. I want my END driver to > > be able to support multiple instances of the NIC. > The > > driver's LOAD and START functions go through > > fine for both the cards that I have put on the > system. I > > managed to call the load for both the cards > > by putting 2 entries in the configNet.h file with > the unit > > number as 0 and 1. After the > > initialisation when I tried to do an ipAttach it > went through > > fine for unit 0 but for unit 1 it > > failed. > > > > ipAttach(0,"dp") : OK > > ipAttach(1,"dp") : ERROR > > > > After this muxShow and ifShow would only show dp0. > The device > > dp1 is not shown. I tried > > doing an endFindByName("dp",0) which returned a > non-zero > > value but endFindByName("dp",1) would > > return me 0. > > > > All these were tried with the IP_MAX_UNITS = 4, in > configNet.h > > > > Could any body please give me some hint as to why > this is > > happening, is there any known problem with > > Tornado 1.0.1 in this regard? I am using tornado > 1.0.1 on Win > > NT and my target is x86. > > > > Thanks & Regards > > > > Manish > > > > ********** > > This is a user group mailing list for vxWorks > related topics > see http://csg.lbl.gov/vxworks for posting and > other info __________________________________________________ Get personalized email addresses from Yahoo! Mail - only $35 a year! http://personal.mail.yahoo.com/ --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Send signals to task from shell Date: Wed, 31 Jan 2001 11:09:19 -0600 From: "M. Kotiaho" Organization: LMMFC - D Message-ID: <3A7846BF.8EDA4083@nothere.com> References: <3a782e48.27615018@news.belgium.eu.net> "Jean-François Daune" wrote: > Hi all, > > is there a way to send signals to a task (SIGINT, SIGTERM,...) from > the Tornado II shell ?? > > J-F Daune kill (taskId, signalNumber) --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Speeding Up VxWorks NFS transfers. Date: Wed, 31 Jan 2001 17:14:16 +0000 From: Mark Armstrong Organization: Kaon Ltd Message-ID: <3A7847E8.926DEC23@kaon.co.uk> References: <94rm96$mvp$1@overload.lbl.gov> <3a75c1a2$1@news.si.com> vanbaren_gerald wrote: > My guess is that you are being killed by dropped UDP packets and the > UDP timeout/retries are 1+ seconds. I saw this when using TFTP (also > runs over UDP) when we had an unreliable ethernet link (PPC8260 with the > CPM clocked too slow). It would burst a couple of packets, drop one, > and take a second or more to timeout and start transfering data again. > > When we fixed the clocking on the 8260 CPM, everything ran MUCH better. I agree with you as the delays were much reduced by changing the global variables: nfsReXmitSec = 5; nfsReXmitUSec = 0; to nfsReXmitSec = 0; nfsReXmitUSec = 50000; Excuse my ignorance, but when you say: "the unreliable ethernet link (PPC8260 with the > CPM clocked too slow)" What do you mean by CPM? (A clock module of some kind?) Could this be an issue with our board or switch? Regards, Mark Armstrong --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Determining free disk space in VxWorks Date: Wed, 31 Jan 2001 17:09:47 GMT From: Andy Kaiser Organization: Deja.com Message-ID: <959gsh$kjl$1@nnrp1.deja.com> References: <949rik$m8d$1@nnrp1.deja.com> Thanks to all that replied to my problem - statfs did the trick nicely. The only problem I'm having now is that statfs conveniently uses a long to count the free space available on the drive, and with my 18GB partition, it's rolling over several times. We can work around that here, though, I think. Anyway, thanks again for the suggestions offered! Andy Sent via Deja.com http://www.deja.com/ --------------------------- Newsgroups: comp.os.vxworks Subject: VxWorks drivers for idt77222 Date: Wed, 31 Jan 2001 17:12:53 GMT From: jaguar01@my-deja.com Organization: Deja.com Message-ID: <959h2b$koi$1@nnrp1.deja.com> Does anybody know of any VxWorks device drivers for idt77222? Sent via Deja.com http://www.deja.com/ --------------------------- Newsgroups: comp.os.vxworks Subject: Re: c/c++ constructor/destructor problems Date: Wed, 31 Jan 2001 18:11:33 -0000 From: "Dave Korn" Organization: Lumber Cartel (tinlc) Members #2234-2237 (owing to browser refresh) Message-ID: <1uYd6.683$i54.912594@newsr1.u-net.net> References: <3a782073.6222625@192.168.1.214> Kim Putnam wrote in message <3a782073.6222625@192.168.1.214>... >Hi all, I am reposting this under a new subject, in hopes that I will >get a response. > >We are trying to create a vxWorks image that boots via NDDS. We are [snip]. >libPENTIUMgnuvx.a. I am able to download this image, but when i try >to >execute the function via the debugger. The box crashes, and I get the >errors like "can't call constructors on target", and "can' read >constructor >address in target memory". [snip]. Standard advice: 1) Did you use the g++ compiler, not cc ? 2) Did you remember to munch your objects? DaveK - -- They laughed at Galileo. They laughed at Copernicus. They laughed at Columbus. But remember, they also laughed at Bozo the Clown. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: c/c++ constructor/destructor problems Date: Wed, 31 Jan 2001 18:06:59 GMT From: kimp@avtecinc.com (Kim Putnam) Organization: Info Avenue Internet Services Message-ID: <3a7853f1.19404250@192.168.1.214> References: <3a782073.6222625@192.168.1.214> <1uYd6.683$i54.912594@newsr1.u-net.net> Thanks for the response!! I did munch my files, but used the cc compiler and not the g++. Where do I find info on g++? in my gnu doc? K On Wed, 31 Jan 2001 18:11:33 -0000, "Dave Korn" wrote: >Kim Putnam wrote in message <3a782073.6222625@192.168.1.214>... >>Hi all, I am reposting this under a new subject, in hopes that I will >>get a response. >> >>We are trying to create a vxWorks image that boots via NDDS. We are > > [snip]. > >>libPENTIUMgnuvx.a. I am able to download this image, but when i try >>to >>execute the function via the debugger. The box crashes, and I get the >>errors like "can't call constructors on target", and "can' read >>constructor >>address in target memory". > > [snip]. > > Standard advice: > > 1) Did you use the g++ compiler, not cc ? > 2) Did you remember to munch your objects? > > > DaveK >-- >They laughed at Galileo. They laughed at Copernicus. They laughed at >Columbus. But remember, they also laughed at Bozo the Clown. > > --------------------------- Newsgroups: comp.os.vxworks Subject: Spawing C++ task Date: Wed, 31 Jan 2001 18:29:49 GMT From: "Martin Boulay" Organization: Magma Communications Ltd. Message-ID: I have some code written in C++ and tried to load and spawn a routine that uses C++. moduleShow tells me the munched .out file was loaded and lkup told me that the routine is also there but when I do sp cortexTlm I get "cortexTlm undefined symbol". What I doing wrong. I use c++ppc to compile the code and the vxwork kernal has c++ lib included. This same program executes when we use windshell to load and execute the task. --------------------------- Newsgroups: comp.os.vxworks Subject: malloc options Date: Wed, 31 Jan 2001 12:36:50 -0600 From: Kevin Cline Organization: Mayan Networks Message-ID: <3A785B42.38ABB4E@mayannetworks.com> This must be a FAQ, but I'm looking for a drop-in replacement for the pessimal malloc supplied by WRS. Any suggestions? Has anyone tried GNU malloc or a variation thereof? Any other options? Kevin Cline --------------------------- Newsgroups: comp.os.vxworks Subject: [Q] default include directory..? Date: Thu, 1 Feb 2001 05:22:44 +0900 From: "skagus" Organization: KREONet news service Message-ID: <959s6f$6da$1@usenet.kreonet.re.kr> I want use some directory as default include directory. (because I want use like this - --> #include or--> #include "somedir/aaa.h" for example .... dir1 ---- subdir1 `---subdir2 subdir1 is c/cpp directory subdir2 is h directory we can use like this in c/cpp files - ---> #include "../subdir2/blabla.h" but I want this - ---> #include "subdir2/blabla.h" or --> #include Thanks for your favor. Good Luck To You!! --------------------------- Newsgroups: comp.os.vxworks Subject: Re: memShow fields Date: Wed, 31 Jan 2001 20:54:46 GMT From: wv12@my-deja.com Organization: Deja.com Message-ID: <959u2i$1p9$1@nnrp1.deja.com> References: <94kopm$2q6$1@overload.lbl.gov> Total number of bytes allocated so far, including bytes that has been free-d. Will In article <94kopm$2q6$1@overload.lbl.gov>, "Kevin A. Sapp" wrote: > > Hello, > > Does any one know what the cumulative field > of the VxWorks memShow is ? > > The docs show that the free is the free > list stats, the alloc is the currently > allocated list... > > Thank you in advance. > > Kevin > Sent via Deja.com http://www.deja.com/ --------------------------- Newsgroups: comp.os.vxworks Subject: Compiler version skew Date: Wed, 31 Jan 2001 16:00:31 -0600 From: Kevin Cline Organization: Mayan Networks Message-ID: <3A788AFF.AD6B855B@mayannetworks.com> We have Tornado 2.0, and I've just discovered that the target compiler ccarm and the simulator compiler ccsimso don't compile the same language. The target compiler is gcc version 2.7.9, but the simulator compiler is cygnus version 2.7.2. Does anyone have a version of ccsimso that matches my version of ccarm? Any ideas when WRS will deliver target compilers based on 2.9.5? Kevin Cline --------------------------- Newsgroups: comp.os.vxworks Subject: Memory leak tools? Date: Wed, 31 Jan 2001 22:41:06 GMT From: jaguar01@my-deja.com Organization: Deja.com Message-ID: <95a4a1$7rb$1@nnrp1.deja.com> Does anybody knows of a VxWorks memory check utility program to catch memory leaks? Sent via Deja.com http://www.deja.com/ --------------------------- Newsgroups: comp.os.vxworks Subject: How do I build a compressed VxWorks? Date: Wed, 31 Jan 2001 14:12:04 -0800 From: "CICON,JAMES (HP-Vancouver,ex1)" Organization: Lawrence Berkeley National Laboratory Message-ID: <95a6bg$48u$1@overload.lbl.gov> How do I build a compressed VxWorks? Inspection of rules.bsp reveals a build for the following target: # vxWorks.st_rom - rommable standalone version using ROM compression # This builds a standalone VxWorks image that can be put in ROM. # It includes a linked in symbol table So I don't want to link in a symbol table and I am not sure what standalone implies. How is standalone different from normal VxWorks?: # vxWorks - normal vxWorks system # vxWorks.sym - symbol table of vxWorks Jim Cicon Hewlett Packard --------------------------- Newsgroups: comp.os.vxworks Subject: re: What is ROUTE HOST TABLE Date: Wed, 31 Jan 2001 14:58:55 -0800 (PST) From: Kumaraswamy Mudide Organization: Lawrence Berkeley National Laboratory Message-ID: <95a6bh$48v$1@overload.lbl.gov> Hi ALL, But when I am trying to add a sub-net in the routing table with routeAdd() routine it goes to ROUTE HOST TABLE. Here is an example routeAdd "65.29.0.0", "91.0.0.1" Here "65.29.0.0" is basically a sub-net I can have 64K host on this sub-net. So ROUTE HOST TABLE is for both sub-nets and single entity. Please let me know if I am wrong. Regards, swamy - --- Carlos Rimola wrote: > Host routes are routes to hosts (a sinle entity). > Net routes are routes to > networks (multiple entities). > > Carlos Rimola > Praxon > > At 09:12 AM 1/31/2001, the vxWorks Users Group > Exploder wrote: > >Submitted-by vxwexplo-errs@csg.lbl.gov Wed Jan 31 > 09:07:43 2001 > >Submitted-by: Kumaraswamy Mudide > > > > >Hi ALL, > > > > When I execute routeShow() from shell prompt, > it > >shoes two tables one is ROUTE NET TABLE and other > is > >ROUTE HOST TABLE. I would like to know what is the > >difference between them. If anyone knows about this > >please let me know. > > > >Thanks > >swamy > > > > > >--- the vxWorks Users Group Exploder > > wrote: > > > Submitted-by vxwexplo-errs@csg.lbl.gov Tue Jan > 30 > > > 01:56:11 2001 > > > Submitted-by: "Bosis, Ilia" > > > > > > Oops... > > > > > > Yes, unfortunately it's the known bug in the > Tornado > > > 1.0.1. END/MUX > > > implementation. > > > It makes it impossible to create more than 1 > device > > > of same type. > > > I.e. You can have devices dp0 and, say, dc0, but > not > > > dp0 and dp1. > > > > > > Ask WRS for patch or upgrade to Tornado2. > > > Or (if it's an option) use the old BSD 3.4 > driver. > > > It works. > > > > > > Best regards > > > Ilia. > > > +================================= > > > | Bosis Ilia > > > | Senior Software engineer > > > | Terayon Information Systems inc. > > > | Broadband voice business unit. > > > | Work phone:972-3-607-7854 > > > | Fax: 972-3-607-7808 > > > | E-mail: iliab@terayon.com > > > | WWW: http://www.terayon.com > > > +================================== > > > > > > > > > > -----Original Message----- > > > > From: vxwexplo@lbl.gov > [mailto:vxwexplo@lbl.gov] > > > > Sent: Monday, January 29, 2001 2:32 PM > > > > To: vxworks_users@csg.lbl.gov > > > > Subject: Multiple instances of a NIC in > VxWorks > > > > > > > > > > > > Submitted-by vxwexplo-errs@csg.lbl.gov Mon > Jan 29 > > > 04:27:13 2001 > > > > Submitted-by: "Manish Kumar C" > > > > > > > > > > > Hi VxWorks/Tornado gurus, > > > > > > > > I have one problem. This is to do with the > > > multiple card > > > > support. I want my END driver to > > > > be able to support multiple instances of the > NIC. > > > The > > > > driver's LOAD and START functions go through > > > > fine for both the cards that I have put on the > > > system. I > > > > managed to call the load for both the cards > > > > by putting 2 entries in the configNet.h file > with > > > the unit > > > > number as 0 and 1. After the > > > > initialisation when I tried to do an ipAttach > it > > > went through > > > > fine for unit 0 but for unit 1 it > > > > failed. > > > > > > > > ipAttach(0,"dp") : OK > > > > ipAttach(1,"dp") : ERROR > > > > > > > > After this muxShow and ifShow would only show > dp0. > > > The device > > > > dp1 is not shown. I tried > > > > doing an endFindByName("dp",0) which returned > a > > > non-zero > > > > value but endFindByName("dp",1) would > > > > return me 0. > > > > > > > > All these were tried with the IP_MAX_UNITS = > 4, in > > > configNet.h > > > > > > > > Could any body please give me some hint as to > why > > > this is > > > > happening, is there any known problem with > > > > Tornado 1.0.1 in this regard? I am using > tornado > > > 1.0.1 on Win > > > > NT and my target is x86. > > > > > > > > Thanks & Regards > > > > > > > > Manish > > > > > > > > > > ********** > > > > > > This is a user group mailing list for > vxWorks > > > related topics > > > see http://csg.lbl.gov/vxworks for posting > and > > > other info > > > > > >__________________________________________________ > >Get personalized email addresses from Yahoo! Mail - > only $35 > >a year! http://personal.mail.yahoo.com/ > > > >********** > > > > This is a user group mailing list for vxWorks > related topics > > see http://csg.lbl.gov/vxworks for posting and > other info > __________________________________________________ Get personalized email addresses from Yahoo! Mail - only $35 a year! http://personal.mail.yahoo.com/ --------------------------- Newsgroups: comp.os.vxworks Subject: malloc and free in vxworks Date: Wed, 31 Jan 2001 15:02:14 -0800 From: Ashutosh Zade Organization: Lawrence Berkeley National Laboratory Message-ID: <95a6bh$490$1@overload.lbl.gov> Hi Vxworks, Can we take control of malloc() and free() in VxWorks by modifying source code or any other way to tune it as per our requirement? If yes, how? Basically we want this: If malloc requests are coming from one certain application, we want to allocate memory from certain memory addresses. If malloc request is coming from some other application, we want to allocate memory from some other memory addresses. The entire memory map may not fall under contiguous memory addresses. Any idea how we can achieve this? Thanks. Ashutosh Zade --------------------------- Newsgroups: comp.os.vxworks Subject: Re: tt with c++ that shows passed parameters? Date: Wed, 31 Jan 2001 23:09:16 GMT From: Bob Witmer Organization: Deja.com Message-ID: <95a5ul$9bk$1@nnrp1.deja.com> References: <93nei4$mbj$1@nnrp1.deja.com> <8%hd6.567$i54.854034@newsr1.u-net.net> I saw some old posts in the vxWorks archive which showed the argument lists with parameters. I think the following was from a 5.31 system. Our system is 5.40 plus we were using C++, so I wasn't sure if there was some variable I needed to set somewhere to get the parameter values off the call stack. - -> tt task1 cc83c _vxTaskEntry +10 : _writeTest (16d368c, 1388, 0, 0, 0, 0, 0, 0, 0, 0 ) 1750750 _writeTest +66 : _fclose (1682868) afa02 _fclose +5a : __sflush(1682868) e01b2 __swrite +3c : _write([45, 1681f48, 17, 1682868, 16d6f24]) a1b2e _write +10 : _iosWrite ([45, 1681f48, 17, 16d6f04, e01b8]) aa778 _iosWrite +72 : 93c7c ([1e330d4, 1681f48, 17, 16d6eec, a1b34]) Sometimes it is not always possible to use the debugger, especially on a delivered/deployed system. I've seen some guys do some amazing detective work on an up and running system. Some problems take a long time to reproduce and in one case our customer would not allow the system to be rebooted for fear the problem could not be reproduced for several days. In article <8%hd6.567$i54.854034@newsr1.u-net.net>, "Dave Korn" wrote: > Bob Witmer wrote in message <93nei4$mbj$1@nnrp1.deja.com>... > >I am debugging a running system on 5.40 and want to see > >what resource my c++ code is blocking on. When I do a tt on > >the task, I wanted to see the parameters that were on the > >stack, but tt is just producing the function calls without > >the parameters. Any ideas? > > Yes. Use the GDB sourcelevel debugger that comes with T2. > > DaveK > -- > They laughed at Galileo. They laughed at Copernicus. They laughed at > Columbus. But remember, they also laughed at Bozo the Clown. > > - -- rwitmer at my-deja dot com Sent via Deja.com http://www.deja.com/ --------------------------- Newsgroups: comp.os.vxworks Subject: Long filename support on dosFs? Date: Thu, 01 Feb 2001 01:17:24 GMT From: mindtree01@my-deja.com Organization: Deja.com Message-ID: <95ades$g3o$1@nnrp1.deja.com> Can anyone tell me how to enable long file name support on dosFs in VxWorks? The manual says: "Long name support is enabled by setting DOS_OPT_LONGNAMES in the dosvc_options field of the DOS_VOL_CONFIG structure when calling dosFsDevInit( )." I can't figure out where to add this code while building the VxWorks image. My target platform is pcPentium. I am using Tornado-2 to build my VxWorks image. Please help! Sent via Deja.com http://www.deja.com/ --------------------------- Newsgroups: comp.os.vxworks Subject: [Q] Page Fault in case of PCI Device Date: Thu, 01 Feb 2001 01:40:11 GMT From: dongan@my-deja.com Organization: Deja.com Message-ID: <95aepq$h67$1@nnrp1.deja.com> Hi all. When I read content that is pointed by base address of PCI device using pciConfigInLong (PciBus, PciDevice, PciFunc, 0x10, &pciBAR0);, it occur page fault. Function flow as follows. pciFindDevice(vendorID, deviceID, 0, &PciBus, &PciDevice, &PciFunc); pciConfigOutWord (PciBus, PciDevice, PciFunc, 0x0d, 0x80); pciConfigOutWord (PciBus, PciDevice,PciFunc, 0x0c, 1); pciConfigInLong (PciBus, PciDevice, PciFunc, 0x10, &PciBAR0); pciConfigInByte (PciBus, PciDevice, PciFunc, 0x3c, &PciIrq); ret=InitDevice(param_list) Thank you for any help. Have a nice day. Best regards atm Sent via Deja.com http://www.deja.com/ --------------------------- Newsgroups: comp.os.vxworks Subject: [Q] Page Fault in case of PCI Device Date: Thu, 01 Feb 2001 01:41:19 GMT From: atm@icu.ac.kr Organization: Deja.com Message-ID: <95aeru$h6m$1@nnrp1.deja.com> Hi all. When I access content that is pointed by base address [base address is readed by pciConfigInLong(PciBus, PciDevice, PciFunc, 0x10, &PciBAR0);], page fault is occured. I use Tornado 2 and PCI configuration is included. My Initialization steps are as follows pciFindDevice (VendorID, DeviceID, 0, &PciBus, &PciDevice, &PciFunc); pciConfigOutWord (PciBus, PciDevice, PciFunc, 0x0d, 0x80); pciConfigOutWord (PciBus, PciDevice, PciFunc, 0x0c, 1); pciConfigInLong (PciBus, PciDevice, PciFunc, 0x10, &PciBAR0); pciConfigInByte (PciBus, PciDevice, PciFunc, 0x3c, &PciIrq); ret=InitDevice(param_list); printf("content of base address : %d\n", *PciBAR0); <- Page Fault Thank you for any help. have a nice day best regards atm... Sent via Deja.com http://www.deja.com/ --------------------------- Newsgroups: comp.os.vxworks Subject: VxWorks Network Facilities Date: Thu, 01 Feb 2001 02:03:48 GMT From: jay_deleon@my-deja.com Organization: Deja.com Message-ID: <95ag61$i68$1@nnrp1.deja.com> Hi, Aside from the protocols and utilities that are supported by VxWorks network stack, are there any other optional protocols/utilities that WindRiver provides? Are there also other third-party vendors that supplies additional protocols/utilities? Thanks! Jay Sent via Deja.com http://www.deja.com/ --------------------------- Newsgroups: comp.os.vxworks Subject: RE: malloc and free in vxworks Date: Wed, 31 Jan 2001 20:24:42 -0500 From: Venkatesh Varadan Organization: Lawrence Berkeley National Laboratory Message-ID: <95agt0$bec$1@overload.lbl.gov> >If malloc requests are coming from one certain application, we want to >allocate memory from certain memory addresses. If malloc request is coming >from some other application, we want to allocate memory from some other >memory addresses. The entire memory map may not fall under contiguous memory >addresses. >Any idea how we can achieve this? I would suggest you to pre-allocate some "huge" memory chunk for different applications(if they are countable in number and if you know how much memory an application can use atmost). Due to pre-allocation, a customised malloc when called just has to assign memory addresses from these "pre-allocated" chunks. I am not sure whether this will satisfy your requirement.... HTH Venkat --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Mot8260 with Galileo Ethernet Switch Date: Thu, 01 Feb 2001 02:40:05 GMT From: gold@sdsu.edu Organization: Deja.com Message-ID: <95aia4$k09$1@nnrp1.deja.com> References: <9585a1$h0s$1@nnrp1.deja.com> <3a7809c3$1@news.si.com> In article <3a7809c3$1@news.si.com>, vanbaren@falcon.si.com (vanbaren_gerald) wrote: > nickp423n@my-deja.com writes: > > >Hi > > >I am trying to get the VxWorks network stack up and running on a > >Motorola MPC8260. The 8260 is connected directly to a Galileo GT-48315 > >Ethernet switch and is not connected to a standard phy device. Does > >anyone have any experience with this? How do you configure motFcc > >ethernet driver to operate without a Phy device. I've set the > >MOT_FCC_USR_PHY_ISO user flag. Is this sufficient? > > >thanks in advance > >-Nick > > Sorry, no. I think you are going to have to work with WindRiver to figure > this one out. The PHY isolation sends a "PHY isolate" command via the > MII port to the PHY, which is not what you want. You want to ditch the > MII entirely. Your best bet is probably to write a "null" MII driver > which is sufficient to keep the WR ethernet driver happy but doesn't > really do anything. As I said, you will probably have to ask WR for at > least some basic information to accomplish this. If you ask nice enough, > they may provide you with the source to the MotFccEnd and MiiLib drivers. I think you may need some help from your hardware guys too. Is this a custom design or did you actually buy it from someone ? If you've hooked an 8260 FCC/MII directly to an MII port on the Galileo port, I don't think it can work. The MII is a MAC-layer interface, and I don't think you can hook up 2 MACs (via MII) and expect it to work. What do you do about signals that the MII expects from a real phy ? I.e. Carrier Sense and Rx Data Valid ? Carrier Sense (for example) is an input to the MII - who's going to drive it if it's hooked up as an input on the other side ? How about Tx/Rx Clocks ? If you're generating them internally on the 8260 ok, but where is the Galileo part getting them from ? Usually MACs like to get their clocks from the PHY. I've seen this kind of thing done before but with an FPGA in the middle that acts like a 'digital PHY' to both sides. Maybe your hardware guys put a little logic in between the 2 sides to make it work. Then you shouldn't need to make any changes to the software, since the MAC layer will think it's talking to a 'real' PHY. hth, Dan Gold gold@sdsu.edu Sent via Deja.com http://www.deja.com/ --------------------------- Newsgroups: comp.os.vxworks Subject: Re: vxWorks for IPX1200 Date: Thu, 01 Feb 2001 04:37:35 GMT From: A wellwisher Organization: EasyNews http://www.easynews.com - UseNet Made Easy Message-ID: References: <3A75A07A.A51264FA@marconi.com> On Mon, 29 Jan 2001 11:55:22 -0500, Yuri Berfeld wrote: >Does somebody have an expirience with vxWorks on Intel IPX1200 ? Bart and Gecko. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: [Q] default include directory..? Date: Wed, 31 Jan 2001 21:24:10 -0800 From: DrDiags Organization: Posted via Supernews, http://www.supernews.com Message-ID: <3A78F2FA.D65D2F81@flashcom.net> References: <959s6f$6da$1@usenet.kreonet.re.kr> Reply-To: drdiags@flashcom.net Skagus, I imagine you are doing a command line build with GNU Tools. If you have T2 installed, look in Tornado Online Help - GNU Make chapter. What you are looking for is a description on the vpath and VPATH directives. Here is the link to the page on my machine, so replace it with where you install resides: file:///D|/TORNADO2_PPC/docs/gnu/make/make.html#SEC27 If you are building out of the IDE, you may try defining "vpath" as described in the above link. I have not done this myself so check the FAQ for more helpful hints at: http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html skagus wrote: > I want use some directory as default include directory. > > (because I want use like this > --> #include > or--> #include "somedir/aaa.h" > > for example .... > > dir1 ---- subdir1 > `---subdir2 > > subdir1 is c/cpp directory > subdir2 is h directory > > we can use like this in c/cpp files > ---> #include "../subdir2/blabla.h" > > but I want this > ---> #include "subdir2/blabla.h" > or --> #include > > Thanks for your favor. > Good Luck To You!! --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Memory leak tools? Date: Wed, 31 Jan 2001 21:29:12 -0800 From: DrDiags Organization: Posted via Supernews, http://www.supernews.com Message-ID: <3A78F428.69125AE1@flashcom.net> References: <95a4a1$7rb$1@nnrp1.deja.com> Reply-To: drdiags@flashcom.net Jaguar, There is two products I know of: AMC's CodeTEST and RTI's ScopeTools. Besides memory leak checking, they offer code coverage analysis, and other features (in the case of the RTI tools). You could visit their websites or talk to your local WRS rep. There is also some sample code on the Tornado and VxWorks FAQ maintained for this discussion group for tracking memory usage, here is the link if you don't already know it: http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html jaguar01@my-deja.com wrote: > Does anybody knows of a VxWorks memory check utility program to > catch memory leaks? > > Sent via Deja.com > http://www.deja.com/ --------------------------- Newsgroups: comp.os.vxworks Subject: Re: How do I build a compressed VxWorks? Date: Wed, 31 Jan 2001 21:40:58 -0800 From: DrDiags Organization: Posted via Supernews, http://www.supernews.com Message-ID: <3A78F6EA.C01CAAC5@flashcom.net> References: <95a6bg$48u$1@overload.lbl.gov> Reply-To: drdiags@flashcom.net James, In the vxWorks 5.4 Users Guide, in section 8.7.1 there is a discussion on the type of buildable images and what they contain. I have added part of it here and inserted the link after for the section it came from (note the use of Tornado when they speak of images that support the host side tools): 8.7.1 Available VxWorks Images There are three types of VxWorks images. Boot ROM images Downloaded VxWorks images ROMmed VxWorks images Boot ROM images come in 3 flavors: compressed, uncompressed, and ROM-resident. bootrom normal compressed boot ROM bootrom_uncmp uncompressed boot ROM bootrom_res ROM-resident boot ROM Downloaded VxWorks images come in two basic varieties, Tornado and standalone. (Here "Tornado" is a Vxworks image that uses the host-based tools and symbol table.) vxWorks basic Tornado uses host shell and symbol table vxWorks.st standalone image has target shell and symbol table ROMmed VxWorks images: vxWorks_rom Tornado in ROM (uncompressed) vxWorks.st_rom vxWorks.st in ROM (compressed) vxWorks.res_rom vxWorks.st ROM-resident file:///D|/TORNADO2_PPC/docs/vxworks/guide/c-config7.html#86341 "CICON,JAMES (HP-Vancouver,ex1)" wrote: > How do I build a compressed VxWorks? > > Inspection of rules.bsp reveals a build for the following target: > > # vxWorks.st_rom - rommable standalone version using ROM compression > # This builds a standalone VxWorks image that can be put in ROM. > # It includes a linked in symbol table > > So I don't want to link in a symbol table and I am not sure what standalone > implies. How is standalone different from normal VxWorks?: > > # vxWorks - normal vxWorks system > # vxWorks.sym - symbol table of vxWorks > > Jim Cicon > Hewlett Packard --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Speeding Up VxWorks NFS transfers. Date: Wed, 31 Jan 2001 21:55:35 -0800 From: DrDiags Organization: Posted via Supernews, http://www.supernews.com Message-ID: <3A78FA57.B85B9E3B@flashcom.net> References: <94rm96$mvp$1@overload.lbl.gov> <3a75c1a2$1@news.si.com> <3A7847E8.926DEC23@kaon.co.uk> Reply-To: drdiags@flashcom.net Mark, Here is what I think was meant by CPM (MBX860 used as an example): The MBX provides an Ethernet interface by way of the SCC1 port within the Communications Processor Module (CPM) of the MBX8XX device. SCC (Serial Communications Controller/Channel?) Yet another acronym...... Mark Armstrong wrote: > vanbaren_gerald wrote: > > My guess is that you are being killed by dropped UDP packets and the > > UDP timeout/retries are 1+ seconds. I saw this when using TFTP (also > > runs over UDP) when we had an unreliable ethernet link (PPC8260 with the > > CPM clocked too slow). It would burst a couple of packets, drop one, > > and take a second or more to timeout and start transfering data again. > > > > When we fixed the clocking on the 8260 CPM, everything ran MUCH better. > > I agree with you as the delays were much reduced by changing the global > variables: > > nfsReXmitSec = 5; > nfsReXmitUSec = 0; > > to > > nfsReXmitSec = 0; > nfsReXmitUSec = 50000; > > Excuse my ignorance, but when you say: > "the unreliable ethernet link (PPC8260 with the > CPM clocked too slow)" > What do you mean by CPM? (A clock module of some kind?) > Could this be an issue with our board or switch? > > Regards, > > Mark Armstrong --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Spawing C++ task Date: Thu, 01 Feb 2001 05:43:47 GMT From: john_94501@my-deja.com Organization: Deja.com Message-ID: <95at2i$sfa$1@nnrp1.deja.com> References: Hello, The shell uses C conventions (i.e. it doesn't change the symbol, except to add an '_' for some architectures). C++ function names are mangled however, so your symbol won't really be just cortexTlm. The solutions are: 1) Find the mangled name and use that (try lkup "cortexTlm" - it should give you a list of all symbols contain that string; one will be your entry point). 2) Force the routine to be C style using the following around the whole function in the source file: extern "C" { Your function goes in here. } HTH, John... In article , "Martin Boulay" wrote: > I have some code written in C++ and tried to load and spawn a routine that > uses C++. moduleShow tells me the munched .out file was loaded and lkup > told me that the routine is also there but when I do sp cortexTlm I get > "cortexTlm undefined symbol". What I doing wrong. > I use c++ppc to compile the code and the vxwork kernal has c++ lib included. > This same program executes when we use windshell to load and execute the > task. > > Sent via Deja.com http://www.deja.com/ --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Send signals to task from shell Date: Thu, 01 Feb 2001 05:48:29 GMT From: john_94501@my-deja.com Organization: Deja.com Message-ID: <95atbb$sja$1@nnrp1.deja.com> References: <3a782e48.27615018@news.belgium.eu.net> <3A7846BF.8EDA4083@nothere.com> Hello, In article <3A7846BF.8EDA4083@nothere.com>, "M. Kotiaho" wrote: > "Jean-François Daune" wrote: > > > Hi all, > > > > is there a way to send signals to a task (SIGINT, SIGTERM,...) from > > the Tornado II shell ?? > > > > J-F Daune > > kill (taskId, signalNumber) Also note that unless your task has explicitly registered a handler for these signals, sending them will not have much effect. Unlike general purpose operating systems, VxWorks does not automatically catch these types of signals and deal with them. Just in case you were expecting more... John... Sent via Deja.com http://www.deja.com/ --------------------------- Newsgroups: comp.os.vxworks Subject: Re: OS structure Date: Thu, 01 Feb 2001 05:52:23 GMT From: john_94501@my-deja.com Organization: Deja.com Message-ID: <95atil$std$1@nnrp1.deja.com> References: <3A76CF95.59BE99FE@mail.com> Hello, In general it is a good idea to look at the company's home page and search for product information from there: http://www.windriver.com To save you some time, here's a link to the programmer's guide for VxWorks 5.4; this book (and it is a fairly big book so you might want to think twice about printing it!) contains a lot of information about the OS architecture. http://www.windriver.com/pdf/vxworks_guide.pdf HTH, John... In article <3A76CF95.59BE99FE@mail.com>, RK wrote: > I would like to know type of VxWorks OS structure. Is it client-server > or layered operating sistem? > > Where can I get Additional information. > > Thanks > Robert Kodric > > Sent via Deja.com http://www.deja.com/ --------------------------- Newsgroups: comp.os.vxworks Subject: TaskDelay( ) bug? Date: Thu, 1 Feb 2001 09:54:34 +0200 From: "Randall" Message-ID: <3a791610$0$232@hades.is.co.za> Hi, I am writing a small graphics application using Tornado. In order for me to display the graphics, I have to initialise and de-initialise the UGL. I have no problem doing this and the graphics works fine. A second part of the application requires me to implement a infinite loop in some task. I use the TaskDelay() function to delay the task by a few milliseconds. This also seems to work fine when I dont use the graphics. The problem is that as soon as I initialise the UGL to display the graphics and load my device drivers, my code seems to 'hang' in the TaskDelay() function. The task basically stops at the TaskDelay and never seems to run again. The system indicates that the task is still alive and pending but it just wont execute further. I have given the task a higher priority and this doesnt work. The task currently has a priority of 10, I have even gone so low as 3. Does anyone have any ideas or ever experienced this problem? This only happens when I initiliase the UGL. Thanks in advance. Randall Michau rmichau@grintek.com --------------------------- Newsgroups: comp.os.vxworks Subject: managment agent Date: Thu, 1 Feb 2001 10:20:43 +0200 From: "yuval shoham" Organization: Bezeq International Ltd. Message-ID: <3a791c3a@news.bezeqint.net> Hi every one Is any one know about agent compiler for VxWorks ? Thank you yuval --------------------------- Newsgroups: comp.os.vxworks Subject: Re: TaskDelay( ) bug? Date: Thu, 01 Feb 2001 08:35:58 GMT From: "Tri VU KHAC" Organization: KPNQwest customer news service Message-ID: References: <3a791610$0$232@hades.is.co.za> Using WindCross you could see where it stops (may be in Assembler). Copy it and post it in this newsgroup. Cheers. Tri. > > I am writing a small graphics application using Tornado. In order for me to > display > the graphics, I have to initialise and de-initialise the UGL. I have no > problem doing > this and the graphics works fine. > > A second part of the application requires me to implement a infinite loop in > some task. > I use the TaskDelay() function to delay the task by a few milliseconds. This > also seems > to work fine when I dont use the graphics. The problem is that as soon as I > initialise the UGL > to display the graphics and load my device drivers, my code seems to 'hang' > in the TaskDelay() > function. The task basically stops at the TaskDelay and never seems to run > again. The system indicates > that the task is still alive and pending but it just wont execute further. > > I have given the task a higher priority and this doesnt work. The task > currently has a priority of 10, > I have even gone so low as 3. Does anyone have any ideas or ever experienced > this problem? > This only happens when I initiliase the UGL. > > Thanks in advance. > Randall Michau > rmichau@grintek.com > > --------------------------- Newsgroups: comp.os.vxworks Subject: AlphaVME and VxWorks Date: Thu, 1 Feb 2001 10:09:14 +0100 From: "Fred Zwarts" Organization: Kernfysisch Versneller Instituut Message-ID: <95b93p$kk0$1@info.service.rug.nl> Recently we were told by Force Computers that the AlphaVME line has reached "end of life". Further there will be no developments anymore for=20 VxWorks on the Alpha platform, it will stay at 5.3.1. Since our major activity, nuclear physics data-acquisition, is based on VxWorks running on AXPvme and AlphaVME processors, we are now looking for a new platform. Does anyone have a good suggestion? We would prefer a 64-bit, little endian processor again for an easy port of the software, but a completely different hardware and software platform may be considered. F.Z. --------------------------- Newsgroups: comp.os.vxworks Subject: resource name Date: Tue, 30 Jan 2001 15:29:04 +0200 From: "ophir fuchs" Organization: NetVision Israel Message-ID: <956flm$2pd$1@news.netvision.net.il> I am new in VxWorks , in Psos each component (Queueu , Semaphore) has name so when another task wants to reference a component he could get the component Id by calling to function that reterns the Id upon the component name this doesn't exist in Vxworks. My question is does I need to have global variable for each component in order to refer the component from enother module ? or is there another way to do this . --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Tornado/Project location independent Makefile Date: Thu, 01 Feb 2001 10:00:33 GMT From: Dragan Ostojic Organization: Sympatico Message-ID: <3A79349D.9020203@sympatico.ca> References: <3A7719D2.9020904@libertysurf.fr> What you have to do is to go through your wpj file and replace absolute paths with something like -I$(PRJ_DIR)/../bsp etc. for the stuff that relates to your project directory and to something like - -I$(WIND_BASE)/target etc. to the stuff that relates to your Tornado installation. Than re-generate the Makefile (for example by re-building dependencies). Open the Makefile and change PRJ_DIR to something like ../vx_prj where vx_prj is where wpj and Makefile are located. Unfortunately, you have to do that over and again whanever you re-generate Makefile. wpj you don't have to change unless you add a new path. Take care that you have to add absolute paths to wpj through project IDE because otherwise Tornado is unable to find your directories (it searches using absolute paths). If you change PRJ_DIR in the Makefile you won't be able to build from IDE. If you don't you'll be able to build both form IDE and from the command-line. We usually change only before we commit into CVS. Sebastien Fauris wrote: > Hi there, > > the Tornado 2 IDE generates the Makefile used > to compile my application. I would like the project > to be Tornado/Project location independent for the > compilation so anybody can recompile it and this from > the IDE or the DOS command line. > Thanks to some previous postings in the newsgroup, > I am getting close to have the things working. > However, I can still see the following lines location > dependent in the generated Makefile: > > CFLAGS = -g -m486 -ansi -nostdinc -DRW_MULTI_THREAD > -D_REENTRANT -fvolatile -nostdlib -fno-builtin -fno-defer-pop -Wall -I. > -ID:/Tornado/target/h -DCPU=I80486 > CFLAGS_AS = -g -m486 -ansi -nostdinc -fvolatile -nostdlib > -fno-builtin -fno-defer-pop -P -x assembler-with-cpp -I. > -ID:/Tornado/target/h -DCPU=I80486 > > I found out that the "-ID:/Tornado/target/h" is present in the > builds tab -> C/C++ compiler tab. > > How can I replace it by something like "-I$(WIND_BASE)/target/h" ? > This does not work when I make the change in the tab. > > How can do also if I want to add something like "-I../myHeaders/" ? > Tornado automatically generates the Makefile with a > "-Ic:/projects/projectA/myHeaders". > > Thanks a lot, > > Sebastien. - -- Dragan Ostojic mail: ostojic@sympatico.ca voice: +1 416 236 7086 --------------------------- Newsgroups: comp.os.vxworks Subject: Web Hosting only cost HK$50 Date: Sat, 27 Jan 2001 19:03:40 +0800 From: "Cisgate Hosting" Organization: Another Netscape Collabra Server User Message-ID: <94u9ap$jaj69@rain.i-cable.com> Web Hosting only cost HK$50 50Mb storage l5 email account Free Account Setup Unlimited Web/FTP Traffics Unlimited FTP access Unlimited E-Mail Aliases Supported standard (html, SSI, PHP3, Perl5) Your own cgi-bin Free Email Tech Support www.cisgate.com --------------------------- End of New-News digest ********************** From vxwexplo-errs@csg.lbl.gov Thu Feb 1 08:12:22 2001 From: Gene Date: Thu Feb 1 08:12:25 PST 2001 Subject: VxWorks and multicast This is a multi-part message in MIME format. --------------B3379F947E7A553B3DBBF511 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit vxworks/vxWorkers, Does anyone have any experience using multicast in a vxWorks environment? I'm using the Dec 21040 driver which seems to support multicast receives but the "if" routines indicate that multicasting is not supported. Thanks in advance, Gene Small Software Technology, Inc --------------B3379F947E7A553B3DBBF511 Content-Type: text/x-vcard; charset=us-ascii; name="gene.vcf" Content-Transfer-Encoding: 7bit Content-Description: Card for Gene Content-Disposition: attachment; filename="gene.vcf" begin:vcard n:Small;Gene x-mozilla-html:FALSE org:Software Technology, Inc version:2.1 email;internet:gene@xgnt.com title:Software Engineer x-mozilla-cpt:;0 fn:Gene Small end:vcard --------------B3379F947E7A553B3DBBF511-- From vxwexplo-errs@csg.lbl.gov Thu Feb 1 09:39:51 2001 From: "Mike Tarlton" Date: Thu Feb 1 09:39:53 PST 2001 Subject: Re: How do I build a compressed VxWorks? James, You asked: > How do I build a compressed VxWorks? You didn't mention in your email which version of Tornado you're using. Under Tornado 1.0.1 the only (supplied) image you can build giving you a compressed ROMable vxWorks is vxWorks.st_rom. Unfortunately, this brings in all the 'standalone' stuff, which you don't want. In Tornado 2.0 the Project Tool offers you the image vxWorks_romCompress (See the build spec properties sheet, in the rules), and 'standalone' has disappeared into the component configuration choices you can make. > Inspection of rules.bsp reveals a build for the following target: > > # vxWorks.st_rom - rommable standalone version using ROM compression > # This builds a standalone VxWorks image that can be put in ROM. > # It includes a linked in symbol table > > So I don't want to link in a symbol table and I am not sure what standalone > implies. How is standalone different from normal VxWorks?: > > # vxWorks - normal vxWorks system > # vxWorks.sym - symbol table of vxWorks The 'standalone' images are only of interest if you are using target-based facilities (such as the target shell, target loader, etc) which rely on the presence of a target-based symbol table (as opposed to the host-based symbol table used by the target server). The choices for initializing this symbol table are 1. The symbol information is extracted into a separate file and is downloaded separately from the boot device. This is the file vxWorks.sym, which gets built at the same time as vxWorks. OR 2. The symbol information is extracted from vxWorks, turned into C code, compiled and then linked in with VxWorks so that it becomes part of the image. This is (or rather was) the file vxWorks.st. So a 'standalone' image is one which has a symbol table built in, together with all the necessary symbols to fill the table. 'Standalone' has got nothing much to do with 'ROMable'. A standalone image could be in ROM or it could be downloaded by the boot code. If you are using Tornado 1.0.1 and you want a compressed ROMable image without the built-in symbol table, then you have to write a new makefile rule yourself (based on the ones in rules.bsp), or get it from Wind River. There's probably a summary on WindSurf that will help. Bear in mind that if you want to use this image with the host tools then you will have to retain a copy of the executable that gets compressed. Otherwise the symbol information needed by the target server will be lost. I hope this is reasonably clear. Tell me if it isn't. Incidentally, when I worked for Wind River, I raised an SPR entitled, as I recall, "Inconsistent use of the term 'standalone' in documentation and elsewhere" because many customers seemed to experience unnecessary confusion. However, this SPR never made it out into public view. Regards, Mike mike_tarlton@3com.com From vxwexplo-errs@csg.lbl.gov Thu Feb 1 15:43:37 2001 From: "lee" Date: Thu Feb 1 15:43:40 PST 2001 Subject: Re: How do I build a compressed VxWorks? Hi, I am writing a driver under vxWorks. I am running into some problems with the driver entry point. My ioctl call that are installed under vxWorks using iosDrvInstall need to have a variable input as the third argument to support extended functionalities. However, the ioctl defined by vxWorks iosLib is strictly int ioctl(int, int, int) Is there a way i can compile a code using the ioctl with the variable input without the compiler barfing complaining about too few or too many input arguments to ioctl?? thanks, lee vxWorks From vxwexplo-errs@csg.lbl.gov Fri Feb 2 02:27:00 2001 From: Friedrich.Scheurer@t-online.de (Friedrich Scheurer / Goeppingen) Date: Fri Feb 2 02:27:02 PST 2001 Subject: Network problem on MVME2700 T1.0.1 Dear VxWorkers, dear David,

David Laight gave some insight into ethernet negotiation aspects and gave
a hint to problems in network traffic "... network to run 'like a sick pig' ...".

I suppose we're suffering from a problem like this for 1,5 years.

We often loose target-server connection to the target MVME2700
during a debug session. 5 times higher timeout didn't solve the problem really.
Often FTP-transfers from target to host are very slowly (about 1 % of maximum performance).
This almost only occurs during daytime if there's lot of traffic in our
network. Out of standard office times it's much better.
The slowing down is in most cases 'healed' if we're rebooting the target.

WRS couldn't help us only from the symptoms, because it's not reproducable.

The target is attached via Hub/Repeater not a switch.
(David said: Generally workin in half duplex)

How can I fix wether my target tries to run in full duplex
and makes the problems he described in his posting?

TIA, regards

F. Scheurer
  

Submitted-by vxwexplo-errs@csg.lbl.gov  Thu Sep 21 02:38:58 2000
Submitted-by: David Laight 

>> When booting VxWorks I get the message "interrupt: dc0 - no carrier"
>> very frequently.

> Wind River is aware of this problem. It is caused if you are connected
> to a ethernet switch that supports Full Duplex. Ethernet Hubs are
> typically Half Duplex.

> Wind River send me a new dec*.o file which essentially removed the no
> carrier print.

> There is an SPR on Windsurf already about this problem.

Does anyone know if WRS have fixed this properly?

The SPR doesn't say much, but indicates that they think error is "normally 
harmless' (mostly harmless?).

The 'no carrier' error is reported when the ethernet chipset doesn't get a 
receive signal (it is the rx carrier that is missing) during its own transmits.  
This is a test of the rx and tx pairs of the AUI cable to the transceiver (the 
SQE test checks the collision pair).  A coax transceiver will always receive its 
own transmittions...

On twisted pair the transceiver (AUI or MII) reports a collision if there is 
receive data (on the TP cable) during transmit, and returns the tx data as rx to 
stop the chipset reporting 'no carrier'.  The chipset will ignore the rx data 
during transmit.

For full duplex the tx and rx sides are completely separate.  However 4 pieces 
of hardware have to agree to use full duplex.  The two transceivers can (and 
will) auto-negotiate full duplex if permitted.  The etherner controller has to 
interrogate the transceiver to determine what has been negotiated.  This is hard 
and is left to the device driver.

Unfortunately there is no easy way (for the device driver) to determine what the 
transceivers have negotiated.  Especially since it can change at any time.

The easy 'solution' is to change the transceivers 'link advertise register' and 
disable the full duplex modes.  Then you know full duplex will never be 
negotiated.  (Don't EVER fix a port to full duplex - it will only end in tears.)

The effect of having the transceiver running full duplex but the ethernet 
chipset half duplex is that receive packets will be discarded - typically during 
heavy traffic.  This can be hard to detect - but causes the network to run 'like 
a sick pig' at times.

Note that you can change from 10M to 100M without the driver/chipset having to 
do anything at all.  It just works.  Determining the current speed is, of 
course, almost impossible from the standard MII registers.

        David


----------------------------------------------------------------
David Laight                          email: dsl@tadpole.co.uk
Tadpole Technology plc                phone: +44 1223 428 232
Cambridge, UK                         fax:   +44 1223 428 201

--
+-------------------------------------------------+
| Dipl.-Ing. F. Scheurer                          |
| c/o Zentrum Fertigungstechnik Stuttgart         |
| Nobelstrasse 15                                 |
| D-70569 Stuttgart                               |
| Telefon +49 711 13162-41(-0)                    |
| Fax     +49 711 13162-11                        |
| E-Mail  scheurer@zfs.tz.uni-stuttgart.de        |
| WWW     http://www.zfs.tz.uni-stuttgart.de      |
+-------------------------------------------------+
| privat:                  Waldeckstrasse 31      |
|                          D-73035 Goeppingen     |
| Telefon +49 7161 686362  Fax +49 7161 686301    |
| Mobil  +49 170 31 77 539                        |
| E-Mail  Friedrich.Scheurer@t-online.de          |
| oder    Friedrich_Scheurer@comundo.de           |
| oder    Friedrich.Scheurer@lycosmail.com        |
| WWW     http://members.tripod.de/machinist/     |
| WWW     http://www.friedrichscheurer.com/       |
+-------------------------------------------------+
  From vxwexplo-errs@csg.lbl.gov Fri Feb 2 02:31:52 2001 From: Friedrich.Scheurer@t-online.de (Friedrich Scheurer / Goeppingen) Date: Fri Feb 2 02:31:54 PST 2001 Subject: [Fwd: Network problem on MVME2700 T1.0.1] Sorry for the HTML-version! -------- Original Message -------- Dear VxWorkers, dear David, David Laight gave some insight into ethernet negotiation aspects and gave a hint to problems in network traffic "... network to run 'like a sick pig' ...". I suppose we're suffering from a problem like this for 1,5 years. We often loose target-server connection to the target MVME2700 during a debug session. 5 times higher timeout didn't solve the problem really. Often FTP-transfers from target to host are very slowly (about 1 % of maximum performance). This almost only occurs during daytime if there's lot of traffic in our network. Out of standard office times it's much better. The slowing down is in most cases 'healed' if we're rebooting the target. WRS couldn't help us only from the symptoms, because it's not reproducable. The target is attached via Hub/Repeater not a switch. (David said: Generally workin in half duplex) How can I fix wether my target tries to run in full duplex and makes the problems he described in his posting? TIA, regards F. Scheurer > Submitted-by vxwexplo-errs@csg.lbl.gov Thu Sep 21 02:38:58 2000 > Submitted-by: David Laight > > >> When booting VxWorks I get the message "interrupt: dc0 - no carrier" > >> very frequently. > > > Wind River is aware of this problem. It is caused if you are connected > > to a ethernet switch that supports Full Duplex. Ethernet Hubs are > > typically Half Duplex. > > > Wind River send me a new dec*.o file which essentially removed the no > > carrier print. > > > There is an SPR on Windsurf already about this problem. > > Does anyone know if WRS have fixed this properly? > > The SPR doesn't say much, but indicates that they think error is "normally > harmless' (mostly harmless?). > > The 'no carrier' error is reported when the ethernet chipset doesn't get a > receive signal (it is the rx carrier that is missing) during its own transmits. > This is a test of the rx and tx pairs of the AUI cable to the transceiver (the > SQE test checks the collision pair). A coax transceiver will always receive its > own transmittions... > > On twisted pair the transceiver (AUI or MII) reports a collision if there is > receive data (on the TP cable) during transmit, and returns the tx data as rx to > stop the chipset reporting 'no carrier'. The chipset will ignore the rx data > during transmit. > > For full duplex the tx and rx sides are completely separate. However 4 pieces > of hardware have to agree to use full duplex. The two transceivers can (and > will) auto-negotiate full duplex if permitted. The etherner controller has to > interrogate the transceiver to determine what has been negotiated. This is hard > and is left to the device driver. > > Unfortunately there is no easy way (for the device driver) to determine what the > transceivers have negotiated. Especially since it can change at any time. > > The easy 'solution' is to change the transceivers 'link advertise register' and > disable the full duplex modes. Then you know full duplex will never be > negotiated. (Don't EVER fix a port to full duplex - it will only end in tears.) > > The effect of having the transceiver running full duplex but the ethernet > chipset half duplex is that receive packets will be discarded - typically during > heavy traffic. This can be hard to detect - but causes the network to run 'like > a sick pig' at times. > > Note that you can change from 10M to 100M without the driver/chipset having to > do anything at all. It just works. Determining the current speed is, of > course, almost impossible from the standard MII registers. > > David > > > ---------------------------------------------------------------- > David Laight email: dsl@tadpole.co.uk > Tadpole Technology plc phone: +44 1223 428 232 > Cambridge, UK fax: +44 1223 428 201 > > -- +-------------------------------------------------+ | Dipl.-Ing. F. Scheurer | | c/o Zentrum Fertigungstechnik Stuttgart | | Nobelstrasse 15 | | D-70569 Stuttgart | | Telefon +49 711 13162-41(-0) | | Fax +49 711 13162-11 | | E-Mail scheurer@zfs.tz.uni-stuttgart.de | | WWW http://www.zfs.tz.uni-stuttgart.de | +-------------------------------------------------+ | privat: Waldeckstrasse 31 | | D-73035 Goeppingen | | Telefon +49 7161 686362 Fax +49 7161 686301 | | Mobil +49 170 31 77 539 | | E-Mail Friedrich.Scheurer@t-online.de | | oder Friedrich_Scheurer@comundo.de | | oder Friedrich.Scheurer@lycosmail.com | | WWW http://members.tripod.de/machinist/ | | WWW http://www.friedrichscheurer.com/ | +-------------------------------------------------+ From vxwexplo-errs@csg.lbl.gov Fri Feb 2 04:03:09 2001 From: Vxworks Exploder Date: Fri Feb 2 04:03:13 PST 2001 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Fri Feb 2 04:03:05 PST 2001 Subject: Serial port interrupts Subject: Releasing bound socket Subject: Q: Starting Targetserver automaticly Subject: VxWorks and multicast Subject: Q. Is there any existing library to compress a RAM file in VxWorks? Subject: Re: malloc options Subject: Target/Tornado shell welcome message Subject: Re: Serial port interrupts Subject: Re: c/c++ constructor/destructor problems Subject: Re: tt with c++ that shows passed parameters? Subject: Re: VxWorks and multicast Subject: re: How do I build a compressed VxWorks? Subject: Re: TaskDelay( ) bug? Subject: dosFS 2.0 ??? Subject: Serial port in vxSim Subject: Re: Output file-format Subject: Cisco Opportunity in Embedded Systems Subject: Re: tt with c++ that shows passed parameters? Subject: Re: Target/Tornado shell welcome message Subject: Re: [Q] Page Fault in case of PCI Device Subject: Re: Q. Is there any existing library to compress a RAM file in VxWorks? Subject: Re: Telnet Subject: pmc750 works? Subject: Re: dosFS 2.0 ??? Subject: Re: Need a C++ Compiler Subject: re: How do I build a compressed VxWorks? Subject: How to "clean up" special test procedures currently run through the Subject: tools.h++ Subject: Re: Telnet Subject: Re: Need a C++ Compiler Subject: Re: Cisco Opportunity in Embedded Systems Subject: Where can I get the old mails? Subject: any books on VxWorks? Subject: Re: Telnet Subject: Re: Where can I get the old mails? Subject: Additional supported PCMCIA networkcards Subject: Re: VxWorks and multicast Subject: generated dependencies Subject: Re: Telnet Subject: Re: dosFS 2.0 ??? Subject: Re: any books on VxWorks? Subject: Re: Where can I get the old mails? Subject: RPC UDP Sockets Problem Subject: How Do I make an interrup handling routine for the LCD on an assabet? Subject: Network problem on MVME2700 T1.0.1 Subject: [Fwd: Network problem on MVME2700 T1.0.1] ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Serial port interrupts Date: Thu, 1 Feb 2001 14:27:52 +0200 From: "Randall" Message-ID: <3a795611$0$229@helios.is.co.za> Hi, I am writing some code that reads data from the serial port and then stores it a ringbuffer. At present I am polling the serial port using a timer. On the timeout, I read all data in the port Rx buffer and write this to my ringbuffer. This seems a clumsy way of doing it since the polling of the port happens even when there is no data received on the port. I therefore want to read data received on the port using some sort of interrupt. This means when any data is received, a interrupt should be generated. How would I implement this? Thanks in advance. Randall Rmichau@grintek.com --------------------------- Newsgroups: comp.os.vxworks Subject: Releasing bound socket Date: Thu, 01 Feb 2001 12:34:50 GMT From: roh_youngjin@hotmail.com Organization: Deja.com Message-ID: <95bl5b$et6$1@nnrp1.deja.com> Hello all, In VxWorks, I bound a UDP socket for waiting packets. In the development, I kill(td) the task for uploading new app. After then the bound socket is not removed in the kernel, so I have to reboot the system to bind the same port again. "inetstatShow" tells the port is bound. Is there any way to release bound socket? I tried to use signal to close the task safely, but the signal doesn't works well like signals in Unix machine. Any help or reference for above problems? Thanks you in advance... Youngjin Sent via Deja.com http://www.deja.com/ --------------------------- Newsgroups: comp.os.vxworks Subject: Q: Starting Targetserver automaticly Date: Thu, 01 Feb 2001 15:01:55 GMT From: S.Flache@signaal.de (Stefan Flache) Organization: News from ezri Message-ID: <3a7979d4.26433048@10.1.1.1> Reply-To: S.Flache@signaal.de Hello, is there a possibility to start all targetservers on a Windows NT machine at the end of the user-login, similar to start the registry while using the autostart-folder ? Thanks for help, Stefan --------------------------- Newsgroups: comp.os.vxworks Subject: VxWorks and multicast Date: Thu, 01 Feb 2001 11:11:34 -0500 From: Gene Organization: Lawrence Berkeley National Laboratory Message-ID: <95c240$o4m$1@overload.lbl.gov> This is a multi-part message in MIME format. - --------------B3379F947E7A553B3DBBF511 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit vxworks/vxWorkers, Does anyone have any experience using multicast in a vxWorks environment? I'm using the Dec 21040 driver which seems to support multicast receives but the "if" routines indicate that multicasting is not supported. Thanks in advance, Gene Small Software Technology, Inc - --------------B3379F947E7A553B3DBBF511 Content-Type: text/x-vcard; charset=us-ascii; name="gene.vcf" Content-Transfer-Encoding: 7bit Content-Description: Card for Gene Content-Disposition: attachment; filename="gene.vcf" begin:vcard n:Small;Gene x-mozilla-html:FALSE org:Software Technology, Inc version:2.1 email;internet:gene@xgnt.com title:Software Engineer x-mozilla-cpt:;0 fn:Gene Small end:vcard - --------------B3379F947E7A553B3DBBF511-- --------------------------- Newsgroups: comp.os.vxworks Subject: Q. Is there any existing library to compress a RAM file in VxWorks? Date: Thu, 1 Feb 2001 11:22:25 -0500 From: "Yin, Yong" Organization: Nortel Message-ID: <95c2dd$a91$1@bmerhc5e.ca.nortel.com> Is there any existing library to compress a RAM file in VxWorks? We want to store something in a RAM file, compress this file and transfer this file to a host machine. I know I can use some gnu codes to do this. But I want to know if there is some existing library running on VxWorks to do this job. Many thanks Yong --------------------------- Newsgroups: comp.os.vxworks Subject: Re: malloc options Date: Thu, 01 Feb 2001 16:28:33 -0000 From: bob schulman Organization: Posted via Supernews, http://www.supernews.com Message-ID: References: <3A785B42.38ABB4E@mayannetworks.com> Kevin Cline wrote: > This must be a FAQ, but I'm looking > for a drop-in replacement for the pessimal > malloc supplied by WRS. Any suggestions? Has anyone > tried GNU malloc or a variation thereof? > Any other options? Seaweed Systems has a written-from-scratch memLib replacement. It has all the functionality which Wind River's memLib has...and more! It has better fragmentation behavior, memory leak detection, sanity checks at malloc/realloc/free time, user-callable memory arena sanity checker, constant-time allocation, callbacks for various low-memory conditions, and more! A real ginzu knife of an app. Please call or email for information: bob@seaweed.com +1 425.895-1721 --------------------------- Newsgroups: comp.os.vxworks Subject: Target/Tornado shell welcome message Date: Thu, 01 Feb 2001 13:52:51 -0300 From: Javier Argomedo Organization: European Southern Observatory Message-ID: <3A799463.5D56D524@eso.org> Dear all, Just a simple question, I want a short text message to be shown when one enters to the vx shell via rlogin/telnet or using tornado's launcher shell, I think it should be possible, is that so ? thanks in advance, Javier. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Serial port interrupts Date: Thu, 01 Feb 2001 16:46:07 GMT From: Bob Houston Organization: Deja.com Message-ID: <95c3sc$sor$1@nnrp1.deja.com> References: <3a795611$0$229@helios.is.co.za> In article <3a795611$0$229@helios.is.co.za>, "Randall" wrote: > Hi, > > I am writing some code that reads data from the serial port and then > stores it a ringbuffer. At present I am polling the serial port using a > timer. > On the timeout, I read all data in the port Rx buffer and write this to my > ringbuffer. > > This seems a clumsy way of doing it since the polling of the port happens Randall, If your serial port is supported by your bsp, you should be able to access it vi /dev/tyCo/N (open, ioctl, read, etc). Do a 'devs' command. Refer to tyLib for the ioctl for checking for new data (FNIOREAD??) Also the size of /dev/tyCo's buffer can be altered if needed. - Bob Sent via Deja.com http://www.deja.com/ --------------------------- Newsgroups: comp.os.vxworks Subject: Re: c/c++ constructor/destructor problems Date: Thu, 1 Feb 2001 17:43:53 -0000 From: "Dave Korn" Organization: Lumber Cartel (tinlc) Members #2234-2237 (owing to browser refresh) Message-ID: <89he6.722$i54.941726@newsr1.u-net.net> References: <3a782073.6222625@192.168.1.214> <1uYd6.683$i54.912594@newsr1.u-net.net> <3a7853f1.19404250@192.168.1.214> Kim Putnam wrote in message <3a7853f1.19404250@192.168.1.214>... >Thanks for the response!! > >I did munch my files, but used the cc compiler and not the >g++. > >Where do I find info on g++? in my gnu doc? Dunno exactly where it is in the docs, but just try running it instead of cc. It's almost exactly the same as cc but passes different options to the compiler sub-programs because it knows it's being used for c++. DaveK - -- They laughed at Galileo. They laughed at Copernicus. They laughed at Columbus. But remember, they also laughed at Bozo the Clown. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: tt with c++ that shows passed parameters? Date: Thu, 1 Feb 2001 17:47:31 -0000 From: "Dave Korn" Organization: Lumber Cartel (tinlc) Members #2234-2237 (owing to browser refresh) Message-ID: References: <93nei4$mbj$1@nnrp1.deja.com> <8%hd6.567$i54.854034@newsr1.u-net.net> <95a5ul$9bk$1@nnrp1.deja.com> Bob Witmer wrote in message <95a5ul$9bk$1@nnrp1.deja.com>... >I saw some old posts in the vxWorks archive which showed the argument >lists with parameters. I think >the following was from a 5.31 system. Our system is 5.40 plus we were >using C++, so I wasn't sure >if there was some variable I needed to set somewhere to get the >parameter values off the call stack. > >-> tt task1 >cc83c _vxTaskEntry +10 : _writeTest (16d368c, 1388, 0, 0, 0, 0, >0, 0, 0, 0 ) >1750750 _writeTest +66 : _fclose (1682868) > afa02 _fclose +5a : __sflush(1682868) > e01b2 __swrite +3c : _write([45, 1681f48, 17, 1682868, >16d6f24]) > a1b2e _write +10 : _iosWrite ([45, 1681f48, 17, >16d6f04, e01b8]) > aa778 _iosWrite +72 : 93c7c ([1e330d4, 1681f48, 17, >16d6eec, a1b34]) I could be remembering wrong but I think that the difference is whether you're using host shell or target shell. I think it's the host shell that gives that kind of detail, so you'd need a target server running on the host and a target (Wdb) agent running on the target. >Sometimes it is not always possible to use the debugger, especially on a >delivered/deployed system. I've seen some guys do some amazing >detective work on an up and running system Yeah, at times like that the Vx command shell with it's built in symbol table is really worth it's weight in gold DaveK - -- They laughed at Galileo. They laughed at Copernicus. They laughed at Columbus. But remember, they also laughed at Bozo the Clown. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: VxWorks and multicast Date: Thu, 01 Feb 2001 18:01:42 GMT From: John Fusco Organization: Deja.com Message-ID: <95c8a0$15f$1@nnrp1.deja.com> References: <95c240$o4m$1@overload.lbl.gov> I have the opposite problem. I can't turn MULTICAST off on the "dc" driver. Using T2.0 on an MVME-2304, ifShow says: dc (unit number 0): Flags: (0x8063) UP BROADCAST RUNNING ARP MULTICAST From what WRS says, this is feature hardwired on or off in the object code. You have no control over it since they don't ship the source code. We had to get a special version of the dec driver object module from WRS to turn off multicast. I wonder if they changed the default to off? Maybe you need to contact WRS. In article <95c240$o4m$1@overload.lbl.gov>, Gene wrote: > This is a multi-part message in MIME format. > --------------B3379F947E7A553B3DBBF511 > Content-Type: text/plain; charset=us-ascii > Content-Transfer-Encoding: 7bit > > vxworks/vxWorkers, > > Does anyone have any experience using multicast in a vxWorks environment? > > I'm using the Dec 21040 driver which seems to support multicast receives but > the "if" routines indicate that multicasting is not supported. > Sent via Deja.com http://www.deja.com/ --------------------------- Newsgroups: comp.os.vxworks Subject: re: How do I build a compressed VxWorks? Date: Thu, 1 Feb 2001 17:35:03 +0000 From: "Mike Tarlton" Organization: Lawrence Berkeley National Laboratory Message-ID: <95c950$24k$1@overload.lbl.gov> James, You asked: > How do I build a compressed VxWorks? You didn't mention in your email which version of Tornado you're using. Under Tornado 1.0.1 the only (supplied) image you can build giving you a compressed ROMable vxWorks is vxWorks.st_rom. Unfortunately, this brings in all the 'standalone' stuff, which you don't want. In Tornado 2.0 the Project Tool offers you the image vxWorks_romCompress (See the build spec properties sheet, in the rules), and 'standalone' has disappeared into the component configuration choices you can make. > Inspection of rules.bsp reveals a build for the following target: > > # vxWorks.st_rom - rommable standalone version using ROM compression > # This builds a standalone VxWorks image that can be put in ROM. > # It includes a linked in symbol table > > So I don't want to link in a symbol table and I am not sure what standalone > implies. How is standalone different from normal VxWorks?: > > # vxWorks - normal vxWorks system > # vxWorks.sym - symbol table of vxWorks The 'standalone' images are only of interest if you are using target-based facilities (such as the target shell, target loader, etc) which rely on the presence of a target-based symbol table (as opposed to the host-based symbol table used by the target server). The choices for initializing this symbol table are 1. The symbol information is extracted into a separate file and is downloaded separately from the boot device. This is the file vxWorks.sym, which gets built at the same time as vxWorks. OR 2. The symbol information is extracted from vxWorks, turned into C code, compiled and then linked in with VxWorks so that it becomes part of the image. This is (or rather was) the file vxWorks.st. So a 'standalone' image is one which has a symbol table built in, together with all the necessary symbols to fill the table. 'Standalone' has got nothing much to do with 'ROMable'. A standalone image could be in ROM or it could be downloaded by the boot code. If you are using Tornado 1.0.1 and you want a compressed ROMable image without the built-in symbol table, then you have to write a new makefile rule yourself (based on the ones in rules.bsp), or get it from Wind River. There's probably a summary on WindSurf that will help. Bear in mind that if you want to use this image with the host tools then you will have to retain a copy of the executable that gets compressed. Otherwise the symbol information needed by the target server will be lost. I hope this is reasonably clear. Tell me if it isn't. Incidentally, when I worked for Wind River, I raised an SPR entitled, as I recall, "Inconsistent use of the term 'standalone' in documentation and elsewhere" because many customers seemed to experience unnecessary confusion. However, this SPR never made it out into public view. Regards, Mike mike_tarlton@3com.com --------------------------- Newsgroups: comp.os.vxworks Subject: Re: TaskDelay( ) bug? Date: Thu, 01 Feb 2001 18:31:57 GMT From: john_94501@my-deja.com Organization: Deja.com Message-ID: <95ca2t$2v1$1@nnrp1.deja.com> References: <3a791610$0$232@hades.is.co.za> Hello, Sounds like it could be an issue with your video device driver, perhaps it is spending a lot of time in an interrupt routine. Which one are you using? Also, I don't understand why you think you need to initialise and de-initialise the UGL stack to display things; why not just initialise it and leave it on? Oh, one final thing, have you checked with i(), ti() or taskShow() whether the task is still delayed? It might be out of the delay but pended for some other reason... Regards, John... In article <3a791610$0$232@hades.is.co.za>, "Randall" wrote: > Hi, > > I am writing a small graphics application using Tornado. In order for me to > display > the graphics, I have to initialise and de-initialise the UGL. I have no > problem doing > this and the graphics works fine. > > A second part of the application requires me to implement a infinite loop in > some task. > I use the TaskDelay() function to delay the task by a few milliseconds. This > also seems > to work fine when I dont use the graphics. The problem is that as soon as I > initialise the UGL > to display the graphics and load my device drivers, my code seems to 'hang' > in the TaskDelay() > function. The task basically stops at the TaskDelay and never seems to run > again. The system indicates > that the task is still alive and pending but it just wont execute further. > > I have given the task a higher priority and this doesnt work. The task > currently has a priority of 10, > I have even gone so low as 3. Does anyone have any ideas or ever experienced > this problem? > This only happens when I initiliase the UGL. > > Thanks in advance. > Randall Michau > rmichau@grintek.com > > Sent via Deja.com http://www.deja.com/ --------------------------- Newsgroups: comp.os.vxworks Subject: dosFS 2.0 ??? Date: Thu, 01 Feb 2001 13:50:55 -0500 From: Al Johnston Organization: MindSpring Enterprises Message-ID: <3A79B00F.8176CC8C@mindspring.com> Can some tell me what dosFS 2.0 is (why it is different from the dosFS that is shipped with vxW 5.4), Where it is. How I can get it. You would think I could get this type of question answered by WRS in a timely manner.... well I guess you know better. thanks - -al --------------------------- Newsgroups: comp.os.vxworks Subject: Serial port in vxSim Date: Thu, 1 Feb 2001 19:27:42 +0100 From: "Andrea Bigagli" Organization: Simtel Message-ID: <3a79aaab.0@news.dada.it> Reply-To: "Andrea Bigagli" Hi vxWorkers, I absolutely need to test my software on my winNT host and this software drives a modem via a normal serial port (57600, 8N1). I'd like to know if there's some trick to be done to have it works under vxSim, because actually it doesnt't work. Is there a way to redirect the normal write and read through the COM1 port of my PC? Thank you to anyone pointing me in the right direction.... Bye, Andrea. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Output file-format Date: Thu, 01 Feb 2001 19:23:18 GMT From: "Chad Augustine" Organization: EarthLink Inc. -- http://www.EarthLink.net Message-ID: References: <3A6DA86D.19A1856@raytheon.com> The output file is a binary format - you can convert it with the objcopyxxx function... K. Stuerz wrote in message news:3A6DA86D.19A1856@raytheon.com... > Hi, > > I am newbie in VxWorks. > My Question is: > > What is the output-format in Tornado I (after you have build a *.out > -File) > Is it a kind of "Motorola S-Record" ? > Or is there a possbility to choose the format. > > We have a PPC DY4-Board as Target with BSP 0.2. > > Thanks...... > > Konstantin > > > --------------------------- Newsgroups: comp.os.vxworks Subject: Cisco Opportunity in Embedded Systems Date: Thu, 01 Feb 2001 19:31:17 GMT From: maitru@my-deja.com Organization: Deja.com Message-ID: <95cdi5$6en$1@nnrp1.deja.com> Hello, Cisco Systems is currently looking for embedded engineers with background in telecom. More specifically we are looking for talent with experience in microcode development and device drivers. For more info, please email @mmadabhu@cisco.com Note for the forum coordinator: Please let me know if it ok to post job opportunities. If not, please direct me to the right forum. Thank you. Sent via Deja.com http://www.deja.com/ --------------------------- Newsgroups: comp.os.vxworks Subject: Re: tt with c++ that shows passed parameters? Date: Thu, 01 Feb 2001 13:36:22 -0600 From: Andrew Johnson Organization: APS, Argonne National Laboratory Message-ID: <3A79BAB6.3F5195EC@aps.anl.gov> References: <93nei4$mbj$1@nnrp1.deja.com> <8%hd6.567$i54.854034@newsr1.u-net.net> <95a5ul$9bk$1@nnrp1.deja.com> Dave Korn wrote: > > Bob Witmer wrote in message <95a5ul$9bk$1@nnrp1.deja.com>... > >I saw some old posts in the vxWorks archive which showed the argument > >lists with parameters. I think ... > I could be remembering wrong but I think that the difference is whether > you're using host shell or target shell. I think it's the host shell that > gives that kind of detail, so you'd need a target server > running on the host and a target (Wdb) agent running on the target. I suspect it's actually associated with the target CPU family that you're using, because I've seen both but never use the host shell. From my experience on 68k family CPUs the traces show parameters, but PowerPC CPUs don't. I've never worked out what the square brackets in the 68k output were for - anybody know? - - Andrew - -- Every great idea appears crazy to start with. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Target/Tornado shell welcome message Date: Thu, 01 Feb 2001 13:48:11 -0600 From: Andrew Johnson Organization: APS, Argonne National Laboratory Message-ID: <3A79BD7B.3A27C08C@aps.anl.gov> References: <3A799463.5D56D524@eso.org> Javier Argomedo wrote: > > Just a simple question, I want a short text message to be shown when one > enters to the vx shell via rlogin/telnet or using tornado's launcher > shell, I think it should be possible, is that so ? I think the nearest you can get is to set the login prompt - look at the reference manual for loginLib and loginStringSet. This requires that you have login security turned on in your vxWorks build (INCLUDE_SECURITY). Alternatively would changing the target shell prompt be sufficient? - see shellPromptSet. - - Andrew - -- Every great idea appears crazy to start with. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: [Q] Page Fault in case of PCI Device Date: Thu, 01 Feb 2001 21:02:44 +0100 From: Patrick Deiber Organization: PatDB Message-ID: <3A79C0E4.240BF08B@wanadoo.fr> References: <95aeru$h6m$1@nnrp1.deja.com> Reply-To: patrick.deiber@wanadoo.fr > When I access content that is pointed by base address > [base address is readed by pciConfigInLong(PciBus, PciDevice, PciFunc, > 0x10, > &PciBAR0);], page fault is occured. > > I use Tornado 2 and PCI configuration is included. > > My Initialization steps are as follows > > pciFindDevice (VendorID, DeviceID, 0, &PciBus, &PciDevice, &PciFunc); > > pciConfigOutWord (PciBus, PciDevice, PciFunc, 0x0d, 0x80); > > pciConfigOutWord (PciBus, PciDevice, PciFunc, 0x0c, 1); > > pciConfigInLong (PciBus, PciDevice, PciFunc, 0x10, &PciBAR0); > > pciConfigInByte (PciBus, PciDevice, PciFunc, 0x3c, &PciIrq); > > ret=InitDevice(param_list); > > printf("content of base address : %d\n", *PciBAR0); <- Page Fault > > Your code try to access to an area that is not defined in MMU descriptor. What is the value of PciBAR0 ? The PCI space of your device must be added in the table sysPhysMemDesc[] before MMU initialization. See function sysMmuAdd() of your BSP to perform this dynamically or create statically an entry in the table sysPhysMemDesc[]. HTH Patrick --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Q. Is there any existing library to compress a RAM file in VxWorks? Date: 1 Feb 2001 15:11:38 -0500 From: vanbaren@falcon.si.com (vanbaren_gerald) Organization: Smiths Industries Message-ID: <3a79c2fa$1@news.si.com> References: <95c2dd$a91$1@bmerhc5e.ca.nortel.com> "Yin, Yong" writes: > Is there any existing library to compress a RAM file in VxWorks? >We want to store something in a RAM file, compress this file and transfer >this file to a host machine. I know I can use some gnu codes to do this. But >I want to know if there is some existing library running on VxWorks to do >this job. >Many thanks >Yong vxWorks uses inflate() to decompress their compresses ROM images. The deflate it in the makefile procedures. It doesn't look like they provide a deflate() function in their libraries. gvb - -- +---------------------------------------------------------------------------+ | Jerry Van Baren / vanbaren_gerald@si.com / Grand Rapids Mi / 616-241-7973 | | My employer is a company. Companies are artifacts of a legal system. | |________________Artifacts are incapable of having opinions.________________| --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Telnet Date: Thu, 01 Feb 2001 14:11:41 -0700 From: Les Hammer Organization: Agilent Technologies Message-ID: <3A79D10D.6E3A22DB@agilent.com> References: <3A76E3D7.32656955@agilent.com> <3A7764DA.7F16D24A@agilent.com> <3A777CE7.A44CEA41@flashcom.net> DrDiags wrote: > If you had routines that erased Flash or the like, these routines could be > accessed and executed. Thanks. One more question. Is there a way to change the telnet port from the default to a more obscure one, such that a hacker would have to try: telnet That way I could use telnet to get to the shell for debug, but the other guys would use the default port and get an interface that I design. I found the telnetInit() command, but it doesn't take any parameters (like desired port number). Thanks --------------------------- Newsgroups: comp.os.vxworks Subject: pmc750 works? Date: Thu, 01 Feb 2001 21:25:03 GMT From: "Kelly Hornsby" Organization: Excite@Home - The Leader in Broadband http://home.com/faster Message-ID: Hello all I have a PrPMC750 from Motorola that I want to make work with a MVM2603 system (short term). Has anyone performed any integration with a setup like this? Any information, or similiar experience, will be appreciated. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: dosFS 2.0 ??? Date: Thu, 01 Feb 2001 21:54:38 GMT From: adrianmichaud@my-deja.com Organization: Deja.com Message-ID: <95clup$etu$1@nnrp1.deja.com> References: <3A79B00F.8176CC8C@mindspring.com> dosFS 2.0 supports FAT32 + has un-TAR API, chkdsk, and some other goodies. Adrian In article <3A79B00F.8176CC8C@mindspring.com>, Al Johnston wrote: > Can some tell me what dosFS 2.0 is (why it is different from the dosFS > that is shipped with vxW 5.4), Where it is. How I can get it. > > You would think I could get this type of question answered by WRS > in a timely manner.... well I guess you know better. > > thanks > > -al > Sent via Deja.com http://www.deja.com/ --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Need a C++ Compiler Date: Thu, 01 Feb 2001 16:29:16 -0600 From: Robert Norton Organization: Gilson Inc. Message-ID: <3A79E33C.9DF0D01F@gilson.com> References: <92spv2$uim$1@nnrp1.deja.com> <3A5277BB.131C67B0@bellsouth.net> <92uqd0$l5c$1@nnrp1.deja.com> <3A533D3F.7439933A@lmco.com> Reply-To: rnorton@gilson.com Joe Durusau wrote: > Ah, so! If you want to do real work on a target with > VxWorks, you have to PAY! As Johan observes elsewhere, you don't get > the libs and suchlike for your target with the eval system. Open your > pocketbook (as widely as possible) and your problems will be solved. We have used an in-house real time OS with the Intel 80C196 family for a long time, and would like to upgrade to a stronger CPU and better real-time OS. VxWorks and an ARM look good to me, but what are the licensing fees like? There seems to be no pricing info on the web pages at all, and that scares me a bit. We would probably want C++ and Java as well as the bare OS. Is WindRiver licensing on a one-time fee, or on per unit sold or what? TIA, Bob Norton. --------------------------- Newsgroups: comp.os.vxworks Subject: re: How do I build a compressed VxWorks? Date: Thu, 1 Feb 2001 17:51:12 -0600 From: "lee" Organization: Lawrence Berkeley National Laboratory Message-ID: <95cu7v$eg4$1@overload.lbl.gov> Hi, I am writing a driver under vxWorks. I am running into some problems with the driver entry point. My ioctl call that are installed under vxWorks using iosDrvInstall need to have a variable input as the third argument to support extended functionalities. However, the ioctl defined by vxWorks iosLib is strictly int ioctl(int, int, int) Is there a way i can compile a code using the ioctl with the variable input without the compiler barfing complaining about too few or too many input arguments to ioctl?? thanks, lee vxWorks --------------------------- Newsgroups: comp.os.vxworks Subject: How to "clean up" special test procedures currently run through the shell? Date: Thu, 01 Feb 2001 16:55:10 -0800 From: Christopher A Leddy Organization: Raytheon Company Message-ID: <3A7A056E.16ABA773@west.raytheon.com> I have a system with 5 unique custom PPC boards (each with ethernet connection) running VxWorks 5.3.1. When developing, the software engineers do the standard drill of connecting target servers to the board, attaching shells and/or debuggers, loading code, executing subroutines, changing variables, etc. We have been using this same method to do special alignment and test of various subsystems by loading special test routines, changing and reading variables, and in some cases even loading special versions of code. All this works marginally in the engineering lab with 5 engineers available to connect the servers and shells, run the scripts, and make sense of the returns, but it won't fly when we go to production. I need to automate the process, and provide a spiffy production GUI interface on top of it, and run the whole thing under Windows NT (no heckles, I have other applications to connect to, already running on NT). So what are my options? I do want to try to run the actual tactical code and add a few alignment/test routines and change a few variables like we do now, because I am afraid of code divergence as the software mutates. The possibilities I have so far are: 1. Add support for a UDP connection into the target and develop a protocol to trigger the test alignment routines and connect to the target with a VC++ host with buttons etc. (Already did it on one processor, painful and inflexible; I don't like this because I have to do it 5 times on 5 different processors, and I make the tactical code drag around all the garbage I add). 2. Figure out how to direct connect into the WDB target agent and use a VC++ host, again with buttons etc. This seems painful because, from the docs, it looks like I need to figure out how to do an RPC client, and then I have to parse out elf/dwarf symbol tables on the host etc. After that, I need to do the WDB command dance using the locations I strip off the symbol table. It would be so much cleaner on the user interface side since I wouldn't need all the servers and shells floating around and connected. (Connecting 5 target servers + 5 shells) == failure, half the time). 3. Try to tie in some type of GUI TCL into tornado to control all the shells and servers? I don't have a feel for how hard this is and whether you can pass data into and out of the shell, or whether TCL will work with 5 different shells open at once, etc. This method also requires Tornado, in all its glory, and all the shells and servers successfully connected; a painful and messy proposition. Comments, Suggestions, Experiences? Thanks, Chris --------------------------- Newsgroups: comp.os.vxworks Subject: tools.h++ Date: Fri, 02 Feb 2001 00:48:15 GMT From: jayantadas@my-deja.com Organization: Deja.com Message-ID: <95d04c$ol7$1@nnrp1.deja.com> Hi, Does any one know how can I have tools.h++ application component configured into vxWorks? I guess it is INCLUDE_CPLUS_TOOLS. Is not this component part of standard vxWorks? Thanks for the help. Sent via Deja.com http://www.deja.com/ --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Telnet Date: Fri, 02 Feb 2001 01:16:16 GMT From: muscou Organization: Sympatico Message-ID: <3A7A0CE7.36C256F9@nospamsympatico.ca> References: <3A76E3D7.32656955@agilent.com> <3A7764DA.7F16D24A@agilent.com> <3A777CE7.A44CEA41@flashcom.net> <3A79D10D.6E3A22DB@agilent.com> To restrict the function available to the shell , parse the generated symTbl.c to keep only the needed function before compiling it out ( change the makefile rules, easy to do unless you are using the project facility...). Also, define INCLUDE_SECURITY when compiling and add a password and username in order to be able to access the telnet port. Les Hammer wrote: > DrDiags wrote: > > > If you had routines that erased Flash or the like, these routines could be > > accessed and executed. > > Thanks. One more question. Is there a way to change the telnet port from the > default to a more obscure one, such that a hacker would have to try: > telnet > That way I could use telnet to get to the shell for debug, but the other > guys would use the default port and get an interface that I design. > I found the telnetInit() command, but it doesn't take any parameters > (like desired port number). > > Thanks --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Need a C++ Compiler Date: Thu, 01 Feb 2001 20:42:47 -0500 From: Joe Durusau Message-ID: <3A7A1097.83CDDDF5@bellsouth.net> References: <92spv2$uim$1@nnrp1.deja.com> <3A5277BB.131C67B0@bellsouth.net> <92uqd0$l5c$1@nnrp1.deja.com> <3A533D3F.7439933A@lmco.com> <3A79E33C.9DF0D01F@gilson.com> There is a large up-front fee for the system, an anual maintenance fee, and a per-CPU target license required. The fees are somewhat negotiable, so you have to call WRS to get a figure. Essentially all of the windxxxx you see talked about are separate products. If you want windview, that's so much extra. If you want multi-cpu capability, that's again so much extra. Be sure you make it absolutely clear to the WRS rep exactly what you want, and if you have to have a central purchasing dept do the actual buy, don't let them change anything for any reason! Speaking only for myself, Joe Durusau Robert Norton wrote: > > Joe Durusau wrote: > > > Ah, so! If you want to do real work on a target with > > VxWorks, you have to PAY! As Johan observes elsewhere, you don't get > > the libs and suchlike for your target with the eval system. Open your > > pocketbook (as widely as possible) and your problems will be solved. > > We have used an in-house real time OS with the Intel 80C196 family for a long > time, and would like to upgrade to a stronger CPU and better real-time OS. > VxWorks and an ARM look good to me, but what are the licensing fees like? > There seems to be no pricing info on the web pages at all, and that scares me > a bit. We would probably want C++ and Java as well as the bare OS. Is > WindRiver licensing on a one-time fee, or on per unit sold or what? > > TIA, Bob Norton. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Cisco Opportunity in Embedded Systems Date: Fri, 02 Feb 2001 01:47:14 GMT From: bmcgonig@my-deja.com Organization: Deja.com Message-ID: <95d3j0$rii$1@nnrp1.deja.com> References: <95cdi5$6en$1@nnrp1.deja.com> Hi All: I would be interested in the opinions of present and former Cisco employees. Would you recommend working for Cisco? In article <95cdi5$6en$1@nnrp1.deja.com>, maitru@my-deja.com wrote: > Hello, > Cisco Systems is currently looking for embedded engineers with > background in telecom. > More specifically we are looking for talent with experience in microcode > development and device drivers. > For more info, please email @mmadabhu@cisco.com > > Note for the forum coordinator: > Please let me know if it ok to post job opportunities. If not, please > direct me to the right forum. > > Thank you. > > Sent via Deja.com > http://www.deja.com/ > Sent via Deja.com http://www.deja.com/ --------------------------- Newsgroups: comp.os.vxworks Subject: Where can I get the old mails? Date: Fri, 2 Feb 2001 10:59:50 +0800 From: "James" Organization: Motorola Message-ID: <95d7r9$n7j$1@newshost.mot.com> Hi, All, Could you tell me where the old mail collection is stored? Thanks a lot Gao Jianwei --------------------------- Newsgroups: comp.os.vxworks Subject: any books on VxWorks? Date: Fri, 02 Feb 2001 04:10:54 GMT From: ashu_123@my-deja.com Organization: Deja.com Message-ID: <95dc09$2hq$1@nnrp1.deja.com> hi, Can anyone tell me if there are any books available on VxWorks? if so then please let me know their details. Where can i get detailed information about the VxWorks API? is it similar to the Windows API? Sent via Deja.com http://www.deja.com/ --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Telnet Date: Thu, 01 Feb 2001 21:56:20 -0800 From: DrDiags Organization: Posted via Supernews, http://www.supernews.com Message-ID: <3A7A4C04.F7A8D91D@flashcom.net> References: <3A76E3D7.32656955@agilent.com> <3A7764DA.7F16D24A@agilent.com> <3A777CE7.A44CEA41@flashcom.net> <3A79D10D.6E3A22DB@agilent.com> Reply-To: drdiags@flashcom.net Les, Unless you have source, it appears the the telnet daemon port is hardwired to port 23. Muscou suggestion is a reasonable one. Les Hammer wrote: > DrDiags wrote: > > > If you had routines that erased Flash or the like, these routines could be > > accessed and executed. > > Thanks. One more question. Is there a way to change the telnet port from the > default to a more obscure one, such that a hacker would have to try: > telnet > That way I could use telnet to get to the shell for debug, but the other > guys would use the default port and get an interface that I design. > I found the telnetInit() command, but it doesn't take any parameters > (like desired port number). > > Thanks --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Where can I get the old mails? Date: Fri, 02 Feb 2001 07:03:37 GMT From: john_94501@my-deja.com Organization: Deja.com Message-ID: <95dm47$a6l$1@nnrp1.deja.com> References: <95d7r9$n7j$1@newshost.mot.com> Hello, deja.com has a lot of them, in a searchable form; not sure if there is an archive anywhere else. HTH, John... In article <95d7r9$n7j$1@newshost.mot.com>, "James" wrote: > Hi, All, > > Could you tell me where the old mail collection is stored? > > Thanks a lot > > Gao Jianwei > > Sent via Deja.com http://www.deja.com/ --------------------------- Newsgroups: comp.os.vxworks Subject: Additional supported PCMCIA networkcards Date: Fri, 02 Feb 2001 08:31:27 +0100 From: uwe foss Organization: Siemens Inc. Message-ID: <3A7A624F.C8EBF276@dit.siemens.de> Dear all, i am using a 486BSP T1 and T2 with pcmcia support. Actually the standard and the *only* pcmcia network driver supported by windriver systems is the 3C509-family from 3COM Company. So the3C589D, 3CCE589ET cards are compatilbe, but the actually 3CCFE574BT networkcard doesn't work. Are there any other pcmcia networking cards / drivers under the Tornado pcccard framework available? thanks in advance Uwe Foss Siemens AG --------------------------- Newsgroups: comp.os.vxworks Subject: Re: VxWorks and multicast Date: Fri, 02 Feb 2001 08:42:03 +0100 From: Albert Wijnja Organization: MECO Equipment Engineers BV Message-ID: <3A7A64CB.A5EB0EA6@meco.nl> References: <95c240$o4m$1@overload.lbl.gov> Reply-To: Albert.Wijnja@meco.nl Hello Gene, I use the DEC 21040 card with multicast. It runs smoothly. Don't forget to add your a multicast route to your interface, e.g: routeAdd("224.0.0.0", "10.0.0.33"); if the IP address of your card is 10.0.0.33. Also, make sure that you have MCAST_ROUTING included in your VxWorks image (INCLUDE_MCAST_ROUTING). Albert Wijnja Gene wrote: > vxworks/vxWorkers, > > Does anyone have any experience using multicast in a vxWorks environment? > > I'm using the Dec 21040 driver which seems to support multicast receives but > the "if" routines indicate that multicasting is not supported. > > Thanks in advance, > Gene Small > Software Technology, Inc > --------------------------- Newsgroups: comp.os.vxworks Subject: generated dependencies Date: Fri, 02 Feb 2001 10:04:59 +0100 From: Joern Tietjen Organization: =?iso-8859-1?Q?Dr=E4ger?= Medizintechnik GmbH Message-ID: <3A7A783B.4AE76F03@draeger.com> Hi, is anybody using the dependencies generated thru T2 (Win NT) ? To me it seems that this feature is highly unreliable. Dependencies get updated only occasionally after you requested it. Sometimes they use invalid pathnames while filenames are correct. Is there a known rule behind this ? Is it a problem of accessing files via NFS or Samba ? TIA, Joern --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Telnet Date: 2 Feb 2001 09:30:05 GMT From: borkhuis@agere.com (Johan Borkhuis) Organization: Lucent Technologies Message-ID: References: <3A76E3D7.32656955@agilent.com> <3A7764DA.7F16D24A@agilent.com> <3A777CE7.A44CEA41@flashcom.net> <3A79D10D.6E3A22DB@agilent.com> Les Hammer wrote: >Thanks. One more question. Is there a way to change the telnet port >from the default to a more obscure one, such that a hacker would have to >try: > telnet >I found the telnetInit() command, but it doesn't take any parameters >(like desired port number). For this you need the unsupported Telnet server. It can be obtained from tech support. You can configure the port and the function that should be executed when accessing this port. 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: dosFS 2.0 ??? Date: 2 Feb 2001 09:35:20 GMT From: borkhuis@agere.com (Johan Borkhuis) Organization: Lucent Technologies Message-ID: References: <3A79B00F.8176CC8C@mindspring.com> Al Johnston wrote: >Can some tell me what dosFS 2.0 is (why it is different from the dosFS >that is shipped with vxW 5.4), Where it is. How I can get it. It is an update/replacement for the old DosFS. You can request it from Windriver through your FAE or sales representative. It is a free update if you have a support contract. 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: any books on VxWorks? Date: 2 Feb 2001 09:44:02 GMT From: borkhuis@agere.com (Johan Borkhuis) Organization: Lucent Technologies Message-ID: References: <95dc09$2hq$1@nnrp1.deja.com> ashu_123@my-deja.com wrote: >hi, >Can anyone tell me if there are any books available on VxWorks? if so >then please let me know their details. Not that I know of. >Where can i get detailed information about the VxWorks API? is it >similar to the Windows API? That is included with VxWorks/Tornado. If you want to look at the API, download the Tornado Prototyper, as this includes the complete documentation. Groeten, Johan - -- o o o o o o o . . . _____________________________ o _____ || Johan Borkhuis | .][__n_n_|DD[ ====_____ | borkhuis@agere.com | >(________|__|_[_________]_|__________________________| _/oo OOOOO oo` ooo ooo 'o!o!o o!o!o` === VxWorks FAQ: http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html === --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Where can I get the old mails? Date: 2 Feb 2001 09:41:35 GMT From: borkhuis@agere.com (Johan Borkhuis) Organization: Lucent Technologies Message-ID: References: <95d7r9$n7j$1@newshost.mot.com> "James" wrote: >Could you tell me where the old mail collection is stored? http://www-csg.lbl.gov/vxworks/ 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: RPC UDP Sockets Problem Date: Fri, 02 Feb 2001 09:42:45 +0000 From: Graham Baxter Organization: Graham Baxter (Software) Limited Message-ID: <3A7A8115.72EE874B@NOSPAM.bcs.org.uk> VxWorkers, We have a problem after connecting greater than 19 RPC connections. When connecting our 20th connection we receive RPC 4 errors meaning we can't receive. We are running RPC over UDP and I believe that reason that this error is occurring is that the UDP is running out of some resources (ports, buffers etc) . The number of UDP ports being used at this point is 43. Does anyone know if UDP has maximum port threshold which we can configure? Or can anyone advise on any other solutions to this problem? Thanks for any help. Graham Baxter gbaxter@NOSPAMdsl.ftel.co.uk +44 (0)1223 495200 --------------------------- Newsgroups: comp.os.vxworks Subject: How Do I make an interrup handling routine for the LCD on an assabet? Date: Fri, 2 Feb 2001 11:48:25 +0100 From: "Carsten Schwartz" Organization: Posted Courtesy of Tele Danmark or one of its European Subsidiaries Message-ID: <95e3b7$26t$1@news.inet.tele.dk> Reply-To: "Carsten Schwartz" I have tried using the IP12 bit in ICPR but no interrupt are generated, any ideas ? --------------------------- Newsgroups: comp.os.vxworks Subject: Network problem on MVME2700 T1.0.1 Date: Fri, 02 Feb 2001 11:28:37 +0100 From: Friedrich.Scheurer@t-online.de (Friedrich Scheurer / Goeppingen) Organization: Lawrence Berkeley National Laboratory Message-ID: <95e4tf$gf$1@overload.lbl.gov> Dear VxWorkers, dear David,

David Laight gave some insight into ethernet negotiation aspects and gave
a hint to problems in network traffic "... network to run 'like a sick pig' ...".

I suppose we're suffering from a problem like this for 1,5 years.

We often loose target-server connection to the target MVME2700
during a debug session. 5 times higher timeout didn't solve the problem really.
Often FTP-transfers from target to host are very slowly (about 1 % of maximum performance).
This almost only occurs during daytime if there's lot of traffic in our
network. Out of standard office times it's much better.
The slowing down is in most cases 'healed' if we're rebooting the target.

WRS couldn't help us only from the symptoms, because it's not reproducable.

The target is attached via Hub/Repeater not a switch.
(David said: Generally workin in half duplex)

How can I fix wether my target tries to run in full duplex
and makes the problems he described in his posting?

TIA, regards

F. Scheurer
  

Submitted-by vxwexplo-errs@csg.lbl.gov  Thu Sep 21 02:38:58 2000
Submitted-by: David Laight 

>> When booting VxWorks I get the message "interrupt: dc0 - no carrier"
>> very frequently.

> Wind River is aware of this problem. It is caused if you are connected
> to a ethernet switch that supports Full Duplex. Ethernet Hubs are
> typically Half Duplex.

> Wind River send me a new dec*.o file which essentially removed the no
> carrier print.

> There is an SPR on Windsurf already about this problem.

Does anyone know if WRS have fixed this properly?

The SPR doesn't say much, but indicates that they think error is "normally 
harmless' (mostly harmless?).

The 'no carrier' error is reported when the ethernet chipset doesn't get a 
receive signal (it is the rx carrier that is missing) during its own transmits.  
This is a test of the rx and tx pairs of the AUI cable to the transceiver (the 
SQE test checks the collision pair).  A coax transceiver will always receive its 
own transmittions...

On twisted pair the transceiver (AUI or MII) reports a collision if there is 
receive data (on the TP cable) during transmit, and returns the tx data as rx to 
stop the chipset reporting 'no carrier'.  The chipset will ignore the rx data 
during transmit.

For full duplex the tx and rx sides are completely separate.  However 4 pieces 
of hardware have to agree to use full duplex.  The two transceivers can (and 
will) auto-negotiate full duplex if permitted.  The etherner controller has to 
interrogate the transceiver to determine what has been negotiated.  This is hard 
and is left to the device driver.

Unfortunately there is no easy way (for the device driver) to determine what the 
transceivers have negotiated.  Especially since it can change at any time.

The easy 'solution' is to change the transceivers 'link advertise register' and 
disable the full duplex modes.  Then you know full duplex will never be 
negotiated.  (Don't EVER fix a port to full duplex - it will only end in tears.)

The effect of having the transceiver running full duplex but the ethernet 
chipset half duplex is that receive packets will be discarded - typically during 
heavy traffic.  This can be hard to detect - but causes the network to run 'like 
a sick pig' at times.

Note that you can change from 10M to 100M without the driver/chipset having to 
do anything at all.  It just works.  Determining the current speed is, of 
course, almost impossible from the standard MII registers.

        David


- ----------------------------------------------------------------
David Laight                          email: dsl@tadpole.co.uk
Tadpole Technology plc                phone: +44 1223 428 232
Cambridge, UK                         fax:   +44 1223 428 201

--
+-------------------------------------------------+
| Dipl.-Ing. F. Scheurer                          |
| c/o Zentrum Fertigungstechnik Stuttgart         |
| Nobelstrasse 15                                 |
| D-70569 Stuttgart                               |
| Telefon +49 711 13162-41(-0)                    |
| Fax     +49 711 13162-11                        |
| E-Mail  scheurer@zfs.tz.uni-stuttgart.de        |
| WWW     http://www.zfs.tz.uni-stuttgart.de      |
+-------------------------------------------------+
| privat:                  Waldeckstrasse 31      |
|                          D-73035 Goeppingen     |
| Telefon +49 7161 686362  Fax +49 7161 686301    |
| Mobil  +49 170 31 77 539                        |
| E-Mail  Friedrich.Scheurer@t-online.de          |
| oder    Friedrich_Scheurer@comundo.de           |
| oder    Friedrich.Scheurer@lycosmail.com        |
| WWW     http://members.tripod.de/machinist/     |
| WWW     http://www.friedrichscheurer.com/       |
+-------------------------------------------------+
  --------------------------- Newsgroups: comp.os.vxworks Subject: [Fwd: Network problem on MVME2700 T1.0.1] Date: Fri, 02 Feb 2001 11:33:37 +0100 From: Friedrich.Scheurer@t-online.de (Friedrich Scheurer / Goeppingen) Organization: Lawrence Berkeley National Laboratory Message-ID: <95e4tg$gg$1@overload.lbl.gov> Sorry for the HTML-version! - -------- Original Message -------- Dear VxWorkers, dear David, David Laight gave some insight into ethernet negotiation aspects and gave a hint to problems in network traffic "... network to run 'like a sick pig' ...". I suppose we're suffering from a problem like this for 1,5 years. We often loose target-server connection to the target MVME2700 during a debug session. 5 times higher timeout didn't solve the problem really. Often FTP-transfers from target to host are very slowly (about 1 % of maximum performance). This almost only occurs during daytime if there's lot of traffic in our network. Out of standard office times it's much better. The slowing down is in most cases 'healed' if we're rebooting the target. WRS couldn't help us only from the symptoms, because it's not reproducable. The target is attached via Hub/Repeater not a switch. (David said: Generally workin in half duplex) How can I fix wether my target tries to run in full duplex and makes the problems he described in his posting? TIA, regards F. Scheurer > Submitted-by vxwexplo-errs@csg.lbl.gov Thu Sep 21 02:38:58 2000 > Submitted-by: David Laight > > >> When booting VxWorks I get the message "interrupt: dc0 - no carrier" > >> very frequently. > > > Wind River is aware of this problem. It is caused if you are connected > > to a ethernet switch that supports Full Duplex. Ethernet Hubs are > > typically Half Duplex. > > > Wind River send me a new dec*.o file which essentially removed the no > > carrier print. > > > There is an SPR on Windsurf already about this problem. > > Does anyone know if WRS have fixed this properly? > > The SPR doesn't say much, but indicates that they think error is "normally > harmless' (mostly harmless?). > > The 'no carrier' error is reported when the ethernet chipset doesn't get a > receive signal (it is the rx carrier that is missing) during its own transmits. > This is a test of the rx and tx pairs of the AUI cable to the transceiver (the > SQE test checks the collision pair). A coax transceiver will always receive its > own transmittions... > > On twisted pair the transceiver (AUI or MII) reports a collision if there is > receive data (on the TP cable) during transmit, and returns the tx data as rx to > stop the chipset reporting 'no carrier'. The chipset will ignore the rx data > during transmit. > > For full duplex the tx and rx sides are completely separate. However 4 pieces > of hardware have to agree to use full duplex. The two transceivers can (and > will) auto-negotiate full duplex if permitted. The etherner controller has to > interrogate the transceiver to determine what has been negotiated. This is hard > and is left to the device driver. > > Unfortunately there is no easy way (for the device driver) to determine what the > transceivers have negotiated. Especially since it can change at any time. > > The easy 'solution' is to change the transceivers 'link advertise register' and > disable the full duplex modes. Then you know full duplex will never be > negotiated. (Don't EVER fix a port to full duplex - it will only end in tears.) > > The effect of having the transceiver running full duplex but the ethernet > chipset half duplex is that receive packets will be discarded - typically during > heavy traffic. This can be hard to detect - but causes the network to run 'like > a sick pig' at times. > > Note that you can change from 10M to 100M without the driver/chipset having to > do anything at all. It just works. Determining the current speed is, of > course, almost impossible from the standard MII registers. > > David > > > ---------------------------------------------------------------- > David Laight email: dsl@tadpole.co.uk > Tadpole Technology plc phone: +44 1223 428 232 > Cambridge, UK fax: +44 1223 428 201 > > - -- +-------------------------------------------------+ | Dipl.-Ing. F. Scheurer | | c/o Zentrum Fertigungstechnik Stuttgart | | Nobelstrasse 15 | | D-70569 Stuttgart | | Telefon +49 711 13162-41(-0) | | Fax +49 711 13162-11 | | E-Mail scheurer@zfs.tz.uni-stuttgart.de | | WWW http://www.zfs.tz.uni-stuttgart.de | +-------------------------------------------------+ | privat: Waldeckstrasse 31 | | D-73035 Goeppingen | | Telefon +49 7161 686362 Fax +49 7161 686301 | | Mobil +49 170 31 77 539 | | E-Mail Friedrich.Scheurer@t-online.de | | oder Friedrich_Scheurer@comundo.de | | oder Friedrich.Scheurer@lycosmail.com | | WWW http://members.tripod.de/machinist/ | | WWW http://www.friedrichscheurer.com/ | +-------------------------------------------------+ --------------------------- End of New-News digest ********************** From vxwexplo-errs@csg.lbl.gov Fri Feb 2 06:49:19 2001 From: Jim Way Date: Fri Feb 2 06:49:21 PST 2001 Subject: How to compare two vxWorks hex files (TSR 149779) About a year ago, I had a question about how to compare two vxWorks HEX files. This was recorded by WindRiver as TSR 149779. The problem was that we use the "romCompress" option. Date/time stamp information is different between the two builds, causing the file compression algorithm to generate radically different results. After some time, I've finally determined a way to do the comparison that is convincing enough for me. (Now that I've seen it, it seems painfully obvious.) The solution is to go to the VxWorks tab in the project window, click on the Build name, call up the context menu, select Properties, then select the Rules tab. Where it says "vxWorks_romCompress" change the rule to "vxWorks_romResident". Apply that rule and click OK. Now rebuild-all the project (you might be able to just relink, but I'm paranoid). The resulting file will be huge because it's uncompressed. All that's left to do is to use your favorite file comparison tool (we use WinDiff and CodeWright) to compare the files. You should still observe a few bytes that are different. You can minimize these by trying to build both HEX files at about the same time on the same day. The only bytes that *should* be different are the date and time representations. Don't forget to change your rule back to "vxWorks_romCompress" when you're done. WindRiver closed TSR 149779 (over my objections). Here's hoping this explanation gets into their support database. Thanks again to everyone for their support and encouragement! Jim ----------------------------------------- Jim Way, Software Engineer Datum Austin (Austron Inc.) voice: 512.721.4170 fax : 512.990.9712 email: jwayATdatumDOTcom (no spam please) ----------------------------------------- From vxwexplo-errs@csg.lbl.gov Fri Feb 2 08:56:17 2001 From: David Laight Date: Fri Feb 2 08:56:19 PST 2001 Subject: Re: How Do I make an interrupt handling routine for the LCD on an assabet? > I have tried using the IP12 bit in ICPR but no interrupt are generated, any > ideas ? I failed to get the LCD error interrupts working on a rev E SA1100. This is basically the same controller as the SA1110. Intel support reported: It does appear that you have things set up properly. I have found other references to problems with the underrun interrupt. However, I haven't been able to find anything definitive. I have a query in to the chip designers to determine whether there is a problem with the interrupt, or whether there is something that we are missing. I will let you know as soon as I have a response. This was 12 months ago, haven't chased them for progress. David vxWorks ---------------------------------------------------------------- David Laight email: dsl@tadpole.co.uk Tadpole Technology plc phone: +44 1223 428 232 Cambridge, UK fax: +44 1223 428 201 From vxwexplo-errs@csg.lbl.gov Fri Feb 2 09:19:07 2001 From: David Laight Date: Fri Feb 2 09:19:09 PST 2001 Subject: Re: Network problem on MVME2700 T1.0.1 > > Dear VxWorkers, dear David, > > David Laight gave some insight into ethernet negotiation aspects and > gave > a hint to problems in network traffic "... network to run 'like a sick > pig' ...". > > I suppose we're suffering from a problem like this for 1,5 years. > > We often loose target-server connection to the target MVME2700 > during a debug session. 5 times higher timeout didn't solve the problem > really. > Often FTP-transfers from target to host are very slowly (about 1 % of > maximum performance). > This almost only occurs during daytime if there's lot of traffic in our > network. Out of standard office times it's much better. The problem could be an inter-hub link set FDX somewhere, but... > The slowing down is in most cases 'healed' if we're rebooting the target. This makes it look much more like a vxworks problem - the system is somehow remembering 'state'. I would check the network data pools: - netStackDataPoolShow() and netPoolShow( end->pNetPool ) In case buffers are being lost..... Many vxWorks END drivers do not handle transmits that are flow controlled by the MAC layer properly. What the driver should do isn't documented, and the driver I copied didn't do the correct thing - the buffers that could not be transmitted were dumped forever. I recall a posting a while ago that explained whet the driver should be doing, however I had already fixed my driver a different way - so didn't test it. (My driver internally queues transmit packets until the ethernet card has space. I ought (maybe do?) dump the entire queue if we are failing to transmit at all.) David ---------------------------------------------------------------- David Laight email: dsl@tadpole.co.uk Tadpole Technology plc phone: +44 1223 428 232 Cambridge, UK fax: +44 1223 428 201 From vxwexplo-errs@csg.lbl.gov Fri Feb 2 11:39:46 2001 From: "Crozier, Bruce" Date: Fri Feb 2 11:39:48 PST 2001 Subject: Update to vxWorks DHCP Hello Luke, I am currently working on DHCP, and your email to the NewsGroup caught my eye. I would appreciate it if you would tell me what target architecture & Tornado version you were working with, some of the major DHCP errors you found, and maybe the contact at WRS support from whom you got the update. My target is PPC (750 processor), and WRS support in Alameda told me that they had no idea what you might be referring to. I would appreciate any assistance you can provide. Thanks, Bruce Crozier BTW, my attempt to email you directly at Diamand@btinternet.com failed for some reason (unknown local-part "diamand" in domain "btinternet.com") ---------------------------------------------------------------------------- --------- Newsgroups: comp.os.vxworks Subject: Re: DHCP Bind failed Date: 29 Jan 2001 22:53:12 GMT From: Diamand@btinternet.com (Luke Diamand) Organization: UKLINUX NET Message-ID: References: <3A64421F.36327B85@temex.fr> Delsol Thomas(chez Tekelec Temex Telecom) wrote: We found the DHCP client in T2 to be hopelessly broken. We complained to support and got a new one which worked like a dream. HTH! Luke Diamand > > From vxwexplo-errs@csg.lbl.gov Sat Feb 3 04:03:09 2001 From: Vxworks Exploder Date: Sat Feb 3 04:03:11 PST 2001 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Sat Feb 3 04:03:05 PST 2001 Subject: Commercial postings in comp.os.vxworks [was Re: Cisco Opportunity in Embedded Systems] Subject: Re: How do I build a compressed VxWorks? Subject: 860SAR at 66Mhz Subject: null pointer on 8260 Subject: Re: dosFS 2.0 ??? Subject: Re: Need a C++ Compiler Subject: Re: How to "clean up" special test procedures currently run through the shell? Subject: How to compare two vxWorks hex files (TSR 149779) Subject: tNetTask Timing Anomaly Subject: Re: TaskDelay( ) bug? -- Probably not Subject: TSFS ioctl and FIOSYNC Subject: re: How Do I make an interrupt handling routine for the LCD on an assabet? Subject: Re: null pointer on 8260 Subject: Re: tools.h++ Subject: re: Network problem on MVME2700 T1.0.1 Subject: Re: RPC UDP Sockets Problem Subject: Random number generator Subject: Re: How to "clean up" special test procedures currently run through the Subject: Update to vxWorks DHCP Subject: Re: DHCP Bind failed Subject: Re: generated dependencies Subject: ANNOUNCE: New Article on Virtual Serial Ports Subject: ANNOUNCE: New Article on Language Selection Subject: ANNOUNCE: New Articles on TCP/IP, UDP/IP, and ARP Subject: Re: Random number generator Subject: ANNOUNCE: New Article on Software Add-ons Subject: Re: Random number generator Subject: Re: generated dependencies Subject: Re: Update to vxWorks DHCP Subject: vxworks and IBM's microdrive Subject: about zbuf ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Commercial postings in comp.os.vxworks [was Re: Cisco Opportunity in Embedded Systems] Date: Fri, 2 Feb 2001 10:27:36 -0000 From: "Dave Korn" Organization: Lumber Cartel (tinlc) Members #2234-2237 (owing to browser refresh) Message-ID: References: <95cdi5$6en$1@nnrp1.deja.com> maitru@my-deja.com wrote in message <95cdi5$6en$1@nnrp1.deja.com>... >Hello, >Cisco Systems is currently looking for embedded engineers with >background in telecom. >More specifically we are looking for talent with experience in microcode >development and device drivers. >For more info, please email @mmadabhu@cisco.com > >Note for the forum coordinator: >Please let me know if it ok to post job opportunities. If not, please >direct me to the right forum. There is no 'official' forum co-ordinator. The relevant standard to adhere to would be the newsgroup's charter, which was used to establish the guidelines for the group when it was created. Here's the details, copied from the archive at ftp.isc.org that holds all the newsgroup creation/deletion messages: - -----Begin quote------------------- comp.os.vxworks is an unmoderated newsgroup which passed its vote for creation by 295:17 as reported in news.announce.newgroups on 6 Jan 1992. For your newsgroups file: comp.os.vxworks The VxWorks real-time operating system. The charter, culled from the call for votes: comp.os.vxworks will be a forum for discussion of topics related to the VxWorks real-time operating system and development environment. Current and prospective users of VxWorks will be able to discuss technical problems, report experiences with various hardware components, share code, report bugs and fixes, and exchange any information of interest to the VxWorks user community. Relevant new product announcements may be made by commercial participants, subject to a limit - e.g. 1/month - on the frequency of such posts. - ---------End quote--------------- DaveK - -- They laughed at Galileo. They laughed at Copernicus. They laughed at Columbus. But remember, they also laughed at Bozo the Clown. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: How do I build a compressed VxWorks? Date: Fri, 2 Feb 2001 10:28:13 -0000 From: "Dave Korn" Organization: Lumber Cartel (tinlc) Members #2234-2237 (owing to browser refresh) Message-ID: References: <95cu7v$eg4$1@overload.lbl.gov> lee wrote in message <95cu7v$eg4$1@overload.lbl.gov>... >[snip!]. My ioctl call that are installed under vxWorks using iosDrvInstall need to have a variable input as the third argument to support extended functionalities. However, the ioctl defined by vxWorks iosLib is strictly int ioctl(int, int, int) > >Is there a way i can compile a code using the ioctl with the variable input without the compiler barfing complaining about too few or too many input arguments to ioctl?? The standard 'well-engineered' solution to a situation like this would be to pass a pointer to a struct for the third parameter (cast to int at the call, and cast back by the driver). That struct can then contain whatever extra arguments or information you need to pass to the driver to make the command work. DaveK - -- They laughed at Galileo. They laughed at Copernicus. They laughed at Columbus. But remember, they also laughed at Bozo the Clown. --------------------------- Newsgroups: comp.os.vxworks Subject: 860SAR at 66Mhz Date: Fri, 02 Feb 2001 12:42:20 GMT From: gkime@my-deja.com Organization: Deja.com Message-ID: <95e9vb$pg8$1@nnrp1.deja.com> Has anyone tried running an 860 or 860SAR at 66Mhz? If so, what were your results? Thanks, Glenn Sent via Deja.com http://www.deja.com/ --------------------------- Newsgroups: comp.os.vxworks Subject: null pointer on 8260 Date: Fri, 02 Feb 2001 13:32:40 GMT From: pgrangier@my-deja.com Organization: Deja.com Message-ID: <95ecto$rq2$1@nnrp1.deja.com> I work with a MPC8260. My memory map is designed in such a way that the 60x SDRAM start at address 0x00000000. To prevent the use of uninitialized pointers, I want protect the address 0x0. The problem is that the Exceptions are vectored to the physical address 0x000n_nnnn. Maybe it's possible to protect this address with MMU.... If someone has an idea, thanks in advance Pascal Sent via Deja.com http://www.deja.com/ --------------------------- Newsgroups: comp.os.vxworks Subject: Re: dosFS 2.0 ??? Date: Fri, 2 Feb 2001 09:06:24 -0500 From: "Mike Kordik" Message-ID: <%7ze6.2$lA2.24@client> References: <3A79B00F.8176CC8C@mindspring.com> I got DosFS2.0 from our inside salesperson at WRS. I just called, gave them my info and they sent me the package. DosFS2.0 can read paritioned and formatted MSDOS disks. It also supports long filenames (beyond the WRS 40 bytes filenames). When we create our bootable VxWorks disks we partion and format using MSDOS and then use mkboot.o to make it VxWorks bootable. I am sure there are other features but these are the only ones we exploit. Hope this helps. - -Mike "Al Johnston" wrote in message news:3A79B00F.8176CC8C@mindspring.com... > Can some tell me what dosFS 2.0 is (why it is different from the dosFS > that is shipped with vxW 5.4), Where it is. How I can get it. > > You would think I could get this type of question answered by WRS > in a timely manner.... well I guess you know better. > > thanks > > -al --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Need a C++ Compiler Date: Fri, 02 Feb 2001 08:20:29 -0600 From: Robert Norton Organization: Gilson Inc. Message-ID: <3A7AC22D.6A2EABB7@gilson.com> References: <92spv2$uim$1@nnrp1.deja.com> <3A5277BB.131C67B0@bellsouth.net> <92uqd0$l5c$1@nnrp1.deja.com> <3A533D3F.7439933A@lmco.com> <3A79E33C.9DF0D01F@gilson.com> <3A7A1097.83CDDDF5@bellsouth.net> Reply-To: rnorton@gilson.com Joe Durusau wrote: > There is a large up-front fee for the system, an > anual maintenance fee, and a per-CPU target license required. Thanks Joe! Not quite the answer I was hoping for, but it is better to know early. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: How to "clean up" special test procedures currently run through the shell? Date: 2 Feb 2001 10:04:12 -0500 From: vanbaren@falcon.si.com (vanbaren_gerald) Organization: Smiths Industries Message-ID: <3a7acc6c$1@news.si.com> References: <3A7A056E.16ABA773@west.raytheon.com> Christopher A Leddy writes: >I have a system with 5 unique custom PPC boards (each with ethernet >connection) running VxWorks 5.3.1. >When developing, the software engineers do the standard drill of >connecting target servers to the board, attaching shells and/or >debuggers, loading code, executing subroutines, changing variables, etc. >We have been using this same method to do special alignment and test of >various subsystems by loading special test routines, changing and >reading variables, and in some cases even loading special versions of >code. >All this works marginally in the engineering lab with 5 engineers >available to connect the servers and shells, run the scripts, and make >sense of the returns, but it won't fly when we go to production. I need >to automate the process, and provide a spiffy production GUI interface >on top of it, and run the whole thing under Windows NT (no heckles, I >have other applications to connect to, already running on NT). >So what are my options? >I do want to try to run the actual tactical code and add a few >alignment/test routines and change a few variables like we do now, >because I am afraid of code divergence as the software mutates. The >possibilities I have so far are: >1. Add support for a UDP connection into the target and develop a >protocol to trigger the test alignment routines and connect to the >target with a VC++ host with buttons etc. (Already did it on one >processor, painful and inflexible; I don't like this because I have to >do it 5 times on 5 different processors, and I make the tactical code >drag around all the garbage I add). > >2. Figure out how to direct connect into the WDB target agent and use a >VC++ host, again with buttons etc. This seems painful because, from the >docs, it looks like I need to figure out how to do an RPC client, and >then I have to parse out elf/dwarf symbol tables on the host etc. After >that, I need to do the WDB command dance using the locations I strip >off the symbol table. It would be so much cleaner on the user interface >side since I wouldn't need all the servers and shells floating around >and connected. (Connecting 5 target servers + 5 shells) == failure, >half the time). >3. Try to tie in some type of GUI TCL into tornado to control all the >shells and servers? I don't have a feel for how hard this is and whether >you can pass data into and out of the shell, or whether TCL will work >with 5 different shells open at once, etc. This method also requires >Tornado, in all its glory, and all the shells and servers successfully >connected; a painful and messy proposition. >Comments, Suggestions, Experiences? > Thanks, > Chris A missing option is "telnet" and the target shell, assuming your target system will support TCP and the target shell (the pair of them drags a lot of code into the system). If you have telnet enabled, you can telnet into the box and the console I/O (target shell) is redirected to the telnet session. Getting a telnet session running is fairly simple (open the telnet port and get past the option negotiation, assuming the vxWorks telnet server does negotiation). You can then send commands to the target shell using simple ASCII (sprintf()) and get responses. You can parse the responses with sscanf(): it is a very good friend if you invest the time in learning how to use the patterns to parse responses. It is a pretty powerful parser in its own right. gvb - -- +---------------------------------------------------------------------------+ | Jerry Van Baren / vanbaren_gerald@si.com / Grand Rapids Mi / 616-241-7973 | | My employer is a company. Companies are artifacts of a legal system. | |________________Artifacts are incapable of having opinions.________________| --------------------------- Newsgroups: comp.os.vxworks Subject: How to compare two vxWorks hex files (TSR 149779) Date: Fri, 2 Feb 2001 06:43:19 -0800 From: Jim Way Organization: Lawrence Berkeley National Laboratory Message-ID: <95eivf$80c$1@overload.lbl.gov> About a year ago, I had a question about how to compare two vxWorks HEX files. This was recorded by WindRiver as TSR 149779. The problem was that we use the "romCompress" option. Date/time stamp information is different between the two builds, causing the file compression algorithm to generate radically different results. After some time, I've finally determined a way to do the comparison that is convincing enough for me. (Now that I've seen it, it seems painfully obvious.) The solution is to go to the VxWorks tab in the project window, click on the Build name, call up the context menu, select Properties, then select the Rules tab. Where it says "vxWorks_romCompress" change the rule to "vxWorks_romResident". Apply that rule and click OK. Now rebuild-all the project (you might be able to just relink, but I'm paranoid). The resulting file will be huge because it's uncompressed. All that's left to do is to use your favorite file comparison tool (we use WinDiff and CodeWright) to compare the files. You should still observe a few bytes that are different. You can minimize these by trying to build both HEX files at about the same time on the same day. The only bytes that *should* be different are the date and time representations. Don't forget to change your rule back to "vxWorks_romCompress" when you're done. WindRiver closed TSR 149779 (over my objections). Here's hoping this explanation gets into their support database. Thanks again to everyone for their support and encouragement! Jim - ----------------------------------------- Jim Way, Software Engineer Datum Austin (Austron Inc.) voice: 512.721.4170 fax : 512.990.9712 email: jwayATdatumDOTcom (no spam please) - ----------------------------------------- --------------------------- Newsgroups: comp.os.vxworks Subject: tNetTask Timing Anomaly Date: Fri, 2 Feb 2001 15:03:24 GMT From: Lee DeRaud Organization: None to speak of Message-ID: Sender: nntp@news.boeing.com (Boeing NNTP News Access) We're running VxWorks 5.3 on a Radstone CPU46 (33MHz 68040), primary network interface is a Radstone FDDI-1, with this CPU acting as a proxy server to another (MVME177) processor. Nominal network traffic to/from this node consists mostly of broadcast UDP sockets (six or so at 16Hz, a few more at 1Hz or slower, split more-or-less evenly between input and output), plus some asynchronous XWindow stuff. There are two such nodes on the ring, plus three Solaris workstations and a couple of other MVME177 nodes. Everything works very well most of the time and pretty well almost all of the time, but... Occasionally (a few times per day), the network will basically lock up for 400-800ms during which either input and/or output packets simply stop moving. We're traced packets in/out and task switches (both via hook functions) and seen two typical manifestations of the symptoms: 1. Input blocks stop. Starting about 60ms later, no task lower in priority than tNetTask executes, i.e. it appears tNetTask is either process-bound or deadlocked. Normal operations resume 400-800ms later. 2. Input *and* output blocks stop. Concurrently, tNetTask stops executing (presumably blocked on something); a lower-priority X-client task is seen to execute occasionally, but I don't believe it is attempting any I/O during this interval. Normal operations resume 400-800ms later. These events are weakly correlated to high-load operations on the workstations, but we haven't been able to find a way to trigger them at will. Any and all ideas and suggestions would be welcome...thanks in advance. Lee --------------------------- Newsgroups: comp.os.vxworks Subject: Re: TaskDelay( ) bug? -- Probably not Date: Fri, 2 Feb 2001 11:56:39 -0500 From: "Jeff Creem" Organization: Posted via Supernews, http://www.supernews.com Message-ID: References: <3a791610$0$232@hades.is.co.za> <95ca2t$2v1$1@nnrp1.deja.com> Something else that can cause this behavior is any blocking calls in an ISR. (taskDelay, semWait, MailboxReads (with timeouts of course). When code does this I have seen both hung processors as well as ultra-bizarre behavior where certain tasks seem to get stuck in strange states but the rest of the system works fine. > > In article <3a791610$0$232@hades.is.co.za>, > "Randall" wrote: > > Hi, > > > > I am writing a small graphics application using Tornado. In order for > me to > > display > > the graphics, I have to initialise and de-initialise the UGL. I have > no > > problem doing > > this and the graphics works fine. > > > > A second part of the application requires me to implement a infinite > loop in > > some task. > > I use the TaskDelay() function to delay the task by a few > milliseconds. This > > also seems > > to work fine when I dont use the graphics. The problem is that as soon > as I > > initialise the UGL > > to display the graphics and load my device drivers, my code seems to > 'hang' > > in the TaskDelay() > > function. The task basically stops at the TaskDelay and never seems to > run > > again. The system indicates > > that the task is still alive and pending but it just wont execute > further. > > > > I have given the task a higher priority and this doesnt work. The task > > currently has a priority of 10, > > I have even gone so low as 3. Does anyone have any ideas or ever > experienced > > this problem? > > This only happens when I initiliase the UGL. > > > > Thanks in advance. > > Randall Michau > > rmichau@grintek.com > > > > > > > Sent via Deja.com > http://www.deja.com/ --------------------------- Newsgroups: comp.os.vxworks Subject: TSFS ioctl and FIOSYNC Date: Fri, 02 Feb 2001 12:15:05 -0500 From: Susan LoVerso Message-ID: <3A7AEB19.C7BF374@loverso.southborough.ma.us> I sent this message already to WRS support but haven't gotten a reply after nearly 2 weeks. I would think looking at source for a few minutes would immediately give the answer. I am hoping that someone here might know the answer. I have a question about the Target Server File System (running to a Solaris host). It states (on page 222 of the VxWorks 5.4 Programmer's Guide, Edition 1) that "TSFS provides all the I/O features that netDrv provides..." and on page 127 (same book) it states that netDrv provides all the same ioctl() functions as NFS devices EXCEPT for FIOSYNC and FIOREADDIR. The wdbTsfsDrv man page also do not list FIOSYNC as one of the supported ioctl() commands. However, in my program, I put a printf in after my ioctl() call (which runs to DosFS, NFS, etc) and when I run it to TSFS I get: ioctl FIOSYNC returned 0 That is, when run to TSFS, which claims not to support FIOSYNC, I do not get back ERROR. I know that the return value of ioctl is driver- dependent, and that for FIOSYNC, some drivers will return 0 on success, and some will return the number of bytes sync'ed on success. In all cases, however, all drivers (should) return ERROR when an error occurs. I would like to know if: 1. It is not returning ERROR, but a sync does NOT happen. I.e. it is a no-op? 2. It really is sync'ing and it is an undocumented/unsupported feature? 3. The documentation is remiss and it really is supported? 4. Something else I haven't thought of? Basically, I need to know if I can depend on the FIOSYNC completing in the absence of an ERROR return. Thanks, Sue LoVerso --------------------------- Newsgroups: comp.os.vxworks Subject: re: How Do I make an interrupt handling routine for the LCD on an assabet? Date: Fri, 2 Feb 2001 16:57:24 +0000 (GMT) From: David Laight Organization: Lawrence Berkeley National Laboratory Message-ID: <200102021657.QAA11828@dsl-2.tadpole.co.uk> > I have tried using the IP12 bit in ICPR but no interrupt are generated, any > ideas ? I failed to get the LCD error interrupts working on a rev E SA1100. This is basically the same controller as the SA1110. Intel support reported: It does appear that you have things set up properly. I have found other references to problems with the underrun interrupt. However, I haven't been able to find anything definitive. I have a query in to the chip designers to determine whether there is a problem with the interrupt, or whether there is something that we are missing. I will let you know as soon as I have a response. This was 12 months ago, haven't chased them for progress. David vxWorks - ---------------------------------------------------------------- David Laight email: dsl@tadpole.co.uk Tadpole Technology plc phone: +44 1223 428 232 Cambridge, UK fax: +44 1223 428 201 --------------------------- Newsgroups: comp.os.vxworks Subject: Re: null pointer on 8260 Date: Fri, 02 Feb 2001 17:39:35 GMT From: gold@sdsu.edu Organization: Deja.com Message-ID: <95ercl$9rg$1@nnrp1.deja.com> References: <95ecto$rq2$1@nnrp1.deja.com> In article <95ecto$rq2$1@nnrp1.deja.com>, pgrangier@my-deja.com wrote: > I work with a MPC8260. My memory map is designed in such a way that the > 60x SDRAM start at address 0x00000000. To prevent the use of > uninitialized pointers, I want protect the address 0x0. The problem is > that the Exceptions are vectored to the physical address 0x000n_nnnn. > Maybe it's possible to protect this address with MMU.... > > If someone has an idea, thanks in advance > Yes you can protect it with the MMU. There are pros and cons. You can do the following: vmBaseStateSet(0x0, 0x0, 0x1000, 0xc, 0x8) makes memory not writable, or cacheable. or vmBaseStateSet(0x0, 0x0, 0x1000, 0xc, 0x9) makes memory not writable You can check the man page on vmBaseStateSet to see the parameters. The positive side is you can protect memory starting at 0x0. The negative side is: The minimum you can protect is 1 MMU page, which is 4K (0x1000). That means everything from 0-0x1000 is write protected, so you better make sure you've done the vmBaseStateSet() after all the vectors have been initialized. And if you plan on changing vectors after the vmBaseStateSet() call, you better restore the writability of the page. If you make the page non-cacheable, that means things like the decrementer exception(0x900) are not cacheable anymore. Since that exception tends to happen quite often in vxWorks, making it non-cacheable is probably not a good thing :-) Of course all the above assumes you're using PTEs (in sysPhysMemDesc) for your SDRAM MMU, and not a single BAT. If you're using a BAT to cover your entire memory space, it can't be done. hth, Dan Gold gold@ensemblecom.com Sent via Deja.com http://www.deja.com/ --------------------------- Newsgroups: comp.os.vxworks Subject: Re: tools.h++ Date: Fri, 02 Feb 2001 11:46:34 -0600 From: Kevin Cline Organization: Mayan Networks Message-ID: <3A7AF27A.650FF57A@mayannetworks.com> References: <95d04c$ol7$1@nnrp1.deja.com> jayantadas@my-deja.com wrote: > Hi, > > Does any one know how can I have tools.h++ application component > configured into vxWorks? I guess it is INCLUDE_CPLUS_TOOLS. Is not this > component part of standard vxWorks? Are you porting existing code? If not, you may want to consider using the STL instead of tools.h++. --------------------------- Newsgroups: comp.os.vxworks Subject: re: Network problem on MVME2700 T1.0.1 Date: Fri, 2 Feb 2001 17:19:58 +0000 (GMT) From: David Laight Organization: Lawrence Berkeley National Laboratory Message-ID: <200102021719.RAA11849@dsl-2.tadpole.co.uk> > > Dear VxWorkers, dear David, > > David Laight gave some insight into ethernet negotiation aspects and > gave > a hint to problems in network traffic "... network to run 'like a sick > pig' ...". > > I suppose we're suffering from a problem like this for 1,5 years. > > We often loose target-server connection to the target MVME2700 > during a debug session. 5 times higher timeout didn't solve the problem > really. > Often FTP-transfers from target to host are very slowly (about 1 % of > maximum performance). > This almost only occurs during daytime if there's lot of traffic in our > network. Out of standard office times it's much better. The problem could be an inter-hub link set FDX somewhere, but... > The slowing down is in most cases 'healed' if we're rebooting the target. This makes it look much more like a vxworks problem - the system is somehow remembering 'state'. I would check the network data pools: - netStackDataPoolShow() and netPoolShow( end->pNetPool ) In case buffers are being lost..... Many vxWorks END drivers do not handle transmits that are flow controlled by the MAC layer properly. What the driver should do isn't documented, and the driver I copied didn't do the correct thing - the buffers that could not be transmitted were dumped forever. I recall a posting a while ago that explained whet the driver should be doing, however I had already fixed my driver a different way - so didn't test it. (My driver internally queues transmit packets until the ethernet card has space. I ought (maybe do?) dump the entire queue if we are failing to transmit at all.) David - ---------------------------------------------------------------- David Laight email: dsl@tadpole.co.uk Tadpole Technology plc phone: +44 1223 428 232 Cambridge, UK fax: +44 1223 428 201 --------------------------- Newsgroups: comp.os.vxworks Subject: Re: RPC UDP Sockets Problem Date: Fri, 02 Feb 2001 18:32:01 GMT From: John Fusco Organization: Deja.com Message-ID: <95euf1$cq0$1@nnrp1.deja.com> References: <3A7A8115.72EE874B@NOSPAM.bcs.org.uk> Could it be running out of file descriptors? In article <3A7A8115.72EE874B@NOSPAM.bcs.org.uk>, Graham Baxter wrote: > VxWorkers, > > We have a problem after connecting greater than 19 RPC connections. > When connecting our 20th connection we receive RPC 4 errors meaning we > can't receive. > Sent via Deja.com http://www.deja.com/ --------------------------- Newsgroups: comp.os.vxworks,comp.security.firewalls,comp.lang.c,alt.computer.security Subject: Random number generator Date: Fri, 02 Feb 2001 19:48:15 GMT From: milind_gunjal@my-deja.com Organization: Deja.com Message-ID: <95f2ts$hat$1@nnrp1.deja.com> Hi all, Couls someone suggest me a good 64-bit pseudo random number generation algorithm and/or code. Any help is greatly appreciated. Thanks, Milind Sent via Deja.com http://www.deja.com/ --------------------------- Newsgroups: comp.os.vxworks Subject: Re: How to "clean up" special test procedures currently run through the shell? Date: Fri, 02 Feb 2001 12:11:42 -0800 From: Christopher A Leddy Organization: Raytheon Company Message-ID: <3A7B147E.16002866@west.raytheon.com> References: <3A7A056E.16ABA773@west.raytheon.com> <3a7acc6c$1@news.si.com> vanbaren_gerald wrote: > > Christopher A Leddy writes: > > >I have a system with 5 unique custom PPC boards (each with ethernet > >connection) running VxWorks 5.3.1. > > >When developing, the software engineers do the standard drill of > >connecting target servers to the board, attaching shells and/or > >debuggers, loading code, executing subroutines, changing variables, etc. > >We have been using this same method to do special alignment and test of > >various subsystems by loading special test routines, changing and > >reading variables, and in some cases even loading special versions of > >code. > > >All this works marginally in the engineering lab with 5 engineers > >available to connect the servers and shells, run the scripts, and make > >sense of the returns, but it won't fly when we go to production. I need > >to automate the process, and provide a spiffy production GUI interface > >on top of it, and run the whole thing under Windows NT (no heckles, I > >have other applications to connect to, already running on NT). > > >So what are my options? > > >I do want to try to run the actual tactical code and add a few > >alignment/test routines and change a few variables like we do now, > >because I am afraid of code divergence as the software mutates. The > >possibilities I have so far are: > > >1. Add support for a UDP connection into the target and develop a > >protocol to trigger the test alignment routines and connect to the > >target with a VC++ host with buttons etc. (Already did it on one > >processor, painful and inflexible; I don't like this because I have to > >do it 5 times on 5 different processors, and I make the tactical code > >drag around all the garbage I add). > > > >2. Figure out how to direct connect into the WDB target agent and use a > >VC++ host, again with buttons etc. This seems painful because, from the > >docs, it looks like I need to figure out how to do an RPC client, and > >then I have to parse out elf/dwarf symbol tables on the host etc. After > >that, I need to do the WDB command dance using the locations I strip > >off the symbol table. It would be so much cleaner on the user interface > >side since I wouldn't need all the servers and shells floating around > >and connected. (Connecting 5 target servers + 5 shells) == failure, > >half the time). > > >3. Try to tie in some type of GUI TCL into tornado to control all the > >shells and servers? I don't have a feel for how hard this is and whether > >you can pass data into and out of the shell, or whether TCL will work > >with 5 different shells open at once, etc. This method also requires > >Tornado, in all its glory, and all the shells and servers successfully > >connected; a painful and messy proposition. > > >Comments, Suggestions, Experiences? > > > Thanks, > > > Chris > > A missing option is "telnet" and the target shell, assuming your target > system will support TCP and the target shell (the pair of them drags > a lot of code into the system). If you have telnet enabled, you can > telnet into the box and the console I/O (target shell) is redirected to > the telnet session. > > Getting a telnet session running is fairly simple (open the telnet > port and get past the option negotiation, assuming the vxWorks telnet > server does negotiation). You can then send commands to the target shell > using simple ASCII (sprintf()) and get responses. You can parse the > responses with sscanf(): it is a very good friend if you invest the time > in learning how to use the patterns to parse responses. It is a pretty > powerful parser in its own right. > > gvb > > -- > +---------------------------------------------------------------------------+ > | Jerry Van Baren / vanbaren_gerald@si.com / Grand Rapids Mi / 616-241-7973 | > | My employer is a company. Companies are artifacts of a legal system. | > |________________Artifacts are incapable of having opinions.________________| A good suggestion. I don't think my boards have enough room to support telnet and the application code (one has telnet support and the others don't but I don't know why). I will look into it... One other question: Does the target's stdout console output get channeled through the telnet connection also? These boards babble status out using printf's and I am wondering if all that nonsense will get jumbled into the shell output without some modification to the tactical code. Thanks for the response, Chris --------------------------- Newsgroups: comp.os.vxworks Subject: Update to vxWorks DHCP Date: Fri, 2 Feb 2001 11:36:04 -0800 From: "Crozier, Bruce" Organization: Lawrence Berkeley National Laboratory Message-ID: <95f4hv$i4l$1@overload.lbl.gov> Hello Luke, I am currently working on DHCP, and your email to the NewsGroup caught my eye. I would appreciate it if you would tell me what target architecture & Tornado version you were working with, some of the major DHCP errors you found, and maybe the contact at WRS support from whom you got the update. My target is PPC (750 processor), and WRS support in Alameda told me that they had no idea what you might be referring to. I would appreciate any assistance you can provide. Thanks, Bruce Crozier BTW, my attempt to email you directly at Diamand@btinternet.com failed for some reason (unknown local-part "diamand" in domain "btinternet.com") - ---------------------------------------------------------------------------- - --------- Newsgroups: comp.os.vxworks Subject: Re: DHCP Bind failed Date: 29 Jan 2001 22:53:12 GMT From: Diamand@btinternet.com (Luke Diamand) Organization: UKLINUX NET Message-ID: References: <3A64421F.36327B85@temex.fr> Delsol Thomas(chez Tekelec Temex Telecom) wrote: We found the DHCP client in T2 to be hopelessly broken. We complained to support and got a new one which worked like a dream. HTH! Luke Diamand > > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: generated dependencies Date: Fri, 2 Feb 2001 21:31:59 +0100 From: "Martin Raabe" Organization: T-Online Message-ID: <95f5ld$6pq$01$1@news.t-online.com> References: <3A7A783B.4AE76F03@draeger.com> Hello Joern, I recommend to use SNiFF+ in your project, I know, you have SNiFF+ inhouse. The integration of SNiFF+ 3.2.1 for TornadoII is available via ftp at ftp://ftp.takefive.co.at/pub/SNiFF/integrations/TornadoII You find here the V2.0 of the integration, which is out since June 2000. This is a reliable solution, you might look for! Speaking for my own, not for my employer! Ciao - -- Martin Raabe Joern Tietjen schrieb in im Newsbeitrag: 3A7A783B.4AE76F03@draeger.com... > Hi, > > is anybody using the dependencies generated thru T2 (Win NT) ? > To me it seems that this feature is highly unreliable. > Dependencies get updated only occasionally after you requested it. > Sometimes they use invalid pathnames while filenames are correct. > Is there a known rule behind this ? > Is it a problem of accessing files via NFS or Samba ? > > TIA, > > Joern --------------------------- Newsgroups: comp.arch.embedded,comp.realtime,comp.os.linux.embedded,comp.os.vxworks,comp.os.psos Subject: ANNOUNCE: New Article on Virtual Serial Ports Date: Fri, 02 Feb 2001 20:57:06 GMT From: Michael Barr Organization: Netrino, LLC (http://www.netrino.com) Message-ID: <3A7B1F22.B105ED3@netrino.com> I have just posted the article "Virtual Serial Ports" to the Netrino website. It can be found at the following URL: http://www.netrino.com/Connecting/2000-02/ Serial ports are extremely versatile peripherals. They can be used for communication with external equipment, to provide low-cost network connectivity, for data-logging purposes, and in myriad other ways. As a result, it is not uncommon to find several serial ports on each piece of embedded hardware that you may work with. But what if you're faced with a situation where you need a serial port but there are none or all of them are already being used for other purposes? In this article, I show how to create virtual serial ports from other hardware that you may already have at your disposal. This article was published in the February 2000 issue of Embedded Systems Programming. For more information about the magazine, or to sign up for a free subscription, please visit: http://www.embedded.com/ The Netrino Consultants Network encourages its partners and associates to share their technical expertise by writing magazine articles and books and by speaking at industry conferences. Whenever possible we retain the right to publish these materials on our website. An complete index of our publications can be found at the following URL: http://www.netrino.com/Publications/ I hope you will find these resources valuable, and relevant to these newsgroups. Sincerely, Michael Barr --------------------------- Newsgroups: comp.arch.embedded,comp.lang.c,comp.realtime,comp.dsp,comp.os.vxworks Subject: ANNOUNCE: New Article on Language Selection Date: Fri, 02 Feb 2001 20:59:33 GMT From: Michael Barr Organization: Netrino, LLC (http://www.netrino.com) Message-ID: <3A7B1FB5.E9306E25@netrino.com> I have just posted the article "Language Lessons" to the Netrino website. It can be found at the following URL: http://www.netrino.com/Connecting/2000-03/ Have you chosen the best programming language for your current embedded project? Your answer may very well depend on how the phrase "best programming language" is defined. Perhaps you've chosen the language that produces the most efficient code for your particular processor; or maybe you've selected the one that will allow you to finish the coding most quickly or with fewer bugs. Then again, maybe you--like so many of us--are just using the same language you always have. This article was first published in the March 2000 issue of Embedded Systems Programming. For more information about the magazine, or to sign up for a free subscription, please visit: http://www.embedded.com/ The Netrino Consultants Network encourages its partners and associates to share their technical expertise by writing magazine articles and books and by speaking at industry conferences. Whenever possible we retain the right to publish these materials on our website. An complete index of our publications can be found at the following URL: http://www.netrino.com/Publications/ I hope you will find these resources valuable, and relevant to these newsgroups. Sincerely, Michael Barr --------------------------- Newsgroups: comp.arch.embedded,comp.realtime,comp.os.vxworks,comp.os.linux.embedded,comp.os.psos Subject: ANNOUNCE: New Articles on TCP/IP, UDP/IP, and ARP Date: Fri, 02 Feb 2001 21:09:52 GMT From: Michael Barr Organization: Netrino, LLC (http://www.netrino.com) Message-ID: <3A7B2220.9ADBB797@netrino.com> I have just posted a series of articles on UDP, IP, ARP, and other TCP/IP protocols to the Netrino website. Collectively, these articles describe an approach to selective inclusion of only those protocols within the TCP/IP suite that are positively necessary. The result of following this approach can be a network-connected device with the minimum amount of code dedicated to the protocol stack. I regret that I have not had more time to spend on the code itself. I had hoped to release a portable (to any RTOS) open source UDP/IP stack for embedded/real-time systems. However, I have been too busy with other responsibilities and projects to take the thread this far. Here is the URL for the first article in the series: http://www.netrino.com/Connecting/2000-04/ Simply follow the "Next" links to follow the thread to its completion. Sorry for the tangents in the May and August installments. Feel free to skip over those quickly. These article were published from April to September 2000 in Embedded Systems Programming. For more information about the magazine, or to sign up for a free subscription, please visit: http://www.embedded.com/ The Netrino Consultants Network encourages its partners and associates to share their technical expertise by writing magazine articles and books and by speaking at industry conferences. Whenever possible we retain the right to publish these materials on our website. An complete index of our publications can be found at the following URL: http://www.netrino.com/Publications/ I hope you will find these resources valuable, and relevant to these newsgroups. Sincerely, Michael Barr --------------------------- Newsgroups: comp.os.vxworks,comp.security.firewalls,comp.lang.c,alt.computer.security Subject: Re: Random number generator Date: Fri, 2 Feb 2001 13:16:58 -0800 From: "Curtis Gillespie" Message-ID: References: <95f2ts$hat$1@nnrp1.deja.com> Reply-To: "Curtis Gillespie" Greetings, I came across a random number generator algorithm in one of my programming classes called the NSTT Algorithm. /* Function using the NSTT Algorithm for providing a pseudo- * random integer between low and high integers passed to it. */ /* NOTE: Remember to "seed the rand() prior to calling this function . */ int randomInteger(int nLow, int nHigh) { float fRandomFloat; /* Normalize and scale */ fRandomFloat = rand() / ( (float) RAND_MAX + 1) * (nHigh - nLow + 1); /* Truncate, translate, and return */ return ( (int) fRandomFloat + nLow); } I don't know specifically about 64-bit relating to this function. However, I thought someone may be interested in what worked for me. Regards, Curtis Gillespie wrote in message news:95f2ts$hat$1@nnrp1.deja.com... > Hi all, > > Couls someone suggest me a good 64-bit pseudo random number generation > algorithm and/or code. > > Any help is greatly appreciated. > > Thanks, > Milind > > > Sent via Deja.com > http://www.deja.com/ --------------------------- Newsgroups: comp.arch.embedded,comp.realtime,comp.os.linux.embedded,comp.os.vxworks,comp.os.psos Subject: ANNOUNCE: New Article on Software Add-ons Date: Fri, 02 Feb 2001 21:47:35 GMT From: Michael Barr Organization: Netrino, LLC (http://www.netrino.com) Message-ID: <3A7B2AF6.D56B1A49@netrino.com> I have just posted the article "Architecting Embedded Systems for Add-on Software" to the Netrino website. It can be found at the following URL: http://www.netrino.com/Articles/AddOnSoftware/ Software upgradability and extensibility are now considered important features for a wide range of embedded systems, particularly those on a network. This may mean anything from a complete replacement of the memory image to an add-on software module developed by a third party. This article provides an overview of the key hardware and software design issues that you must consider to make upgradability and extensibility possible. This article was first published in the September 1999 issue of Embedded Systems Programming. For more information about the magazine, or to sign up for a free subscription, please visit: http://www.embedded.com/ The Netrino Consultants Network encourages its partners and associates to share their technical expertise by writing magazine articles and books and by speaking at industry conferences. Whenever possible we retain the right to publish these materials on our website. An complete index of our publications can be found at the following URL: http://www.netrino.com/Publications/ I hope you will find these resources valuable, and relevant to these newsgroups. Sincerely, Michael Barr --------------------------- Newsgroups: comp.os.vxworks,comp.security.firewalls,comp.lang.c,alt.computer.security Subject: Re: Random number generator Date: Fri, 02 Feb 2001 21:38:17 GMT From: rawcswi@my-deja.com Organization: Deja.com Message-ID: <95f9c8$ngp$1@nnrp1.deja.com> References: <95f2ts$hat$1@nnrp1.deja.com> In article <95f2ts$hat$1@nnrp1.deja.com>, milind_gunjal@my-deja.com wrote: > > Couls someone suggest me a good 64-bit pseudo random number generation > algorithm and/or code. Try http://www.math.keio.ac.jp/matumoto/emt.html - -- MJSR Sent via Deja.com http://www.deja.com/ --------------------------- Newsgroups: comp.os.vxworks Subject: Re: generated dependencies Date: Fri, 02 Feb 2001 23:07:51 GMT From: wv12@my-deja.com Organization: Deja.com Message-ID: <95fejv$si8$1@nnrp1.deja.com> References: <3A7A783B.4AE76F03@draeger.com> Dependencies are generated by calling the compiler with the -M flag, and feeding it with path(s) to where the headers are. If you want to recreate the dependencies, simply do a "make clean" in the build directory. If you don't get the right path, try adding - -I $(TGT_DIR)/path_to_your_includes to CC_INCLUDE (to your rules.bsp) I don't know how do this using the IDE. I avoid IDE's like the plague. Will In article <3A7A783B.4AE76F03@draeger.com>, Joern Tietjen wrote: > Hi, > > is anybody using the dependencies generated thru T2 (Win NT) ? > To me it seems that this feature is highly unreliable. > Dependencies get updated only occasionally after you requested it. > Sometimes they use invalid pathnames while filenames are correct. > Is there a known rule behind this ? > Is it a problem of accessing files via NFS or Samba ? > > TIA, > > Joern > Sent via Deja.com http://www.deja.com/ --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Update to vxWorks DHCP Date: Sat, 03 Feb 2001 00:25:15 GMT From: john_94501@my-deja.com Organization: Deja.com Message-ID: <95fj55$8h$1@nnrp1.deja.com> References: <95f4hv$i4l$1@overload.lbl.gov> Hello Bruce, Just out of interest, what configuration are you using DHCP in, and what is happening? For example, are you using it in a VxWorks bootrom to boot your system? Do you then use it in the kernel as well to configure the network port? Or are you using a ROM'd version of the kernel, and trying to use DHCP to get a network interface configured for that? With more information about what you're up to, and what isn't working, perhaps others can provide pointers too... Juts a thought, John... In article <95f4hv$i4l$1@overload.lbl.gov>, "Crozier, Bruce" wrote: > > Hello Luke, > > I am currently working on DHCP, and your email to the NewsGroup caught my > eye. > I would appreciate it if you would tell me what target architecture & > Tornado version > you were working with, some of the major DHCP errors you found, and maybe > the > contact at WRS support from whom you got the update. > > My target is PPC (750 processor), and WRS support in Alameda told me that > they had > no idea what you might be referring to. > > I would appreciate any assistance you can provide. > > Thanks, > Bruce Crozier > > BTW, my attempt to email you directly at Diamand@btinternet.com failed > for > some reason (unknown local-part "diamand" in domain "btinternet.com") > > - ------------------------------------------------------------------------ - ---- > --------- > Newsgroups: comp.os.vxworks > Subject: Re: DHCP Bind failed > Date: 29 Jan 2001 22:53:12 GMT > From: Diamand@btinternet.com (Luke Diamand) > Organization: UKLINUX NET > Message-ID: > References: <3A64421F.36327B85@temex.fr> > > Delsol Thomas(chez Tekelec Temex Telecom) wrote: > > We found the DHCP client in T2 to be hopelessly broken. We complained to > support > and got a new one which worked like a dream. > > HTH! > Luke Diamand > > > > > Sent via Deja.com http://www.deja.com/ --------------------------- Newsgroups: comp.os.vxworks Subject: vxworks and IBM's microdrive Date: Sat, 03 Feb 2001 05:37:23 GMT From: jagan Organization: Road Runner Message-ID: <3A7BC2F8.91B80142@mediaone.net> Hi all, Does anyone have experience with using the IBM microdrive with vxworks? I would like to know what the experience was like and if there is any software available for it. thanks, - -jagan. --------------------------- Newsgroups: comp.os.vxworks Subject: about zbuf Date: Sat, 03 Feb 2001 08:27:36 GMT From: pavankumar2311@my-deja.com Organization: Deja.com Message-ID: <95gfdl$lub$1@nnrp1.deja.com> Hi, I am doing a time critical project in vxworks. I am using zbuf. I want to know the time complexity to operate on zbuf. If using zbuf is costly (in terms of time), is there any other way? thanks in advance, pavan. Sent via Deja.com http://www.deja.com/ --------------------------- End of New-News digest ********************** From vxwexplo-errs@csg.lbl.gov Sun Feb 4 04:03:07 2001 From: Vxworks Exploder Date: Sun Feb 4 04:03:09 PST 2001 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Sun Feb 4 04:03:04 PST 2001 Subject: MPC823 Rev A --> Rev B problem Subject: Simple Tornado WTX programming Subject: Re: GDBARM and JEENI embeddedICE, problems, help! Subject: Re: target "forwards" all packets sent to it ??? Subject: Re: pmc750 works? ------------------------------------------------------- Newsgroups: comp.os.vxworks,comp.sys.powerpc.tech Subject: MPC823 Rev A --> Rev B problem Date: Sat, 3 Feb 2001 11:03:14 -0700 From: "Amy Michaelis" Organization: Dimensional Communications Message-ID: Help! Motorola has recently discontinued the Rev A silicon for the 823, and now we can only get the Rev B silicon. This should all be seamless of course, but after some testing with the Rev B chip, we've found a problem. The Rev B chip causes random machine checks in our system. We're using vxWorks as an operating system, and a check with Wind River didn't turn up anything. I've been trying to find out what's going on here, but so far no luck. We're using quite a bit of CPM, and I tend to believe that's were the problem lies. We have the SCC in Ethernet mode, and the SPI and I2C are also being used. These two guys have been relocated using the downloaded microcode patch. My sense is that since Motorola cleaned up a lot of the problems in the I2C driver for the Rev B silicon, something has gone wrong in the downloaded microcode. The problem could also be the new interrupt for SCC3, but we're not using the new SCC on the rev B chip, so I don't know how that would be a problem. Any ideas would be greatly appreciated. Thanks, Matt Michaelis amich@dimensional.com --------------------------- Newsgroups: comp.os.vxworks Subject: Simple Tornado WTX programming Date: Sat, 3 Feb 2001 21:14:52 +0100 From: "Jørgen Jakobsen" Organization: Posted Courtesy of Tele Danmark or one of its European Subsidiaries Message-ID: <95hp08$kl1$1@news.inet.tele.dk> Dear VxWork'ers I'm trying to make a simple tool which monitors the value of selected variables and structures on my target by accessing it through my targetserver utilising a simple WTX based C program. It is basicly my own implementation of the watch window in the Tornado debugger ;) I have written a small piece of code which attaches to the targetserver, looks up symbols in the symboltable and gets the address.using wtxSymListGet(), and I can also read memory from the target using wtxMemRead(). My question is about how I know the type of the variable to probe, the size of memory to read from target, and how I interpret the memory read ? I guess I will then have to cast the block of memory read to the actual type to display the value properly ? Best regards Jorgen --------------------------- Newsgroups: comp.os.vxworks Subject: Re: GDBARM and JEENI embeddedICE, problems, help! Date: 4 Feb 2001 00:50:32 GMT From: Diamand@btinternet.com (Luke Diamand) Organization: UKLINUX NET Message-ID: References: We get pretty good results from canonical gdb from the FSF. Download, configure and make. This is for arm. HTH! Luke Diamand On Sun, 14 Jan 2001 17:28:46 GMT, george najarian wrote: >i am running into the situation where it appears that gdbarm >is generating a null pointer exception because of input >received by the jeeni.dll target. Tech support at Embedded Performance >told me that the dll is generating the correct output, and that >the trap is being generated in gdbarm. > >I am running Win2000. > >Does anybody have any suggestions on how to pursue this issue? I >know that gdbarm is public domain, but doesn't Windriver put in >custom hooks? Where can I get the source and load it into WinIce to debug? >Does Windriver support gdb issues? > >I am very hesitant to contact Windriver again until i have tried everything >i can because of the very arrogant and condescending treatment i received the >last time i talked to them. Any and all comments and suggestions will be >gratefully accepted. > >George Najarian >Principal Engineer >Conet Communications, Inc. - -- --------------------------- Newsgroups: comp.os.vxworks Subject: Re: target "forwards" all packets sent to it ??? Date: 4 Feb 2001 00:50:33 GMT From: Diamand@btinternet.com (Luke Diamand) Organization: UKLINUX NET Message-ID: References: If you've written the ethernet driver yourself, have you checked that the ip portion of the packets (start of ethernet + 14) is word aligned? If it isn't, the IP stack will fail to recognise the packets, and (IIRC) forward them. HTH! Luke Diamand On Mon, 22 Jan 2001 16:09:19 GMT, Stuart J Adams wrote: >We have written a BSP for an SA1110 strongARM board >and are seeing a weird problem with the ethernet >and tcp/ip (Tornado 2.0.1 for ARM) > >For some reason the tcp/ip stack >forwards packets that are sent to the target back >out the same ethernet connection. For example if I try >to ping the target (say 192.168.1.50), the target >receives the packet but for some reasons does not >process it but "forwards" the packet back out the >ethernet connect. The target does respond to arp >requests okay. I think this must be some weird >configuration problem where the target does not >realize that it is 192.168.1.50. > >(using etherInputHookAdd we are able to print all > incoming packets and they look fine. Also, the > network analyzer shows that packets are transmitted > fine by the target too) > >Any ideas ??? > >Thanks, > Stuart > - -- --------------------------- Newsgroups: comp.os.vxworks Subject: Re: pmc750 works? Date: Sun, 04 Feb 2001 11:25:46 GMT From: Bill Dennen Organization: Motorola Computer Group Message-ID: <3A7D3826.1A53D65B@home.com> References: Reply-To: wdennen@home.com Kelly Hornsby wrote: > > Hello all > > I have a PrPMC750 from Motorola that I want to make work with a MVM2603 > system (short term). Has anyone performed any integration with a setup like > this? > > Any information, or similiar experience, will be appreciated. You'll need to obtain the PPMC750 Slave Mode patch for the PrPMC750 BSP from your MCG support representative. This patch only works with the PPMC750 1.2/0 BSP released for Tornado 2. Regards - -- >@< Bill Dennen william.dennen@motorola.com Motorola Computer Group, TOPAZ --------------------------- End of New-News digest ********************** From vxwexplo-errs@csg.lbl.gov Mon Feb 5 02:29:12 2001 From: "R.vander.Meij" Date: Mon Feb 5 02:29:14 PST 2001 Subject: VxWorks nfs setup and rights. 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_01C08F5E.76E830C0 Content-Type: text/plain; charset="iso-8859-1" Help, i am getting desperate! For the Y2K we upgraded from MVME147 to MVME167 boards. We have a Silicon Graphics indigo station connected to the 167 board. For some reason the nfs does not work anymore. mounting a directory on the MVME167 from the SGI. According to me all the rights are setup correct. The man pages are very short on our machine. The nfs devices show up and the exported directories are visible with the devs command. From other machines i can mount on the SGI. I can not read or write to the mounted directories and with the ls command i get an error message. So what can go wrong and can sombody send me a more detailed descripton of the nfs and net commands. Richard van der Meij r.vander.meij@mst.schreiner-fsc.com ------_=_NextPart_001_01C08F5E.76E830C0 Content-Type: text/html; charset="iso-8859-1" VxWorks nfs setup and rights.

Help, i am getting desperate!

For the Y2K we upgraded from MVME147 to MVME167 boards.
We have a Silicon Graphics indigo station connected to the 167 board.
For some reason the nfs does not work anymore. mounting a directory on the MVME167 from the SGI.
According to me all the rights are setup correct. The man pages are very short on our machine.
The nfs devices show up and the exported directories are visible with the devs command.
From other machines i can mount on the SGI.
I can not read or write to the mounted directories and with the ls command i get an error message.
 So what can go wrong and can sombody send me a more detailed descripton
of the nfs and net commands.

Richard van der Meij
r.vander.meij@mst.schreiner-fsc.com



------_=_NextPart_001_01C08F5E.76E830C0-- From vxwexplo-errs@csg.lbl.gov Mon Feb 5 04:03:07 2001 From: Vxworks Exploder Date: Mon Feb 5 04:03:10 PST 2001 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Mon Feb 5 04:03:04 PST 2001 Subject: Re: MPC823 Rev A --> Rev B problem Subject: Re: tt with c++ that shows passed parameters? Subject: memPartCreate/memPartAlloc Subject: fatal WTX error 0x10136, 0x1012f Subject: How do I set up an interrupthandler for the 'User Software definable switches' on assabet... Subject: RTP & vxWorks Subject: VxWorks nfs setup and rights. Subject: Re: Fibre Channel to Raid Speed Subject: Gcc 2.95.3 for VxWorks-PPC available.... Subject: Re: Spawing C++ task Subject: Re: Releasing bound socket Subject: Re: memPartCreate/memPartAlloc ------------------------------------------------------- Newsgroups: comp.os.vxworks,comp.sys.powerpc.tech Subject: Re: MPC823 Rev A --> Rev B problem Date: Sun, 04 Feb 2001 18:26:37 GMT From: john_94501@my-deja.com Organization: Deja.com Message-ID: <95k6so$7i7$1@nnrp1.deja.com> References: Hello, Have you tried checking the Motorola errata for the new device? It can be found here: http://e-www.motorola.com/collateral/823ERRATA.html There are a number of different documents there (all PDF); each covers a different set of versions of the silicon so you'll need to check which version of the 'B' revision you have... HTH, John... In article , "Amy Michaelis" wrote: > Help! > > Motorola has recently discontinued the Rev A silicon for the 823, and now we > can only get the Rev B silicon. This should all be seamless of course, but > after some testing with the Rev B chip, we've found a problem. The Rev B > chip causes random machine checks in our system. > > We're using vxWorks as an operating system, and a check with Wind River > didn't turn up anything. I've been trying to find out what's going on here, > but so far no luck. We're using quite a bit of CPM, and I tend to believe > that's were the problem lies. We have the SCC in Ethernet mode, and the SPI > and I2C are also being used. These two guys have been relocated using the > downloaded microcode patch. My sense is that since Motorola cleaned up a > lot of the problems in the I2C driver for the Rev B silicon, something has > gone wrong in the downloaded microcode. > > The problem could also be the new interrupt for SCC3, but we're not using > the new SCC on the rev B chip, so I don't know how that would be a problem. > > Any ideas would be greatly appreciated. > > Thanks, > > Matt Michaelis > amich@dimensional.com > > Sent via Deja.com http://www.deja.com/ --------------------------- Newsgroups: comp.os.vxworks Subject: Re: tt with c++ that shows passed parameters? Date: Mon, 05 Feb 2001 00:11:10 GMT From: Bob Witmer Organization: Deja.com Message-ID: <95kr2s$l4i$1@nnrp1.deja.com> References: <93nei4$mbj$1@nnrp1.deja.com> <8%hd6.567$i54.854034@newsr1.u-net.net> <95a5ul$9bk$1@nnrp1.deja.com> <3A79BAB6.3F5195EC@aps.anl.gov> Yeah, we are on a PPC604. Thanks for the insight everybody. In article <3A79BAB6.3F5195EC@aps.anl.gov>, Andrew Johnson wrote: > Dave Korn wrote: > > > > Bob Witmer wrote in message <95a5ul$9bk$1@nnrp1.deja.com>... > > >I saw some old posts in the vxWorks archive which showed the argument > > >lists with parameters. I think > ... > > I could be remembering wrong but I think that the difference is whether > > you're using host shell or target shell. I think it's the host shell that > > gives that kind of detail, so you'd need a target server > > running on the host and a target (Wdb) agent running on the target. > > I suspect it's actually associated with the target CPU family that you're > using, because I've seen both but never use the host shell. From my > experience on 68k family CPUs the traces show parameters, but PowerPC CPUs > don't. I've never worked out what the square brackets in the 68k output > were for - anybody know? > > - Andrew > -- > Every great idea appears crazy to start with. > - -- rwitmer at my-deja dot com Sent via Deja.com http://www.deja.com/ --------------------------- Newsgroups: comp.os.vxworks Subject: memPartCreate/memPartAlloc Date: Mon, 5 Feb 2001 00:39:34 -0500 From: "Normand Barriere" Message-ID: Hi All, I'm currently developing a networking application, that requires multiple memory partition. I've tried to use the functions memPartCreate and memPartAlloc. The memPartCreate function appears to work, but as soon as I try to alloc memory in this pool, I always get an error. Memory cannot be allocated. I'm new to this OS, so is there anybody that can help me in setting up those functions correctly. This is the method that I've used to set up my system : mem = malloc( 1024 ) partid = memPartCreate( mem, 1000 ); ..... later on in the code: memptr = memPartAlloc( partid, 64 ); /* result of this operation is a failure */ Is there anything wrong in performing these operations ? If so, can anybody explain me the procedure that should be used in creating/managing memory partitions ? Thanks Normand --------------------------- Newsgroups: comp.os.vxworks Subject: fatal WTX error 0x10136, 0x1012f Date: Mon, 5 Feb 2001 07:53:18 +0100 From: "Markus MItterer" Organization: KPNQwest customer news service Message-ID: Hello VxWorkers, My system configuration looks like this: - - a LAN network with several WindowsNT machines, one of them is my host - - in this host I have 2 network cards, one for the LAN and the other for the peer-to-peer connection to my target I was the only one working with this target, but now I want to share it with somebody else. To do so I configured the target using my host as gateway and set the second NT-machine as host. The connection seems to work, but after a few minutes I get the error message "fatal WTX error 0x10136, 0x1012f" and no synchronisation works. I already set the tornado registry to run on a remote host (second NT ip-address). May this problem be caused of slow network-speed? (but I use 100Mbit network interfaces) We bought 2 licenses from wrs - may this be a problem with the tornado registry (but a direcly connected target to the second host is working properly) Any ideas? regards - -- Markus Mitterer Software-Development, STK mailto: markus_mitterer@sbu1.storkgroup.com --------------------------- Newsgroups: comp.os.vxworks Subject: How do I set up an interrupthandler for the 'User Software definable switches' on assabet... Date: Mon, 5 Feb 2001 10:13:26 +0100 From: "Carsten Schwartz" Organization: Posted Courtesy of Tele Danmark or one of its European Subsidiaries Message-ID: <95lqt9$k20$1@news.inet.tele.dk> Reply-To: "Carsten Schwartz" I can't find any clear documentation in the Development Board User's guide on this subject. Which is te vector i need to use in intConnect?. --------------------------- Newsgroups: comp.os.vxworks Subject: RTP & vxWorks Date: Mon, 5 Feb 2001 12:00:07 +0100 From: "sarefrech" Organization: Guest of France Telecom Oleane's newsreading service Message-ID: <95m13m$iub$1@s1.read.news.oleane.net> References: <3A7BC2F8.91B80142@mediaone.net> Reply-To: "sarefrech" Hi, I'm considering using vxworks to do audio and video streaming over IP. As far as I know vxworks does not include natively the support for RTP/RTCP. Does anyboby have experience about off-the-shelf products or custom implementations of RTP on vxWorks? Thanks sarefrech --------------------------- Newsgroups: comp.os.vxworks Subject: VxWorks nfs setup and rights. Date: Mon, 5 Feb 2001 11:29:05 +0100 From: "R.vander.Meij" Organization: Lawrence Berkeley National Laboratory Message-ID: <95m21f$hum$1@overload.lbl.gov> 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_01C08F5E.76E830C0 Content-Type: text/plain; charset="iso-8859-1" Help, i am getting desperate! For the Y2K we upgraded from MVME147 to MVME167 boards. We have a Silicon Graphics indigo station connected to the 167 board. For some reason the nfs does not work anymore. mounting a directory on the MVME167 from the SGI. According to me all the rights are setup correct. The man pages are very short on our machine. The nfs devices show up and the exported directories are visible with the devs command. From other machines i can mount on the SGI. I can not read or write to the mounted directories and with the ls command i get an error message. So what can go wrong and can sombody send me a more detailed descripton of the nfs and net commands. Richard van der Meij r.vander.meij@mst.schreiner-fsc.com - ------_=_NextPart_001_01C08F5E.76E830C0 Content-Type: text/html; charset="iso-8859-1" VxWorks nfs setup and rights.

Help, i am getting desperate!

For the Y2K we upgraded from MVME147 to MVME167 boards.
We have a Silicon Graphics indigo station connected to the 167 board.
For some reason the nfs does not work anymore. mounting a directory on the MVME167 from the SGI.
According to me all the rights are setup correct. The man pages are very short on our machine.
The nfs devices show up and the exported directories are visible with the devs command.
From other machines i can mount on the SGI.
I can not read or write to the mounted directories and with the ls command i get an error message.
 So what can go wrong and can sombody send me a more detailed descripton
of the nfs and net commands.

Richard van der Meij
r.vander.meij@mst.schreiner-fsc.com



- ------_=_NextPart_001_01C08F5E.76E830C0-- --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Fibre Channel to Raid Speed Date: Mon, 5 Feb 2001 11:02:42 -0000 From: "Dave Korn" Organization: Lumber Cartel (tinlc) Members #2234-2237 (owing to browser refresh) Message-ID: References: <3A6F77F3.FAB5284B@erols.com> <3A70CC93.3444D4EB@west.raytheon.com> "Christopher A Leddy" wrote in message news:3A70CC93.3444D4EB@west.raytheon.com... > Fred Walther wrote: > > > > Does anybody have any experience with Synergy VGM5 PPC and Fibre channel > > PMC? > > I am using DosFs and I am getting only 30 MB/Sec Read and 20 MBytes/Sec > > Write to the raid disks. Has anybody gotten better speeds with VxWorks > > and any other PPC Please let me know. Eventually I will need to use > > DosFs2 > Probably PCI bus speed limitations. I have gotten 55 MB/sec peak rate > on a PPC603, 33Mhz PCI, with a Qlogic 2100 fibre chip, but that was > memory to memory transfers with custom hardware and software. [snip] > In a practical sense, you may be able to double the numbers below for > peak rates, but if you are talking sustained real data transfers at 20 > to 30 MB/sec on 33 Mhz PCI bus host, then you are doing pretty well. PMFBI, but really? I may be missing something obvious here but surely PCI = 33MHz * 4 bytes (32 bits per word) = 132MB/sec bandwidth. Is 23% of capacity really the most that can be expected ? DaveK - -- They laughed at Galileo. They laughed at Copernicus. They laughed at Columbus. But remember, they also laughed at Bozo the Clown. --------------------------- Newsgroups: comp.os.vxworks Subject: Gcc 2.95.3 for VxWorks-PPC available.... Date: Mon, 5 Feb 2001 11:27:32 -0000 From: "Dave Korn" Organization: Lumber Cartel (tinlc) Members #2234-2237 (owing to browser refresh) Keywords: VxWorks Ppc Gcc 2.95.3 fnord! Message-ID: As it says in the subject, I'm making available the latest version of the Gnu compiler, 2.95.3, patched to fix various problems that have arisen over the years and never been fixed in the VxWorks-PPc port. I've been using it at work on and off for the past month or two, and it seems reliable and robust. The most immediately noticeable effect it had on my application was reducing the final downloadable object from 17Mb to 13Mb. It's nice to be able to do without -fvolatile too. I can't promise that problems won't appear, but the basic C and C++ stuff is fine. The areas where I expect some trouble might arise would be with exception handling, c++ rtti, tricky stuff like that. OTOH I don't know that it's broken, either; I haven't tested it too hard yet. I've developed a binary distribution, compiled for i686 cpu; if enough people give me grief about it I might get around to recompiling for i386, but I've only got enough webspace for one copy of the distribution. It drops into place in your existing Tornado installation while preserving all the files it wants to update; once installed, you can freely switch between the new compiler and your original compiler with a batch file, and it also provides an uninstall batch file that will restore your Tornado 2 directory tree to how it was before you installed the new compiler. If you're happy building the compiler from source code, you can download either the entire tarball or just a patchset against the current sources available from the FSF's anonymous CVS. There's a few bits of documentation and stuff there as well. I'll be adding more bits and pieces and expanding on some of what's already there as time goes on. The address is http://www.newgcc4vxworks4ppc.cjb.net/ which is a redirector, so it will remain the same even if I need to move the website at some point. Let me know how you get on! DaveK - -- I wouldn't recommend sex, drugs or insanity for everyone, but they've always worked for me. -- Hunter S. Thompson --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Spawing C++ task Date: Mon, 5 Feb 2001 11:28:36 -0000 From: "Dave Korn" Organization: Lumber Cartel (tinlc) Members #2234-2237 (owing to browser refresh) Message-ID: References: <95at2i$sfa$1@nnrp1.deja.com> wrote in message news:95at2i$sfa$1@nnrp1.deja.com... > Hello, > > The shell uses C conventions (i.e. it doesn't change the symbol, except > to add an '_' for some architectures). C++ function names are mangled > however, so your symbol won't really be just cortexTlm. > > The solutions are: > > 1) Find the mangled name and use that (try lkup "cortexTlm" - it should > give you a list of all symbols contain that string; one will be your > entry point). 1a) Type the beginning of the symbol, "cortextTlm" and press Ctrl+D; the target shell should then complete the name for you automatically (as long as the bit you've typed so far uniquely identifies the function). DaveK - -- > Your spamming dishonors the ghosts of your ancestors. > Su spam deshonra los espíritus de sus antepasados -- From "Learn Spanish with teleline.es" --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Releasing bound socket Date: Mon, 5 Feb 2001 11:28:49 -0000 From: "Dave Korn" Organization: Lumber Cartel (tinlc) Members #2234-2237 (owing to browser refresh) Message-ID: <%7wf6.17357$LQ2.373754@news2-win.server.ntlworld.com> References: <95bl5b$et6$1@nnrp1.deja.com> wrote in message news:95bl5b$et6$1@nnrp1.deja.com... > Hello all, > > In VxWorks, I bound a UDP socket for waiting packets. > In the development, I kill(td) the task for uploading new app. > After then the bound socket is not removed in the kernel, so I have to > reboot the system to bind the same port again. > "inetstatShow" tells the port is bound. > > Is there any way to release bound socket? > I tried to use signal to close the task safely, but the signal doesn't > works well like signals in Unix machine. > > Any help or reference for above problems? VxWorks isn't like Unix; when a task ends or is deleted in VxWorks, none of the task's resources are freed. So open file descriptors are left open, any memory it malloced will remain malloced and never be freed, and as you see above any sockets it has bound will remain bound. Even worse things can happen if you delete a task unexpectedly, it might have taken a semaphore that it will never get a chance to give back, then any other tasks in the system that try to take the same semaphore will wait forever. You have to arrange for all your task's resources to be cleaned up when the task dies. One way to do this is by not killing your task directly, but sending it a message to shut down. When it receives the message it releases all its resources and exits gracefully. This does require that your task is written in such a way that it will notice it has been sent the message and act on it within a reasonable period of time; it might not be suitable for tasks that had lots of long processing loops, because you might not want to put tests in all those loops to see if the message had arrived, but it is good for tasks that are based around an event driven loop. Otherwise you could arrange for the task to hold details of all its resources in global variables, so it could be killed and then another task could release all the resources. You'd need to take care to avoid race conditions where your task might be deleted in between, for example, mallocing some memory and storing the address of that memory somewhere it can be found and freed by the cleanup procedure. That could easily be done by wrapping resource-allocations in taskSafe / taskUnsafe pairs. DaveK - -- They laughed at Galileo. They laughed at Copernicus. They laughed at Columbus. But remember, they also laughed at Bozo the Clown. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: memPartCreate/memPartAlloc Date: Mon, 05 Feb 2001 12:50:18 +0100 From: alain Organization: Nortel Message-ID: <1n4t7t4otafam6phuq5m9qq8jeeu69tani@4ax.com> References: On Mon, 5 Feb 2001 00:39:34 -0500, "Normand Barriere" wrote: >Hi All, > (...) >mem = malloc( 1024 ) >partid = memPartCreate( mem, 1000 ); >..... > >later on in the code: > >memptr = memPartAlloc( partid, 64 ); /* result of this operation is a >failure */ > > >Is there anything wrong in performing these operations ? If so, can anybody >explain me the procedure that should be used in creating/managing memory >partitions ? > >Thanks > >Normand > > Hi, This is the right process and it should work. Try memPartShow(partid) to check partition creation. Alain. --------------------------- End of New-News digest ********************** From vxwexplo-errs@csg.lbl.gov Tue Feb 6 04:03:10 2001 From: Vxworks Exploder Date: Tue Feb 6 04:03:13 PST 2001 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Tue Feb 6 04:03:06 PST 2001 Subject: Cache coherency issue. Subject: BOOTP client cann't receive response. Subject: ftp error Subject: Re: ftp error Subject: load Module and tffs Subject: Re: Gcc 2.95.3 for VxWorks-PPC available.... Subject: Code Relocation Subject: Specify location for code Subject: NFS-Serverproblems Subject: Re: Gcc 2.95.3 for VxWorks-PPC available.... Subject: Re: Spawing C++ task Subject: namespace not support. Subject: RPC Subject: Re: Gcc 2.95.3 for VxWorks-PPC available.... Subject: Re: load Module and tffs Subject: Re: Gcc 2.95.3 for VxWorks-PPC available.... Subject: Re: Cache coherency issue. Subject: Re: How to compare two vxWorks hex files (TSR 149779) Subject: Re: Gcc 2.95.3 for VxWorks-PPC available.... Subject: Re: Cache coherency issue. Subject: Re: Gcc 2.95.3 for VxWorks-PPC available.... Subject: Re: Random number generator Subject: Re: Gcc 2.95.3 for VxWorks-PPC available.... Subject: Re: Gcc 2.95.3 for VxWorks-PPC available.... Subject: Re: load Module and tffs Subject: Re: Gcc 2.95.3 for VxWorks-PPC available.... Subject: Re: Need a C++ Compiler Subject: Re: Specify location for code Subject: Re: MPC823 Rev A --> Rev B problem Subject: Re: Fibre Channel to Raid Speed Subject: LoadModule and compressed file systems Subject: EST MPC8260 SSP Subject: *UPDATED* Gcc 2.95.3 for VxWorks-PPC available.... Subject: excTask on MPC8260 Subject: VxSim: reloc overflow Subject: Re: Gcc 2.95.3 for VxWorks-PPC available.... Subject: dialog with WindRiver - reasonable response Subject: RE: TSR 227541 Subject: RE: TSR 227541 Subject: RE: TSR 227541 Subject: RE: TSR 227541 Subject: RE: TSR 227541 Subject: RE: TSR 227541 Subject: RE: TSR 227541 Subject: RE: TSR 227541 Subject: RE: TSR 227541 Subject: RE: TSR 227541 Subject: RE: TSR 227541 Subject: RE: TSR 227541 Subject: [Help Me] linker script file Subject: Re: *UPDATED* Gcc 2.95.3 for VxWorks-PPC available.... Subject: Re: dialog with WindRiver - reasonable response Subject: ATA configuration & CDFS / DVD ROM drivers ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Cache coherency issue. Date: Mon, 05 Feb 2001 12:16:26 GMT From: gujc@my-deja.com Organization: Deja.com Message-ID: <95m5im$j7n$1@nnrp1.deja.com> I'm developing ATM driver on MPC8260. When the cache on data buffer is disabled(the memory is obtained using cacheDmaMallc),the driver works properly. But, when the cache on date buffer is enabled(the memory is obtained using malloc), there will be errors. We traced the data and found OLD DATA are kept in data buffer on error. When the cache is enabled, GBL(snooping) bit in TCT and RCT are set. Who can explain the mechanism of SNOOPING and the reason of error. Thanks in advance! Gu, Jiangchun Feb. 6, 2000 Sent via Deja.com http://www.deja.com/ --------------------------- Newsgroups: comp.os.vxworks Subject: BOOTP client cann't receive response. Date: Mon, 05 Feb 2001 12:22:39 GMT From: gujc@my-deja.com Organization: Deja.com Message-ID: <95m5ua$jhv$1@nnrp1.deja.com> Hi, everyone: I use the BOOTPclient(bootpMsgSend) provided by vxWorks V5.4 to get IP address on a custom END driver(IPOA), but the client cann't receive the response from the server. I can trace the END driver received the response and delivered it to upper layer. I modified the bootpLib.c and call the bootpInputHook routine directly from the END driver, and the BOOTP client can receive the response. Who can tell me what is the normal procedure for BOOTP client. Thank in advance! Gu, Jiangchun Feb. 6, 2001 Sent via Deja.com http://www.deja.com/ --------------------------- Newsgroups: comp.os.vxworks Subject: ftp error Date: Mon, 5 Feb 2001 13:11:07 -0000 From: "Martin Dowie" Message-ID: <3a7ea4fa$1@pull.gecm.com> Can anyone shed some light on ftp server the error I'm getting when trying to load a vxworks image over a vme backplane? [L 1622] 02/05/01 13:06:35 Connection accepted from 210.1.1.2 [C 1622] 02/05/01 13:06:35 Command "USER mgp" received [C 1622] 02/05/01 13:06:35 PASSword accepted [L 1622] 02/05/01 13:06:35 User mgp logged in. [C 1622] 02/05/01 13:06:35 Command "TYPE I" received [C 1622] 02/05/01 13:06:35 TYPE set to I N [C 1622] 02/05/01 13:06:35 Command "PORT 210,1,1,2,4,64" received [C 1622] 02/05/01 13:06:35 PORT set to 210.1.1.2 - 1088 (4,64) [C 1622] 02/05/01 13:06:35 Command "RETR vxworksmap2" received [C 1622] 02/05/01 13:06:35 RETRieve started on file vxworksmap2 [C 1621] 02/05/01 13:06:45 Transfer aborted [G 1621] 02/05/01 13:06:45 Got file E:\TORNADO2\TARGET\CONFIG\MGP\vxworksmap2 unsuccessfully [! 1621] 02/05/01 13:06:45 Socket error 10054 while sending data --------------------------- Newsgroups: comp.os.vxworks Subject: Re: ftp error Date: Mon, 5 Feb 2001 13:30:40 -0000 From: "Martin Dowie" Message-ID: <3a7ea98f$1@pull.gecm.com> References: <3a7ea4fa$1@pull.gecm.com> I should also have added that there is a machine exception being raised... other images load ok - the only obvious difference in the images is the size (the failing image is the largest at circa 5.8Mbytes). Exception at interrupt level: program Exception current instruction address: 0xeeeeeeec Machine Status Register: 0x00089000 Condition Register: 0x24008000 Regs at 0x277d00 "Martin Dowie" wrote in message news:3a7ea4fa$1@pull.gecm.com... > Can anyone shed some light on ftp server the error I'm getting when trying > to > load a vxworks image over a vme backplane? > > [L 1622] 02/05/01 13:06:35 Connection accepted from 210.1.1.2 > [C 1622] 02/05/01 13:06:35 Command "USER mgp" received > [C 1622] 02/05/01 13:06:35 PASSword accepted > [L 1622] 02/05/01 13:06:35 User mgp logged in. > [C 1622] 02/05/01 13:06:35 Command "TYPE I" received > [C 1622] 02/05/01 13:06:35 TYPE set to I N > [C 1622] 02/05/01 13:06:35 Command "PORT 210,1,1,2,4,64" received > [C 1622] 02/05/01 13:06:35 PORT set to 210.1.1.2 - 1088 (4,64) > [C 1622] 02/05/01 13:06:35 Command "RETR vxworksmap2" received > [C 1622] 02/05/01 13:06:35 RETRieve started on file vxworksmap2 > [C 1621] 02/05/01 13:06:45 Transfer aborted > [G 1621] 02/05/01 13:06:45 Got file > E:\TORNADO2\TARGET\CONFIG\MGP\vxworksmap2 unsuccessfully > [! 1621] 02/05/01 13:06:45 Socket error 10054 while sending data > > > --------------------------- Newsgroups: comp.os.vxworks Subject: load Module and tffs Date: Mon, 05 Feb 2001 16:46:52 +0100 From: Werner Metternich Organization: Raytheon Company Message-ID: <3A7ECAEC.B1329A88@raytheon.com> This is a multi-part message in MIME format. - --------------E829173FC07C6B5A07531AEA Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit I got a problem involving tffs and the loadModule routine of vxworks. Tffs works fine and I can put files in the flash disk working with the stdIO routines, but when I try to load a module *.o / *.out from a file in the flash disk the following happens: 1) the open of the file is successfull 2) the call of loadModule with the file descriptor results in a value 0x0 and the module does not show up with the moduleShow command. Does anybody have/had a similar problem? I work with VxWokrs 5.4, Tornado 2 with a Dy4 SVME 179. Thanks for any help!! Werner. - --------------E829173FC07C6B5A07531AEA Content-Type: text/x-vcard; charset=us-ascii; name="WMetternich.vcf" Content-Transfer-Encoding: 7bit Content-Description: Card for Werner Metternich Content-Disposition: attachment; filename="WMetternich.vcf" begin:vcard n:Metternich;Werner tel;fax:08191 / 94 38 40 tel;home:08191 / 94 30 91 tel;work:08161 / 902 313 x-mozilla-html:FALSE adr:;;;;;; version:2.1 email;internet:WMetternich@raytheon.com x-mozilla-cpt:;29184 fn:Werner Metternich end:vcard - --------------E829173FC07C6B5A07531AEA-- --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Gcc 2.95.3 for VxWorks-PPC available.... Date: Mon, 05 Feb 2001 15:36:17 GMT From: bblank63@my-deja.com Organization: Deja.com Message-ID: <95mh9g$tic$1@nnrp1.deja.com> References: Hi Dave, thanks for your newest gcc-port to vxWorks, but... I'm using Win95 and seems that install.bat does not work: IF EXIST on directories doesn't work there are a lot of "goto copy_err" but only a label err_copy Is it possible to install Gcc2.95.3 on Win95? Thanks in advance Bernd Blank Sent via Deja.com http://www.deja.com/ --------------------------- Newsgroups: comp.os.vxworks Subject: Code Relocation Date: Mon, 05 Feb 2001 15:39:08 GMT From: melloids@my-deja.com Organization: Deja.com Message-ID: <95mher$tl7$1@nnrp1.deja.com> Hi, First, I apologize for the length of this post. I am running VxWorks 5.4, Tornado 2.0, on a Power PC MCP750 board. I am trying to set up a cache-safe memory area and load a certain function into that memory area. So far, I've been able to set up 0x00100000 - 0x00200000 as a cache-safe area. The system loads the OS at address 0x00200000, starts from 0x00200000, and everything seems to run fine. The problem is I can't figure out how to get my new function loaded to address 0x00100000. I've gotten close, but haven't quite got it yet. I've modified the default linker command file and created a new section loaded to 0x00100000. I've specified that my function .o file should be loaded to this area. To avoid a multiple define error, I've had to take my .o file out of prjObjs.lst. When I do this, I get my function loaded at the right address, but the symbols of my function don't show up in the symbol table. Has anyone managed to do something like this successfully? Any suggestions would be appreciated. Thanks, Matt Sent via Deja.com http://www.deja.com/ --------------------------- Newsgroups: comp.os.vxworks Subject: Specify location for code Date: Mon, 05 Feb 2001 15:39:07 GMT From: melloids@my-deja.com Organization: Deja.com Message-ID: <95mhep$tl6$1@nnrp1.deja.com> I'm using VxWorks 5.4, Tornado 2.0 on a PowerPC MCP750 board. I am writing a function that I need the linker to place at a specific address. I've gotten this to work by messing with the linker script, but in order to aviod a multiple define error on the linker, I've had to remove the source file from prjObjs.lst. The result of this is that the symbols for myt function do not appear in the symbol table. Does anyone know how to get tornado 2.0 to locate a function at a specific address? Thanks in advance, Matt Sent via Deja.com http://www.deja.com/ --------------------------- Newsgroups: comp.os.vxworks Subject: NFS-Serverproblems Date: Mon, 05 Feb 2001 15:48:15 GMT From: bblank@init-ka.de Organization: Deja.com Message-ID: <95mhvr$u70$1@nnrp1.deja.com> I've connected two PPC823 machines over Ethernet, running vxWorks 5.4 with NFS-Server/Client. Every time I try to open an existing file for appending on the NFS- Server from the NFS-Client via fp = fopen( , "a"); fopen fails and another NFS-Server-daemon is suspended. Is this a known bug? Is there a workaround? Thanks Bernd Blank Sent via Deja.com http://www.deja.com/ --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Gcc 2.95.3 for VxWorks-PPC available.... Date: Mon, 5 Feb 2001 16:31:35 -0000 From: "Dave Korn" Organization: Lumber Cartel (tinlc) Members #2234-2237 (owing to browser refresh) Message-ID: References: <95mh9g$tic$1@nnrp1.deja.com> wrote in message news:95mh9g$tic$1@nnrp1.deja.com... > Hi Dave, > > thanks for your newest gcc-port to vxWorks, but... > > I'm using Win95 and seems that install.bat does > not work: > IF EXIST on directories doesn't work > there are a lot of "goto copy_err" but only a > label err_copy > > Is it possible to install Gcc2.95.3 on Win95? Oops. I should have been clearer. I'm afraid that all the batch files require the windows NT command line extensions. They could probably be rewritten to work on w95; I didn't know that Tornado would run on it! The err_copy vs. copy_err thing is just a bug by me. It will at least achieve the desired effect of aborting the installation, although not as nicely as I had intended! I guess I'll fix that and reupload. If you can see what to do by reading the batch files, you could always do it manually. The actual executables should still work fine. The steps are: - MAKE A SAFETY BACKUP OF YOUR ENTIRE TORNADO TREE! If you're going to be messing around in there manually, I can't emphasise enough how important this step is! - Backup all the .h files we're going to overwrite by copying them to the same names with .wrs extensions. - Likewise for the directories %WIND_BASE%/host/x86-win32/powerpc-wrs-vxworks/bin and .../lib - And the .exe files in %WIND_BASE%/host/x86-win32/bin. - Move all the .gcc files into their appropriate places: .h files into target/h, target/h/netinet and target/h/arch/ppc, .exe files into WIND_BASE/host/x86-win32/bin, the 2.95.3 subdir goes into WIND_BASE/host/x86-win32/lib/gcc-lib/powepc-wrs-vxworks, and the bin and lib subdirs go into WIND_BASE/host/x86-win32/powerpc-wrs-vxworks. Now all you need is a batch file that deletes the existing version of each file or directory and copies either the .gcc or .wrs equivalent in its place.. If you want to offer win95 compatible versions of the batch files, I'd be glad to include them in the distribution... DaveK - -- > If you do not cease your spamming activities, I will drop a large nuclear weapon on your servers. > Si usted no cesa en sus actividades de spam, caerá una gran arma nuclear en sus servidores - From "Learn spanish with teleline.es" --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Spawing C++ task Date: Mon, 05 Feb 2001 16:46:42 GMT From: "Martin Boulay" Organization: Magma Communications Ltd. Message-ID: References: <95at2i$sfa$1@nnrp1.deja.com> Thank you!!!! Martin Dave Korn wrote in message ... > > wrote in message >news:95at2i$sfa$1@nnrp1.deja.com... >> Hello, >> >> The shell uses C conventions (i.e. it doesn't change the symbol, except >> to add an '_' for some architectures). C++ function names are mangled >> however, so your symbol won't really be just cortexTlm. >> >> The solutions are: >> >> 1) Find the mangled name and use that (try lkup "cortexTlm" - it should >> give you a list of all symbols contain that string; one will be your >> entry point). > > 1a) Type the beginning of the symbol, "cortextTlm" and press Ctrl+D; the >target shell should then complete the name for you automatically (as long as >the bit you've typed so far uniquely identifies the function). > > > DaveK >-- >> Your spamming dishonors the ghosts of your ancestors. >> Su spam deshonra los espíritus de sus antepasados > -- From "Learn Spanish with teleline.es" > > --------------------------- Newsgroups: comp.os.vxworks Subject: namespace not support. Date: Mon, 05 Feb 2001 16:50:36 GMT From: "Martin Boulay" Organization: Magma Communications Ltd. Message-ID: Hi everybody! I usally use in c++: namespace sdm { const double BOOK_VERSION = 2.0; } and when I call my variable I do e.g. x = sdm::const double BOOK_VERSION; but in vxWorks the namespace is not seem to be support. Do you have a suggestion for me? What I can do to replace namespace Thank you! Martin --------------------------- Newsgroups: comp.os.vxworks Subject: RPC Date: Mon, 05 Feb 2001 17:23:49 +0000 From: Graham Baxter Organization: Graham Baxter (Software) Limited Message-ID: <3A7EE1A5.57C8A220@NOSPAM.bcs.org.uk> VxWorkers, Does anyone know where we can find the C source code for RPC? Thanks for any help. Graham Baxter gbaxter@NOSPAMbcs.org.uk --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Gcc 2.95.3 for VxWorks-PPC available.... Date: 05 Feb 2001 10:31:49 -0700 From: taashlo@sandia.gov Organization: Sandia National Laboratories, Albuquerque, NM USA Message-ID: References: <95mh9g$tic$1@nnrp1.deja.com> After installing on an NT host, when I try to build, I get the following: ========================================================================== ccppc -O0 -g -mstrict-align -ansi -nostdinc -DRW_MULTI_THREAD - -D_REENTRANT -fvolatile -fno -builtin -I. -IC:\Tornado\target\h - -DCPU=PPC604 -frepo -fno-rtti -fno-exceptions -D_DEBUG - -c C:\Projects\Aura\src\DACS\Version.cpp ccppc: installation problem, cannot exec `cpp0': No such file or directory make: *** [Version.o] Error 0x1 Done. ========================================================================== Any ideas about "cannot exec `cpp0'"? Thanks, Tad --------------------------- Newsgroups: comp.os.vxworks Subject: Re: load Module and tffs Date: Mon, 05 Feb 2001 17:39:50 GMT From: kimp@avtecinc.com (Kim Putnam) Organization: Info Avenue Internet Services Message-ID: <3a7ee448.16981500@192.168.1.214> References: <3A7ECAEC.B1329A88@raytheon.com> I am having a some what similar problem. We are using Vxworks 5.4, Tornado 2, VMICPCI-7715, and DOC 2000. I am trying to load some runtime libs programatically. I have the files located on the DOC. I call tffsDrv(), usrTffsConfig(), and open the file. I then use the fd for loadModule. I, however, do get a ModuleId back, but still cannot set the module when I do a moduleShow. I then try to execute a call fromthe runtime libs, and crash the box. I read thru the archives on deja.com, and found some info on loadModule. I set the errno to 0 before I call loadModule then get the errno after the call. It is 0, which should indicate all is okay. I am at a lost. I also ran symFindByName for one of the routines, from the shell, but this fails. Any help would be appreciated Kim On Mon, 05 Feb 2001 16:46:52 +0100, Werner Metternich wrote: >This is a multi-part message in MIME format. >--------------E829173FC07C6B5A07531AEA >Content-Type: text/plain; charset=us-ascii >Content-Transfer-Encoding: 7bit > > >I got a problem involving tffs and the loadModule routine of vxworks. >Tffs works fine and I can put files in the flash disk working with the >stdIO routines, but when I try to load a module *.o / *.out from a file >in the flash disk the following happens: >1) the open of the file is successfull >2) the call of loadModule with the file descriptor results in a value >0x0 and the module does not show up with the moduleShow command. > >Does anybody have/had a similar problem? >I work with VxWokrs 5.4, Tornado 2 with a Dy4 SVME 179. > >Thanks for any help!! > Werner. > >--------------E829173FC07C6B5A07531AEA >Content-Type: text/x-vcard; charset=us-ascii; > name="WMetternich.vcf" >Content-Transfer-Encoding: 7bit >Content-Description: Card for Werner Metternich >Content-Disposition: attachment; > filename="WMetternich.vcf" > >begin:vcard >n:Metternich;Werner >tel;fax:08191 / 94 38 40 >tel;home:08191 / 94 30 91 >tel;work:08161 / 902 313 >x-mozilla-html:FALSE >adr:;;;;;; >version:2.1 >email;internet:WMetternich@raytheon.com >x-mozilla-cpt:;29184 >fn:Werner Metternich >end:vcard > >--------------E829173FC07C6B5A07531AEA-- > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Gcc 2.95.3 for VxWorks-PPC available.... Date: Mon, 05 Feb 2001 17:52:31 GMT From: adrianmichaud@my-deja.com Organization: Deja.com Message-ID: <95mp8q$5v3$1@nnrp1.deja.com> References: You are >>THE MAN<<... Thanks :) Adrian In article , "Dave Korn" wrote: > > As it says in the subject, I'm making available the latest version of the > Gnu compiler, 2.95.3, patched to fix various problems that have arisen over > the years and never been fixed in the VxWorks-PPc port. > > I've been using it at work on and off for the past month or two, and it > seems reliable and robust. The most immediately noticeable effect it had on > my application was reducing the final downloadable object from 17Mb to 13Mb. > It's nice to be able to do without -fvolatile too. > > I can't promise that problems won't appear, but the basic C and C++ stuff > is fine. The areas where I expect some trouble might arise would be with > exception handling, c++ rtti, tricky stuff like that. OTOH I don't know > that it's broken, either; I haven't tested it too hard yet. > > I've developed a binary distribution, compiled for i686 cpu; if enough > people give me grief about it I might get around to recompiling for i386, > but I've only got enough webspace for one copy of the distribution. It > drops into place in your existing Tornado installation while preserving all > the files it wants to update; once installed, you can freely switch between > the new compiler and your original compiler with a batch file, and it also > provides an uninstall batch file that will restore your Tornado 2 directory > tree to how it was before you installed the new compiler. > > If you're happy building the compiler from source code, you can download > either the entire tarball or just a patchset against the current sources > available from the FSF's anonymous CVS. > > There's a few bits of documentation and stuff there as well. I'll be > adding more bits and pieces and expanding on some of what's already there as > time goes on. > > The address is http://www.newgcc4vxworks4ppc.cjb.net/ which is a > redirector, so it will remain the same even if I need to move the website at > some point. > > Let me know how you get on! > > DaveK > -- > I wouldn't recommend sex, drugs or insanity for everyone, but they've > always worked for me. > -- Hunter S. Thompson > > Sent via Deja.com http://www.deja.com/ --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Cache coherency issue. Date: Mon, 5 Feb 2001 13:02:17 -0500 From: "Jeff Creem" Organization: Posted via Supernews, http://www.supernews.com Message-ID: References: <95m5im$j7n$1@nnrp1.deja.com> I don't know the details of your particular architecture however if you use malloc to allocate the memory and it gives you a block that is not cache line aligned you can get into what is called the double dirty problem. The data in the cache and the data in memory (from a nearby variable) are different and there is no single cache line action that can be taken to solve the problem. So, if you really want to use malloc for some reason, you need to ask for more memory than you want and then only use the portions that are multiples (in length and start address) of a complete cache line length. (I am assuming here that you are correct and full snooping is really supported in your hardware).. wrote in message news:95m5im$j7n$1@nnrp1.deja.com... > > > I'm developing ATM driver on MPC8260. > > When the cache on data buffer is disabled(the memory is obtained > using cacheDmaMallc),the driver works properly. > But, when the cache on date buffer is enabled(the memory is > obtained using malloc), there will be errors. We traced the data and > found OLD DATA are kept in data buffer on error. When the cache is > enabled, GBL(snooping) bit in TCT and RCT are set. > > Who can explain the mechanism of SNOOPING and the reason of error. > Thanks in advance! > > Gu, Jiangchun Feb. 6, 2000 > > > Sent via Deja.com > http://www.deja.com/ --------------------------- Newsgroups: comp.os.vxworks Subject: Re: How to compare two vxWorks hex files (TSR 149779) Date: 5 Feb 2001 13:19:02 -0500 From: vanbaren@falcon.si.com (vanbaren_gerald) Organization: Smiths Industries Message-ID: <3a7eee96$1@news.si.com> References: <95eivf$80c$1@overload.lbl.gov> Jim Way writes: >About a year ago, I had a question about how to compare two vxWorks HEX >files. This was recorded by WindRiver as TSR 149779. The problem was that we >use the "romCompress" option. Date/time stamp information is different >between the two builds, causing the file compression algorithm to generate >radically different results. >After some time, I've finally determined a way to do the comparison that is >convincing enough for me. (Now that I've seen it, it seems painfully >obvious.) The solution is to go to the VxWorks tab in the project window, >click on the Build name, call up the context menu, select Properties, then >select the Rules tab. Where it says "vxWorks_romCompress" change the >rule to "vxWorks_romResident". Apply that rule and click OK. Now rebuild-all >the project (you might be able to just relink, but I'm paranoid). The >resulting file will be huge because it's uncompressed. >All that's left to do is to use your favorite file comparison tool (we use >WinDiff and CodeWright) to compare the files. You should still observe a few >bytes that are different. You can minimize these by trying to build both HEX >files at about the same time on the same day. The only bytes that *should* >be different are the date and time representations. >Don't forget to change your rule back to "vxWorks_romCompress" when you're >done. >WindRiver closed TSR 149779 (over my objections). Here's hoping this >explanation gets into their support database. >Thanks again to everyone for their support and encouragement! >Jim >----------------------------------------- >Jim Way, Software Engineer >Datum Austin (Austron Inc.) >voice: 512.721.4170 >fax : 512.990.9712 >email: jwayATdatumDOTcom (no spam please) >----------------------------------------- Jim: It seems to me that you are working too hard. I assume you are using Tornado2 (I'm not familiar with T-1). T2 builds your application, named vxWorks. It then converts the application into assembly language "define bytes" (compressing it first, if you specified romCompressed). It then builds the startup code, re-assembles the "define bytes" application lump, and links the start up with the application lump. On start up, the startup code copies the application lump from ROM to RAM (decompressing it, if applicable), and then jumps to it. The bottom line: if you want to compare the actual code you wrote (i.e. you don't care about the startup code), all you have to compare are the files named vxWorks. It is still laying around in the "default" subdirectory when your build completes. This is quite close to your method (above), the difference is that, with your method, you compare the start up code as well as your application code (sans the start up decompression code). If you have modified the start up (BSP) code, the vxWorks comparison won't show those differences. You could improve this to compare the start up code in the vxWorks_romCompress module (just the first N bytes, N needs to be determined from the symbol table every time), but that would be a lot more work. gvb - -- +---------------------------------------------------------------------------+ | Jerry Van Baren / vanbaren_gerald@si.com / Grand Rapids Mi / 616-241-7973 | | My employer is a company. Companies are artifacts of a legal system. | |________________Artifacts are incapable of having opinions.________________| --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Gcc 2.95.3 for VxWorks-PPC available.... Date: Mon, 5 Feb 2001 18:20:51 -0000 From: "Dave Korn" Organization: Lumber Cartel (tinlc) Members #2234-2237 (owing to browser refresh) Message-ID: References: <95mh9g$tic$1@nnrp1.deja.com> wrote in message news:uk875829m.fsf@sandia.gov... > After installing on an NT host, when I try to build, I get the > following: > > ========================================================================== > ccppc -O0 -g -mstrict-align -ansi -nostdinc -DRW_MULTI_THREAD > -D_REENTRANT -fvolatile -fno -builtin -I. -IC:\Tornado\target\h > -DCPU=PPC604 -frepo -fno-rtti -fno-exceptions -D_DEBUG > -c C:\Projects\Aura\src\DACS\Version.cpp > ccppc: installation problem, cannot exec `cpp0': No such file or directory > make: *** [Version.o] Error 0x1 > > Done. > ========================================================================== > > Any ideas about "cannot exec `cpp0'"? > > Thanks, > Tad Try setting GCC_EXEC_PREFIX=%WIND_BASE%\host\x86-win32\lib\gcc-lib in your My Computer / environment option settings. If it still doesn't work, try the command again but add '-v' to the command line, and post the results here. DaveK - -- They laughed at Galileo. They laughed at Copernicus. They laughed at Columbus. But remember, they also laughed at Bozo the Clown. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Cache coherency issue. Date: Mon, 05 Feb 2001 18:19:41 GMT From: gold@sdsu.edu Organization: Deja.com Message-ID: <95mqrh$7e4$1@nnrp1.deja.com> References: <95m5im$j7n$1@nnrp1.deja.com> In article <95m5im$j7n$1@nnrp1.deja.com>, gujc@my-deja.com wrote: > > > I'm developing ATM driver on MPC8260. > > When the cache on data buffer is disabled(the memory is obtained > using cacheDmaMallc),the driver works properly. > But, when the cache on date buffer is enabled(the memory is > obtained using malloc), there will be errors. We traced the data and > found OLD DATA are kept in data buffer on error. When the cache is > enabled, GBL(snooping) bit in TCT and RCT are set. > > Who can explain the mechanism of SNOOPING and the reason of error. > Thanks in advance! > We're using ATM on the 8260 (dcache enabled) with snooping enabled (GBL set in TCT/RCT) and it works fine, we haven't had any problems. When you allocate memory with cacheDmaMalloc(), it's always aligned on an MMU_PPC_PAGESIZE boundary. When you call malloc(), I don't think it's aligned on anything in particular (besides maybe word aligned). If you call malloc(), are you sure all your buffer addresses in the BDs are properly aligned ? Also, if you just use malloc(), I don't think you can guarrantee that your address is on a cache line boundary, so you may be sharing your data buffer with other stuff in the cache line, although I haven't looked at memPartAlloc() to be sure. You might try memalign() with the page size and see if that makes a difference. Have you looked at the RCT/TCT tables to make sure the bit is set ? Have you tried using malloc() but added explicit cacheFlush() and cacheInvalidate() to your driver ? hth, Dan Gold gold@ensemblecom.com Sent via Deja.com http://www.deja.com/ --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Gcc 2.95.3 for VxWorks-PPC available.... Date: Mon, 05 Feb 2001 10:40:42 -0800 From: Melvin Gardipee <"> Organization: Raytheon Company Message-ID: References: <95mh9g$tic$1@nnrp1.deja.com> I am seeing the same failure from within Tornado 2. I have already tried adding GCC_EXEC_PREFIX to my environment without success. Just executing CC from within Tornado 2 yields: > ccppc: installation problem, cannot exec `cpp0': No such file or > directory while executing > "exec ccppc -E -P -xc -DPPC604 -Dgnu > C:\Tornado\target\config\comps\vxWorks\00bsp.cdf > C:\Tornado\target\config\comps\vxWorks\00network.cdf > C:\Tornado\t..." Executing ccppc -v yields: > Using builtin specs > gcc version 2.95.3 20010125 (prerelease) Mel Dave Korn wrote: > > wrote in message news:uk875829m.fsf@sandia.gov... > > After installing on an NT host, when I try to build, I get the > > following: > > > > ========================================================================== > > ccppc -O0 -g -mstrict-align -ansi -nostdinc -DRW_MULTI_THREAD > > -D_REENTRANT -fvolatile -fno -builtin -I. -IC:\Tornado\target\h > > -DCPU=PPC604 -frepo -fno-rtti -fno-exceptions -D_DEBUG > > -c C:\Projects\Aura\src\DACS\Version.cpp > > ccppc: installation problem, cannot exec `cpp0': No such file or directory > > make: *** [Version.o] Error 0x1 > > > > Done. > > ========================================================================== > > > > Any ideas about "cannot exec `cpp0'"? > > > > Thanks, > > Tad > > Try setting GCC_EXEC_PREFIX=%WIND_BASE%\host\x86-win32\lib\gcc-lib in your > My Computer / environment option settings. > > If it still doesn't work, try the command again but add '-v' to the > command line, and post the results here. > > DaveK > -- > They laughed at Galileo. They laughed at Copernicus. They laughed at > Columbus. But remember, they also laughed at Bozo the Clown. --------------------------- Newsgroups: comp.os.vxworks,comp.security.firewalls,comp.lang.c,alt.computer.security Subject: Re: Random number generator Date: Sun, 4 Feb 2001 18:52:36 -0000 From: "Sam Simpson" Organization: ntl Cablemodem News Service Message-ID: References: <95f2ts$hat$1@nnrp1.deja.com> Use RC4 (or the Open Source implementation ARC4). Or use any decent block cipher in CFB mode with random key, IV and plaintext. What programming language do you need? - -- Regards, Sam http://www.scramdisk.clara.net/ wrote in message news:95f2ts$hat$1@nnrp1.deja.com... > Hi all, > > Couls someone suggest me a good 64-bit pseudo random number generation > algorithm and/or code. > > Any help is greatly appreciated. > > Thanks, > Milind > > > Sent via Deja.com > http://www.deja.com/ --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Gcc 2.95.3 for VxWorks-PPC available.... Date: Mon, 5 Feb 2001 18:51:42 -0000 From: "Dave Korn" Organization: Lumber Cartel (tinlc) Members #2234-2237 (owing to browser refresh) Message-ID: References: <95mh9g$tic$1@nnrp1.deja.com> "Melvin Gardipee >" <" wrote in message news:JuCf6.114$3d3.2372@dfw-service2.ext.raytheon.com... > I am seeing the same failure from within Tornado 2. I have already > tried adding GCC_EXEC_PREFIX to my environment without success. > > Just executing CC from within Tornado 2 yields: > > > ccppc: installation problem, cannot exec `cpp0': No such file or > > directory while executing > > "exec ccppc -E -P -xc -DPPC604 -Dgnu > > C:\Tornado\target\config\comps\vxWorks\00bsp.cdf > > C:\Tornado\target\config\comps\vxWorks\00network.cdf > > C:\Tornado\t..." > > Executing ccppc -v yields: > > Using builtin specs > > gcc version 2.95.3 20010125 (prerelease) Sorry, I meant to add the -v flag to the options on the command line that failed, so we can see where it's looking for cpp0. Can you give that a try and post the output here please ? DaveK --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Gcc 2.95.3 for VxWorks-PPC available.... Date: Mon, 05 Feb 2001 11:12:46 -0800 From: Melvin Gardipee <"> Organization: Raytheon Company Message-ID: References: <95mh9g$tic$1@nnrp1.deja.com> OK, here is the output when I try to compile: > ccppc -O2 -v -mstrict-align -nostdinc -Wall -Wbad-function-cast -Wmissing-prototypes -Wnes > ted-externs -Wmissing-declarations -DRW_MULTI_THREAD -D_REENTRANT -DNDEBUG -fvolatile -fno > -builtin -fno-for-scope -funsigned-bitfields -I. -IC:\Tornado\target\h -IC:\Tornado\target > \config\mv2304 -IC:\FMPS\VxWorks\lib1553\source\common\Library -IC:\FMPS\VxWorks\lib1553\s > ource\M1553\Library -IC:\FMPS\fmps -DCPU=PPC604 -c C:\FMPS\VxWorks\fmps\scramnet.c > Using builtin specs. > gcc version 2.95.3 20010125 (prerelease) > cpp0 -lang-c -nostdinc -v -I. -IC:\Tornado\target\h -IC:\Tornado\target\config\mv2304 -IC > :\FMPS\VxWorks\lib1553\source\common\Library -IC:\FMPS\VxWorks\lib1553\source\M1553\Librar > y -IC:\FMPS\fmps -iprefix \Tornado\host\x86-win32\lib\gcc-libpowerpc-wrs-vxworks\2.95.3\ - > D__GNUC__=2 -D__GNUC_MINOR__=95 -D__vxworks -D__PPC__ -D__VXVARGS__ -D__vxworks -D__PPC__ > -D__VXVARGS__ -Asystem(vxworks) -Asystem(embedded) -Acpu(powerpc) -Amachine(powerpc) -D__C > HAR_UNSIGNED__ -D__OPTIMIZE__ -Wall -Wbad-function-cast -Wmissing-prototypes -Wnested-exte > rns -Wmissing-declarations -D_CALL_SYSV -D__BIG_ENDIAN__ -Amachine(bigendian) -D_ARCH_PPC > -DRW_MULTI_THREAD -D_REENTRANT -DNDEBUG -DCPU=PPC604 C:\FMPS\VxWorks\fmps\scramnet.c \TEMP > \cc5WuXM9.i Mel Dave Korn wrote: > > "Melvin Gardipee >" <" wrote in > message news:JuCf6.114$3d3.2372@dfw-service2.ext.raytheon.com... > > I am seeing the same failure from within Tornado 2. I have already > > tried adding GCC_EXEC_PREFIX to my environment without success. > > > > Just executing CC from within Tornado 2 yields: > > > > > ccppc: installation problem, cannot exec `cpp0': No such file or > > > directory while executing > > > "exec ccppc -E -P -xc -DPPC604 -Dgnu > > > C:\Tornado\target\config\comps\vxWorks\00bsp.cdf > > > C:\Tornado\target\config\comps\vxWorks\00network.cdf > > > C:\Tornado\t..." > > > > Executing ccppc -v yields: > > > Using builtin specs > > > gcc version 2.95.3 20010125 (prerelease) > > Sorry, I meant to add the -v flag to the options on the command line that > failed, so we can see where it's looking for cpp0. Can you give that a try > and post the output here please ? > > DaveK --------------------------- Newsgroups: comp.os.vxworks Subject: Re: load Module and tffs Date: Mon, 05 Feb 2001 19:34:49 GMT From: kimp@avtecinc.com (Kim Putnam) Organization: Info Avenue Internet Services Message-ID: <3a7f0015.24098234@192.168.1.214> References: <3A7ECAEC.B1329A88@raytheon.com> <3a7ee448.16981500@192.168.1.214> Okay, I admit I can be an idot at times. In this case the loadModule does work. I was failing to recognize the difference between the target shell, and the target server shell. When I ran moduleShow from the target I did see my modules, and was able to execute functions. K On Mon, 05 Feb 2001 17:39:50 GMT, kimp@avtecinc.com (Kim Putnam) wrote: >I am having a some what similar problem. We are using Vxworks 5.4, >Tornado 2, VMICPCI-7715, and DOC 2000. > >I am trying to load some runtime libs programatically. I have the >files located on the DOC. > >I call tffsDrv(), usrTffsConfig(), and open the file. I then use the >fd for loadModule. I, however, do get a ModuleId back, but still >cannot set the module when I do a moduleShow. > >I then try to execute a call fromthe runtime libs, and crash the box. > >I read thru the archives on deja.com, and found some info on >loadModule. I set the errno to 0 before I call loadModule then get the >errno after the call. It is 0, which should indicate all is okay. > >I am at a lost. I also ran symFindByName for one of the routines, from >the shell, but this fails. > >Any help would be appreciated > >Kim > >On Mon, 05 Feb 2001 16:46:52 +0100, Werner Metternich > wrote: > >>This is a multi-part message in MIME format. >>--------------E829173FC07C6B5A07531AEA >>Content-Type: text/plain; charset=us-ascii >>Content-Transfer-Encoding: 7bit >> >> >>I got a problem involving tffs and the loadModule routine of vxworks. >>Tffs works fine and I can put files in the flash disk working with the >>stdIO routines, but when I try to load a module *.o / *.out from a file >>in the flash disk the following happens: >>1) the open of the file is successfull >>2) the call of loadModule with the file descriptor results in a value >>0x0 and the module does not show up with the moduleShow command. >> >>Does anybody have/had a similar problem? >>I work with VxWokrs 5.4, Tornado 2 with a Dy4 SVME 179. >> >>Thanks for any help!! >> Werner. >> >>--------------E829173FC07C6B5A07531AEA >>Content-Type: text/x-vcard; charset=us-ascii; >> name="WMetternich.vcf" >>Content-Transfer-Encoding: 7bit >>Content-Description: Card for Werner Metternich >>Content-Disposition: attachment; >> filename="WMetternich.vcf" >> >>begin:vcard >>n:Metternich;Werner >>tel;fax:08191 / 94 38 40 >>tel;home:08191 / 94 30 91 >>tel;work:08161 / 902 313 >>x-mozilla-html:FALSE >>adr:;;;;;; >>version:2.1 >>email;internet:WMetternich@raytheon.com >>x-mozilla-cpt:;29184 >>fn:Werner Metternich >>end:vcard >> >>--------------E829173FC07C6B5A07531AEA-- >> > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Gcc 2.95.3 for VxWorks-PPC available.... Date: Mon, 5 Feb 2001 19:58:13 -0000 From: "Dave Korn" Organization: Lumber Cartel (tinlc) Members #2234-2237 (owing to browser refresh) Message-ID: References: <95mh9g$tic$1@nnrp1.deja.com> OK, there's a problem with the path searching in the top level gcc driver. There's a simple workaround: add the directories %WIND_BASE%\host\x86-win32\powerpc-wrs-vxworks\bin and %WIND_BASE%\host\x86-win32\lib\gcc-lib\2.95.3 to the beginning of your PATH environment variable in My Computer/properties/environment variables. This should fix all the 'cpp0 not found' and 'as not found' problems. DaveK --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Need a C++ Compiler Date: Mon, 05 Feb 2001 15:34:57 -0600 From: Luc Message-ID: <3A7F1C81.A3776584@qwest.net> References: <92spv2$uim$1@nnrp1.deja.com> <3A5277BB.131C67B0@bellsouth.net> <92uqd0$l5c$1@nnrp1.deja.com> <3A533D3F.7439933A@lmco.com> <3A79E33C.9DF0D01F@gilson.com> it's better and it's free......... http://www.redhat.com/embedded/technologies/ecos/ Robert Norton wrote: > Joe Durusau wrote: > > > Ah, so! If you want to do real work on a target with > > VxWorks, you have to PAY! As Johan observes elsewhere, you don't get > > the libs and suchlike for your target with the eval system. Open your > > pocketbook (as widely as possible) and your problems will be solved. > > We have used an in-house real time OS with the Intel 80C196 family for a long > time, and would like to upgrade to a stronger CPU and better real-time OS. > VxWorks and an ARM look good to me, but what are the licensing fees like? > There seems to be no pricing info on the web pages at all, and that scares me > a bit. We would probably want C++ and Java as well as the bare OS. Is > WindRiver licensing on a one-time fee, or on per unit sold or what? > > TIA, Bob Norton. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Specify location for code Date: Mon, 05 Feb 2001 22:50:46 GMT From: john_94501@my-deja.com Organization: Deja.com Message-ID: <95nao1$n8a$1@nnrp1.deja.com> References: <95mhep$tl6$1@nnrp1.deja.com> Hello, You could always use the dynamic loader (target loader); use the loadModuleAt() to specify addresses for text, data and bss segments (or any subset of those three). Of course, this only works if you are willing to load the module at runtime, and don't want it linked at a specific address as part of the build process. HTH, John... PS Out of interest, why do you need the function placed at a specific address? In article <95mhep$tl6$1@nnrp1.deja.com>, melloids@my-deja.com wrote: > I'm using VxWorks 5.4, Tornado 2.0 on a PowerPC > MCP750 board. I am writing a function that I > need the linker to place at a specific address. > I've gotten this to work by messing with the > linker script, but in order to aviod a multiple > define error on the linker, I've had to remove > the source file from prjObjs.lst. The result of > this is that the symbols for myt function do not > appear in the symbol table. > > Does anyone know how to get tornado 2.0 to locate > a function at a specific address? > > Thanks in advance, > Matt > > Sent via Deja.com > http://www.deja.com/ > Sent via Deja.com http://www.deja.com/ --------------------------- Newsgroups: comp.os.vxworks,comp.sys.powerpc.tech Subject: Re: MPC823 Rev A --> Rev B problem Date: Mon, 5 Feb 2001 18:20:40 -0500 From: "Gary Sinde" Organization: Net Direct, Inc. Internet Access - Indianapolis, IN Message-ID: <3a7f35b1$0$7170$a05e1490@news.netdirect.net> References: I just got done putting the new I2C/SPI patch in my code. I could never get motorola to state why I needed to change the patch but it has worked in both Rev-A and Rev-B chips. My next revision of the board has several Rev-A's and several Rev-B's to check for differences. I also am using ethernet, I2C, SPI, and both SMC's. Gary "Amy Michaelis" wrote in message news:SGXe6.1708$5s2.60221@wormhole.dimensional.com... > Help! > > Motorola has recently discontinued the Rev A silicon for the 823, and now we > can only get the Rev B silicon. This should all be seamless of course, but > after some testing with the Rev B chip, we've found a problem. The Rev B > chip causes random machine checks in our system. > > We're using vxWorks as an operating system, and a check with Wind River > didn't turn up anything. I've been trying to find out what's going on here, > but so far no luck. We're using quite a bit of CPM, and I tend to believe > that's were the problem lies. We have the SCC in Ethernet mode, and the SPI > and I2C are also being used. These two guys have been relocated using the > downloaded microcode patch. My sense is that since Motorola cleaned up a > lot of the problems in the I2C driver for the Rev B silicon, something has > gone wrong in the downloaded microcode. > > The problem could also be the new interrupt for SCC3, but we're not using > the new SCC on the rev B chip, so I don't know how that would be a problem. > > Any ideas would be greatly appreciated. > > Thanks, > > Matt Michaelis > amich@dimensional.com > > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Fibre Channel to Raid Speed Date: Mon, 05 Feb 2001 18:30:01 -0800 From: Christopher A Leddy Organization: Raytheon Company Message-ID: <3A7F61A9.ABCC795D@west.raytheon.com> References: <3A6F77F3.FAB5284B@erols.com> <3A70CC93.3444D4EB@west.raytheon.com> Dave Korn wrote: > > "Christopher A Leddy" wrote in message > news:3A70CC93.3444D4EB@west.raytheon.com... > > Fred Walther wrote: > > > > > > Does anybody have any experience with Synergy VGM5 PPC and Fibre channel > > > PMC? > > > I am using DosFs and I am getting only 30 MB/Sec Read and 20 MBytes/Sec > > > Write to the raid disks. Has anybody gotten better speeds with VxWorks > > > and any other PPC Please let me know. Eventually I will need to use > > > DosFs2 > > > Probably PCI bus speed limitations. I have gotten 55 MB/sec peak rate > > on a PPC603, 33Mhz PCI, with a Qlogic 2100 fibre chip, but that was > > memory to memory transfers with custom hardware and software. > > [snip] > > > In a practical sense, you may be able to double the numbers below for > > peak rates, but if you are talking sustained real data transfers at 20 > > to 30 MB/sec on 33 Mhz PCI bus host, then you are doing pretty well. > > PMFBI, but really? I may be missing something obvious here but surely PCI > = 33MHz * 4 bytes (32 bits per word) = 132MB/sec bandwidth. Is 23% of > capacity really the most that can be expected ? > > DaveK > -- > They laughed at Galileo. They laughed at Copernicus. They laughed at > Columbus. But remember, they also laughed at Bozo the Clown. They also laughed at 132MB/sec PCI transfers in the real world. By the way the 132MB/sec is nonsense because you are forced to release (and reinitiate) transfers after a fixed number of transfers anyhow. Some of the best claim numbers like 120-125 MB/sec. Let's see - what is missing is: PCI chip set details (go to motorola/intel to get some idea of variability here, and a lot of cards use custom PCI cores that suck a whole lot worse than these) PCI Config Settings (Optimal settings for a real system != settings for fibrechannel only system) Other Users of the Bus (like the very same fibre card using local memory mapped on the PCI bus to keep track of it's own transfers, filling and draining buffers, plus timers,serial ports, ethernet cards,all dancing on your bus, etc) So Yes, if you 1. Pre-fill all your memory buffers with data 2. Transfer only very large block sizes 3. Optimize all PCI config settings for your transfers at the expense of all others 4. Do no processing with the data and keep track of nothing related to the data 5. Pick/Design your PCI controllers on both sides very carefully 6. Ignore all required arbitration on the fibrechannel bus (which is insignificant for the artificially large test blocks but adds up for real size blocks) Then you will see peak rates above 100Mbytes per second, and you have created a beautiful benchmark for the fibrechannel vendor. This number still tells you nothing about sustained transfer performance on a real system running an OS, doing file transfers over a fibrechannel link. Chris --------------------------- Newsgroups: comp.os.vxworks Subject: LoadModule and compressed file systems Date: Mon, 5 Feb 2001 21:16:22 -0500 From: "James Robinson" Message-ID: <3a7f5eb7_1@news3.buffnet.net> Hello, I didn't see my post so here goes again..... I am using flashLib V2.0 with an Intel 28f160 Flash chip. I have tested my flashDrv with the drvtest.c program and it passes all of the test. When I use loadModule it takes about 10 seconds to load a 47K file. What I think is going on is that loadModule is seeking within the file. This causes a problem in the fact that every time the file seeks I think it restarts from the beginning of the file. The filesystem uses inflate and deflate. I don't think there is a way to jump into a compressed datastream. So... here is the big question, is there an other way of loading a file besides loadModule, or is there a way to tell loadModule NOT to seek, but do a single pass load.... We are using the following. flashLib V2.0 flashDrv for INTEL 28f160 Tornado 2 for MIPS (RC32364) BSP - idts134 Thanks, James Robinson jrobinson@clearwire.com --------------------------- Newsgroups: comp.os.vxworks Subject: EST MPC8260 SSP Date: Mon, 5 Feb 2001 18:50:38 -0800 From: "D. H. Everett" Organization: Posted via Supernews, http://www.supernews.com Message-ID: Does WindRiver supply an SSP for the EST MPC8260 Reference Design? I see they have one for the MBC860. I've seen references in this newsgroup to the MPC8260, so I was wondering if an SSP is available (before I go through a porting effort). I don't see one on the WindSurf site, BTW. Thanks, Dave --------------------------- Newsgroups: comp.os.vxworks Subject: *UPDATED* Gcc 2.95.3 for VxWorks-PPC available.... Date: Tue, 6 Feb 2001 03:52:50 -0000 From: "Dave Korn" Organization: Lumber Cartel (tinlc) Members #2234-2237 (owing to browser refresh) Keywords: VxWorks Ppc Gcc 2.95.3 fnord! Message-ID: Hi everyone, My apologies for the false start. I've now tracked down the bug with ccppc not finding various executables: it was my Cygwin installation getting in the way that stopped me from experiencing it myself. I've now fixed the bugs and verified that the compiler works by removing my Cygwin installation and registry settings; it still ran fine, so I believe the code is correct now. I've now fixed the ccppc driver so it works out all the paths from your environment settings of WIND_BASE and WIND_HOST_TYPE. There's no more need to mess with your PATH setting or to define GCC_EXEC_PREFIX. I've also fixed the version of ldppc that I included so that it recognizes the '@' parameter used in WRS makefile to indicate a file that contains a list of files to be linked. The format of the sizeppc command's output had changed, causing the vxsize utility to return false results. I've fixed that by making the distribution reuse the old WRS version. Finally I've recompiled the whole thing for i386, so it should run on any CPU! I haven't updated the individual patches yet but I will update the overall patch set in the next few minutes. I hope you'll all be a lot happier with this new installation :-) DaveK - -- They laughed at Galileo. They laughed at Copernicus. They laughed at Columbus. But remember, they also laughed at Bozo the Clown. --------------------------- Newsgroups: comp.os.vxworks Subject: excTask on MPC8260 Date: Tue, 06 Feb 2001 05:33:43 GMT From: muscou Organization: Sympatico Message-ID: <3A7F8F52.5FFBCFAF@NOSPAMsympatico.ca> Hi, The excTask read messages from a queue ( 28 bytes in size, 4 for the function pointer and 6 parameters of 4 bytes each). Most of the time, it handle all those logMsg being called from interrupt... In our case, the function reboot end up on the queue. At that point, we would expect a message to be written at the sysExcMsg pointer... but it is empty, only zero. If an exception had happen, I would have also expect our exception handler ( attached with excHookAdd) to run, but it does not( breakpoint is not hit). My theory is that there is a fatal kernel error but since the sysExcMsg is empty, I have no idea what it can be. Does anybody have an idea as to what kernel function could put the reboot fct. on the excTask queue without generating an appropriate message at the sysExcMsg pointer location? I have ask WRS support but it did not really help much. I have also try to get windview running but can't reproduce the problem in that case as the timing seems to be modified. LA won't really work either ( or I disable the cache or I generate an exception every time a branch is taken, this would really screw up the timing...) Does anybody have ever seen a problem like this? Any idea has to what to look for? I am now in the process of removing all kernel calls from all ISR, I might be able to narrow down the problem, any other suggestions? Thanks, --------------------------- Newsgroups: comp.os.vxworks Subject: VxSim: reloc overflow Date: Tue, 06 Feb 2001 06:29:50 GMT From: "rt" Organization: EarthLink Inc. -- http://www.EarthLink.net Message-ID: I get reloc overflow message during the link of the large project. The compilation flag -mlongcall is not supported by ccsimpc compiler. Following is the error message: vxrm C:\VxWorks\WxWorks_Tt_Lib\prjObjs.lst Generating C:\VxWorks\WxWorks_Tt_Lib\prjObjs.lst... ccsimpc -nostdlib -r -Wl,-X -Wl,@C:\VxWorks\WxWorks_Tt_Lib\prjObjs.lst -o partialImage.o \Tornado\host\x86-win32\i386-pc-mingw32\bin\ld.exe: partialImage.o: reloc overflow: 0x1835 7 > 0xffff partialImage.o: final close failed: File truncated make: *** [tt_lib.out] Error 0x1 Has anyone solve this problem ? Thanks, RT rt@uniphone.com --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Gcc 2.95.3 for VxWorks-PPC available.... Date: 06 Feb 2001 01:58:12 -0500 From: "Paul D. Smith" Organization: Bay Networks, Inc. / Billerica, MA Message-ID: References: Reply-To: psmith@baynetworks.com Hey Dave; this looks very cool. It's also fortuitous because I'm having a problem related to this. Our product currently builds with the old GCC 2.7.2 PPC stuff from WRS. It's almost all C with a few token C++ files thrown in. I'm working on an "advanced technology" initiative and it's done in ISO C++ (no exceptions, thank goodness, and I don't think it uses RTTI, but it uses STL all over). Needless to say this doesn't even come close to compiling with GCC 2.7.2. I got a copy of binutils 2.10.1 and GCC 2.95.2 and built those, and was able to get all the code to compile and link. However, it will no longer boot (with or without the new code) :-( :-( :-(. It actually starts booting, and gets into some VxWorks code (usrConfig.c:usrRoot - --admittedly this is pretty early going :-/), then it tries to call sprintf() but winds up in the middle of the putenv() function :-P. If I disassemble the line, sure enough the last instruction is a branchl to . Weird. So: 1) Any thoughts? I suppose it must be a link problem, and this app does do some funky link-time antics. Since this is the only vxworks app I've used, I'm not sure if this is normal. I'm also suspicious of the problems addressed by your varargs patch, since sprintf() is a variadic function, and at least few other, non-variadic functions _appear_ to be referenced OK. However, I don't _think_ we use any long doubles (or even normal doubles) in our code. Not sure though. 2) First I should say that my host platform is Solaris, not Windows. Any extra thoughts or advice on building your patched GCC for Solaris? 3) What version of binutils did you use? Did you just keep the original binutils from WRS, and simply build a new GCC? Maybe that was my problem... Thanks... - -- - ------------------------------------------------------------------------------- Paul D. Smith HASMAT--HA Software Methods & 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: dialog with WindRiver - reasonable response Date: Tue, 06 Feb 2001 08:29:21 GMT From: george najarian Organization: najay engineering Message-ID: Here is a dialog with WindRiver regarding a problem I am having with the tools - can anybody tell me if I was being reasonable in my request, or not? Thanks - email dialog follows: - --------------------------------- I am limited in my ability to help you. 1) You are using a 3rd party-custom BSP 2) The gdbarm works without the EPI jenni jtag.-Implies that the gdbarm works. 3) I do not know how gdbarm is built. (I am not part of the development team that enhanced gdbarm) My question is how did EPI obtain source code for gdbarm? (Especially the custom hooks) Thanks Jack - -----Original Message----- From: Najarian, George [mailto:najay@conetcomm.com] Sent: Monday, February 05, 2001 2:19 PM To: 'Chow, Jack' Subject: RE: TSR 227541 so you don't plan on helping me? - -----Original Message----- From: Chow, Jack [mailto:Jack.Chow@windriver.com] Sent: Monday, February 05, 2001 11:46 AM To: 'Najarian, George' Subject: RE: TSR 227541 I do not have that type of information on hand. (build of gdbarm) Also, your board does not appear to be supported by Wind River (3rd party or custom bsp) The bsp was created by another vendor. Why don't you use the suggestion posted by Luke Diamond ? He responded to your post in the comp.os.vxworks to get the canonical gdb from FSF. Thanks Jack - -----Original Message----- From: Najarian, George [mailto:najay@conetcomm.com] Sent: Monday, February 05, 2001 1:47 PM To: 'Chow, Jack' Subject: RE: TSR 227541 I am using a v1.2 BSP based on this BSP : /* config.h - ARM PID configuration header */ /* Copyright 1996-1998 Wind River Systems, Inc. */ /* modification history - -------------------- 01a,05oct98,vnk  written from pid7t bsp by mBedded Innovations Inc. */ /* This module contains the configuration parameters for the ARM 6910 based Hydrogen Eval board. */ It has been modified by Intel to support their BC6911 based product. What am I supposed to do if the version of gdbarm that you released is proprietary and you will not help me debug it? Can you at least tell me how you build your copy of gdbarm? What tools and build environment to I use? What do I need to do to capture (and hopefully fix) the problem I am seeing? I want to let you know I am keeping logs of this and am going to post them to the appropriate newsgroups - I need to to get some opinions on whether you are violating the GPL or not by not releasing the source. I don't think I am asking too much ... just for the level of support that we already paid for. - -----Original Message----- From: Chow, Jack [mailto:Jack.Chow@windriver.com] Sent: Monday, February 05, 2001 11:18 AM To: 'Najarian, George' Subject: RE: TSR 227541 Please get the gdbarm from GPL. Wind River did make some enhancements off of the freeware version. Some portion of wind river's gdbarm is considered intellectual property. Thus, I can not give you the source. Thanks Jack - -----Original Message----- From: Najarian, George [mailto:najay@conetcomm.com] Sent: Monday, February 05, 2001 1:17 PM To: 'Chow, Jack' Subject: RE: TSR 227541 As far as I am concerned, the TSR is still open. Here is the text from my original request. - ------ Tornado license #: 104022. I am currently trying to find out why the gdbarm debugger is causing an internal exception when I try to debug my arm target with the JEENI JTAG EmbeddedIce debugger. The techs at EPI helped me with the log file that their dll generated and told me this is a gdb problem, most likely with an unitialized string. I can send detailed information of the failure mode if you are willing tohelp me. If you are not, then can you help me build a debugging version of gdbarm so I can trace the exception when it happens? I would need the source for gdbarm and instructions on how to rebuild it. Thank you for your prompt attention. - ------ What I am looking for is a debug version of the gdbarm that you supply with your toolset, or the source and build instructions so I can do it myself. Since it is GPL freeware, there should not be any issues with you releasing this information. Thanks. - -----Original Message----- From: Chow, Jack [mailto:Jack.Chow@windriver.com] Sent: Monday, February 05, 2001 8:38 AM To: 'Najarian, George' Subject: RE: TSR 227541 Can you please let me know what is the current status of this TSR? Thanks Jack - -----Original Message----- From: Najarian, George [mailto:najay@conetcomm.com] Sent: Thursday, February 01, 2001 4:32 PM To: 'Chow, Jack' Subject: RE: TSR 227541 I understand that. What I need and originally requested was instructions (including the tools used) on how to build gdb for the windriver environment. Can you help me with this? - -----Original Message----- From: Chow, Jack [mailto:Jack.Chow@windriver.com] Sent: Thursday, February 01, 2001 1:55 PM To: 'Najarian, George' Subject: RE: TSR 227541 Actually, I believe that gdbarm is freeware protected by the GNU General Publifc License (GPL) and you can probably grab the source somewhere. (maybe Cygnus Solutions). Since the debugger works without the JENNI JTAG, this is not really a tornado tools problem. Thanks Jack - -----Original Message----- From: Najarian, George [mailto:najay@conetcomm.com] Sent: Thursday, February 01, 2001 3:49 PM To: 'Chow, Jack' Subject: RE: TSR 227541 I recompile in the JEENI support when I need it. I only run one debugger environment at a time. - -----Original Message----- From: Chow, Jack [mailto:Jack.Chow@windriver.com] Sent: Thursday, February 01, 2001 1:37 PM To: 'Najarian, George' Subject: RE: TSR 227541 Just wondering why you are using two separate debuggers. I did not know that was possible. Thanks Jack - -----Original Message----- From: Najarian, George [mailto:najay@conetcomm.com] Sent: Thursday, February 01, 2001 3:45 PM To: 'Chow, Jack' Subject: RE: TSR 227541 Yes. I do not have any problems when I am running the wdbrpc interface. I am attaching a couple of .log files from the error dump when using the JENNI dll interface. - -----Original Message----- From: Chow, Jack [mailto:Jack.Chow@windriver.com] Sent: Thursday, February 01, 2001 1:30 PM To: 'najay@conetcomm.com' Subject: RE: TSR 227541 Good afternoon, My name is Jack and I have been assigned your TSR. Does the gdbarm work without the JENNI JTAG EmbeddedICE debugger attached? Thanks Jack   --------------------------- Newsgroups: comp.os.vxworks Subject: [Help Me] linker script file Date: Tue, 06 Feb 2001 08:54:36 GMT From: "Kim, Jeong-Hwan" Organization: Dacom Message-ID: I'm a novice in VxWorks In tornado for PPC, is there linker script file explicitly ? Where can I find the linker script file ? --------------------------- Newsgroups: comp.os.vxworks Subject: Re: *UPDATED* Gcc 2.95.3 for VxWorks-PPC available.... Date: Tue, 6 Feb 2001 09:12:28 GMT From: hoh@invalid.invalid (Goran Larsson) Organization: [1] + 5934 done /bin/rm -rf ~/ & Keywords: VxWorks Ppc Gcc 2.95.3 fnord! Message-ID: <2001Feb6.091228.6806@lorelei.approve.se> References: Sender: hoh@lorelei.approve.se.NO_JUNK_EMAIL (Goran Larsson) In article , Dave Korn wrote: > Finally I've recompiled the whole thing for i386, so it should run on any > CPU! Any CPU? We cross compile our PPC code on UltraSPARC and I'm sure that the UltraSPARC isn't going to like the 4004 derived opcodes. :-) - -- Göran Larsson Senior Systems Analyst hoh AT approve DOT se --------------------------- Newsgroups: comp.os.vxworks Subject: Re: dialog with WindRiver - reasonable response Date: Tue, 06 Feb 2001 10:09:59 +0100 From: Jan =?iso-8859-1?Q?Dj=E4rv?= Organization: Ericsson Erisoft AB Message-ID: <3A7FBF67.AD49A900@lu.erisoft.se> References: > -----Original Message----- > From: Chow, Jack [mailto:Jack.Chow@windriver.com] > Sent: Monday, February 05, 2001 11:18 AM > To: 'Najarian, George' > Subject: RE: TSR 227541 > Please get the gdbarm from GPL. > Wind River did make some enhancements off of the freeware version. > Some portion of wind river's gdbarm is considered intellectual property. > Thus, I can not give you the source. > Thanks > Jack If the "enhancements" are linked in the gdbarm executable and if WindRiver distributed this enhanced gdbarm to you, not giving you the source is clearly a violation of the GPL. I suggest you take this up with the copyright holder of GDB proper, which is FSF. Jan D. --------------------------- Newsgroups: comp.os.vxworks Subject: ATA configuration & CDFS / DVD ROM drivers Date: Tue, 06 Feb 2001 11:46:22 GMT From: Alex Holland Organization: Deja.com Message-ID: <95oo6c$s10$1@nnrp1.deja.com> Hi, I have a PPC target device, which has an EIDE(ATAPI) hardware interface. Can anyone point me in the direction of how to configure the OS for my hardware, and anyone who does CDFS (ISO9660) and DVD-ROM file system handlers for VXWorks?? - -- Cheers Alex Holland Sent via Deja.com http://www.deja.com/ --------------------------- End of New-News digest ********************** From vxwexplo-errs@csg.lbl.gov Tue Feb 6 19:13:49 2001 From: sachithanandam karthikselvan Date: Tue Feb 6 19:13:51 PST 2001 Subject: build problem Hi all, 1) I included some c files and then tried to built it. It says "no such files or directroy". 2) I checked the makefile. Instead full path, it has file name only. ex. rnlog.h. So I gave full path ex. c:\sssafe\embedd\rnlog.h in the makefile itself. 3) Now I said make from the dos prompt.Again it says that file not found. But it is able to comple other vxworks related files like romstart.c 4) Can anyone throw some lights?? Thanks in advance karthikselvan ____________________________________________________________________ Get free email and a permanent address at http://www.netaddress.com/?N=1 From vxwexplo-errs@csg.lbl.gov Wed Feb 7 04:03:09 2001 From: Vxworks Exploder Date: Wed Feb 7 04:03:12 PST 2001 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Wed Feb 7 04:03:05 PST 2001 Subject: Re: dialog with WindRiver - reasonable response Subject: Re: *UPDATED* Gcc 2.95.3 for VxWorks-PPC available.... Subject: Re: Gcc 2.95.3 for VxWorks-PPC available.... Subject: Problems getting tRdbTask up and running... Subject: Re: Gcc 2.95.3 for VxWorks-PPC available.... Subject: SM network with a non-VxWorks system controller Subject: TFFS alternatives Subject: Software Engineer (VxWorks) with Ericsson Subject: VxWorks performance Subject: memShow output: invalid block at 0xfffd000 deleted Subject: Re: SM network with a non-VxWorks system controller Subject: setjmp unresolved in VxSim Subject: unknown error message Subject: command line build help Subject: Re: command line build help - solved - don't bother Subject: Re: Specify location for code Subject: How can i install and use VxVMI? Subject: Telecom Jobs Electro-Optics Software Project Leader - Ottawa Subject: Telecom Jobs Electro-Optical Sensor Software Engineers - Ottawa Subject: Telecom Jobs Real-time Embedded Electro-Optics Software Engineers - Ottawa Subject: Re: Specify location for code Subject: Re: [Help Me] linker script file Subject: Re: VxWorks performance Subject: Re: How can i install and use VxVMI? Subject: libstdc++ for vxworks + gcc 2.95.3? Subject: Task total running time. Subject: Re: libstdc++ for vxworks + gcc 2.95.3? Subject: Re: libstdc++ for vxworks + gcc 2.95.3? Subject: build problem Subject: Re: dialog with WindRiver - reasonable response Subject: sbrk equivalent in VxWorks Subject: Re: VxSim: reloc overflow Subject: Re: *UPDATED* Gcc 2.95.3 for VxWorks-PPC available.... Subject: Re: malloc and free in vxworks Subject: Interrup trouble on Assabet SA-1110? Subject: Re: unknown error message ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: dialog with WindRiver - reasonable response Date: Tue, 6 Feb 2001 10:17:47 -0000 From: "Dave Korn" Organization: Lumber Cartel (tinlc) Members #2234-2237 (owing to browser refresh) Message-ID: References: george najarian wrote in message ... >Here is a dialog with WindRiver regarding a problem I am having with >the tools - can anybody tell me if I was being reasonable in my request, >or not? Thanks - You are being absolutely reasonable, but you're talking to the wrong monkey. You *can* get a cd with all the source code for WRS' modified toolchain; they are of course *obliged* to make it available to you under the terms of the GPL. I believe you'll have more luck if you speak to a sales monkey than a support monkey. >-----Original Message----- >From: Chow, Jack [mailto:Jack.Chow@windriver.com] >Sent: Monday, February 05, 2001 11:18 AM >To: 'Najarian, George' >Subject: RE: TSR 227541 >Please get the gdbarm from GPL. >Wind River did make some enhancements off of the freeware version. >Some portion of wind river's gdbarm is considered intellectual property. >Thus, I can not give you the source. >Thanks >Jack This guy is an idiot, a complete gibbering fool. Technically, this statement places WRS in breach of the terms of the GPL, and the FSF could sue them. Fortunately I don't believe this represents company policy, just an under-trained support monkey. You should ask to speak to his management and explain that they *really* don't want their employees making illegal claims like that. DaveK - -- They laughed at Galileo. They laughed at Copernicus. They laughed at Columbus. But remember, they also laughed at Bozo the Clown. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: *UPDATED* Gcc 2.95.3 for VxWorks-PPC available.... Date: Tue, 6 Feb 2001 10:18:09 -0000 From: "Dave Korn" Organization: Lumber Cartel (tinlc) Members #2234-2237 (owing to browser refresh) Message-ID: References: <2001Feb6.091228.6806@lorelei.approve.se> Goran Larsson wrote in message <2001Feb6.091228.6806@lorelei.approve.se>... >In article , >Dave Korn wrote: > >> Finally I've recompiled the whole thing for i386, so it should run on any >> CPU! > >Any CPU? We cross compile our PPC code on UltraSPARC and I'm sure that >the UltraSPARC isn't going to like the 4004 derived opcodes. :-) No really, it'll run on any CPU, but only one architecture: xxx-pc-cygwin32. Do you have a copy of cygwin dll for UltraSPARC? OK, I meant any CPU that might be used in a PC compatible. I know all the world is not an IBM really. Cor, anyone who just wandered in here would think we're a bunch of pedantic nerdy computer engineers. DaveK - -- They laughed at Galileo. They laughed at Copernicus. They laughed at Columbus. But remember, they also laughed at Bozo the Clown. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Gcc 2.95.3 for VxWorks-PPC available.... Date: Tue, 6 Feb 2001 11:40:17 -0000 From: "Dave Korn" Organization: Lumber Cartel (tinlc) Members #2234-2237 (owing to browser refresh) Message-ID: References: Paul D. Smith wrote in message ... >Hey Dave; this looks very cool. Even cooler now I've fixed the file-path bug.. >STL all over). Needless to say this doesn't even come close to >compiling with GCC 2.7.2. Hey, 2.7.2 is only five years old, you don't expect WRS to have updated it yet? They'd look almost awake if they did that! >I got a copy of binutils 2.10.1 and GCC 2.95.2 and built those, and was >able to get all the code to compile and link. However, it will no >longer boot (with or without the new code) :-( :-( :-(. It actually >starts booting, and gets into some VxWorks code (usrConfig.c:usrRoot >--admittedly this is pretty early going :-/), then it tries to call >sprintf() but winds up in the middle of the putenv() function :-P. If I >disassemble the line, sure enough the last instruction is a branchl to >. Weird. > >So: > > 1) Any thoughts? I suppose it must be a link problem, and this app > does do some funky link-time antics. Since this is the only vxworks > app I've used, I'm not sure if this is normal. Nope, it's not a link problem.... > I'm also suspicious of the problems addressed by your varargs patch, > since sprintf() is a variadic function, and at least few other, > non-variadic functions _appear_ to be referenced OK. > > However, I don't _think_ we use any long doubles (or even normal > doubles) in our code. Not sure though. This problem *is* the varargs problem, which is kind of over-and-above the alignment of doubles problem. (Don't forget that owing to default type promotion, you *do* use doubles in your code the moment you pass a floating point number to a variadic function). What goes wrong with varargs is that the 2.7.2 header files use these obsolete compiler builtins called __builtin_args_info to get details of the arguments that were passed to the varargs function. One of these functions returns an incorrect value under 2.95.x, since they aren't supposed to be used any more, and the result is th at the variadic function goes looking at some random and potentially non-word-aligned part of the stack for its args. > 2) First I should say that my host platform is Solaris, not Windows. > Any extra thoughts or advice on building your patched GCC for > Solaris? Erm.... hmm. My patch that fixes the path problem may not be a valid solution for Solaris... the best approach would probably be to remove the last three #defines in gcc/config/rs6000/vxppc.h: #define HAVE_DOS_BASED_FILESYSTEM #define FORCE_DOS_PATH_STYLE 1 #define SYNTHETIC_VXWORKS_PREFIXES I should probably have put them into the cygwin-host-dependent file instead I guess. This only applies if you have the code dated Jan 6th from the website; the slightly older version with the 'paths bug' will actually work correctly I think. Apart from that, building and installing should proceed as normal. The windows binary distribution I've made simply renames files to the WRS names and drops them into the right places in the Tornado directory tree. > 3) What version of binutils did you use? Did you just keep the > original binutils from WRS, and simply build a new GCC? Maybe that > was my problem... I used a snapshot of binutils dated 20000722, but it seems to be fairly similar to 2.10. I had to make the original distribution use the WRS ldppc, since the binutils version doesn't understand the '@' file list option. I've now made a patched version of ld that does understand it, I'll upload a patch file to the website as soon as I get time. DaveK - -- They laughed at Galileo. They laughed at Copernicus. They laughed at Columbus. But remember, they also laughed at Bozo the Clown. --------------------------- Newsgroups: comp.os.vxworks Subject: Problems getting tRdbTask up and running... Date: Tue, 6 Feb 2001 15:10:15 +0100 From: "Eelco Chaudron" Message-ID: <981468786.22576.0.pluto.d4ee6f20@news.demon.nl> Hi, I'm trying to get my vxworks kernel to startup the tRdbTask so I can use GDB to debug it. I did include INCLUDE_RDB, but my system is not starting the debug task. I also made sure that INCLUDE_RPC, _DEBUG, _LOADER and _UNLOADER where included. Any idea which calls I have to make and where to get it working? It's for a MIPS based platform. Thanks, Eelco --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Gcc 2.95.3 for VxWorks-PPC available.... Date: Tue, 06 Feb 2001 06:24:14 -0800 From: Matthew Semersky Organization: Iospan Wireless, Inc. Message-ID: <3A80090E.5617A92B@blee.com> References: Dave Korn wrote: > > > I'm also suspicious of the problems addressed by your varargs patch, > > since sprintf() is a variadic function, and at least few other, > > non-variadic functions _appear_ to be referenced OK. > > > > However, I don't _think_ we use any long doubles (or even normal > > doubles) in our code. Not sure though. > > This problem *is* the varargs problem, which is kind of over-and-above the > alignment of doubles problem. (Don't forget that owing to default type > promotion, you *do* use doubles in your code the moment you pass a floating > point number to a variadic function). What goes wrong with varargs is that > the 2.7.2 header files use these obsolete compiler builtins called > __builtin_args_info to get details of the arguments that were passed to the > varargs function. One of these functions returns an incorrect value under > 2.95.x, since they aren't supposed to be used any more, and the result is th > at the variadic function goes looking at some random and potentially > non-word-aligned part of the stack for its args. This sounds a bit odd to me. The varargs ABI incompatibility shouldn't come into question when calling sprintf().... vsprintf() sure, but I don't see why sprintf would be using any of the compiler builtins. I hacked up my own copy of the vxWorks libraries to get around the ABI change... and since sprintf() is in the 2.7.2 library and a prototype like sprintf( char *, const char *, ... ) doesn't invoke any of the compiler __builtin's.... you may have another problem entirely. Assuming you hack up a 2.7.2 version of vxWorks libraries and put your own stdio calls in for vsprintf, vprintf, etc.... you should be able to use the stock ABI in 2.95.2. Of course, that can be more trouble than it is worth. Matt Semersky --------------------------- Newsgroups: comp.os.vxworks Subject: SM network with a non-VxWorks system controller Date: 06 Feb 2001 15:07:27 +0000 From: Charlie Wood Organization: Spider Software Limited Message-ID: <78elxb4zps.fsf@orchy.spider.com> I have several '860 based cards, housed in a cPCI chassis, running VxWorks 5.4. None of the cards has it's own network interface. With a system controller also running VxWorks 5.4, which has a network interface, I can use Tornado with these cards by running a virtual shared memory network over the backplane. If have now to use a Sparc card running Solaris 8 as my system contoller, is there any way (other than a serial link to the '860 cards) to continue using Tornado? Is it possible for the Solaris S/C to host the shared memory network? - -- charlie wood Spider Software Limited +44 (0)131 475 7028 o/ http://www.spider.com/ ___#___ Veni Vidi Natavi ~~~~~~~~~~/~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ --------------------------- Newsgroups: comp.os.vxworks Subject: TFFS alternatives Date: Tue, 06 Feb 2001 17:06:38 GMT From: bblank@init-ka.de Organization: Deja.com Message-ID: <95paul$cmo$1@nnrp1.deja.com> I'm looking for a TrueFFS alternative (e.g. JFFS), runnning on vxWorks and NT. Has anybody an idea or even made a port to vxWorks? Is there a TFFS running on NT (we use our own flash devices)? Thanx Bernd Blank Sent via Deja.com http://www.deja.com/ --------------------------- Newsgroups: comp.os.vxworks Subject: Software Engineer (VxWorks) with Ericsson Date: Tue, 06 Feb 2001 17:16:08 GMT From: Mike DeCellis Organization: Deja.com Message-ID: <95pbgc$d9q$1@nnrp1.deja.com> I noticed some other job postings in this group with needs for embedded engineers, so I thought I might post some applicable needs...but if this is not wanted or wrong, I apologize, and please let me know. I am an internal recruiter with Ericsson in Raleigh (RTP), North Carolina, and have been tasked in finding software engineers. Our position entails the use of object-oriented and/or structured design methods, C, C++, UNIX, DSP, NT and real-time embedded software, real- time operating systems to support our Base Station and DWOS products. We are currently using VxWorks as our RTOS. We have intermediate and senior level positions available. Senior positions will consist of recommending actions, modifying software design, assisting and mentoring other engineers. These individuals would be working with our research and development facility in Raleigh, NC, specifically in our DWOS (wireless office) and Base Station development groups. Interested individuals should have 2-7 years of related work experience with all phases of software development. Experience in one or more of the following technologies is needed: C, C++, UNIX, NT, DSP, real-time embedded software, real-time operating systems (VxWorks) and software design/development. Experience with data or voice communications development and protocols (TDMA, GSM, T1, E1, etc.) are a big plus, preferably for wireless applications. Knowledge of object-oriented design is a plus. Any help that you can give me, in either locating qualified individuals for the roles, or potential places to look, would be most appreciated. Additionally, I would welcome an opportunity to speak with you directly if you are interested yourself. Thank you very much and Best Regards, - -- Mike DeCellis Ericsson Recruiter RTP, NC (Raleigh-Durham) Phone: 919-472-6541 Fax: 919-472-6510 Mobile: 919-522-2815 mike.decellis@ericsson.com Sent via Deja.com http://www.deja.com/ --------------------------- Newsgroups: comp.os.vxworks Subject: VxWorks performance Date: Tue, 6 Feb 2001 18:40:36 +0100 From: "r1p3" Organization: tp.internet - http://www.tpi.pl Message-ID: <95pcqh$soi$1@news.tpi.pl> Hi I'm working on my diploma thesis and I need information about VxWorks/Tornado: - - typical context switching time, - - maximum interrupt latency, - - system clock's minimum resolution, - - kernel ROM minimum size, - - and price. Can you help me? Thanks in advance... r1p3 --------------------------- Newsgroups: comp.os.vxworks Subject: memShow output: invalid block at 0xfffd000 deleted Date: Tue, 06 Feb 2001 19:02:57 +0100 From: Michael Fischer / 30050 Organization: Swisscom IP+ (post doesn't reflect views of Swisscom) Message-ID: <3A803C51.22AD49F7@acterna.com> This is a multi-part message in MIME format. - --------------22FC953AA0711C07D557D6A2 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit If I use the memShow on the target shell, I get this: - -> - -> memShow invalid block at 0xfffd000 deleted value = -1 = 0xffffffff = _VxWorksIPaddress + 0xf0000d27 - -> "invalid block" doesn't sound very good, is this ok or an indication of a problem in our system? We are using Tornado I and the image has been built with INCLUDE_SHOW_ROUTINES. I also tried to use memShowInit before memShow, I get the same output. Thanks in advance. Michael Fischer - --------------22FC953AA0711C07D557D6A2 Content-Type: text/x-vcard; charset=us-ascii; name="michael.fischer.vcf" Content-Transfer-Encoding: 7bit Content-Description: Card for Michael Fischer / 30050 Content-Disposition: attachment; filename="michael.fischer.vcf" begin:vcard n:Fischer;Michael tel;fax:++41 76 355 67 52 tel;work:++41 1 355 67 39 x-mozilla-html:FALSE org:Acterna version:2.1 email;internet:michael.fischer@acterna.com title:Software Engineer adr;quoted-printable:;;Foerrlibuckstr. 62=0D=0APostfach 74=0D=0A;Zuerich;;CH-8037;Switzerland x-mozilla-cpt:;29488 fn:Michael Fischer end:vcard - --------------22FC953AA0711C07D557D6A2-- --------------------------- Newsgroups: comp.os.vxworks Subject: Re: SM network with a non-VxWorks system controller Date: Tue, 06 Feb 2001 19:42:10 +0100 From: Sebastien Fauris Message-ID: <3A804582.7030708@libertysurf.fr> References: <78elxb4zps.fsf@orchy.spider.com> Charlie, if your 860 target is a Force Computer board, you can get busNet from your support contract which is the Force implementation of the sm driver but supported with different OSes (VxWorks, Solaris, NT, ...). Regards, Sebastien. Charlie Wood wrote: > I have several '860 based cards, housed in a cPCI chassis, running > VxWorks 5.4. None of the cards has it's own network interface. > > With a system controller also running VxWorks 5.4, which has a network > interface, I can use Tornado with these cards by running a virtual > shared memory network over the backplane. > > If have now to use a Sparc card running Solaris 8 as my system > contoller, is there any way (other than a serial link to the '860 > cards) to continue using Tornado? Is it possible for the Solaris S/C > to host the shared memory network? --------------------------- Newsgroups: comp.os.vxworks Subject: setjmp unresolved in VxSim Date: 6 Feb 2001 19:03:21 GMT From: Juergen Naumann Organization: Uni Dortmund, Physik Message-ID: <95phpp$e7n$1@nx6.HRZ.Uni-Dortmund.DE> Reply-To: naumann@Physik.Uni-Dortmund.DE Dear all I have problems using setjmp/longjmp with vxSim built in T2 on Windows NT4.0. Loading the program to the simulator leads to Unresolved symbols encountered while downloading: _longjmp _setjmp I built my own simulator which doesn't solve the problem. Furthermore I didn't find the place where to include the setjmp-Library, because you may include all ANSI C stuff under - -> operating system components -> ANSI C components (libc) but not the setjmp stuff. Doing ar -t Tornado/target/lib/libSIMNTgnuvx.a shows that "setjmp" is not there, but "longjmp" is (!?), scanning this file with an editor shows an entry for "_longjmp" and "setjmp" (without an underscore). The Browser shows an entry for "setjmp" (again without an underscore), but none for "_longjmp". Any idea? - -- Juergen e----------------------------->H1<---------------------------p --------------------------- Newsgroups: comp.os.vxworks Subject: unknown error message Date: Tue, 06 Feb 2001 19:00:46 GMT From: kimp@avtecinc.com (Kim Putnam) Organization: Info Avenue Internet Services Message-ID: <3a804305.21210562@192.168.1.214> I need to locate any information on bootLoadModule(). I am getting the following error "S_bootLoadLib_ROUTINE_NOT_INSTALLED" kim --------------------------- Newsgroups: comp.os.vxworks Subject: command line build help Date: Tue, 06 Feb 2001 19:42:04 GMT From: bpuli@my-deja.com Organization: Deja.com Message-ID: <95pk2b$m06$1@nnrp1.deja.com> hi all: a simple question from a configuration manager. bear with me if i get some of the terminology wrong. i want to be able to build from the command line. all the developers use the tornado gui - so i can depend on the .wpj files. i have been able to generate a makefile from a .wpj file by using the makeGen.tcl file (a hacked configGen.tcl).and used it successfully. however this does not work for bsp builds. the makefile that makeGen.tcl is generating is subtantially different from the one that the GUI is generating (from the same .wpj file) i guess, the question would be, what tcl function does the gui call to generate the makefiles, if not the one that makeGen.tcl is calling (prj_vxApp_hidden::makeGen)? any pointers appreciated thanks in advance bp - -- bp Sent via Deja.com http://www.deja.com/ --------------------------- Newsgroups: comp.os.vxworks Subject: Re: command line build help - solved - don't bother Date: Tue, 06 Feb 2001 19:56:10 GMT From: bpuli@my-deja.com Organization: Deja.com Message-ID: <95pksg$n4g$1@nnrp1.deja.com> References: <95pk2b$m06$1@nnrp1.deja.com> the same function in prj_vxWorks*.tcl should have looked a little better ;-) bp In article <95pk2b$m06$1@nnrp1.deja.com>, bpuli@my-deja.com wrote: > hi all: > > a simple question from a configuration manager. bear with me if i get > some of the terminology wrong. > > i want to be able to build from the command line. all the developers > use the tornado gui - so i can depend on the .wpj files. > > i have been able to generate a makefile from a .wpj file by using the > makeGen.tcl file (a hacked configGen.tcl).and used it successfully. > > however this does not work for bsp builds. the makefile that > makeGen.tcl is generating is subtantially different from the one that > the GUI is generating (from the same .wpj file) > > i guess, the question would be, what tcl function does the gui call to > generate the makefiles, if not the one that makeGen.tcl is calling > (prj_vxApp_hidden::makeGen)? > > any pointers appreciated > > thanks in advance > > bp > > -- > bp > > Sent via Deja.com > http://www.deja.com/ > - -- bp Sent via Deja.com http://www.deja.com/ --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Specify location for code Date: Tue, 06 Feb 2001 20:41:00 GMT From: melloids@my-deja.com Organization: Deja.com Message-ID: <95pngq$pr5$1@nnrp1.deja.com> References: <95mhep$tl6$1@nnrp1.deja.com> <95nao1$n8a$1@nnrp1.deja.com> Hi John, Thanks for the response. I had been planning on building my function into the vxWorks build, but I'll give the option of loading it with loadModuleAt some thought. I need to locate the code at a specific location because I'm trying to write a cache-loading routine to pre-load code into the instruction cache. The cache-loading routine itself needs to be in non-cacheable memory so it doesn't end up getting loaded into the instruction cache. So I've set up the MMU to create a cache- inhibited memory area, and I want the linker to stick my cache-loading routine into that area. Matt In article <95nao1$n8a$1@nnrp1.deja.com>, john_94501@my-deja.com wrote: > Hello, > > You could always use the dynamic loader (target loader); use the > loadModuleAt() to specify addresses for text, data and bss segments (or > any subset of those three). > > Of course, this only works if you are willing to load the module at > runtime, and don't want it linked at a specific address as part of the > build process. > > HTH, > > John... > > PS Out of interest, why do you need the function placed at a specific > address? > > In article <95mhep$tl6$1@nnrp1.deja.com>, > melloids@my-deja.com wrote: > > I'm using VxWorks 5.4, Tornado 2.0 on a PowerPC > > MCP750 board. I am writing a function that I > > need the linker to place at a specific address. > > I've gotten this to work by messing with the > > linker script, but in order to aviod a multiple > > define error on the linker, I've had to remove > > the source file from prjObjs.lst. The result of > > this is that the symbols for myt function do not > > appear in the symbol table. > > > > Does anyone know how to get tornado 2.0 to locate > > a function at a specific address? > > > > Thanks in advance, > > Matt > > > > Sent via Deja.com > > http://www.deja.com/ > > > > Sent via Deja.com > http://www.deja.com/ > Sent via Deja.com http://www.deja.com/ --------------------------- Newsgroups: comp.os.vxworks Subject: How can i install and use VxVMI? Date: Mon, 5 Feb 2001 19:26:59 +0800 From: "Nison" Organization: Utstarcom Message-ID: <95m2pn$ea0$1@news.cz.js.cn> Reply-To: "Nison" Hi all: I want to add virtual memory support in my own program,but i can't add VxVMI module in my kernel,why? How to install it and use it,thanks for your help... - --------------- Best Regards! Nison 02/05/01 --------------------------- Newsgroups: comp.os.vxworks,comp.realtime Subject: Telecom Jobs Electro-Optics Software Project Leader - Ottawa Date: Tue, 6 Feb 2001 17:52:20 -0500 From: "PROCOM" Organization: IGS - Information Gateway Services Message-ID: <95pv7o$5al$2@news.igs.net> Our client is currently seeking people who wish to join a team of other professionals providing software development, programming and systems analysis for our Infrared Search and Track (IRST) development program. Applicants must have a formal education in computer science or electrical engineering with at least 8 years of experience developing software in a structured development environment. In addition to possessing initiative and drive, the successful candidate is required to have had experience integrating and testing complex software, as well as the ability for working independently and collaborating in teams. Highly desired are programming skills and significant experience in a broad range of languages (C, C++, Fortran, IDL) and operating systems (Windows NT, Unix, VxWorks). Also preferred is experience working on structured military software development programs (Mil-Std-2167A, Mil-Std-498), knowledge of the infrared application domain (environments, models, targets), real-time development experience, team and leadership experience on software development teams, exposure to object-oriented design methodologies, preferably Booch and UML, and knowledge of or exposure to DSP. Dutch or Canadian citizenship is a requirement, due to the requirements of the program. Candidates with an active Canadian security clearance also will be considered. Please indicate that you meet this requirement on the subject line of your cover letter in order to be considered for this opportunity. Compensation: Open ended Contract and Full time PROCOM: Established in 1978, Professional Computer Consultants Group Ltd. (Procom) is a national leader in the provision of Computer personnel on a contract and full-time basis. Our clients are comprised of the largest national and international corporations that utilize technical resources extensively across a wide range of disciplines. In the Financial Post (March 1999) Procom was ranked as the 6th largest professional Services Company in Canada. In November of 1999, Procom was named a Regional finalist in Canada' s 50 Best Managed Private companies. Our track record is proven with more than 180 consultants servicing Ottawa's high tech community and more than 1600 Procom consultants currently on assignment throughout. North America. For further information on this and other opportunities please visit our web site at www.procom.ca. Derek Weber PROCOM 300 March Rd Suite 600 Kanata, Ontario K2K-2E2 613-270-9339 x231 613-270-9449 (FAX) derekw@procom.ca www.procom.ca --------------------------- Newsgroups: comp.os.vxworks,comp.realtime Subject: Telecom Jobs Electro-Optical Sensor Software Engineers - Ottawa Date: Tue, 6 Feb 2001 17:52:09 -0500 From: "PROCOM" Organization: IGS - Information Gateway Services Message-ID: <95pv7n$5al$1@news.igs.net> Our client is currently seeking Electro-Optical Sensor Systems Software Engineers to join there development program. Those selected will work with other professionals performing software development, programming and systems analysis. Applicants should have a formal education in computer science or electrical engineering, with at least 3 years of experience in real-time software development. In addition to possessing initiative and drive, successful candidates must have significant experience developing software in C and/or C++ on a real-time operating system, and the ability to read and understand a complex SRS. Knowledge of the infrared application domain (environments, models and targets)and experience working on a military software development program (MIL-STD-2167A or MIL-STD-498) are highly desired. Also desired are experience in object-oriented design, C++ and VxWorks, plus team and leadership experience on software development teams. Dutch or Canadian citizenship is required. Candidates with an active Canadian security clearance also will be considered. Applicants must indicate that they meet this requirement on the subject line of cover letters in order to be considered for this position. Compensation: Open ended Contract and Full time PROCOM: Established in 1978, Professional Computer Consultants Group Ltd. (Procom) is a national leader in the provision of Computer personnel on a contract and full-time basis. Our clients are comprised of the largest national and international corporations that utilize technical resources extensively across a wide range of disciplines. In the Financial Post (March 1999) Procom was ranked as the 6th largest professional Services Company in Canada. In November of 1999, Procom was named a Regional finalist in Canada' s 50 Best Managed Private companies. Our track record is proven with more than 180 consultants servicing Ottawa's high tech community and more than 1600 Procom consultants currently on assignment throughout. North America. For further information on this and other opportunities please visit our web site at www.procom.ca. Derek Weber PROCOM 300 March Rd Suite 600 Kanata, Ontario K2K-2E2 613-270-9339 x231 613-270-9449 (FAX) derekw@procom.ca www.procom.ca --------------------------- Newsgroups: comp.os.vxworks,comp.realtime Subject: Telecom Jobs Real-time Embedded Electro-Optics Software Engineers - Ottawa Date: Tue, 6 Feb 2001 17:52:34 -0500 From: "PROCOM" Organization: IGS - Information Gateway Services Message-ID: <95pv7o$5al$3@news.igs.net> Our client is searching for entry, mid-level and senior Software Engineers to develop applications to operate on real-time embedded CPU hardware platforms for military display and radar systems. The successful applicants will be responsible for all phases of software development. Knowledge of C, C++ and UNIX is required. A B.S.C.S. or B.S.E.E. degree also is required, and applicants must be eligible for a government Secret-level clearance. Our client utilizes advanced commercial technology for cost-effective solutions in the design and manufacture of a variety of air-, land- and sea-based advanced defense electronics systems, including combat display systems, portable workstations, trainer and emulation systems, coastal surveillance systems and radar systems. Compensation: Open ended Contract and Full time PROCOM: Established in 1978, Professional Computer Consultants Group Ltd. (Procom) is a national leader in the provision of Computer personnel on a contract and full-time basis. Our clients are comprised of the largest national and international corporations that utilize technical resources extensively across a wide range of disciplines. In the Financial Post (March 1999) Procom was ranked as the 6th largest professional Services Company in Canada. In November of 1999, Procom was named a Regional finalist in Canada' s 50 Best Managed Private companies. Our track record is proven with more than 180 consultants servicing Ottawa's high tech community and more than 1600 Procom consultants currently on assignment throughout. North America. For further information on this and other opportunities please visit our web site at www.procom.ca. Derek Weber PROCOM 300 March Rd Suite 600 Kanata, Ontario K2K-2E2 613-270-9339 x231 613-270-9449 (FAX) derekw@procom.ca www.procom.ca --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Specify location for code Date: Tue, 06 Feb 2001 23:11:16 GMT From: john_94501@my-deja.com Organization: Deja.com Message-ID: <95q0ac$2jm$1@nnrp1.deja.com> References: <95mhep$tl6$1@nnrp1.deja.com> <95nao1$n8a$1@nnrp1.deja.com> <95pngq$pr5$1@nnrp1.deja.com> Hello Matt, OK, I have another suggestion then... You could force the alignment of your pre-load code such that it is on a page boundary (4Kb); this is easy with assembler code, perhaps not so easy with C code. Having done that, you can use vmBaseStateSet() to mark the page as uncacheable at runtime rather than worrying about a fixed location. One solution for aligning C code might be to place it in a section of its own, and modify/write a linker script to align that section (see the GNU manual for details on specifying section names for functions, and aligning them using linker scripts). HTH, John... > Thanks for the response. I had been planning on building my function > into the vxWorks build, but I'll give the option of loading it > with loadModuleAt some thought. > > I need to locate the code at a specific location because I'm trying > to write a cache-loading routine to pre-load code into the > instruction cache. The cache-loading routine itself needs to be > in non-cacheable memory so it doesn't end up getting loaded into > the instruction cache. So I've set up the MMU to create a cache- > inhibited memory area, and I want the linker to stick my cache-loading > routine into that area. > > Matt > > In article <95nao1$n8a$1@nnrp1.deja.com>, > john_94501@my-deja.com wrote: > > Hello, > > > > You could always use the dynamic loader (target loader); use the > > loadModuleAt() to specify addresses for text, data and bss segments > (or > > any subset of those three). > > > > Of course, this only works if you are willing to load the module at > > runtime, and don't want it linked at a specific address as part of the > > build process. > > > > HTH, > > > > John... > > > > PS Out of interest, why do you need the function placed at a specific > > address? > > > > In article <95mhep$tl6$1@nnrp1.deja.com>, > > melloids@my-deja.com wrote: > > > I'm using VxWorks 5.4, Tornado 2.0 on a PowerPC > > > MCP750 board. I am writing a function that I > > > need the linker to place at a specific address. > > > I've gotten this to work by messing with the > > > linker script, but in order to aviod a multiple > > > define error on the linker, I've had to remove > > > the source file from prjObjs.lst. The result of > > > this is that the symbols for myt function do not > > > appear in the symbol table. > > > > > > Does anyone know how to get tornado 2.0 to locate > > > a function at a specific address? > > > > > > Thanks in advance, > > > Matt > > > > > > Sent via Deja.com > > > http://www.deja.com/ > > > > > > > Sent via Deja.com > > http://www.deja.com/ > > > > Sent via Deja.com > http://www.deja.com/ > Sent via Deja.com http://www.deja.com/ --------------------------- Newsgroups: comp.os.vxworks Subject: Re: [Help Me] linker script file Date: 6 Feb 2001 23:41:44 GMT From: borkhuis@agere.com (Johan Borkhuis) Organization: Lucent Technologies Message-ID: References: "Kim, Jeong-Hwan" wrote: >I'm a novice in VxWorks >In tornado for PPC, is there linker script file explicitly ? >Where can I find the linker script file ? N6, there is no linker command file, except for the implicit one. If you want to know how this looks like try to create a link-file. The linker command file is in there. Groeten, Johan - -- o o o o o o o . . . _____________________________ o _____ || Johan Borkhuis | .][__n_n_|DD[ ====_____ | borkhuis@agere.com | >(________|__|_[_________]_|__________________________| _/oo OOOOO oo` ooo ooo 'o!o!o o!o!o` === VxWorks FAQ: http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html === --------------------------- Newsgroups: comp.os.vxworks Subject: Re: VxWorks performance Date: 6 Feb 2001 23:43:56 GMT From: borkhuis@agere.com (Johan Borkhuis) Organization: Lucent Technologies Message-ID: References: <95pcqh$soi$1@news.tpi.pl> "r1p3" wrote: >Hi > >I'm working on my diploma thesis and I need information about >VxWorks/Tornado: > >- typical context switching time, >- maximum interrupt latency, >- system clock's minimum resolution, >- kernel ROM minimum size, >- and price. > >Can you help me? Thanks in advance... Contact your sales representative or an FAE. They can provide you with this information. 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: How can i install and use VxVMI? Date: Tue, 06 Feb 2001 23:48:33 GMT From: john_94501@my-deja.com Organization: Deja.com Message-ID: <95q2ge$4j6$1@nnrp1.deja.com> References: <95m2pn$ea0$1@news.cz.js.cn> Hello, VxVMI is an optional product that you must buy separately to use (assuming that it is available for your CPU). That said, now you might be better off looking at the new VxWorks AE product if you need the features offered by VxVMI. HTH, John... In article <95m2pn$ea0$1@news.cz.js.cn>, "Nison" wrote: > Hi all: > > I want to add virtual memory support in my own program,but i can't add > VxVMI module in my kernel,why? How to install it and use it,thanks for your > help... > > --------------- > Best Regards! > Nison > 02/05/01 > > Sent via Deja.com http://www.deja.com/ --------------------------- Newsgroups: comp.os.vxworks Subject: libstdc++ for vxworks + gcc 2.95.3? Date: 06 Feb 2001 19:50:34 -0500 From: "Paul D. Smith" Organization: Bay Networks, Inc. / Billerica, MA Message-ID: Reply-To: psmith@baynetworks.com I'm using davek's patched 2.95.3 and it seems to work well, but my app can't link because it can't find libstdc++.a. Sure enough, it's not in the gcc-lib/powerpc-wrs-vxworks/2.95.3 directory. I saw a note somewhere on his site saying that libstdc++ doesn't build for vxworks? I have to have this library, as my code is using C++ STL, etc.! I built 2.95.2 from GNU sources for vxworks and libstdc++ built fine there--is this a know 2.95.3 thing? Or a known build issue? Or...? Thx. - -- - ------------------------------------------------------------------------------- Paul D. Smith HASMAT--HA Software Methods & 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: Task total running time. Date: Tue, 06 Feb 2001 19:58:18 -0500 From: viv Organization: Lucent Technologies, Columbus, Ohio Message-ID: <3A809DAA.C3C01C4B@lucent.com> How can I determine the total running time for a task(in READY state) from the time a the task has started? How can one determine/develop this? Thanks, viv --------------------------- Newsgroups: comp.os.vxworks Subject: Re: libstdc++ for vxworks + gcc 2.95.3? Date: Wed, 7 Feb 2001 01:19:12 -0000 From: "Dave Korn" Organization: Lumber Cartel (tinlc) Members #2234-2237 (owing to browser refresh) Message-ID: References: "Paul D. Smith" wrote in message news:p5elxb8gf9.fsf@lemming.engeast.baynetworks.com... > I'm using davek's patched 2.95.3 and it seems to work well, but my app > can't link because it can't find libstdc++.a. Sure enough, it's not in > the gcc-lib/powerpc-wrs-vxworks/2.95.3 directory. ...yep > I saw a note somewhere on his site saying that libstdc++ doesn't build > for vxworks? I have to have this library, as my code is using C++ STL, > etc.! It's built into the OS libraries - when you add C++ language support in the Tornado GUI, that's what you get. > I built 2.95.2 from GNU sources for vxworks and libstdc++ built fine > there--is this a know 2.95.3 thing? Or a known build issue? Or...? Are you sure? I've build 2.95.2 from GNU a few times myself and it never worked for me! I'll give it another go; maybe it is indeed a regression in 2.95.3 DaveK --------------------------- Newsgroups: comp.os.vxworks Subject: Re: libstdc++ for vxworks + gcc 2.95.3? Date: 06 Feb 2001 21:10:29 -0500 From: "Paul D. Smith" Organization: Bay Networks, Inc. / Billerica, MA Message-ID: References: Reply-To: psmith@baynetworks.com %% "Dave Korn" writes: >> I saw a note somewhere on his site saying that libstdc++ doesn't build >> for vxworks? I have to have this library, as my code is using C++ STL, >> etc.! dk> It's built into the OS libraries - when you add C++ language dk> support in the Tornado GUI, that's what you get. I don't have Tornado, or a GUI. I have VxWorks 5.3.1 and the build tools for Solaris... and that's it. >> I built 2.95.2 from GNU sources for vxworks and libstdc++ built fine >> there--is this a know 2.95.3 thing? Or a known build issue? Or...? dk> Are you sure? I've build 2.95.2 from GNU a few times myself and dk> it never worked for me! I'll give it another go; maybe it is dk> indeed a regression in 2.95.3 I definitely did it. The entire build completed without any errors; I had to change a few files IIRC. Many others have too, on the crossgcc list. I found a post and a web site by folks who'd done it, and followed their directions. I'll have to dig back through my notes... Oh, right. I remember now. The build failure on fdmatch.c is easily overcome by adding an #include into sys/stat.h. Works like a charm. I followed the directions posted to crossgcc by Tony Farrell on 20 Jul 2000; see this link: http://sources.redhat.com/ml/crossgcc/2000-07/msg00112.html This is m68k but I did it with ppc with mostly the same steps. HTH... - -- - ------------------------------------------------------------------------------- Paul D. Smith HASMAT--HA Software Methods & 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: build problem Date: 6 Feb 2001 20:13:47 MST From: sachithanandam karthikselvan Organization: Lawrence Berkeley National Laboratory Message-ID: <95qeld$9aq$1@overload.lbl.gov> Hi all, 1) I included some c files and then tried to built it. It says "no such files or directroy". 2) I checked the makefile. Instead full path, it has file name only. ex. rnlog.h. So I gave full path ex. c:\sssafe\embedd\rnlog.h in the makefile itself. 3) Now I said make from the dos prompt.Again it says that file not found. But it is able to comple other vxworks related files like romstart.c 4) Can anyone throw some lights?? Thanks in advance karthikselvan ____________________________________________________________________ Get free email and a permanent address at http://www.netaddress.com/?N=1 --------------------------- Newsgroups: comp.os.vxworks Subject: Re: dialog with WindRiver - reasonable response Date: Tue, 06 Feb 2001 20:18:04 -0800 From: DrDiags Organization: Posted via Supernews, http://www.supernews.com Message-ID: <3A80CC7C.9F9E4DDC@flashcom.net> References: Reply-To: drdiags@flashcom.net George, Dave is correct. There is a GNU Source CD - TDK-12878-ZC-00 which should contain the source required to do what you want. If you speak to your Account Manager, he will be able to get this for you. The only disclaimer from what I understand is that you cannot expect support to build the tools. So, once you obtain source, you may have to figure out how to build for your own environment. I just started looking at trying to do this on a Windows machine and there are some pre-requisites. The first is that you need to have Microsoft Visual C++ 5.0 tools to build for Windows, but the paths in the makefile seem to point to systems and tools that have to be modified. If I ever get something going, I will post here and to you privately. It is just going to be time consuming and we all are busy right?....... As far as Jack Chow of WRS, he may be working under assumptions that need to be made clearer to him. I don't hold in total disdain anyone who has to try to be an expert in matters for which he didn't have a hand in developing nor given the tools to support. So try the avenue of obtaining the source, which is available from your AM or FAE, who have a vested interest in keeping the flames cool and let us know how things turn out. HTH. --------------------------- Newsgroups: comp.os.vxworks Subject: sbrk equivalent in VxWorks Date: Wed, 07 Feb 2001 04:59:59 GMT From: haribalram@my-deja.com Organization: Deja.com Message-ID: <95qkoc$jpi$1@nnrp1.deja.com> Hi I am designing a memory manager for a protocol stack. This memory manager wld maintain a pool of different sized memory chunks and allocate the chunks to the requested module/appln. When the pool doesn't hv enough chunks to allocate to, it does sbrk to get more memory from the os. sbrk() is used to change dynamically the amount of space allocated for the calling process's data segment. This implementation works fine in solaris. I need to port this code to VxWorks. I wld like to know if there is an equivalent of "sbrk" in VxWorks? If its not there, can anybody suggest how to implement sbrk in VxWorks? Thanx in advance. - - regards, Haribalram. Sent via Deja.com http://www.deja.com/ --------------------------- Newsgroups: comp.os.vxworks Subject: Re: VxSim: reloc overflow Date: 07 Feb 2001 05:20:49 GMT From: fnordz@aol.com (Fnordz) Organization: AOL http://www.aol.com Message-ID: <20010207002049.09920.00001152@ng-cc1.aol.com> References: Hello, The relocation overflow complaint from your linker is not likely to be solved by using the -mlongcall flag, as the problem is not related to an out of range branch target. Rather, the problem lies in the COFF - specifically PE-COFF in your case - representation of relocation entries. The COFF OMF uses 16-bit fields to hold a value which specifies the number of relocation entries for any particular section. Try the linker flag, --split-by-reloc , where is specifies the maximum number of relocation entries for a section. There are a couple of things I don't recall off-hand w.r.t. this flag: 1) I don't remember whether value must be base-16, base-10, or either. 2) I'm not sure that the compiler driver can pass it down to the linker via the -Wl, flag... try out because this seems like a reasonable thing to do. Possible pitfalls?? Always. Be wary of where ld tacks on the extra sections in the resulting object file. Use objdump -h myfile.o to check out the arrangement. You may have to use a custom linker script to get the sections layed out right. You can see what the internal linker scripts are doing by printing them out; here's how: ld --verbose prints the internal link script used on full links ls -r --verbose prints the internal link script used on partial links cheers, paul --------------------------- Newsgroups: comp.os.vxworks Subject: Re: *UPDATED* Gcc 2.95.3 for VxWorks-PPC available.... Date: 7 Feb 2001 05:39:59 GMT From: yucao@alumnae.caltech.edu (Yu Cao) Organization: California Institute of Technology, Pasadena Message-ID: <95qn3f$4p9@gap.cco.caltech.edu> References: <2001Feb6.091228.6806@lorelei.approve.se> Thanks Dave! I've installed your gcc package and managed to build bootrom_uncmp.bin. This is really encouraging. However when I tried to "make ESTWdbNet.vxWorks", I got some "undefined reference" errors (see below). This used to build with WRS supplied compiler (gcc 2.7.2, VxWorks 5.4). Any ideas what I might be doing wrong? Thanks. - --Yu Cao ccppc -BC:\Tornado/host/x86-win32/lib/gcc-lib/ -c -mcpu=860 -ansi -nostdinc -O -fvolatile -fno-builtin -fno-for-scope -Wall -I/h -I. -IC:\Tornado\target\config\all -IC:\Tornado\target/h -IC:\Tornado\target/src/config -IC:\Tornado\target/src/drv -DCPU=PPC860 -DMDP_BASIC -g -O0 -DROM_BASED -o version.o C:\Tornado\target\config\all/version.c ccppc -BC:\Tornado/host/x86-win32/lib/gcc-lib/ -nostdlib -r -Wl,-X \ -o vxWorks.tmp sysALib.o sysLib.o usrConfig.o version.o ../app/bin.PPC860/liblexus.a C:\Tornado\target/lib/libPPC860gnuvx.a nmppc vxWorks.tmp | wtxtcl C:\Tornado/host/src/hutils/munch.tcl -asm ppc > ctdt.c /Tornado/host/x86-win32/bin/make CC_COMPILER="-fdollars-in-identifiers" ctdt.o make[1]: Entering directory `C:\home\yucao\work\lexus\target\bsp' ccppc -BC:\Tornado/host/x86-win32/lib/gcc-lib/ -mcpu=860 -fdollars-in-identifiers -O -fvolatile -fno-builtin -fno-for-scope -Wall -I/h -I. -IC:\Tornado\target\config\all -IC:\Tornado\target/h -IC:\Tornado\target/src/config -IC:\Tornado\target/src/drv -DCPU=PPC860 -DMDP_BASIC -g -O0 -DROM_BASED -c ctdt.c make[1]: Leaving directory `C:\home\yucao\work\lexus\target\bsp' ldppc -X -N -e _sysInit -Ttext 00100000 \ -o vxWorks dataSegPad.o vxWorks.tmp ctdt.o vxWorks.tmp(.text+0x834c):C:\Tornado\target\config\all/version.c: undefined reference to `__start_cp_handler' vxWorks.tmp(.text+0xa444):C:\Tornado\target\config\all/version.c: undefined reference to `__eh_alloc' vxWorks.tmp(.text+0xa634):C:\Tornado\target\config\all/version.c: undefined reference to `__eh_alloc' vxWorks.tmp(.text+0xa80c):C:\Tornado\target\config\all/version.c: undefined reference to `__eh_alloc' vxWorks.tmp(.text+0xa9e4):C:\Tornado\target\config\all/version.c: undefined reference to `__eh_alloc' vxWorks.tmp(.text+0xabd4):C:\Tornado\target\config\all/version.c: undefined reference to `__eh_alloc' vxWorks.tmp(.text+0xaddc):C:\Tornado\target\config\all/version.c: more undefined references to `__eh_alloc' follow vxWorks.tmp(.text+0x1cdc0):C:\Tornado\target\config\all/version.c: undefined reference to `__start_cp_handler' vxWorks.tmp(.text+0x1cefc):C:\Tornado\target\config\all/version.c: undefined reference to `__start_cp_handler' vxWorks.tmp(.text+0x2da28):C:\Tornado\target\config\all/version.c: undefined reference to `__eh_alloc' vxWorks.tmp(.text+0x2dc60):C:\Tornado\target\config\all/version.c: undefined reference to `__eh_alloc' vxWorks.tmp(.text+0x2dec0):C:\Tornado\target\config\all/version.c: undefined reference to `__eh_alloc' vxWorks.tmp(.text+0x2e0e8):C:\Tornado\target\config\all/version.c: undefined reference to `__eh_alloc' vxWorks.tmp(.text+0x2e304):C:\Tornado\target\config\all/version.c: undefined reference to `__eh_alloc' vxWorks.tmp(.text+0x2e4b0):C:\Tornado\target\config\all/version.c: more undefined references to `__eh_alloc' follow vxWorks.tmp(.text+0x55a54):C:\Tornado\target\config\all/version.c: undefined reference to `__start_cp_handler' vxWorks.tmp(.text+0x55b30):C:\Tornado\target\config\all/version.c: undefined reference to `__start_cp_handler' vxWorks.tmp(.text+0x595e8):C:\Tornado\target\config\all/version.c: undefined reference to `__eh_alloc' vxWorks.tmp: In function `sysSerialHwInit': /home/yucao/work/lexus/target/bsp/sysSerial.c:109: undefined reference to `__eh_alloc' vxWorks.tmp: In function `ppc860IntrInit': C:/Tornado/target/src/drv/intrCtl/ppc860Intr.c:149: undefined reference to `__eh_alloc' vxWorks.tmp: In function `ppc860IntConnect': C:/Tornado/target/src/drv/intrCtl/ppc860Intr.c:268: undefined reference to `__eh_alloc' vxWorks.tmp: In function `ppc860CpmIntrDeMux': C:/Tornado/target/src/drv/intrCtl/ppc860Intr.c:412: undefined reference to `__eh_alloc' vxWorks.tmp:/home/yucao/work/lexus/target/bsp/ppcDecTimer.c:181: more undefined references to `__eh_alloc' follow make: *** [vxWorks] Error 0x1 --------------------------- Newsgroups: comp.os.vxworks Subject: Re: malloc and free in vxworks Date: Wed, 07 Feb 2001 07:02:02 GMT From: cwein@my-deja.com Organization: Deja.com Message-ID: <95qrt9$p1m$1@nnrp1.deja.com> References: <95a6bh$490$1@overload.lbl.gov> In article <95a6bh$490$1@overload.lbl.gov>, Ashutosh Zade wrote: > > Hi Vxworks, > > Can we take control of malloc() and free() in VxWorks by modifying source > code or any other way to tune it as per our requirement? If yes, how? > > Basically we want this: > If malloc requests are coming from one certain application, we want to > allocate memory from certain memory addresses. If malloc request is coming > from some other application, we want to allocate memory from some other > memory addresses. The entire memory map may not fall under contiguous memory > addresses. VxWorks has support for memory partitions. Each partition is internally contiguous but partitions may not necessarily be contiguous. Check out memPartCreate, memPartAlloc. Also note that there is one partition defined at startup, the system partition and that malloc (and free) are really just wrappers around sysMemPartAlloc which is a memPartAlloc(sysMemPartId...) or something close to that. Sent via Deja.com http://www.deja.com/ --------------------------- Newsgroups: comp.os.vxworks Subject: Interrup trouble on Assabet SA-1110? Date: Wed, 7 Feb 2001 09:31:49 +0100 From: "Carsten Schwartz" Organization: Posted Courtesy of Tele Danmark or one of its European Subsidiaries Message-ID: <95r17d$2fh$1@news.inet.tele.dk> Reply-To: "Carsten Schwartz" I am making an interactive application on the assabet evaluation board. I need to handle the touch screen and the switches that can be 'defined in software' as mentioned in the user guide. I have tried to make a simple interrupt handler with the intConnect and intEnable like this but nothing happens, does any body have example code or experience with this ? This code should handle switch 2 on the board but the global counter is not incemented.... #define LCD_INT GPIO_1 // Definition of interrupt bit mask (1<<1)... #define IRQ_LCD_INT 1 // Definition of interrupt level... int g_iInterruptCount = 0 // Global counter to increase for every interrupt... void LCDInterruptHandler(int iCounter) { g_iInterruptCount++; REG_WRITE(GEDR, LCD_INT); /* Clear edge detection interrupt - done by writing a 1 to the register */ } void EmptyInterruptHandler(int iCounter) { } void ConnectInt() { int iLockLevel = intLock (); printf("Lock Level = 0x%08x\n", iLockLevel); intUnlock(iLockLevel); if (intConnect(INUM_TO_IVEC (IRQ_LCD_INT), (VOIDFUNCPTR) LCDInterruptHandler, g_iInterruptCount) == OK) { GPIODirectionSet(LCD_INT, IN); // Enable LCD Interrupt to CPU - set irq-pin as input SetBit(GRER, LCD_INT); // Enable int edge-detect on rising edge ClearBit(GFER, LCD_INT); // Disable int edge-detect on falling edge REG_WRITE(GEDR, LCD_INT); // Clear edge detection interrupt - done by writing a 1 to the register if (OK == intEnable(IRQ_LCD_INT)) // Enable interrupt { printf("Interrupt Connection Successful\n"); } else { printf("Interrupt Enabling Failed\n"); } } else { printf("Interrupt Connection Failed\n"); } } DisableInt() { intDisable(IRQ_LCD_INT); intConnect(INUM_TO_IVEC (IRQ_LCD_INT), (VOIDFUNCPTR) EmptyInterruptHandler, 0); } --------------------------- Newsgroups: comp.os.vxworks Subject: Re: unknown error message Date: Wed, 07 Feb 2001 07:02:37 -0800 From: Joe Durusau Organization: Lockheed Martin Corporation Message-ID: <3A81638D.4921BF7B@lmco.com> References: <3a804305.21210562@192.168.1.214> Suggest that you specify version of vxWorks, platform, and BSP. There is no mention of this routine in the standard documentation. The error would see to mean that the bootLoadModule routine is a dummy whic does not do anything. Speaking only for myself, Joe Durusau Kim Putnam wrote: > > I need to locate any information on bootLoadModule(). I am getting the > following error "S_bootLoadLib_ROUTINE_NOT_INSTALLED" > > kim --------------------------- End of New-News digest ********************** From vxwexplo-errs@csg.lbl.gov Thu Feb 8 04:03:11 2001 From: Vxworks Exploder Date: Thu Feb 8 04:03:14 PST 2001 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Thu Feb 8 04:03:06 PST 2001 Subject: Building a bootrom with C++ support Subject: Re: Building a bootrom with C++ support Subject: Re: MPC823 Rev A --> Rev B problem Subject: Re: unknown error message Subject: Many questions.... Help required.... Subject: Hi guys.. Where can I find vxworks 5.4 OS? Subject: SSH for VxWorks Subject: ttyDrv and Polled Mode Serial Subject: Re: dialog with WindRiver - reasonable response Subject: Ethernet driver for Intel PRO/100+ Subject: Access to CR0 register in X86 Subject: Re: Ethernet driver for Intel PRO/100+ Subject: Re: unknown error message Subject: Re: Hi guys.. Where can I find vxworks 5.4 OS? Subject: Re: build problem Subject: Re: dialog with WindRiver - reasonable response Subject: Re: *UPDATED* Gcc 2.95.3 for VxWorks-PPC available.... Subject: Re: *UPDATED* Gcc 2.95.3 for VxWorks-PPC available.... Subject: Re: libstdc++ for vxworks + gcc 2.95.3? Subject: Re: VxSim: reloc overflow Subject: Re: *UPDATED* Gcc 2.95.3 for VxWorks-PPC available.... Subject: Re: Building a bootrom with C++ support Subject: Re: Hi guys.. Where can I find vxworks 5.4 OS? Subject: I Need Your Help Please!!! Subject: Re: I Need Your Help Please!!! Subject: Re: Fibre Channel to Raid Speed Subject: Re: I Need Your Help Please!!! Subject: Re: Random number generator Subject: Re: Random number generator Subject: Re: Random number generator Subject: Re: Random number generator Subject: Re: Random number generator Subject: Re: Hi guys.. Where can I find vxworks 5.4 OS? Subject: MVME162-20 == MVME162LX ???? Subject: LXT971 MDIO Problem Subject: Re: Ethernet driver for Intel PRO/100+ Subject: Re: sbrk equivalent in VxWorks Subject: Re: *UPDATED* Gcc 2.95.3 for VxWorks-PPC available.... Subject: Re: build problem Subject: Re: *UPDATED* Gcc 2.95.3 for VxWorks-PPC available.... Subject: Re: Hi guys.. Where can I find vxworks 5.4 OS? Subject: Re: MVME162-20 == MVME162LX ???? Subject: WTX Error 0x100C8 Subject: about configuring IP at booting time.. Subject: Raven MPIC timer interrupt question Subject: Re: Hi guys.. Where can I find vxworks 5.4 OS? Subject: Can't download big .out file to VXSIM Subject: Please help with C++ examples for VxWorks Subject: Re: unknown error message Subject: Re: *UPDATED* Gcc 2.95.3 for VxWorks-PPC available.... Subject: Re: I Need Your Help Please!!! Subject: Re: about configuring IP at booting time.. ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Building a bootrom with C++ support Date: Wed, 07 Feb 2001 13:04:38 GMT From: [RMRL]BSR Organization: Deja.com Message-ID: <95rh53$99p$1@nnrp1.deja.com> Hello, I'm having problems building a bootrom for one of our processors (a Motorola Scout card). The original BSP builds OK, but I want to add more functionality to it. This additional functionality has been coded in C++ but when I try to link the bootrom I get unresolved external references (see below). The additional functionality has been added successfully to a bootable VxWorks image. Does anyone know whether it is possible to link C++ code with the bootrom and if so what I need to do? Cheers, Paul Crowley ERRORS: ccppc -BC:\TORNADO\host\x86-win32\lib\gcc-lib\ -mstrict-align -ansi - nostdinc -DRW_MULTI_T HREAD -D_REENTRANT -O2 -fvolatile -fno-builtin -fno-for-scope -I\h - I. -IC:\TORNADO\tar get\config\all -IC:\TORNADO\target\h -IC:\TORNADO\target\src\config - IC:\TORNADO\target\sr c\drv -DCPU=PPC604 -DSCOUT -c FlashInterface.cpp ccppc -BC:\TORNADO\host\x86-win32\lib\gcc-lib\ -c -mstrict-align -ansi - nostdinc -O2 -fvol atile -fno-builtin -fno-for-scope -Wall -I\h -I. - IC:\TORNADO\target\config\all -IC:\TOR NADO\target\h -IC:\TORNADO\target\src\config - IC:\TORNADO\target\src\drv -DCPU=PPC604 -DS COUT -gdwarf -O0 -DROM_BASED -o version.o C:\TORNADO\target\config\all\version.c ldppc -X -N -e _romInit -Ttext 00F00000 \ -o bootrom_uncmp romInit.o bootInit_uncmp.o version.o bootConfig.o \ sysALib.o sysLib.o miiLib.obj FlashDriver.o FlashInterface.o C:\TORNADO\target\lib\ libPPC604gnuvx.a C:\TORNADO\target\lib\libPPC604gnuvx.a(cplusGlob.o): In function `cplusCtorsLink': cplusGlob.o(.text+0x22): undefined reference to `_ctors' cplusGlob.o(.text+0x26): undefined reference to `_ctors' C:\TORNADO\target\lib\libPPC604gnuvx.a(cplusGlob.o): In function `cplusDtorsLink': cplusGlob.o(.text+0x56): undefined reference to `_dtors' cplusGlob.o(.text+0x5a): undefined reference to `_dtors' make.exe: *** [bootrom_uncmp] Error 0x1 Done. Sent via Deja.com http://www.deja.com/ --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Building a bootrom with C++ support Date: Wed, 07 Feb 2001 13:07:36 GMT From: kimp@avtecinc.com (Kim Putnam) Organization: Info Avenue Internet Services Message-ID: <3a814850.2152812@192.168.1.214> References: <95rh53$99p$1@nnrp1.deja.com> We are tyring to do the something similar. I resolved the kind of errors you had by munching the files. STill not able to ge the bootrom up though. Kim On Wed, 07 Feb 2001 13:04:38 GMT, [RMRL]BSR wrote: >Hello, > >I'm having problems building a bootrom for one of our processors (a >Motorola Scout card). The original BSP builds OK, but I want to add >more functionality to it. This additional functionality has been coded >in C++ but when I try to link the bootrom I get unresolved external >references (see below). > >The additional functionality has been added successfully to a bootable >VxWorks image. > >Does anyone know whether it is possible to link C++ code with the >bootrom and if so what I need to do? > >Cheers, > >Paul Crowley > >ERRORS: > >ccppc -BC:\TORNADO\host\x86-win32\lib\gcc-lib\ -mstrict-align -ansi - >nostdinc -DRW_MULTI_T >HREAD -D_REENTRANT -O2 -fvolatile -fno-builtin -fno-for-scope -I\h - >I. -IC:\TORNADO\tar >get\config\all -IC:\TORNADO\target\h -IC:\TORNADO\target\src\config - >IC:\TORNADO\target\sr >c\drv -DCPU=PPC604 -DSCOUT -c FlashInterface.cpp >ccppc -BC:\TORNADO\host\x86-win32\lib\gcc-lib\ -c -mstrict-align -ansi - >nostdinc -O2 -fvol >atile -fno-builtin -fno-for-scope -Wall -I\h -I. - >IC:\TORNADO\target\config\all -IC:\TOR >NADO\target\h -IC:\TORNADO\target\src\config - >IC:\TORNADO\target\src\drv -DCPU=PPC604 -DS >COUT -gdwarf -O0 -DROM_BASED -o version.o >C:\TORNADO\target\config\all\version.c >ldppc -X -N -e _romInit -Ttext 00F00000 \ > -o bootrom_uncmp romInit.o bootInit_uncmp.o version.o bootConfig.o \ > sysALib.o sysLib.o miiLib.obj FlashDriver.o FlashInterface.o >C:\TORNADO\target\lib\ >libPPC604gnuvx.a >C:\TORNADO\target\lib\libPPC604gnuvx.a(cplusGlob.o): In function >`cplusCtorsLink': >cplusGlob.o(.text+0x22): undefined reference to `_ctors' >cplusGlob.o(.text+0x26): undefined reference to `_ctors' >C:\TORNADO\target\lib\libPPC604gnuvx.a(cplusGlob.o): In function >`cplusDtorsLink': >cplusGlob.o(.text+0x56): undefined reference to `_dtors' >cplusGlob.o(.text+0x5a): undefined reference to `_dtors' >make.exe: *** [bootrom_uncmp] Error 0x1 > >Done. > > >Sent via Deja.com >http://www.deja.com/ --------------------------- Newsgroups: comp.os.vxworks,comp.sys.powerpc.tech Subject: Re: MPC823 Rev A --> Rev B problem Date: Wed, 7 Feb 2001 06:27:10 -0700 From: "Amy Michaelis" Organization: Dimensional Communications Message-ID: <%%bg6.1757$5s2.63018@wormhole.dimensional.com> References: <3a7f35b1$0$7170$a05e1490@news.netdirect.net> Gary, Thanks for the input. At least I've got some confidence that this can work. As far as the new I2C/SPI patch is concerned, if you download the one from the MPC860 web site, it has a bit of revision information which has been left off by the MPC823 version. Basically, using the SCC in Ethernet mode can cause erratic operation with the SPI bus under the old patch. Matt Michaelis Gary Sinde wrote in message news:3a7f35b1$0$7170$a05e1490@news.netdirect.net... > I just got done putting the new I2C/SPI patch in my code. I could never > get motorola to state why I needed to change the patch but it has worked in > both Rev-A and Rev-B chips. My next revision of the board has several > Rev-A's and several Rev-B's to check for differences. I also am using > ethernet, I2C, SPI, and both SMC's. > > Gary > > > "Amy Michaelis" wrote in message > news:SGXe6.1708$5s2.60221@wormhole.dimensional.com... > > Help! > > > > Motorola has recently discontinued the Rev A silicon for the 823, and now > we > > can only get the Rev B silicon. This should all be seamless of course, > but > > after some testing with the Rev B chip, we've found a problem. The Rev B > > chip causes random machine checks in our system. > > > > We're using vxWorks as an operating system, and a check with Wind River > > didn't turn up anything. I've been trying to find out what's going on > here, > > but so far no luck. We're using quite a bit of CPM, and I tend to > believe > > that's were the problem lies. We have the SCC in Ethernet mode, and the > SPI > > and I2C are also being used. These two guys have been relocated using the > > downloaded microcode patch. My sense is that since Motorola cleaned up a > > lot of the problems in the I2C driver for the Rev B silicon, something has > > gone wrong in the downloaded microcode. > > > > The problem could also be the new interrupt for SCC3, but we're not using > > the new SCC on the rev B chip, so I don't know how that would be a > problem. > > > > Any ideas would be greatly appreciated. > > > > Thanks, > > > > Matt Michaelis > > amich@dimensional.com > > > > > > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: unknown error message Date: Wed, 07 Feb 2001 13:33:48 GMT From: Urban Lindberg Organization: Deja.com Message-ID: <95rirs$ams$1@nnrp1.deja.com> References: <3a804305.21210562@192.168.1.214> Hi! The error you get is because you have not included support for the object format you are trying to load. Try with including ELF support when you build the bootrom. objdump -a filename tells you what format the file is. HTH Urban In article <3a804305.21210562@192.168.1.214>, kimp@avtecinc.com (Kim Putnam) wrote: > I need to locate any information on bootLoadModule(). I am getting the > following error "S_bootLoadLib_ROUTINE_NOT_INSTALLED" > > kim > Sent via Deja.com http://www.deja.com/ --------------------------- Newsgroups: comp.os.vxworks Subject: Many questions.... Help required.... Date: Wed, 07 Feb 2001 09:46:50 -0500 From: haroldt.creacom@netcom.ca Organization: Crea.Com Inc. Message-ID: <3A815FDA.5BB91ACA@netcom.ca> Reply-To: htomlins@certicom.com - --------------AF8E3E11E2B91F40419797E2 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit I am trying to port some code to VxWorks and having some problems. The code appears to run, but runs endlessly. I tried adding even more printf's to see where the program is hanging/busy/whatever, but the output appears to be buffered. Okay, it is buffered. It has to be. The very first thing the program does is spew a message saying it has started. This takes some time to appear and appears only when there is more to output. I tried flushing the output with fflush, but it still appears buffered. The docs indicate that printf is not part of the Formatted/Buffered IO so that simple printing could be done unbuffered without loading the entire library. Could it be my X-windows or the network connection that is buffering the output? Since I need to know where the program is having problems in order to fix it, I ran it in the debugger and, once it ran for a number of minutes with no apparent progress, I paused the execution. It was in a "SemWait" routine called indirectly by printf. What would it be waiting for? More importantly, why? Still more importantly, how do I fix that? The configuration is as follows: VxWorks is installed on Pentium class computer in a desk-top box. It has a console, keyboard, and mouse, and is connected via the network to a Sparc running Solaris. I connect using eXceed and run the wind shell to load and execute the programs. One of the unusual behavioural aspects is that the X-windows will periodically die. I do not know if it is the server or client. I'm assuming it is the server as it is eXceed running on Windows Nice Try (aka Windows NT). It will simply stop responding, stop repainting, and require restarting. The shell exhibits an odd behaviour as well. I can create two shell windows which do connect to the VxWorks box, but, once I am running the application in one, any commands issued to the other (like d or i) result in time-outs and a restart of the shell. Of course, the shell restart also times out. The same is true of the debugger if I try to launch it after starting an application. Attempting to attach to a running process is pointless. So, now that you have the background to my dilemma, here are my questions: 0) Is the time-out that I am experiencing due to the network drivers not getting cycles? 1) Is the OS incorrectly configured so that an application can take up all of the cycles? 2) How do we relinquish cycles? (I tried taskWait(0), no change.) 3) Why would printf result in a semaphore wait? What is it fighting against? How do we write debugging information without printf? or, How do we correct printf to not end up in this race condition? 4) Is there a way to run the applications and/or debugger directly on the console instead of over the network? This would illiminate the network from the list of usual suspects. 5) Is there a profiling facility? 6) What is the best way to benchmark the system? 7) What is the best way to implement a heartbeat? I tried wdCreate, but need to do more work with that to get it running. Suggestions? 8) How do we flush the output? If fflush works, could it be the network that is buffering, or perhaps, simply not getting enough cpu time? Answers would be awesome, suggestions would be great, and any pointers to other sources (yes, I've seen the vxworks faq) would be welcomed. Please reply to htomlins@certicom.com as I cannot read the group from work. Thanks. - -- Harold Tomlinson Software Design Crea.Com Inc. Training & Documentation haroldt.creacom@netcom.ca Photographic Arts 416-879-2415 (voice) 416-879-2606 (fax) - --------------AF8E3E11E2B91F40419797E2 Content-Type: text/html; charset=us-ascii Content-Transfer-Encoding: 7bit  
  I am trying to port some code to VxWorks and having some problems.  The code appears to run,
but runs endlessly.  I tried adding even more printf's to see where the program is
hanging/busy/whatever, but the output appears to be buffered.  Okay, it is buffered.  It has to be.
The very first thing the program does is spew a message saying it has started.  This takes some
time to appear and appears only when there is more to output.

  I tried flushing the output with fflush, but it still appears buffered.  The docs indicate that
printf is not part of the Formatted/Buffered IO so that simple printing could be done unbuffered
without loading the entire library.  Could it be my X-windows or the network connection that
is buffering the output?

  Since I need to know where the program is having problems in order to fix it, I ran it in
the debugger and, once it ran for a number of minutes with no apparent progress, I paused the
execution.  It was in a "SemWait" routine called indirectly by printf.  What would it be waiting
for?  More importantly, why?  Still more importantly, how do I fix that?

  The configuration is as follows:   VxWorks is installed on Pentium class computer in a
desk-top box.  It has a console, keyboard, and mouse, and is connected via the network to
a Sparc running Solaris.  I connect using eXceed and run the wind shell to load and execute
the programs.

  One of the unusual behavioural aspects is that the X-windows will periodically die.  I do
not know if it is the server or client.  I'm assuming it is the server as it is eXceed running
on Windows Nice Try (aka Windows NT).  It will simply stop responding, stop repainting,
and require restarting.

  The shell exhibits an odd behaviour as well.  I can create two shell windows which do
connect to the VxWorks box, but, once I am running the application in one, any commands
issued to the other (like d or i) result in time-outs and a restart of the shell.  Of course, the
shell restart also times out.   The same is true of the debugger if I try to launch it after
starting an application.  Attempting to attach to a running process is pointless.

  So, now that you have the background to my dilemma, here are my questions:

    0) Is the time-out that I am experiencing due to the network drivers not getting cycles?
    1) Is the OS incorrectly configured so that an application can take up all of the cycles?
    2) How do we relinquish cycles?  (I tried taskWait(0), no change.)
    3) Why would printf result in a semaphore wait?  What is it fighting against?  How do
    we write debugging information without printf?  or, How do we correct printf to not
    end up in this race condition?
    4) Is there a way to run the applications and/or debugger directly on the console instead
    of over the network?  This would illiminate the network from the list of usual suspects.
    5) Is there a profiling facility?
    6) What is the best way to benchmark the system?
    7) What is the best way to implement a heartbeat?  I tried wdCreate, but need to do
    more work with that to get it running.  Suggestions?
    8) How do we flush the output?  If fflush works, could it be the network that is
    buffering, or perhaps, simply not getting enough cpu time?

  Answers would be awesome, suggestions would be great, and any pointers to other
sources (yes, I've seen the vxworks faq) would be welcomed.

  Please reply to htomlins@certicom.com as I cannot read the group from work.  Thanks. 




-- 
Harold Tomlinson                      Software Design
Crea.Com Inc.                         Training & Documentation
haroldt.creacom@netcom.ca             Photographic Arts
416-879-2415 (voice)                  416-879-2606 (fax)

 

- --------------AF8E3E11E2B91F40419797E2--


---------------------------

Newsgroups: comp.os.vxworks
Subject: Hi guys.. Where can I find vxworks 5.4 OS?
Date: Wed, 7 Feb 2001 16:34:07 +0200
From: "SentinuS" 
Organization: TTnet News Server 
Message-ID: <95rj2b$5ih238@baran22.ttnet.net.tr>
Reply-To: "SentinuS" 





---------------------------

Newsgroups: comp.os.vxworks
Subject: SSH for VxWorks
Date: Wed, 07 Feb 2001 15:15:30 GMT
From: dukenana@my-deja.com
Organization: Deja.com
Message-ID: <95roqa$g5m$1@nnrp1.deja.com>

We would like to implement SSH server on VxWorks. Any suggestions for
that?

Tx


Sent via Deja.com
http://www.deja.com/

---------------------------

Newsgroups: comp.os.vxworks
Subject: ttyDrv and Polled Mode Serial
Date: Wed, 7 Feb 2001 15:37:48 -0000
From: "Kelvin Lawson" 
Message-ID: <95rptl$i9hrt$1@ID-56276.news.dfncis.de>

WRS/VxWorks documentation suggests that polled mode serial is used for debug
communication while interrupt mode serial is used with ttyDrv.

Does anyone know if it's possible to use a polled mode SIO driver with
ttyDrv ? We don't have any interrupts available on this processor.

Cheers,
Kelvin.



---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: dialog with WindRiver - reasonable response
Date: Wed, 07 Feb 2001 16:09:09 GMT
From: Bill Pringlemeir 
Organization: Factory of the mind
Message-ID: 
References:  <3A80CC7C.9F9E4DDC@flashcom.net>
Sender: bpringlemeir@DeadDuck

>>>>> "DD" == DrDiags   writes:

 DD> George, Dave is correct. There is a GNU Source CD -
[snip]
 DD> are busy right?.......  As far as Jack Chow of WRS, he may be
 DD> working under assumptions that need to be made clearer to him. I
 DD> don't hold in total disdain anyone who has to try to be an expert
 DD> in matters for which he didn't have a hand in developing nor
 DD> given the tools to support. So try the avenue of obtaining the
 DD> source, which is available from your AM or FAE, who have a vested
 DD> interest in keeping the flames cool and let us know how things
 DD> turn out. HTH.

I should note that some people at WRS are really helpful.  Some of
them could even be developers for WRS who post helpful messages here
on occasion.  At any rate, I don't think it is bad to `flame' their
technical support.  It is pretty atrocious.  I had asked if anyone
here had positive comments on it.  I didn't hear any, but maybe that
is apathy.

Flaming is sometimes deserved.  It is difficult to recommend vxWorks
because of the quality of tech support alone.  Several smaller
embedded companies have great tech support; why cann't WRS?

fwiw,
Bill

- -- 
Ft.  Meade terrorist Mossad explosion Kennedy DES CIA arrangements
Delta Force Peking $400 million in gold bullion Soviet KGB BATF Nazi

---------------------------

Newsgroups: comp.os.vxworks
Subject: Ethernet driver for Intel PRO/100+
Date: Wed, 07 Feb 2001 16:14:33 GMT
From: chafmich@my-deja.com
Organization: Deja.com
Message-ID: <95rs8u$jht$1@nnrp1.deja.com>

I am new to VxWorks development and I need to get a 100Mbps NIC working
on VxWorks that runs on an x86 platform.  I have an Intel Pro/100+ NIC
which is based on the Intel 82559 chip.  I have read that the if_fei
driver supports the Intel 82557, but I assume it does not support the
newer 82559.

I have downloaded the eev driver from Intel's web page that is part of
the 82559ER driver zip file.  I am going to try to use that driver.

I am not sure if I am going down the right path, so I am looking for
some reassurance that what I'm doing is the right thing to do.

Thanks,

Mike


Sent via Deja.com
http://www.deja.com/

---------------------------

Newsgroups: comp.os.vxworks
Subject: Access to CR0 register in X86
Date: Wed, 07 Feb 2001 16:22:21 +0000
From: James Marshall 
Organization: Agilent Technologies UK Ltd
Message-ID: <3A81763D.DF9B7F6E@agilent.com>

Hi,

How can I read the control register CR0 using GDB or the VxWorks shell? I've
tried 'info registers' but that only lists the standard regs. Any ideas?

James Marshall
Agilent Technologies
Edinburgh, Scotland



---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: Ethernet driver for Intel PRO/100+
Date: Wed, 07 Feb 2001 16:32:58 +0000
From: James Marshall 
Organization: Agilent Technologies UK Ltd
Message-ID: <3A8178BA.27D7D052@agilent.com>
References: <95rs8u$jht$1@nnrp1.deja.com>

Intel's driver has no support ("see WindRiver" they say) and we've found
problems using it on our 82559ER hardware - stick to WindRiver's 'fei'.

A few weeks ago I posted a fix (search back through DejaNews for 559ER) which
adds a single line to code that searches for the 557 so it also matches the
559ER. If you use a faster than usual system tick rate (ours is 1kHz) then
you'll also need a fix to the driver code. I got WRS support to give me a
modified driver object file. The driver uses tick delays without scaling them
against sysClkRateGet(). Hopefully by now they have an SPR against it.

By the way, we're using the END driver ('fei') not the BSD driver ('if_fei') -
the info above assumes you use the END driver.

James.


chafmich@my-deja.com wrote:

> I am new to VxWorks development and I need to get a 100Mbps NIC working
> on VxWorks that runs on an x86 platform.  I have an Intel Pro/100+ NIC
> which is based on the Intel 82559 chip.  I have read that the if_fei
> driver supports the Intel 82557, but I assume it does not support the
> newer 82559.
>
> I have downloaded the eev driver from Intel's web page that is part of
> the 82559ER driver zip file.  I am going to try to use that driver.
>
> I am not sure if I am going down the right path, so I am looking for
> some reassurance that what I'm doing is the right thing to do.
>
> Thanks,
>
> Mike
>
> Sent via Deja.com
> http://www.deja.com/


---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: unknown error message
Date: Wed, 07 Feb 2001 16:30:20 GMT
From: kimp@avtecinc.com (Kim Putnam)
Organization: Info Avenue Internet Services
Message-ID: <3a81779b.6423375@192.168.1.214>
References: <3a804305.21210562@192.168.1.214> <95rirs$ams$1@nnrp1.deja.com>

I ran objdump386 -a and agot a.out....

I believe INCLUDE_AOUT is defined. 

K
On Wed, 07 Feb 2001 13:33:48 GMT, Urban Lindberg 
wrote:

>Hi!
>
>The error you get is because you have not included support for the
>object format you are trying to load. Try with including ELF support
>when you build the bootrom. objdump -a filename tells you what
>format the file is.
>
>HTH
>
>Urban
>
>In article <3a804305.21210562@192.168.1.214>,
>  kimp@avtecinc.com (Kim Putnam) wrote:
>> I need to locate any information on bootLoadModule(). I am getting the
>> following error "S_bootLoadLib_ROUTINE_NOT_INSTALLED"
>>
>> kim
>>
>
>
>Sent via Deja.com
>http://www.deja.com/


---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: Hi guys.. Where can I find vxworks 5.4 OS?
Date: Wed, 07 Feb 2001 11:56:36 -0800
From: Joe Durusau 
Organization: Lockheed Martin Corporation
Message-ID: <3A81A874.9ED81BB8@lmco.com>
References: <95rj2b$5ih238@baran22.ttnet.net.tr>

www.wrs.com for the loc of the nearest sales office.

Speaking only for myself,

Joe Durusau


SentinuS wrote:

---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: build problem
Date: 7 Feb 2001 17:17:29 GMT
From: borkhuis@agere.com (Johan Borkhuis)
Organization: Lucent Technologies
Message-ID: 
References: <95qeld$9aq$1@overload.lbl.gov>

sachithanandam karthikselvan  wrote:

>Hi all,
>
>1) I included some c files and then tried to built it.
>It says "no such files or directroy".
>
>2) I checked the makefile. Instead full path, it has file name only.
>ex. rnlog.h. So I gave full path ex. c:\sssafe\embedd\rnlog.h in the
>makefile itself.
>
>3) Now I said make from the dos prompt.Again it says that file not
>found. But it is able to comple other vxworks related files like
>romstart.c 
>
>4) Can anyone throw some lights??

You have to add the VPATH variable to your macros. Have a look at my web 
page for more information.

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: dialog with WindRiver - reasonable response
Date: 7 Feb 2001 17:22:10 GMT
From: borkhuis@agere.com (Johan Borkhuis)
Organization: Lucent Technologies
Message-ID: 
References:  <3A80CC7C.9F9E4DDC@flashcom.net> 

Bill Pringlemeir  wrote:
>I should note that some people at WRS are really helpful.  Some of
>them could even be developers for WRS who post helpful messages here
>on occasion.  At any rate, I don't think it is bad to `flame' their
>technical support.  It is pretty atrocious.  I had asked if anyone
>here had positive comments on it.  I didn't hear any, but maybe that
>is apathy.

Actually I did not have any negative experiences with the tech support. I 
had several SPR's and TSR's, but they were all resolved quickly.
The FAE's I have ben in contact with were always very helpfull.

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: *UPDATED* Gcc 2.95.3 for VxWorks-PPC available....
Date: Wed, 07 Feb 2001 17:22:00 GMT
From: bblank@init-ka.de
Organization: Deja.com
Message-ID: <95s07b$nkg$1@nnrp1.deja.com>
References:  <2001Feb6.091228.6806@lorelei.approve.se> 

Hi Dave,

I have the same problems as Yu Cao:

After successfuly compiling my ExceptionTest.cpp and building
Exception_Test.out the target-download says:

Errors while downloading
W:/USER/BLK/T2proj/ExctTest/PPC860gnu/Exception_Test.out:
__start_cp_handler
__eh_alloc

any ideas?

Thanks
Bernd Blank


Sent via Deja.com
http://www.deja.com/

---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: *UPDATED* Gcc 2.95.3 for VxWorks-PPC available....
Date: Wed, 7 Feb 2001 16:59:45 -0000
From: "Dave Korn" 
Organization: Lumber Cartel (tinlc) Members #2234-2237 (owing to browser refresh)
Message-ID: 
References:  <2001Feb6.091228.6806@lorelei.approve.se>  <95qn3f$4p9@gap.cco.caltech.edu>

Yu Cao wrote in message <95qn3f$4p9@gap.cco.caltech.edu>...
>
>Thanks Dave! I've installed your gcc package and managed to build
>bootrom_uncmp.bin. This is really encouraging.
>
>However when I tried to "make ESTWdbNet.vxWorks", I got some "undefined
>reference" errors (see below). This used to build with WRS supplied
>compiler (gcc 2.7.2, VxWorks 5.4). Any ideas what I might be doing wrong?

>vxWorks.tmp(.text+0xaddc):C:\Tornado\target\config\all/version.c: more
undefined references to `__eh_alloc' follow
>vxWorks.tmp(.text+0x1cdc0):C:\Tornado\target\config\all/version.c:
undefined reference to `__start_cp_handler'
  [and loads more similar]

  I'm not sure if I didn't already post a reply to this from one of my other
accounts,
but at the risk of repeating myself, the answer is to either:

>ccppc -BC:\Tornado/host/x86-win32/lib/gcc-lib/ -nostdlib -r -Wl,-X \
>    -o vxWorks.tmp sysALib.o sysLib.o    usrConfig.o version.o
>../app/bin.PPC860/liblexus.a C:\Tornado\target/lib/libPPC860gnuvx.a

  ...remove the -nostdlib flag from the compiler flags, or...

>ldppc -X -N -e _sysInit -Ttext 00100000   \
>    -o vxWorks dataSegPad.o vxWorks.tmp ctdt.o

  ...add "-lgcc" to the final link stage flags.

        DaveK
- --
They laughed at Galileo.  They laughed at Copernicus.  They laughed at
Columbus. But remember, they also laughed at Bozo the Clown.



---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: libstdc++ for vxworks + gcc 2.95.3?
Date: Wed, 7 Feb 2001 16:48:57 -0000
From: "Dave Korn" 
Organization: Lumber Cartel (tinlc) Members #2234-2237 (owing to browser refresh)
Message-ID: 
References:   

Paul D. Smith wrote in message ...
>%% "Dave Korn"  writes:

>  dk> It's built into the OS libraries - when you add C++ language
>  dk> support in the Tornado GUI, that's what you get.
>
>I don't have Tornado, or a GUI.  I have VxWorks 5.3.1 and the build
>tools for Solaris... and that's it.

  Well, then you have to add a #define INCLUDE_ something-or-other to your
config.h (or whatever it's called) file.  I haven't explored the
complications of manually editing all those configuration defines, since I
have a nice GUI that does it all for me!

>  >> I built 2.95.2 from GNU sources for vxworks and libstdc++ built fine
>  >> there--is this a know 2.95.3 thing?  Or a known build issue?  Or...?
>
>  dk> Are you sure?  I've build 2.95.2 from GNU a few times myself and
>  dk> it never worked for me!  I'll give it another go; maybe it is
>  dk> indeed a regression in 2.95.3
>
>I definitely did it.  The entire build completed without any errors; I
>had to change a few files IIRC.  Many others have too, on the crossgcc
>list.  I found a post and a web site by folks who'd done it, and
>followed their directions.  I'll have to dig back through my notes...
>
>Oh, right.  I remember now.  The build failure on fdmatch.c is easily
>overcome by adding an #include  into sys/stat.h.
>Works like a charm.
>
>I followed the directions posted to crossgcc by Tony Farrell on 20 Jul
>2000; see this link:
>
>  http://sources.redhat.com/ml/crossgcc/2000-07/msg00112.html
>
>This is m68k but I did it with ppc with mostly the same steps.

   Cheers for the pointer!  I'll add an extra fixinclude script to take care
of that problem and see how it goes.  I've made a few minor fixes in my
source code, which should have the effect of not trying to use DOS paths on
a Solaris x-compiler any more - moved some stuff from vxppc.h to xm-cygwin32
(IIRC), so I'll want to do an update in the next few days anyway.  It would
be great to have libstdc++ build OK, although I'm a little worried it might
duplicate some of the OS functionality in a non-compatible way.

        DaveK
- --
They laughed at Galileo.  They laughed at Copernicus.  They laughed at
Columbus. But remember, they also laughed at Bozo the Clown.



---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: VxSim: reloc overflow
Date: Wed, 7 Feb 2001 17:01:27 -0000
From: "Dave Korn" 
Organization: Lumber Cartel (tinlc) Members #2234-2237 (owing to browser refresh)
Message-ID: 
References:  <20010207002049.09920.00001152@ng-cc1.aol.com>

Fnordz wrote in message <20010207002049.09920.00001152@ng-cc1.aol.com>...
>Hello,

   HAH!  I CAN SEE THE FNORDS!!! I CAN *SEE* THE FNORDS!!!

 

          DaveK
- --
They laughed at Galileo.  They laughed at Copernicus.  They laughed at
Columbus. But remember, they also laughed at Bozo the Clown.



---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: *UPDATED* Gcc 2.95.3 for VxWorks-PPC available....
Date: Wed, 7 Feb 2001 18:04:38 -0000
From: "Dave Korn" 
Organization: Lumber Cartel (tinlc) Members #2234-2237 (owing to browser refresh)
Message-ID: 
References:  <2001Feb6.091228.6806@lorelei.approve.se>  <95s07b$nkg$1@nnrp1.deja.com>

bblank@init-ka.de wrote in message <95s07b$nkg$1@nnrp1.deja.com>...
>Hi Dave,
>
>I have the same problems as Yu Cao:
>
>After successfuly compiling my ExceptionTest.cpp and building
>Exception_Test.out the target-download says:
>
>Errors while downloading
>W:/USER/BLK/T2proj/ExctTest/PPC860gnu/Exception_Test.out:
>__start_cp_handler
>__eh_alloc
>
>any ideas?

  Same answer: link with libgcc, see

news:

        DaveK
- --
They laughed at Galileo.  They laughed at Copernicus.  They laughed at
Columbus. But remember, they also laughed at Bozo the Clown.



---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: Building a bootrom with C++ support
Date: Wed, 7 Feb 2001 18:10:07 -0000
From: "Dave Korn" 
Organization: Lumber Cartel (tinlc) Members #2234-2237 (owing to browser refresh)
Message-ID: 
References: <95rh53$99p$1@nnrp1.deja.com>

[RMRL]BSR wrote in message <95rh53$99p$1@nnrp1.deja.com>...
>Hello,
>
>I'm having problems building a bootrom for one of our processors (a
>Motorola Scout card).  The original BSP builds OK, but I want to add
>more functionality to it.  This additional functionality has been coded
>in C++ but when I try to link the bootrom I get unresolved external
>references (see below).


>ccppc -BC:\TORNADO\host\x86-win32\lib\gcc-lib\ -mstrict-align -ansi -
>nostdinc -DRW_MULTI_T
>HREAD -D_REENTRANT -O2 -fvolatile -fno-builtin -fno-for-scope  -I\h   -
>I. -IC:\TORNADO\tar
>get\config\all -IC:\TORNADO\target\h -IC:\TORNADO\target\src\config -
>IC:\TORNADO\target\sr
>c\drv -DCPU=PPC604  -DSCOUT    -c FlashInterface.cpp
>ccppc -BC:\TORNADO\host\x86-win32\lib\gcc-lib\ -c -mstrict-align -ansi -
>nostdinc -O2 -fvol
>atile -fno-builtin -fno-for-scope -Wall -I\h   -I. -
>IC:\TORNADO\target\config\all -IC:\TOR
>NADO\target\h -IC:\TORNADO\target\src\config -
>IC:\TORNADO\target\src\drv -DCPU=PPC604  -DS
>COUT  -gdwarf -O0 -DROM_BASED  -o version.o
>C:\TORNADO\target\config\all\version.c
>ldppc -X -N  -e _romInit -Ttext 00F00000   \
>    -o bootrom_uncmp romInit.o bootInit_uncmp.o version.o bootConfig.o \
>    sysALib.o sysLib.o  miiLib.obj FlashDriver.o FlashInterface.o
>C:\TORNADO\target\lib\
>libPPC604gnuvx.a
>C:\TORNADO\target\lib\libPPC604gnuvx.a(cplusGlob.o): In function
>`cplusCtorsLink':
>cplusGlob.o(.text+0x22): undefined reference to `_ctors'
>cplusGlob.o(.text+0x26): undefined reference to `_ctors'
>C:\TORNADO\target\lib\libPPC604gnuvx.a(cplusGlob.o): In function
>`cplusDtorsLink':
>cplusGlob.o(.text+0x56): undefined reference to `_dtors'
>cplusGlob.o(.text+0x5a): undefined reference to `_dtors'
>make.exe: *** [bootrom_uncmp] Error 0x1


  I don't spot any munching being done there.  Presumably the C++ support
contains some static objects that need to be constructed at init-time;
that's what the _ctors and _dtors symbols are for, they refer to lists of
pointers-to-functions that are called to construct all the static objects,
and munch is responsible for building them (into a file called ctdt.c).

  Try adding the munch step, that should help.

       DaveK
- --
They laughed at Galileo.  They laughed at Copernicus.  They laughed at
Columbus. But remember, they also laughed at Bozo the Clown.



---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: Hi guys.. Where can I find vxworks 5.4 OS?
Date: Wed, 7 Feb 2001 18:11:01 -0000
From: "Dave Korn" 
Organization: Lumber Cartel (tinlc) Members #2234-2237 (owing to browser refresh)
Message-ID: 
References: <95rj2b$5ih238@baran22.ttnet.net.tr>

SentinuS wrote in message <95rj2b$5ih238@baran22.ttnet.net.tr>...
>

  Look in %WIND_BASE%, it's generally there somewhere!

      DaveK
- --
They laughed at Galileo.  They laughed at Copernicus.  They laughed at
Columbus. But remember, they also laughed at Bozo the Clown.




---------------------------

Newsgroups: comp.os.vxworks
Subject: I Need Your Help Please!!! 
Date: Wed, 07 Feb 2001 18:20:52 GMT
From: "Jim" 
Organization: Magma Communications Ltd.
Message-ID: <8mgg6.820$f5.158250@news>

Hello all,
I am running into a small oddity here with two different bsp's ... and i am
wondering if you could help ... our ping (from windriver) may have a mem
leak ...

At the command your shell:
>memShow
>ping 
let it run for about 10 or 20 .. or 100's if you want then ctrl-c out of it
>memShow

Then email me and tell me if you alloc'd amount of mem has increase ...
repeat if you want and tell me if its consistent ...

Thanks ... I really appreciate help with this one ;-)

Cheers
Jim



---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: I Need Your Help Please!!! 
Date: Wed, 07 Feb 2001 18:30:42 GMT
From: "Jim" 
Organization: Magma Communications Ltd.
Message-ID: 
References: <8mgg6.820$f5.158250@news>

Here is what I see ... fyi

- -> memShow
 status   bytes    blocks   avg block  max block
 ------ --------- -------- ---------- ----------
current
   free  13912184        3    4637394  13898432
  alloc    460536      293       1571        -
cumulative
  alloc    463728      301       1540        -
value = 0 = 0x0
- -> ping "192.168.103.70"
PING 192.168.103.70: 56 data bytes
64 bytes from mars (192.168.103.70): icmp_seq=0. time=0. ms
64 bytes from mars (192.168.103.70): icmp_seq=1. time=0. ms
64 bytes from mars (192.168.103.70): icmp_seq=2. time=0. ms
64 bytes from mars (192.168.103.70): icmp_seq=3. time=0. ms
64 bytes from mars (192.168.103.70): icmp_seq=4. time=0. ms
64 bytes from mars (192.168.103.70): icmp_seq=5. time=0. ms
64 bytes from mars (192.168.103.70): icmp_seq=6. time=0. ms
64 bytes from mars (192.168.103.70): icmp_seq=7. time=0. ms
64 bytes from mars (192.168.103.70): icmp_seq=8. time=0. ms
64 bytes from mars (192.168.103.70): icmp_seq=9. time=0. ms
64 bytes from mars (192.168.103.70): icmp_seq=10. time=0. ms
64 bytes from mars (192.168.103.70): icmp_seq=11. time=0. ms
64 bytes from mars (192.168.103.70): icmp_seq=12. time=0. ms
64 bytes from mars (192.168.103.70): icmp_seq=13. time=0. ms
64 bytes from mars (192.168.103.70): icmp_seq=14. time=0. ms
64 bytes from mars (192.168.103.70): icmp_seq=15. time=0. ms
64 bytes from mars (192.168.103.70): icmp_seq=16. time=0. ms
64 bytes from mars (192.168.103.70): icmp_seq=17. time=0. ms
64 bytes from mars (192.168.103.70): icmp_seq=18. time=0. ms
64 bytes from mars (192.168.103.70): icmp_seq=19. time=0. ms
64 bytes from mars (192.168.103.70): icmp_seq=20. time=0. ms
^[c64 bytes from mars (192.168.103.70): icmp_seq=21. time=0. ms
64 bytes from mars (192.168.103.70): icmp_seq=22. time=0. ms
^[64 bytes from mars (192.168.103.70): icmp_seq=23. time=0. ms
 e38e0 vxTaskEntry    +5c : shell ()
 b612c shell          +180: b6158 ()
 b6368 shell          +3bc: execute ()
 b64e8 execute        +d4 : yyparse ()
103da8 yyparse        +790: 101d64 ()
101ed4 yystart        +878: 90b2c (54)
 90b2c ping           +478: recvfrom ()
 926bc recvfrom       +48 : bsdRecvfrom ()
 87b7c bsdRecvfrom    +b8 : soreceive ()
 fb49c soreceive      +318: sbwait ()
 843f0 sbwait         +2c : ksleep ()
 851a4 ksleep         +68 : semTake ()
 cad00 semTake        +134: semBTake ()
- ----192.168.103.70 PING Statistics----
24 packets transmitted, 24 packets received, 0% packet loss
round-trip (ms)  min/avg/max = 0/0/0
tShell restarted.

- -> memShow
 status   bytes    blocks   avg block  max block
 ------ --------- -------- ---------- ----------
current
   free  13909144        3    4636381  13898432
  alloc    463576      296       1566        -
cumulative
  alloc    488312      312       1565        -
value = 0 = 0x0
- ->




"Jim"  wrote in message
news:8mgg6.820$f5.158250@news...
> Hello all,
> I am running into a small oddity here with two different bsp's ... and i
am
> wondering if you could help ... our ping (from windriver) may have a mem
> leak ...
>
> At the command your shell:
> >memShow
> >ping 
> let it run for about 10 or 20 .. or 100's if you want then ctrl-c out of
it
> >memShow
>
> Then email me and tell me if you alloc'd amount of mem has increase ...
> repeat if you want and tell me if its consistent ...
>
> Thanks ... I really appreciate help with this one ;-)
>
> Cheers
> Jim
>
>



---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: Fibre Channel to Raid Speed
Date: Wed, 07 Feb 2001 13:01:49 -0600
From: Bob Irwin 
Message-ID: <3A819B9D.61E6D1BC@sis-dev.com>
References: <3A6F77F3.FAB5284B@erols.com> <3A70CC93.3444D4EB@west.raytheon.com>  <3A7F61A9.ABCC795D@west.raytheon.com>


> > > Fred Walther wrote:
> > > >
> > > > Does anybody have any experience with Synergy VGM5 PPC and Fibre channel
> > > > PMC?
> > > > I am using DosFs and I am getting only 30 MB/Sec Read and 20 MBytes/Sec
> > > > Write to the raid disks. Has anybody gotten better speeds with VxWorks
> > > > and any other PPC Please let me know. Eventually I will need to use
> > > > DosFs2


Sounds to me like the disks are probably the bottleneck. 20 MBytes/sec
(sustained
is excellent for most disks) in writing.  check out the xfr specs for your
drives.  if you dont already, try 'striping' in your raid array.  Also,
i think you'll find that dosFs2 will give you better performance.

  - bob

---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: I Need Your Help Please!!! 
Date: Wed, 07 Feb 2001 19:16:56 GMT
From: "Jim" 
Organization: Magma Communications Ltd.
Message-ID: 
References: <8mgg6.820$f5.158250@news> 

Seems like ctrl-c doesn't clean up mem in shell ...

When i try ping ,10 it seems to use up less memory ...

Jim

"Jim"  wrote in message
news:mvgg6.821$f5.158625@news...
> Here is what I see ... fyi
>
> -> memShow
>  status   bytes    blocks   avg block  max block
>  ------ --------- -------- ---------- ----------
> current
>    free  13912184        3    4637394  13898432
>   alloc    460536      293       1571        -
> cumulative
>   alloc    463728      301       1540        -
> value = 0 = 0x0
> -> ping "192.168.103.70"
> PING 192.168.103.70: 56 data bytes
> 64 bytes from mars (192.168.103.70): icmp_seq=0. time=0. ms
> 64 bytes from mars (192.168.103.70): icmp_seq=1. time=0. ms
> 64 bytes from mars (192.168.103.70): icmp_seq=2. time=0. ms
> 64 bytes from mars (192.168.103.70): icmp_seq=3. time=0. ms
> 64 bytes from mars (192.168.103.70): icmp_seq=4. time=0. ms
> 64 bytes from mars (192.168.103.70): icmp_seq=5. time=0. ms
> 64 bytes from mars (192.168.103.70): icmp_seq=6. time=0. ms
> 64 bytes from mars (192.168.103.70): icmp_seq=7. time=0. ms
> 64 bytes from mars (192.168.103.70): icmp_seq=8. time=0. ms
> 64 bytes from mars (192.168.103.70): icmp_seq=9. time=0. ms
> 64 bytes from mars (192.168.103.70): icmp_seq=10. time=0. ms
> 64 bytes from mars (192.168.103.70): icmp_seq=11. time=0. ms
> 64 bytes from mars (192.168.103.70): icmp_seq=12. time=0. ms
> 64 bytes from mars (192.168.103.70): icmp_seq=13. time=0. ms
> 64 bytes from mars (192.168.103.70): icmp_seq=14. time=0. ms
> 64 bytes from mars (192.168.103.70): icmp_seq=15. time=0. ms
> 64 bytes from mars (192.168.103.70): icmp_seq=16. time=0. ms
> 64 bytes from mars (192.168.103.70): icmp_seq=17. time=0. ms
> 64 bytes from mars (192.168.103.70): icmp_seq=18. time=0. ms
> 64 bytes from mars (192.168.103.70): icmp_seq=19. time=0. ms
> 64 bytes from mars (192.168.103.70): icmp_seq=20. time=0. ms
> ^[c64 bytes from mars (192.168.103.70): icmp_seq=21. time=0. ms
> 64 bytes from mars (192.168.103.70): icmp_seq=22. time=0. ms
> ^[64 bytes from mars (192.168.103.70): icmp_seq=23. time=0. ms
>  e38e0 vxTaskEntry    +5c : shell ()
>  b612c shell          +180: b6158 ()
>  b6368 shell          +3bc: execute ()
>  b64e8 execute        +d4 : yyparse ()
> 103da8 yyparse        +790: 101d64 ()
> 101ed4 yystart        +878: 90b2c (54)
>  90b2c ping           +478: recvfrom ()
>  926bc recvfrom       +48 : bsdRecvfrom ()
>  87b7c bsdRecvfrom    +b8 : soreceive ()
>  fb49c soreceive      +318: sbwait ()
>  843f0 sbwait         +2c : ksleep ()
>  851a4 ksleep         +68 : semTake ()
>  cad00 semTake        +134: semBTake ()
> ----192.168.103.70 PING Statistics----
> 24 packets transmitted, 24 packets received, 0% packet loss
> round-trip (ms)  min/avg/max = 0/0/0
> tShell restarted.
>
> -> memShow
>  status   bytes    blocks   avg block  max block
>  ------ --------- -------- ---------- ----------
> current
>    free  13909144        3    4636381  13898432
>   alloc    463576      296       1566        -
> cumulative
>   alloc    488312      312       1565        -
> value = 0 = 0x0
> ->
>
>
>
>
> "Jim"  wrote in message
> news:8mgg6.820$f5.158250@news...
> > Hello all,
> > I am running into a small oddity here with two different bsp's ... and i
> am
> > wondering if you could help ... our ping (from windriver) may have a mem
> > leak ...
> >
> > At the command your shell:
> > >memShow
> > >ping 
> > let it run for about 10 or 20 .. or 100's if you want then ctrl-c out of
> it
> > >memShow
> >
> > Then email me and tell me if you alloc'd amount of mem has increase ...
> > repeat if you want and tell me if its consistent ...
> >
> > Thanks ... I really appreciate help with this one ;-)
> >
> > Cheers
> > Jim
> >
> >
>
>



---------------------------

Newsgroups: comp.os.vxworks,comp.security.firewalls,comp.lang.c,alt.computer.security
Subject: Re: Random number generator
Date: Wed, 07 Feb 2001 20:18:35 GMT
From: Christer Palm 
Organization: chello broadband
Message-ID: <3A81BBAC.5FB0DBC2@nogui.se>
References: <95f2ts$hat$1@nnrp1.deja.com>

This is a multi-part message in MIME format.
- --------------E5FC7FF4BFF0D181BE94FAC6
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit

milind_gunjal@my-deja.com wrote:
> 
> Couls someone suggest me a good 64-bit pseudo random number generation
> algorithm and/or code.
> 
What constitutes a "good" pseudo random number generator also depends on
what you want to use it for. Do you only need good random distribution
(i.e. for statistical simulations, etc.), do you need repeatability, or
do you want "cryptographically strong" random numbers?

For a great, although somewhat outdated, introduction and discussion
about pseudo random number generation, I suggest the chapter on random
numbers in Donald Knuth's "The art of computer programming", vol 2 -
Seminumerical Algorithms.

Either way, don't use rand()... It usually sucks.
- --------------E5FC7FF4BFF0D181BE94FAC6
Content-Type: text/x-vcard; charset=us-ascii;
 name="palm.vcf"
Content-Transfer-Encoding: 7bit
Content-Description: Card for Christer Palm
Content-Disposition: attachment;
 filename="palm.vcf"

begin:vcard 
n:Palm;Christer
tel;cell:+46 (70) 327 30 23
tel;fax:+46 (8) 23 49 01
tel;work:+46 (8) 23 49 00
x-mozilla-html:FALSE
url:http://www.nogui.se
org:NOGUI AB
adr:;;Palmfeltsvägen 21;JOHANNESHOV;;SE-121 62;SWEDEN
version:2.1
email;internet:palm@nogui.se
title:UNIX-Specialist
note:OBS! Ny adress!!
x-mozilla-cpt:;-31296
fn:Christer Palm
end:vcard

- --------------E5FC7FF4BFF0D181BE94FAC6--


---------------------------

Newsgroups: comp.os.vxworks,comp.security.firewalls,comp.lang.c,alt.computer.security
Subject: Re: Random number generator
Date: Wed, 07 Feb 2001 20:45:26 GMT
From: Martin Ambuhl 
Organization: EarthLink Inc. -- http://www.EarthLink.net
Message-ID: <3A81B432.30C536F5@earthlink.net>
References: <95f2ts$hat$1@nnrp1.deja.com> <3A81BBAC.5FB0DBC2@nogui.se>

Christer Palm wrote:

> Either way, don't use rand()... It usually sucks.

This is a QoI issue.  There is no way for anyone to know that someone
else's implementation of rand() 'sucks' without examining it or its
behavior.  Not even Christer Palm has such omniscience.

---------------------------

Newsgroups: comp.os.vxworks,comp.security.firewalls,comp.lang.c,alt.computer.security
Subject: Re: Random number generator
Date: Wed, 07 Feb 2001 20:52:28 GMT
From: Christer Palm 
Organization: chello broadband
Message-ID: <3A81C39D.6590FDEB@nogui.se>
References: <95f2ts$hat$1@nnrp1.deja.com> <3A81BBAC.5FB0DBC2@nogui.se> <3A81B432.30C536F5@earthlink.net>

This is a multi-part message in MIME format.
- --------------63FBA2632A635678C9E46942
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit

Martin Ambuhl wrote:
> 
> Christer Palm wrote:
> 
> > Either way, don't use rand()... It usually sucks.
> 
> This is a QoI issue.  There is no way for anyone to know that someone
> else's implementation of rand() 'sucks' without examining it or its
> behavior.  Not even Christer Palm has such omniscience.

That's correct, and that's also why I wrote "usually"...
- --------------63FBA2632A635678C9E46942
Content-Type: text/x-vcard; charset=us-ascii;
 name="palm.vcf"
Content-Transfer-Encoding: 7bit
Content-Description: Card for Christer Palm
Content-Disposition: attachment;
 filename="palm.vcf"

begin:vcard 
n:Palm;Christer
tel;cell:+46 (70) 327 30 23
tel;fax:+46 (8) 23 49 01
tel;work:+46 (8) 23 49 00
x-mozilla-html:FALSE
url:http://www.nogui.se
org:NOGUI AB
adr:;;Palmfeltsvägen 21;JOHANNESHOV;;SE-121 62;SWEDEN
version:2.1
email;internet:palm@nogui.se
title:UNIX-Specialist
note:OBS! Ny adress!!
x-mozilla-cpt:;-31296
fn:Christer Palm
end:vcard

- --------------63FBA2632A635678C9E46942--


---------------------------

Newsgroups: comp.os.vxworks,comp.security.firewalls,comp.lang.c,alt.computer.security
Subject: Re: Random number generator
Date: Wed, 07 Feb 2001 20:46:20 GMT
From: kaz@ashi.footprints.net (Kaz Kylheku)
Organization: Psycho-Neurotic Institute for the Very, Very Nervous
Message-ID: 
References: <95f2ts$hat$1@nnrp1.deja.com> <3A81BBAC.5FB0DBC2@nogui.se> <3A81B432.30C536F5@earthlink.net>
Reply-To: kaz@ashi.footprints.net

On Wed, 07 Feb 2001 20:45:26 GMT, Martin Ambuhl  wrote:
>Christer Palm wrote:
>
>> Either way, don't use rand()... It usually sucks.
>
>This is a QoI issue.  There is no way for anyone to know that someone
>else's implementation of rand() 'sucks' without examining it or its
>behavior.  Not even Christer Palm has such omniscience.

If you want to write a portable C program which can only meet its
requirements if it has a PRNG that is good for some purpose, then you
cannot use rand(), precisely because the quality varies. What sucks is
the ANSI C interface specification for the function, which does not
state any requirements on the PRNG behavior; this ``suckiness'' exists
independently of the behavior of any implementation.

---------------------------

Newsgroups: comp.os.vxworks,comp.security.firewalls,comp.lang.c,alt.computer.security
Subject: Re: Random number generator
Date: 07 Feb 2001 13:33:36 -0800
From: Micah Cowan 
Organization: Transmeta Corporation
Message-ID: 
References: <95f2ts$hat$1@nnrp1.deja.com> <3A81BBAC.5FB0DBC2@nogui.se>
Sender: mcowan@dyn-26-123.transmeta.com

Christer Palm  writes:

> Content-Type: multipart/mixed;
>	boundary="------------E5FC7FF4BFF0D181BE94FAC6"

Please don't do this... Text-only... blah blah...

Micah

---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: Hi guys.. Where can I find vxworks 5.4 OS?
Date: Thu, 8 Feb 2001 00:18:19 +0200
From: "TTNet" 
Organization: TTnet News Server 
Message-ID: <95se39$9ov33@baran22.ttnet.net.tr>
References: <95rj2b$5ih238@baran22.ttnet.net.tr> 
Reply-To: "TTNet" 

        Where can I find you stupid ????

"Dave Korn"  wrote in message
news:A6gg6.34$Zg4.14913@newsr1.u-net.net...
> SentinuS wrote in message <95rj2b$5ih238@baran22.ttnet.net.tr>...
> >
>
>   Look in %WIND_BASE%, it's generally there somewhere!
>
>       DaveK
> --
> They laughed at Galileo.  They laughed at Copernicus.  They laughed at
> Columbus. But remember, they also laughed at Bozo the Clown.
>
>
>



---------------------------

Newsgroups: comp.os.vxworks
Subject: MVME162-20 == MVME162LX ????
Date: Wed, 07 Feb 2001 15:41:04 -0700
From: Pete Willemsen 
Organization: School of Computing, University of Utah
Message-ID: <3A81CF00.3DBA5783@cs.utah.edu>

Hi.

We've got a MVME 162-20 board and a MVME 162-020A board that we're using
for an application.  We also have an updated ROM for each of these
boards, but the names printed on the ROM packages are MVME 162LX and
MVME 162/LC/FX.

What I have been unable to determine is which of these ROMs go with our
boards.  I've found some information on the Motorola site that would
lead me to believe that the following equivalencies are true:

   MVME 162-20   ==  MVME 162LX
   MVME 162-020A ==  MVME 162/LC/FX

but I'm not for sure.  Can anyone verify this for me?  Or point me to
some documentation?  At first I thought that the LC designated the
68040LC processor, but the processor on the 020A is not a 68040LC
processor (at least it isn't printed on the chip as such).

Thanks for your help.

Pete Willemsen
School of Computing
University of Utah

---------------------------

Newsgroups: comp.os.vxworks
Subject: LXT971 MDIO Problem
Date: Wed, 07 Feb 2001 23:32:53 GMT
From: Shahid Mahmood 
Organization: Deja.com
Message-ID: <95slv6$dhd$1@nnrp1.deja.com>



This is what I have:
vxWorks 5.4/T2
MPC860P
BSP for Fads860T (with lxt970a)
Custom board: 860P with lxt971
Driver source for motFecEnd

Problem:
Cannot access lxt registers.
Details:
During initialization the driver code writes a read command
(0x60020000) to the MPC860 MII_DATA register (IMMR+0x0e80). the speed
register (IMMR + 0xe84) is 0x14 (2.5 MHz @ 50MHz system clk). Then it
waits for semaphore, which comes. Now the contents of MII_DATA (IMMR +
0x0e80) should be 0x600x. Instead It reads 0xFFFFFFFF
Any attempt to write anything to MII_DATA register has no effect. ie.
the contents remain 0xFFFFFFFF and MDIO pind show no activity. However,
MDC gives appropriate clock and I can change it with MII_SPEED.

The contents of MII_DATA can be changed if I write 0 to the MII_SPEED
register. But the moment a valid number is written to the mii_speed
register, the mii_data becomes 0xffffffff.

While on FADS board, I can write data to MII_DATA anytime and it
intantly appears on the MDIO pins. Reading the MII_DATA also shows
meaningful numbers.

any help ?









Sent via Deja.com
http://www.deja.com/

---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: Ethernet driver for Intel PRO/100+
Date: Thu, 08 Feb 2001 00:19:38 GMT
From: vlam@JasmineNetworks.com
Organization: Deja.com
Message-ID: <95somj$fo5$1@nnrp1.deja.com>
References: <95rs8u$jht$1@nnrp1.deja.com> <3A8178BA.27D7D052@agilent.com>

Intel 82559ER will not be in production, Intel had recalled all ER or
ver 9 chip.

In article <3A8178BA.27D7D052@agilent.com>,
  James Marshall  wrote:
> Intel's driver has no support ("see WindRiver" they say) and we've
found
> problems using it on our 82559ER hardware - stick to
WindRiver's 'fei'.
>
> A few weeks ago I posted a fix (search back through DejaNews for
559ER) which
> adds a single line to code that searches for the 557 so it also
matches the
> 559ER. If you use a faster than usual system tick rate (ours is 1kHz)
then
> you'll also need a fix to the driver code. I got WRS support to give
me a
> modified driver object file. The driver uses tick delays without
scaling them
> against sysClkRateGet(). Hopefully by now they have an SPR against it.
>
> By the way, we're using the END driver ('fei') not the BSD driver
('if_fei') -
> the info above assumes you use the END driver.
>
> James.
>
> chafmich@my-deja.com wrote:
>
> > I am new to VxWorks development and I need to get a 100Mbps NIC
working
> > on VxWorks that runs on an x86 platform.  I have an Intel Pro/100+
NIC
> > which is based on the Intel 82559 chip.  I have read that the if_fei
> > driver supports the Intel 82557, but I assume it does not support
the
> > newer 82559.
> >
> > I have downloaded the eev driver from Intel's web page that is part
of
> > the 82559ER driver zip file.  I am going to try to use that driver.
> >
> > I am not sure if I am going down the right path, so I am looking for
> > some reassurance that what I'm doing is the right thing to do.
> >
> > Thanks,
> >
> > Mike
> >
> > Sent via Deja.com
> > http://www.deja.com/
>
>


Sent via Deja.com
http://www.deja.com/

---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: sbrk equivalent in VxWorks
Date: Thu, 08 Feb 2001 00:21:38 GMT
From: vlam@JasmineNetworks.com
Organization: Deja.com
Message-ID: <95soqa$fqv$1@nnrp1.deja.com>
References: <95qkoc$jpi$1@nnrp1.deja.com>

If you have Diab compiler there is an sbrk source code that you can
used.
In article <95qkoc$jpi$1@nnrp1.deja.com>,
  haribalram@my-deja.com wrote:
> Hi
> I am designing a memory manager for a protocol stack.  This memory
> manager wld maintain a pool of different sized memory chunks and
> allocate the chunks to the requested module/appln.  When the pool
> doesn't hv enough chunks to allocate to, it does sbrk to get more
> memory from the os.  sbrk() is used to change dynamically the amount
of
> space allocated for the calling process's data segment. This
> implementation works fine in solaris.  I need to port this code to
> VxWorks.
>
> I wld like to know if there is an equivalent of "sbrk" in VxWorks?  If
> its not there, can anybody suggest how to implement sbrk in VxWorks?
>
> Thanx in advance.
>
> -
> regards,
> Haribalram.
>
> Sent via Deja.com
> http://www.deja.com/
>


Sent via Deja.com
http://www.deja.com/

---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: *UPDATED* Gcc 2.95.3 for VxWorks-PPC available....
Date: 8 Feb 2001 01:56:03 GMT
From: yucao@alumnae.caltech.edu (Yu Cao)
Organization: California Institute of Technology, Pasadena
Message-ID: <95subj$fft@gap.cco.caltech.edu>
References:  <2001Feb6.091228.6806@lorelei.approve.se>  <95s07b$nkg$1@nnrp1.deja.com> 

Hi Dave. I tried to follow your instructions and took out -nostdlib
at the partial link step. However, that alone did not work, as certain
C++ related things are defined both in libgcc.a and libPPC860gnuvx.a,
which caused many conflicts.

So I tried the following:

1. Ran "arppc -x" on libPPC860gnuvx.a in a temp directory;

2. Removed a few .o files: new.o, exception.o, tinfo.o, and cplusInit.o;

3. Ran "arppc crus libPPC860gnuvx.a" and replaced old libPPC860gnuvx.a;

4. Commented out cplusLibInit() in usrConfig.c.

Then things would link using the new libgcc.a. However, I have no idea
what the effect is by taking out cplusLibInit(), as I could not find the
source code for cplusInit.o. I will try it out on our board in a few
minutes.

More hints would be appreciated. Or maybe I should check your website
again and see if you've already provided a canned libPPC860gnuvx.a ...

- --Yu


"Dave Korn"  writes:

>bblank@init-ka.de wrote in message <95s07b$nkg$1@nnrp1.deja.com>...
>>Hi Dave,
>>
>>I have the same problems as Yu Cao:
>>
>>After successfuly compiling my ExceptionTest.cpp and building
>>Exception_Test.out the target-download says:
>>
>>Errors while downloading
>>W:/USER/BLK/T2proj/ExctTest/PPC860gnu/Exception_Test.out:
>>__start_cp_handler
>>__eh_alloc
>>
>>any ideas?

>  Same answer: link with libgcc, see

>news:

>        DaveK
>--
>They laughed at Galileo.  They laughed at Copernicus.  They laughed at
>Columbus. But remember, they also laughed at Bozo the Clown.



---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: build problem
Date: Thu, 08 Feb 2001 02:46:59 GMT
From: "GuruDev" 
Organization: Road Runner
Message-ID: 
References: <95qeld$9aq$1@overload.lbl.gov>

which 'make' you are using? It may be due to "/" versus "\".
also try with relative path name like
../../../sssafe/embedd/rnlog.h  or whatever it is.
- -Gurudev

"sachithanandam karthikselvan"  wrote in message
news:95qeld$9aq$1@overload.lbl.gov...
> Hi all,
>
> 1) I included some c files and then tried to built it.
> It says "no such files or directroy".
>
> 2) I checked the makefile. Instead full path, it has file name only.
> ex. rnlog.h. So I gave full path ex. c:\sssafe\embedd\rnlog.h in the
makefile
> itself.
>
> 3) Now I said make from the dos prompt.Again it says that file not found.
> But it is able to comple other vxworks related files like romstart.c
>
> 4) Can anyone throw some lights??
>
> Thanks in advance
> karthikselvan
>
>
>
>
> ____________________________________________________________________
> Get free email and a permanent address at http://www.netaddress.com/?N=1



---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: *UPDATED* Gcc 2.95.3 for VxWorks-PPC available....
Date: Thu, 8 Feb 2001 03:17:32 -0000
From: "Dave Korn" 
Organization: Lumber Cartel (tinlc) Members #2234-2237 (owing to browser refresh)
Message-ID: 
References:  <2001Feb6.091228.6806@lorelei.approve.se>  <95s07b$nkg$1@nnrp1.deja.com>  <95subj$fft@gap.cco.caltech.edu>


"Yu Cao"  wrote in message
news:95subj$fft@gap.cco.caltech.edu...
> Hi Dave. I tried to follow your instructions and took out -nostdlib
> at the partial link step. However, that alone did not work, as certain
> C++ related things are defined both in libgcc.a and libPPC860gnuvx.a,
> which caused many conflicts.
>
> So I tried the following:
>
> 1. Ran "arppc -x" on libPPC860gnuvx.a in a temp directory;
>
> 2. Removed a few .o files: new.o, exception.o, tinfo.o, and cplusInit.o;
>
> 3. Ran "arppc crus libPPC860gnuvx.a" and replaced old libPPC860gnuvx.a;
>
> 4. Commented out cplusLibInit() in usrConfig.c.
>
> Then things would link using the new libgcc.a. However, I have no idea
> what the effect is by taking out cplusLibInit(), as I could not find the
> source code for cplusInit.o. I will try it out on our board in a few
> minutes.
>
> More hints would be appreciated. Or maybe I should check your website
> again and see if you've already provided a canned libPPC860gnuvx.a ...
>
> --Yu

    Hi again, Yu!

  I'm afraid I haven't achieved any miracles yet!  Now, maybe the best
solution would be for me to fix the gcc configuration so it builds libgcc
but leaves those modules out.  On the other hand, I've just succeeded in
getting libstdc++ to build, so maybe I should go down the road of replacing
all the OS C++ functionality completely.  This will need a bit of trial and
error to find out.

  I don't know what cplusLibInit does either, but it's probably important.
My guess would be it calls static constructors for objects,  perhaps it
makes entries into the VxWorks symbol table as well.

  Anyway, if you've got a log of the compiler output lying around that shows
a list of all the conflicting functions, that might come in handy.  There's
an email link on the website (mind the spamblock :) or you could post it
here.

    cheers,
         DaveK






---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: Hi guys.. Where can I find vxworks 5.4 OS?
Date: Thu, 8 Feb 2001 03:27:17 -0000
From: "Dave Korn" 
Organization: Lumber Cartel (tinlc) Members #2234-2237 (owing to browser refresh)
Message-ID: 
References: <95rj2b$5ih238@baran22.ttnet.net.tr>  <95se39$9ov33@baran22.ttnet.net.tr>


"TTNet"  wrote in message
news:95se39$9ov33@baran22.ttnet.net.tr...
>         Where can I find you stupid ????
>

     I'm over here !!1!


        DaveK
- --
They laughed at Galileo.  They laughed at Copernicus.  They laughed at
Columbus. But remember, they also laughed at Bozo the Clown.

> "Dave Korn"  wrote in message
> news:A6gg6.34$Zg4.14913@newsr1.u-net.net...
> > SentinuS wrote in message <95rj2b$5ih238@baran22.ttnet.net.tr>...
> > >

                                                    Now I'm over here!!!1


                                                              DaveK
- --
They laughed at Galileo.  They laughed at Copernicus.  They laughed at
Columbus. But remember, they also laughed at Bozo the Clown.
> >
> >   Look in %WIND_BASE%, it's generally there somewhere!
> >
> >       DaveK
> > --
> > They laughed at Galileo.  They laughed at Copernicus.  They laughed at
> > Columbus. But remember, they also laughed at Bozo the Clown.
> >

       Now I'm
       down
          here

> >
> >
>
>
                                                              ha! you'll never catch me!!!!!1

        DaveK
- --
They laughed at Galileo.  They laughed at Copernicus.  They laughed at
Columbus. But remember, they also laughed at Bozo the Clown.










                                    look now i hace hidden in the basement
                                      ooooo it's dark in here....






---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: MVME162-20 == MVME162LX ????
Date: Thu, 08 Feb 2001 04:22:57 GMT
From: Bill Dennen 
Organization: Motorola Computer Group
Message-ID: <3A821ACB.C666CDE3@home.com>
References: <3A81CF00.3DBA5783@cs.utah.edu>
Reply-To: william.dennen@motorola.com

Pete Willemsen wrote:
> 
> Hi.
> 
> We've got a MVME 162-20 board and a MVME 162-020A board that we're using
> for an application.  We also have an updated ROM for each of these
> boards, but the names printed on the ROM packages are MVME 162LX and
> MVME 162/LC/FX.
> 
> What I have been unable to determine is which of these ROMs go with our
> boards.  I've found some information on the Motorola site that would
> lead me to believe that the following equivalencies are true:
> 
>    MVME 162-20   ==  MVME 162LX
>    MVME 162-020A ==  MVME 162/LC/FX
> 
> but I'm not for sure.  Can anyone verify this for me?  Or point me to
> some documentation?  At first I thought that the LC designated the
> 68040LC processor, but the processor on the 020A is not a 68040LC
> processor (at least it isn't printed on the chip as such).
> 
> Thanks for your help.
> 
> Pete Willemsen
> School of Computing
> University of Utah
I believe you are correct in your research but since we marketed a wide
variety of these boards over time I cannot be certain.  A quick rule of
thumb is that if it has 2 IP sites and all front panel IO it's an LX, if
there are 4 IP sites its an FX.  If you can supply the engineering part
numbers of the board I can identify it precisely.  That number can be
found
on the solder side of the board and begins 01-W.  Be sure to include the
contents of the sticker found nearby which identifies the revision.

Regards
- -- 
>@<
Bill Dennen		william.dennen@motorola.com
Motorola Computer Group, TOPAZ

---------------------------

Newsgroups: comp.os.vxworks
Subject: WTX Error 0x100C8
Date: Wed, 7 Feb 2001 23:36:20 -0500
From: "Normand Barriere" 
Message-ID: 

Hi All,

Currently I'm experiencing a strange problem with Tornado. I have a piece of
software that is fairly modular, and that I'm starting to build and link
 eventually debug ) each layer one at a time. Today after adding a small
software block, I'm no longer able to download the VxWorks image without
this error
Error : load failed with error 0x100C8 ( AGENT_UNSUPPORTED_REQUEST)

Does anybody know what this error mean ? Has anybody experienced this error,
and how they were able to resolve it ?


Thanks

Normand





---------------------------

Newsgroups: comp.os.vxworks
Subject: about configuring IP at booting time..
Date: Thu, 8 Feb 2001 13:43:23 +0900
From: "½Å°æ" 
Message-ID: 

 is there any good ways to set the IP address of target system at booting
time?

 I designed target system with motorola powerPC 860SAR and wanna set the IP
at boot time through serial port(RS-232). There seems to be some functions
to do these things. But I cannnot find anything on many manuals from wrs,
including reference manual, programmer's guide and network guide.

 Any guy knows how to do it. plz help.



---------------------------

Newsgroups: comp.os.vxworks
Subject: Raven MPIC timer interrupt question
Date: Thu, 08 Feb 2001 04:57:23 GMT
From: wazkero@my-deja.com
Organization: Deja.com
Message-ID: <95t8vh$t5q$1@nnrp1.deja.com>

Hi guys!

Maybe you could help me with this one!  :)

I'm using one of the Raven MPIC timers (on a MCP750 board) that has to
run once in a while to provide me with a precise time offset.  In order
to acheive this, after the timer is properly setup, I fire it up at
some point in my code.  I would then disable it from within it's own
associated ISR using something like this:

sysPciWrite32(MPIC_ADDR(MPIC_TIMER1_BASE_CT_REG),MPIC_TIMER_DISABLE);
EIEIO;       /* synchronize */

This seems to work just fine, stopping the count, but the next time I
start this timer using something like this:

sysPciWrite32(MPIC_ADDR(MPIC_TIMER1_BASE_CT_REG),
                             (offset) & ~MPIC_TIMER_BASE_CT_CI));

the ISR gets called right away, even though 'offset' has a value
corresponding to almost one second.  Bizarre thing is it will work once
out of two, meaning if I setup this timer again, it'll work properly...

I made a test using a dummy (empty) ISR function: instead of disabling
the timer, I would switch the handler to this dummy function like this:

sysIntTbl[(int)TIMER1_INT_VEC]->vec = dummy_isr_function;

then switch it back to my regular ISR function & resetting the timer
when I am ready to use it for real and this seemed to work just fine.

So... is there something wrong with disabling the timer inside it's own
associated ISR?  Or am I missing something?

Your thoughts will be appreciated!

Thanks!


Sent via Deja.com
http://www.deja.com/

---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: Hi guys.. Where can I find vxworks 5.4 OS?
Date: Thu, 8 Feb 2001 09:41:02 +0200
From: "SentinuS" 
Organization: TTnet News Server 
Message-ID: <95tf7s$jkj2@baran22.ttnet.net.tr>
References: <95rj2b$5ih238@baran22.ttnet.net.tr>  <95se39$9ov33@baran22.ttnet.net.tr> 
Reply-To: "SentinuS" 

    heh... You are really stupid.. don't matter..

"Dave Korn"  wrote in message
news:jnog6.22648$Ee3.579496@news6-win.server.ntlworld.com...
>
> "TTNet"  wrote in message
> news:95se39$9ov33@baran22.ttnet.net.tr...
> >         Where can I find you stupid ????
> >
>
>      I'm over here !!1!
>
>
>         DaveK
> --
> They laughed at Galileo.  They laughed at Copernicus.  They laughed at
> Columbus. But remember, they also laughed at Bozo the Clown.
>
> > "Dave Korn"  wrote in message
> > news:A6gg6.34$Zg4.14913@newsr1.u-net.net...
> > > SentinuS wrote in message <95rj2b$5ih238@baran22.ttnet.net.tr>...
> > > >
>
>                                                     Now I'm over here!!!1
>
>
>                                                               DaveK
> --
> They laughed at Galileo.  They laughed at Copernicus.  They laughed at
> Columbus. But remember, they also laughed at Bozo the Clown.
> > >
> > >   Look in %WIND_BASE%, it's generally there somewhere!
> > >
> > >       DaveK
> > > --
> > > They laughed at Galileo.  They laughed at Copernicus.  They laughed at
> > > Columbus. But remember, they also laughed at Bozo the Clown.
> > >
>
>        Now I'm
>        down
>           here
>
> > >
> > >
> >
> >
>                                                               ha! you'll
never catch me!!!!!1
>
>         DaveK
> --
> They laughed at Galileo.  They laughed at Copernicus.  They laughed at
> Columbus. But remember, they also laughed at Bozo the Clown.
>
>
>
>
>
>
>
>
>
>
>                                     look now i hace hidden in the basement
>                                       ooooo it's dark in here....
>
>
>
>
>



---------------------------

Newsgroups: comp.os.vxworks
Subject: Can't download big .out file to VXSIM
Date: Thu, 08 Feb 2001 08:14:12 GMT
From: "rt" 
Organization: EarthLink Inc. -- http://www.EarthLink.net
Message-ID: 

 can't download big  .out file (~20 MB) to VxWorks simulator.   Simulator
memory  is growing up to 3 MB  and then memory grows very slow 2K per 10
mins. I didn't succeed to download the image of 20Mb..

I use Tornado 2 (VxWorks 5.4) and Windows 2000 simulator. I run VXSIM from
the command line and provide the options to allocate 100M of Ram for VXSIM.

Did anyone face same problem and what could be the solution for the big
binary images?

Thanks,
Rt



---------------------------

Newsgroups: comp.os.vxworks
Subject: Please help with C++ examples for VxWorks
Date: Thu, 08 Feb 2001 08:20:54 GMT
From: talrostoker@my-deja.com
Organization: Deja.com
Message-ID: <95tkt5$5rt$1@nnrp1.deja.com>

Hi Guys,

I need your help with C++ examples for VxWorks.
Things like spawning task that exacute a class function (I would like
this class to be active).
I would be very happy with any C++ code for VxWorks.

TIA,

Tal Rostoker


Sent via Deja.com
http://www.deja.com/

---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: unknown error message
Date: Thu, 08 Feb 2001 08:42:09 GMT
From: Urban Lindberg 
Organization: Deja.com
Message-ID: <95tm51$7i5$1@nnrp1.deja.com>
References: <3a804305.21210562@192.168.1.214> <95rirs$ams$1@nnrp1.deja.com> <3a81779b.6423375@192.168.1.214>

Hi!

I saw in another post that you were trying to compile C++ support into
the bootrom, I am not sure if this is possible. If I remember correctly
vxworks resets almost the entire memory and therefor no constructors can
be called before vxworks has started. There are probably a couple of
more reasons why it will not work. I am not an expert on C++, but trying
to use C++ before the OS has started is maybe not a good idea. If you
get it to work, please let me know how. Is it possible to load vxworks
if you remove the C++ parts? If so, then at least you know it was not
the aout support that was missing. Btw. INCLUDE_AOUT, thats Tornado 1.x
configuration? It's always a good idea to say what Tornado version and
which arch. (BSP) you have, when posting.

Regards

Urban

In article <3a81779b.6423375@192.168.1.214>,
  kimp@avtecinc.com (Kim Putnam) wrote:
> I ran objdump386 -a and agot a.out....
>
> I believe INCLUDE_AOUT is defined.
>
> K
> On Wed, 07 Feb 2001 13:33:48 GMT, Urban Lindberg 
> wrote:
>
> >Hi!
> >
> >The error you get is because you have not included support for the
> >object format you are trying to load. Try with including ELF support
> >when you build the bootrom. objdump -a filename tells you what
> >format the file is.
> >
> >HTH
> >
> >Urban
> >
> >In article <3a804305.21210562@192.168.1.214>,
> >  kimp@avtecinc.com (Kim Putnam) wrote:
> >> I need to locate any information on bootLoadModule(). I am getting
the
> >> following error "S_bootLoadLib_ROUTINE_NOT_INSTALLED"
> >>
> >> kim
> >>
> >
> >
> >Sent via Deja.com
> >http://www.deja.com/
>
>


Sent via Deja.com
http://www.deja.com/

---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: *UPDATED* Gcc 2.95.3 for VxWorks-PPC available....
Date: Thu, 08 Feb 2001 10:03:24 +0100
From: Roger 
Message-ID: <3A8260DC.875C3C84@here.i.am>
References: 

Dave Korn wrote:
> 
>      Hi everyone,
> 
>   My apologies for the false start.  I've now tracked down the bug with
> ccppc not finding various executables: it was my Cygwin installation getting
> in the way that stopped me from experiencing it myself.  I've now fixed the
> bugs and verified that the compiler works by removing my Cygwin installation
> and registry settings; it still ran fine, so I believe the code is correct
> now.

I'm new here. I just got a VxWorks box as part of an image processing
system.
Where can I download this version of gcc? Also, where is the FAQ (if
any) for
this group. Finally, I would be quite happe for a few pointers to a
source of
vxWorks software. I tried a web search and did not find much. So, where
should
I have looked?

I'm looking forward to getting a handle on VxWorks. Can anyone suggest a
good
book? I have been told that vxWorks does not lend itself to a user's
guide, but
I think that must be odd if true.

Please respond here or to roger @ opq . se (no spaces)

TIA

---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: I Need Your Help Please!!!
Date: Thu, 08 Feb 2001 10:55:03 GMT
From: Urban Lindberg 
Organization: Deja.com
Message-ID: <95ttu5$d27$1@nnrp1.deja.com>
References: <8mgg6.820$f5.158250@news>  

Hi!

This has been explained before (maybe something for the FAQ ?). The
shell has to allocate memory in the target for the text string (remember
it has to pass the address of the string to the function). The shell can
not determine when/if it is safe to free this allocated memory, so it is
better to leave it. It can not reuse the string next time either,
because the string might have changed, so it will allocate some more
memory next time you call ping. You could try to use a variable:

pingaddress = "192.168.103.70"
memShow
ping pingaddress
memShow


Urban

In article ,
  "Jim"  wrote:
> Seems like ctrl-c doesn't clean up mem in shell ...
>
> When i try ping ,10 it seems to use up less memory ...
>
> Jim
>
> "Jim"  wrote in message
> news:mvgg6.821$f5.158625@news...
> > Here is what I see ... fyi
> >
> > -> memShow
> >  status   bytes    blocks   avg block  max block
> >  ------ --------- -------- ---------- ----------
> > current
> >    free  13912184        3    4637394  13898432
> >   alloc    460536      293       1571        -
> > cumulative
> >   alloc    463728      301       1540        -
> > value = 0 = 0x0
> > -> ping "192.168.103.70"
> > PING 192.168.103.70: 56 data bytes
> > 64 bytes from mars (192.168.103.70): icmp_seq=0. time=0. ms
> > 64 bytes from mars (192.168.103.70): icmp_seq=1. time=0. ms
> > 64 bytes from mars (192.168.103.70): icmp_seq=2. time=0. ms
> > 64 bytes from mars (192.168.103.70): icmp_seq=3. time=0. ms
> > 64 bytes from mars (192.168.103.70): icmp_seq=4. time=0. ms
> > 64 bytes from mars (192.168.103.70): icmp_seq=5. time=0. ms
> > 64 bytes from mars (192.168.103.70): icmp_seq=6. time=0. ms
> > 64 bytes from mars (192.168.103.70): icmp_seq=7. time=0. ms
> > 64 bytes from mars (192.168.103.70): icmp_seq=8. time=0. ms
> > 64 bytes from mars (192.168.103.70): icmp_seq=9. time=0. ms
> > 64 bytes from mars (192.168.103.70): icmp_seq=10. time=0. ms
> > 64 bytes from mars (192.168.103.70): icmp_seq=11. time=0. ms
> > 64 bytes from mars (192.168.103.70): icmp_seq=12. time=0. ms
> > 64 bytes from mars (192.168.103.70): icmp_seq=13. time=0. ms
> > 64 bytes from mars (192.168.103.70): icmp_seq=14. time=0. ms
> > 64 bytes from mars (192.168.103.70): icmp_seq=15. time=0. ms
> > 64 bytes from mars (192.168.103.70): icmp_seq=16. time=0. ms
> > 64 bytes from mars (192.168.103.70): icmp_seq=17. time=0. ms
> > 64 bytes from mars (192.168.103.70): icmp_seq=18. time=0. ms
> > 64 bytes from mars (192.168.103.70): icmp_seq=19. time=0. ms
> > 64 bytes from mars (192.168.103.70): icmp_seq=20. time=0. ms
> > ^[c64 bytes from mars (192.168.103.70): icmp_seq=21. time=0. ms
> > 64 bytes from mars (192.168.103.70): icmp_seq=22. time=0. ms
> > ^[64 bytes from mars (192.168.103.70): icmp_seq=23. time=0. ms
> >  e38e0 vxTaskEntry    +5c : shell ()
> >  b612c shell          +180: b6158 ()
> >  b6368 shell          +3bc: execute ()
> >  b64e8 execute        +d4 : yyparse ()
> > 103da8 yyparse        +790: 101d64 ()
> > 101ed4 yystart        +878: 90b2c (54)
> >  90b2c ping           +478: recvfrom ()
> >  926bc recvfrom       +48 : bsdRecvfrom ()
> >  87b7c bsdRecvfrom    +b8 : soreceive ()
> >  fb49c soreceive      +318: sbwait ()
> >  843f0 sbwait         +2c : ksleep ()
> >  851a4 ksleep         +68 : semTake ()
> >  cad00 semTake        +134: semBTake ()
> > ----192.168.103.70 PING Statistics----
> > 24 packets transmitted, 24 packets received, 0% packet loss
> > round-trip (ms)  min/avg/max = 0/0/0
> > tShell restarted.
> >
> > -> memShow
> >  status   bytes    blocks   avg block  max block
> >  ------ --------- -------- ---------- ----------
> > current
> >    free  13909144        3    4636381  13898432
> >   alloc    463576      296       1566        -
> > cumulative
> >   alloc    488312      312       1565        -
> > value = 0 = 0x0
> > ->
> >
> >
> >
> >
> > "Jim"  wrote in message
> > news:8mgg6.820$f5.158250@news...
> > > Hello all,
> > > I am running into a small oddity here with two different bsp's ...
and i
> > am
> > > wondering if you could help ... our ping (from windriver) may have
a mem
> > > leak ...
> > >
> > > At the command your shell:
> > > >memShow
> > > >ping 
> > > let it run for about 10 or 20 .. or 100's if you want then ctrl-c
out of
> > it
> > > >memShow
> > >
> > > Then email me and tell me if you alloc'd amount of mem has
increase ...
> > > repeat if you want and tell me if its consistent ...
> > >
> > > Thanks ... I really appreciate help with this one ;-)
> > >
> > > Cheers
> > > Jim
> > >
> > >
> >
> >
>
>


Sent via Deja.com
http://www.deja.com/

---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: about configuring IP at booting time..
Date: 8 Feb 2001 12:09:13 +0100
From: jochen@siemens.com.br ("Jochen Mielke de Lima")
Organization: Mailgate.ORG Server - http://www.Mailgate.ORG
Message-ID: <00a001c091c7$c62fba70$4850d681@siemens.com.br>
References: 

You can do the following :
start your terminal program (like tip, or hyperterminal) and reboot your
target.
You=B4ll see a message like "Press any key to stop auto-boot..." on your
terminal.
Press any key. Then you=B4ll get a prompt. In this prompt, type "c", then
enter.
You=B4ll then have access to the boards boot parameters, and among them y=
ou=B4ll
find the target IP address+netmask (I guess the parameter is "inet
address").
Just press enter to bypass the other parameters. Once you=B4re done, type=
 "@"
and enter.
That=B4s it.


- -----------------------------------------------------------------
Jochen Mielke de Lima - ICN TR ES
Software Development Engineer
CITS / Siemens
+55 41 341 6861
- -----------------------------------------------------------------


- -- 
Posted from [200.230.253.37] 
via Mailgate.ORG Server - http://www.Mailgate.ORG

---------------------------

End of New-News digest
**********************


From vxwexplo-errs@csg.lbl.gov  Thu Feb  8 07:52:10 2001
From: "Sinn, Larry" 
Date: Thu Feb  8 07:52:12 PST 2001
Subject: Re: sbrk equivalent in VxWorks

Haribalram:

sbrk() is the Unix routine to request more memory from the OS memory pool.
If your memory manager only lives in your protocol stack,  just replace it
with a malloc() call.
If your memory manager is going to replace all the malloc()/free() you are
doing, use memPartFindMax() followed by malloc(), but leave enough room for
WRS routines to also get some memory.

Larry.


vlam@JasmineNetworks.com wrote:

If you have Diab compiler there is an sbrk source code that you can
used.
In article <95qkoc$jpi$1@nnrp1.deja.com>,
  haribalram@my-deja.com wrote:
> Hi
> I am designing a memory manager for a protocol stack.  This memory
> manager wld maintain a pool of different sized memory chunks and
> allocate the chunks to the requested module/appln.  When the pool
> doesn't hv enough chunks to allocate to, it does sbrk to get more
> memory from the os.  sbrk() is used to change dynamically the amount
of
> space allocated for the calling process's data segment. This
> implementation works fine in solaris.  I need to port this code to
> VxWorks.
>
> I wld like to know if there is an equivalent of "sbrk" in VxWorks?  If
> its not there, can anybody suggest how to implement sbrk in VxWorks?
>
> Thanx in advance.
>
> -
> regards,
> Haribalram.


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  Thu Feb  8 10:19:32 2001
From: Jim Way 
Date: Thu Feb  8 10:19:34 PST 2001
Subject: Re: dialog with WindRiver - reasonable response

Bill wrote:

> I should note that some people at WRS are really helpful.  Some of
> them could even be developers for WRS who post helpful messages here
> on occasion.  At any rate, I don't think it is bad to `flame' their
> technical support.  It is pretty atrocious.  I had asked if anyone
> here had positive comments on it.  I didn't hear any, but maybe that
> is apathy.

I just want to amplify what you've said about support. The phone-in support
does seem to be less than what I would expect, considering the cost of the
software, licenses, etc. HOWEVER, my experiences with my local sales team
including the FAE (Field Application Engineer) have been positive and
helpful. But my best resource for support has been this user group. Up until
a couple months ago, I was the only person in my department with any vxWorks
experience. This group has become my lifeline.

You are also correct that sometimes WRS developers do respond to posts in
this group.

FWIW, I've been on the receiving side of customer support calls. I
understand how difficult it can be to extract useful information from a
caller. But WRS doesn't make the process any easier by requiring you to
first validate your license, then get a call-back telling you you're
validated, then wait hours/days for a callback from someone to try to help
with your problem. I suspect that they have a two-tiered system where one
group is exclusively phone handlers and the other group is exclusively
development engineering. Perhaps they could benefit from rotating their
development staff through the customer support function (in very short
shifts, of course, of no more than one week at a time). I also don't have
much sense that they divide the incoming calls up between different host
platforms or different target processors, although that *could* be happening
and I just don't see it.

WRS could also help their customers by developing user-focused
documentation. What they have at the moment I describe as "menu
documentation". Here's a list of everything you can do. But what I want is
documentation that answers the question, "How do I ...?". That's the
question I bring to the manuals, and I'm always frustrated in my quest for
knowledge. I know (again from personal experience) that this documentation
takes a long time to develop. I also know that once the customers find it
and learn to trust it, it can dramatically reduce the amount of tech support
that needs to be provided in real-time.

Repeating a complaint (suggestion?) I've made here before, a great start
would be to create a Master Index that spans all the documentation. Too
often I see people chastising to "RTFM". Without a Master Index, how do you
know *which* FM?

The bottom line is, we are smart, motivated, hardworking people who are
trying to use vxWorks as a tool to do our jobs. We don't need to be
spoonfed. We don't need to be belittled or intimidated either.

My $0.02,
Jim
-----------------------------------------
Jim Way, Software Engineer
Datum Austin (Austron Inc.)
voice: 512.721.4170
fax  : 512.990.9712
email: jwayATdatumDOTcom (no spam please)
-----------------------------------------
Tornado vxWorks


From vxwexplo-errs@csg.lbl.gov  Thu Feb  8 10:49:29 2001
From: rnaismit@squared.com
Date: Thu Feb  8 10:49:31 PST 2001
Subject: Re: C++ in BootRom


Original Question:

---------------

Newsgroups: comp.os.vxworks
Subject: Building a bootrom with C++ support
Date: Wed, 07 Feb 2001 13:04:38 GMT
From: [RMRL]BSR 
Organization: Deja.com
Message-ID: <95rh53$99p$1@nnrp1.deja.com>

Hello,

I'm having problems building a bootrom for one of our processors (a
Motorola Scout card).  The original BSP builds OK, but I want to add
more functionality to it.  This additional functionality has been coded
in C++ but when I try to link the bootrom I get unresolved external
references (see below).

The additional functionality has been added successfully to a bootable
VxWorks image.

Does anyone know whether it is possible to link C++ code with the
bootrom and if so what I need to do?

Cheers,

Paul Crowley

------


I have C++ in my BootRom.  I had to modify the rule in rules.bsp, as follows:

You need to add the step to create the ctdt.o and link it in.  The ctdt.o has
the constructors/destructors.

Good Luck,
Ron

bootrom_uncmp : depend.$(BSP_NAME) bootInit_uncmp.o romInit.o \
               bootConfig.o $(MACH_DEP) $(LDDEPS) $(LIBS)
               - @ $(RM) $@
               - @ $(RM) $(IMI)
               - @ $(RM) $(IBR)
               - @ $(RM) version.o
               - @ $(RM) ctmp.o ctdt.c ctdt.o
               $(CC) -c $(CFLAGS) -o version.o $(CONFIG_ALL)/version.c
               $(LD) -o ctmp.o $(LDFLAGS) $(LD_PARTIAL_FLAGS) bootConfig.o \
               $(MACH_DEP) version.o $(LIBS)
# Here is one added line to create the ctdt.o
          $(NM) ctmp.o | $(MUNCH) >ctdt.c
     $(MAKE) CC_COMPILER="-traditional" ctdt.o
# Add the ctdt.o in the link in the following line

     $(LD) $(LDFLAGS) $(ROM_LDFLAGS) -e $(ROM_ENTRY) $(LD_HIGH_FLAGS) \
     -o $@ romInit.o bootInit_uncmp.o  ctdt.o version.o bootConfig.o \
     $(MACH_DEP) $(LIBS)

     $(BINHEX) $(HEX_FLAGS) $(SECT_SPEC)bootrom_uncmp$(TXT_OFFSET) \
     $(IMI_SPEC)$(IMI)$(IMI_OFFSET) > $@.hex
     -@ $(MV_BIN)
     $(CONVERT) $@.hex -a -l f0000100 -u f03fffff

     $(CONVERT) $@ -c -m arm
     $(NM) -A -n -r -t x $@ > map
     $(ROMSIZEPROG) -b $(ROM_SIZE) $@
     $(LDOUT_CONV) $@












From vxwexplo-errs@csg.lbl.gov  Thu Feb  8 16:41:45 2001
From: "CICON,JAMES (HP-Vancouver,ex1)" 
Date: Thu Feb  8 16:41:48 PST 2001
Subject: WindView and portMapper

When ever I launch WindView the vxWorks system task 'tPortmapd' stack
overflows.  

WindView continues to run and returns trace data that looks fine at first
glance until you notice that it is not recording any context switches.

The trace shows that the tWdbTask is always executing, that the tNetTask is
ready and that all other tasks are either pended or delayed.  This is false
because other tasks are running, for example we know that Tornado is
attached and passing network traffic to the target.  Also LEDs are flashing
and the console continues to work.

My questions are:
1) Why is WindView causing tPortmapd to overflow its stack space.
2) Is there away to increase tPortmapd stack size
3) Why is Windview not reporting any context switches?

Jim Cicon
Inkjet Products Division
Hewlett Packard Company




VxWorks Tornado


From vxwexplo-errs@csg.lbl.gov  Thu Feb  8 21:22:59 2001
From: Fiederling.Ulf@dornier.dasa.de
Date: Thu Feb  8 21:23:01 PST 2001
Subject: Antwort: Re: dialog with WindRiver - reasonable response


Hi VxWorks users

I tried different times to get support from Wind River.
The result was always very very poor.

They were not able to help me getting code for using the RTC on a X86
platform, (they even coplained they couldnt)  i had to use a unix code,
adapted by myself.
They dont inform us about the GNU Compiler uptade with Pentium support,
although we have support contract, and they know about our performance
problems with pentium processors.
They dont react to now about the problem of getting disalignment,
downloading software with double variables, (-Malign-double switch in
makefile set, natually).

Every question for support is going into "nirvana"

Very good support we get by vxwexplo


Many thanks about that

Ulf



From vxwexplo-errs@csg.lbl.gov  Thu Feb  8 22:05:53 2001
From: Rashmi_Topno 
Date: Thu Feb  8 22:05:55 PST 2001
Subject: RE: Please help with C++ examples for VxWorks

Hi Tal,

A related question was also bothering me a lot.

As far as spawning a class member function, the following code works well
(on vxSim).

class cOne{
	public:
		void  some_func();
};

then u can spawn it by

	taskSpawn ("func", 100, 0, 3000, (FUNCPTR)one_obj.some_finc,
.......);  // where     cOne    one_obj

This will solve what you want.

Here, my question is, if you see the defintion of FUNCPTR, it goes like
this....

typedef int (*FUNCPTR) (...);

SO in my view, it takes the address to a "C" like function, i.e. the offset
of the function in object layout. Then how is it able to spawn a "C++"
function. Somewhere I have read that we should not pass a pinter to a C++
function to a place where compiler is expecting a C function.

Rashmi

Tal Rostoker wrote --->

Hi Guys,

I need your help with C++ examples for VxWorks.
Things like spawning task that exacute a class function (I would like
this class to be active).
I would be very happy with any C++ code for VxWorks.

TIA,

Tal Rostoker




From vxwexplo-errs@csg.lbl.gov  Fri Feb  9 04:03:11 2001
From: Vxworks Exploder 
Date: Fri Feb  9 04:03:13 PST 2001
Subject: comp.os.vxworks newsdigest

Comp.Os.Vxworks Daily Digest    Fri Feb  9 04:03:06 PST 2001

        Subject: Re: unknown error message
        Subject: MPC860T Rev.D motFecEnd() problem. ?
        Subject: SAMBA and VxWorks
        Subject: Re: SAMBA and VxWorks
        Subject: WTX C API: how do you look up the symbol type?
        Subject: Re: Access to CR0 register in X86
        Subject: Re: ttyDrv and Polled Mode Serial
        Subject: Re: Can't download big .out file to VXSIM
        Subject: Re: Building a bootrom with C++ support
        Subject: Source code for ether_output()
        Subject: Timers
        Subject: re: sbrk equivalent in VxWorks
        Subject: Re: Task total running time.
        Subject: Re: unknown error message
        Subject: Re: WTX C API: how do you look up the symbol type?
        Subject: loadModule from bootConfig.c
        Subject: Re: unknown error message
        Subject: Re: loadModule from bootConfig.c
        Subject: Re: loadModule from bootConfig.c
        Subject: Problems with the Intel eeV driver
        Subject: the format of .out file
        Subject: Re: MVME162-20 == MVME162LX ????
        Subject: Re: the format of .out file
        Subject: re: dialog with WindRiver - reasonable response
        Subject: re: C++ in BootRom
        Subject: Building a bootrom with C++ support
        Subject: Can't download big file.out to Vxsim
        Subject: Can't download big XXX.out file to VxWorks simulator
        Subject: Re: Building a bootrom with C++ support
        Subject: Re: Please help with C++ examples for VxWorks
        Subject: Impact of using C++ Multiple Inheritance
        Subject: Vx task stack protect
        Subject: Sporadic reset
        Subject: Sporadic reboots
        Subject: Sporadic reboots
        Subject: Sporadic rebbots
        Subject: Re: Impact of using C++ Multiple Inheritance
        Subject: Re: LXT971 MDIO Problem
        Subject: Re: Timers
        Subject: WindView and portMapper
        Subject: Re: Impact of using C++ Multiple Inheritance
        Subject: Multicast Data Reception App.
        Subject: tgtsvr : segmentation fault
        Subject: Re: the format of .out file
        Subject: PowerPC 7400 Altivec and VxWorks
        Subject: How can I check for Ethernet connection is OK.
        Subject: Re: Vx task stack protect
        Subject: Re: Impact of using C++ Multiple Inheritance
        Subject: Re: I Need Your Help Please!!!
        Subject: Re: Please help with C++ examples for VxWorks - xxx.zip (0/1)
        Subject: Antwort: re: dialog with WindRiver - reasonable response
        Subject: RE: Please help with C++ examples for VxWorks
        Subject: Test ignore it please
        Subject: Re: Problems with the Intel eeV driver
        Subject: Re: Multicast Data Reception App.

-------------------------------------------------------

Newsgroups: comp.os.vxworks
Subject: Re: unknown error message
Date: Thu, 08 Feb 2001 12:27:30 GMT
From: kimp@avtecinc.com (Kim Putnam)
Organization: Info Avenue Internet Services
Message-ID: <3a828ff5.262984@192.168.1.214>
References: <3a804305.21210562@192.168.1.214> <95rirs$ams$1@nnrp1.deja.com> <3a81779b.6423375@192.168.1.214> <95tm51$7i5$1@nnrp1.deja.com>

I am using Tornado 2.0, and VMICPCI-7715.  

I've taken out the c++ code from the bootrom. I have put the runtime
libs on flash, as well as the c++ code. I am now trying to use
loadModule to load these.  Unfortunately I am getting an error of not
insalled.  Not sure if I am missing some kind of init function or if
this is even possible.

K

On Thu, 08 Feb 2001 08:42:09 GMT, Urban Lindberg 
wrote:

>Hi!
>
>I saw in another post that you were trying to compile C++ support into
>the bootrom, I am not sure if this is possible. If I remember correctly
>vxworks resets almost the entire memory and therefor no constructors can
>be called before vxworks has started. There are probably a couple of
>more reasons why it will not work. I am not an expert on C++, but trying
>to use C++ before the OS has started is maybe not a good idea. If you
>get it to work, please let me know how. Is it possible to load vxworks
>if you remove the C++ parts? If so, then at least you know it was not
>the aout support that was missing. Btw. INCLUDE_AOUT, thats Tornado 1.x
>configuration? It's always a good idea to say what Tornado version and
>which arch. (BSP) you have, when posting.
>
>Regards
>
>Urban
>
>In article <3a81779b.6423375@192.168.1.214>,
>  kimp@avtecinc.com (Kim Putnam) wrote:
>> I ran objdump386 -a and agot a.out....
>>
>> I believe INCLUDE_AOUT is defined.
>>
>> K
>> On Wed, 07 Feb 2001 13:33:48 GMT, Urban Lindberg 
>> wrote:
>>
>> >Hi!
>> >
>> >The error you get is because you have not included support for the
>> >object format you are trying to load. Try with including ELF support
>> >when you build the bootrom. objdump -a filename tells you what
>> >format the file is.
>> >
>> >HTH
>> >
>> >Urban
>> >
>> >In article <3a804305.21210562@192.168.1.214>,
>> >  kimp@avtecinc.com (Kim Putnam) wrote:
>> >> I need to locate any information on bootLoadModule(). I am getting
>the
>> >> following error "S_bootLoadLib_ROUTINE_NOT_INSTALLED"
>> >>
>> >> kim
>> >>
>> >
>> >
>> >Sent via Deja.com
>> >http://www.deja.com/
>>
>>
>
>
>Sent via Deja.com
>http://www.deja.com/


---------------------------

Newsgroups: comp.os.vxworks
Subject: MPC860T Rev.D motFecEnd() problem. ?
Date: Thu, 8 Feb 2001 21:47:05 +0900
From: "Hideaki Izumi" 
Organization: NTT Communications Co.(OCN)
Message-ID: <95u4ce$q7d$1@nn-tk103.ocn.ad.jp>

Hi, Dear.

Since the variable of the argument of motFecRxCtrlRegValGet() function is
not initialized to 0,
the contents of a stack will be set to a RX_CTRL register.
Consequently, FEC operation mode and PHY mediation result become
inharmonious.
The same thing occurs in the part which is calling motFecTxCtrlRegValGet().
Is my indication right?


+----------------------------------------------------------------+
TAKASAGO,LTD.  2nd Development Dept. Solution 1.
1-24-16 Mizonokuchi Takatsu-ku Kawasaki Kanagawa 213-8558 Japan
Hideaki Izumi
E-Mail:izumi@tss.trd.tmg.nec.co.jp



---------------------------

Newsgroups: comp.os.vxworks
Subject: SAMBA and VxWorks
Date: Thu, 8 Feb 2001 13:56:19 +0100
From: "Jorgen Jakobsen" 
Organization: Posted Courtesy of Tele Danmark or one of its European Subsidiaries
Message-ID: <95u52t$qv3$1@news.inet.tele.dk>

Dear VxWork'ers

Anybody know of a VxWorks port of SAMBA (GNU SMB) ?

/Jørgen



---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: SAMBA and VxWorks
Date: Thu, 8 Feb 2001 13:12:26 -0000
From: "Dave Korn" 
Organization: Lumber Cartel (tinlc) Members #2234-2237 (owing to browser refresh)
Message-ID: 
References: <95u52t$qv3$1@news.inet.tele.dk>

Jorgen Jakobsen wrote in message <95u52t$qv3$1@news.inet.tele.dk>...
>Dear VxWork'ers
>
>Anybody know of a VxWorks port of SAMBA (GNU SMB) ?
>
>/Jørgen

  I asked this same question a while ago, and nobody had any answers;
websearches don't turn anything up either.

       DaveK
- --
They laughed at Galileo.  They laughed at Copernicus.  They laughed at
Columbus. But remember, they also laughed at Bozo the Clown.




---------------------------

Newsgroups: comp.os.vxworks
Subject: WTX C API: how do you look up the symbol type?
Date: Thu, 8 Feb 2001 14:33:13 +0100
From: "Carsten Schwartz" 
Organization: Posted Courtesy of Tele Danmark or one of its European Subsidiaries
Message-ID: <95u78j$98n$1@news.inet.tele.dk>
Reply-To: "Carsten Schwartz" 

It is mentioned in the documentation under wtxSymListGet( ) that you can use
the 'typeMask' from theWTX_SYMBOL struct for lookup in the description of
the structure members:

    UINT8		type;		/* Symbol type                 */
    UINT8		typeMask;	/* Symbol type mask for lookup */

How can you determine if a symbol is a short, int, double, struct, function
or whatever?

Basicly I need a translation table...

Carsten Schwartz - Any help appreciated.



---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: Access to CR0 register in X86
Date: Thu, 8 Feb 2001 09:39:13 -0000
From: "ncrnews.daytonoh.ncr.com" 
Message-ID: <3a826943$1@rpc1284.daytonoh.ncr.com>
References: <3A81763D.DF9B7F6E@agilent.com>



"James Marshall"  wrote in message
news:3A81763D.DF9B7F6E@agilent.com...
> Hi,
>
> How can I read the control register CR0 using GDB or the VxWorks shell?
I've
> tried 'info registers' but that only lists the standard regs. Any ideas?
>
> James Marshall
> Agilent Technologies
> Edinburgh, Scotland
>
>

I haven't tried this myself but...you could try try the following (from
frc386 BSP) :

Add this routine to your sysALib.s file.

/***************************************************************************
****
*
* sysRegGet - get the control and test registers
*
* This routine get the control and test registers
*
* NOMANUAL

* int sysRegGet (register)
*      int register;  /@ kind of the control or test registers @/

*/

 .align 4,0x90
_sysRegGet:
 movl SP_ARG1(%esp),%eax
 cmpl $ CR0,%eax
 je sysRegGetCR0
 cmpl $ CR2,%eax
 je sysRegGetCR2
 cmpl $ CR3,%eax
 je sysRegGetCR3
 cmpl $ TR6,%eax
 je sysRegGetTR6
 cmpl $ TR7,%eax
 je sysRegGetTR7
sysRegGetEnd:
 ret

sysRegGetCR0:
 movl %cr0,%eax
 jmp sysRegGetEnd
sysRegGetCR2:
 movl %cr2,%eax
 jmp sysRegGetEnd
sysRegGetCR3:
 movl %cr3,%eax
 jmp sysRegGetEnd
sysRegGetTR6:
 movl %tr6,%eax
 jmp sysRegGetEnd
sysRegGetTR7:
 movl %tr7,%eax
 jmp sysRegGetEnd


The definitions for CR0 etc are in the regs.h file for the i86 processor
architecture.

Hope this works for you.


- --

Niall Leonard
___________________________________________________
Senior Software Engineer,
NCR Financial Solutions Group Ltd,
Kingsway West, Dundee, DD2 3XX, Scotland
Tel:  +44 (0)1382 592866 (Direct)
Fax: +44 (0)1382 622243
e-mail: Niall.Leonard@Scotland.NCR.com







---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: ttyDrv and Polled Mode Serial
Date: Thu, 8 Feb 2001 14:14:19 -0000
From: "Kelvin Lawson" 
Message-ID: <95u9d2$igiqa$1@ID-56276.news.dfncis.de>
References: <95rptl$i9hrt$1@ID-56276.news.dfncis.de>

For those that are interested, the simple answer was to bring up the SIO
driver in interrupt mode but create a task which polls the interrupt
handler.


"Kelvin Lawson"  wrote

> WRS/VxWorks documentation suggests that polled mode serial is used for
debug
> communication while interrupt mode serial is used with ttyDrv.
>
> Does anyone know if it's possible to use a polled mode SIO driver with
> ttyDrv ? We don't have any interrupts available on this processor.
>
> Cheers,
> Kelvin.
>
>



---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: Can't download big .out file to VXSIM
Date: 8 Feb 2001 14:28:02 GMT
From: borkhuis@agere.com (Johan Borkhuis)
Organization: Lucent Technologies
Message-ID: 
References: 

"rt"  wrote:

> can't download big  .out file (~20 MB) to VxWorks simulator.  
> Simulator 
>memory  is growing up to 3 MB  and then memory grows very slow 2K per 10
>mins. I didn't succeed to download the image of 20Mb..
>
>I use Tornado 2 (VxWorks 5.4) and Windows 2000 simulator. I run VXSIM
>from the command line and provide the options to allocate 100M of Ram
>for VXSIM. 
>
>Did anyone face same problem and what could be the solution for the big
>binary images?

You can change the memory using the /r switch when starting up the 
simulator.

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: Building a bootrom with C++ support
Date: Thu, 08 Feb 2001 14:31:44 GMT
From: pcrowley 
Organization: Deja.com
Message-ID: <95uakg$n9i$1@nnrp1.deja.com>
References: <95rh53$99p$1@nnrp1.deja.com> 

Thanks Dave and Kim.

I modified the BSP bootrom makefile to munch the C++ files and this
eliminated the unresolved external references (phew).

However, the static objects do not seem to be created by the bootrom.
I think I probably need to call some function which will iterate
through the list of static objects (created by the munching stage).
Can anyone confirm that this is the case and if so, which function I
need to call?

Cheers,

Paul


Sent via Deja.com
http://www.deja.com/

---------------------------

Newsgroups: comp.os.vxworks
Subject: Source code for ether_output()
Date: Thu, 8 Feb 2001 10:09:16 -0500
From: "Stefan Chiris" 
Organization: Nortel
Message-ID: <95ucj3$afr$1@bmerhc5e.ca.nortel.com>

Hi everybody,

I have the source code for the ether_output() function that comes with
vxWorks 5.3.X using a BSD 4.3 stack.  I am looking for a copy of this source
code for the BSD 4.4 stack (ie. Tornado for SENS).  I'd really appreciate
any help in locating a copy.

Regards,
Stefan



---------------------------

Newsgroups: comp.os.vxworks
Subject: Timers
Date: Thu, 08 Feb 2001 10:59:51 -0500
From: viv 
Organization: Lucent Technologies, Columbus, Ohio
Message-ID: <3A82C277.7C6A626B@lucent.com>

Can someone explain the relation between the hardware timer, the vxWorks
timer, the sysClkRateSet(and the ticks) and how does all of this work
together?

Thanks,
viv

---------------------------

Newsgroups: comp.os.vxworks
Subject: re: sbrk equivalent in VxWorks
Date: Thu, 8 Feb 2001 07:47:12 -0800
From: "Sinn, Larry" 
Organization: Lawrence Berkeley National Laboratory
Message-ID: <95ugnt$2ic$1@overload.lbl.gov>

Haribalram:

sbrk() is the Unix routine to request more memory from the OS memory pool.
If your memory manager only lives in your protocol stack,  just replace it
with a malloc() call.
If your memory manager is going to replace all the malloc()/free() you are
doing, use memPartFindMax() followed by malloc(), but leave enough room for
WRS routines to also get some memory.

Larry.


vlam@JasmineNetworks.com wrote:

If you have Diab compiler there is an sbrk source code that you can
used.
In article <95qkoc$jpi$1@nnrp1.deja.com>,
  haribalram@my-deja.com wrote:
> Hi
> I am designing a memory manager for a protocol stack.  This memory
> manager wld maintain a pool of different sized memory chunks and
> allocate the chunks to the requested module/appln.  When the pool
> doesn't hv enough chunks to allocate to, it does sbrk to get more
> memory from the os.  sbrk() is used to change dynamically the amount
of
> space allocated for the calling process's data segment. This
> implementation works fine in solaris.  I need to port this code to
> VxWorks.
>
> I wld like to know if there is an equivalent of "sbrk" in VxWorks?  If
> its not there, can anybody suggest how to implement sbrk in VxWorks?
>
> Thanx in advance.
>
> -
> regards,
> Haribalram.


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


---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: Task total running time.
Date: Thu, 8 Feb 2001 17:28:23 +0100
From: "Romano Signorelli" 
Organization: Centro Servizi Interbusiness
Message-ID: <95ugom$29f$1@fe2.cs.interbusiness.it>
References: <3A809DAA.C3C01C4B@lucent.com>

Take a look at timexLib
Bye

"viv"  ha scritto nel messaggio
news:3A809DAA.C3C01C4B@lucent.com...
> How can I determine the total running time for a task(in READY state)
> from the time a the task has started? How can one determine/develop
> this?
>
> Thanks,
> viv



---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: unknown error message
Date: Thu, 08 Feb 2001 16:36:54 GMT
From: Urban Lindberg 
Organization: Deja.com
Message-ID: <95uhv5$ugu$1@nnrp1.deja.com>
References: <3a804305.21210562@192.168.1.214> <95rirs$ams$1@nnrp1.deja.com> <3a81779b.6423375@192.168.1.214> <95tm51$7i5$1@nnrp1.deja.com> <3a828ff5.262984@192.168.1.214>

Assuming that you at least got vxworks running, do a lkup "cplusLibInit"
in the tornado shell. If you get the address to cplusLibInit, then I
don't have a clue what's wrong (unless you're trying to access the Wind
Foundation Classes and haven't bought them). If not, then you probably
haven't included the C++ support. I don't know how experienced you are
with vxworks, but it's almost always a bad configuration of vxworks if
you get this kind of problems. If you still have problem I suggest you
start a new thread with another subject and describe some more details
about what you're doing and the error messages and the value of errno.

You could try to put:

fd = open("you_filename", 0666);
if(ERROR != fd)
{
    md = loadModule(fd, LOAD_ALL_SYMBOLS);
    if(NULL == md);
        printErrno();
}

in a c file, compile, load and execute this. It might say something more
than "not installed".

My knowledge in C++ in combination with vxworks is almost non existant,
so try to get in contact with someone that has experience in that
combination.

Regards Urban


In article <3a828ff5.262984@192.168.1.214>,
  kimp@avtecinc.com (Kim Putnam) wrote:
> I am using Tornado 2.0, and VMICPCI-7715.
>
> I've taken out the c++ code from the bootrom. I have put the runtime
> libs on flash, as well as the c++ code. I am now trying to use
> loadModule to load these.  Unfortunately I am getting an error of not
> insalled.  Not sure if I am missing some kind of init function or if
> this is even possible.
>
> K
>
> On Thu, 08 Feb 2001 08:42:09 GMT, Urban Lindberg 
> wrote:
>
> >Hi!
> >
> >I saw in another post that you were trying to compile C++ support
into
> >the bootrom, I am not sure if this is possible. If I remember
correctly
> >vxworks resets almost the entire memory and therefor no constructors
can
> >be called before vxworks has started. There are probably a couple of
> >more reasons why it will not work. I am not an expert on C++, but
trying
> >to use C++ before the OS has started is maybe not a good idea. If you
> >get it to work, please let me know how. Is it possible to load
vxworks
> >if you remove the C++ parts? If so, then at least you know it was not
> >the aout support that was missing. Btw. INCLUDE_AOUT, thats Tornado
1.x
> >configuration? It's always a good idea to say what Tornado version
and
> >which arch. (BSP) you have, when posting.
> >
> >Regards
> >
> >Urban
> >
> >In article <3a81779b.6423375@192.168.1.214>,
> >  kimp@avtecinc.com (Kim Putnam) wrote:
> >> I ran objdump386 -a and agot a.out....
> >>
> >> I believe INCLUDE_AOUT is defined.
> >>
> >> K
> >> On Wed, 07 Feb 2001 13:33:48 GMT, Urban Lindberg

> >> wrote:
> >>
> >> >Hi!
> >> >
> >> >The error you get is because you have not included support for the
> >> >object format you are trying to load. Try with including ELF
support
> >> >when you build the bootrom. objdump -a filename tells you
what
> >> >format the file is.
> >> >
> >> >HTH
> >> >
> >> >Urban
> >> >
> >> >In article <3a804305.21210562@192.168.1.214>,
> >> >  kimp@avtecinc.com (Kim Putnam) wrote:
> >> >> I need to locate any information on bootLoadModule(). I am
getting
> >the
> >> >> following error "S_bootLoadLib_ROUTINE_NOT_INSTALLED"
> >> >>
> >> >> kim
> >> >>
> >> >
> >> >
> >> >Sent via Deja.com
> >> >http://www.deja.com/
> >>
> >>
> >
> >
> >Sent via Deja.com
> >http://www.deja.com/
>
>


Sent via Deja.com
http://www.deja.com/

---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: WTX C API: how do you look up the symbol type?
Date: Thu, 08 Feb 2001 09:07:58 -0800
From: Lori Fraleigh 
Organization: Real-Time Innovations
Message-ID: <3A82D26E.9209BE02@rti.com>
References: <95u78j$98n$1@news.inet.tele.dk>


The "type" in the WTX_SYMBOL structure is a bit mask including one or more
of the following (from symbol.h):

/* symbol types */

#define SYM_UNDF        0x0     /* undefined */
#define SYM_LOCAL       0x0     /* local */
#define SYM_GLOBAL      0x1     /* global (external) (ORed) */
#define SYM_ABS         0x2     /* absolute */
#define SYM_TEXT        0x4     /* text */
#define SYM_DATA        0x6     /* data */
#define SYM_BSS         0x8     /* bss */
#define SYM_COMM        0x12    /* common symbol */
#define SYM_SDA         0x40    /* symbols related to a PowerPC SDA section */
#define SYM_SDA2        0x80    /* symbols related to a PowerPC SDA2 section
*/
#define SYM_THUMB       0x40    /* Thumb function */

The type is not the actual variable type.  The target server symbol table
does not keep track of or parse debugging information to store the types
of symbols beyond what is above.

Regards,
Lori


Carsten Schwartz wrote:
> 
> It is mentioned in the documentation under wtxSymListGet( ) that you can use
> the 'typeMask' from theWTX_SYMBOL struct for lookup in the description of
> the structure members:
> 
>     UINT8               type;           /* Symbol type                 */
>     UINT8               typeMask;       /* Symbol type mask for lookup */
> 
> How can you determine if a symbol is a short, int, double, struct, function
> or whatever?
> 
> Basicly I need a translation table...
> 
> Carsten Schwartz - Any help appreciated.

- -- 
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: loadModule from bootConfig.c
Date: Thu, 08 Feb 2001 17:14:49 GMT
From: kimp@avtecinc.com (Kim Putnam)
Organization: Info Avenue Internet Services
Message-ID: <3a82d11a.16939390@192.168.1.214>

I am running on a VMICPCI-7751 board, with Tornado 2.0.  We are using
NDDS multicast for communications. We are currenty creating a
bootrom_uncmp, using tftp to load vxworks.  What we would like to do
is use the NDDS facitilites to load vxworks. (Not even sure this is
possible).

I have loaded the NDDS shared libs required onto flash. I have create
a  filesrv.out and loaded this onto flash as well. This contains code
to init NDDS, and get the vxworks file. 

I modified bootConfig.c to not call netLoad, but instead call a
function that opens a file on flash, and then perform  the following:

errnoSet(0);

if((id = loadModule(fd,LOAD_ALL_SYMBOLS)) == 0)
{
        err = errnoGet();
        printErr("Error loading module errno= %x\n",err);

        return(ERROR);
}

The bootrom_uncmp loads, and I get the following error from loadModule
0xe0001. 

I figure that I am missing some kind of config, or init function, or
cannot execute this from bootConfig.c

Any ideas?

Kim 






---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: unknown error message
Date: Thu, 08 Feb 2001 17:18:55 GMT
From: kimp@avtecinc.com (Kim Putnam)
Organization: Info Avenue Internet Services
Message-ID: <3a82d430.17729937@192.168.1.214>
References: <3a804305.21210562@192.168.1.214> <95rirs$ams$1@nnrp1.deja.com> <3a81779b.6423375@192.168.1.214> <95tm51$7i5$1@nnrp1.deja.com> <3a828ff5.262984@192.168.1.214> <95uhv5$ugu$1@nnrp1.deja.com>

I am relatively new to vxworks,,,ugh almost hate to say that after
reading some of the posts, and going thru the archives.

I have included cplusLibInit, as well as cplusDemanglerInit.  I
figured this was a config problem, and have been trying to track it
down. I am trying to execute loadModule from bootConfig...and am not
sure there is enough in the kernel at this point to even execute this
function.  

My code is similar to what you have below and I am getting the
following error 0xe0001. I get S_loadLib_ROUTINE_NOT_INSTALLED when I
execute printErrno.

I will start a new thread and appreciate all the help.

K
On Thu, 08 Feb 2001 16:36:54 GMT, Urban Lindberg 
wrote:

>Assuming that you at least got vxworks running, do a lkup "cplusLibInit"
>in the tornado shell. If you get the address to cplusLibInit, then I
>don't have a clue what's wrong (unless you're trying to access the Wind
>Foundation Classes and haven't bought them). If not, then you probably
>haven't included the C++ support. I don't know how experienced you are
>with vxworks, but it's almost always a bad configuration of vxworks if
>you get this kind of problems. If you still have problem I suggest you
>start a new thread with another subject and describe some more details
>about what you're doing and the error messages and the value of errno.
>
>You could try to put:
>
>fd = open("you_filename", 0666);
>if(ERROR != fd)
>{
>    md = loadModule(fd, LOAD_ALL_SYMBOLS);
>    if(NULL == md);
>        printErrno();
>}
>
>in a c file, compile, load and execute this. It might say something more
>than "not installed".
>
>My knowledge in C++ in combination with vxworks is almost non existant,
>so try to get in contact with someone that has experience in that
>combination.
>
>Regards Urban
>
>
>In article <3a828ff5.262984@192.168.1.214>,
>  kimp@avtecinc.com (Kim Putnam) wrote:
>> I am using Tornado 2.0, and VMICPCI-7715.
>>
>> I've taken out the c++ code from the bootrom. I have put the runtime
>> libs on flash, as well as the c++ code. I am now trying to use
>> loadModule to load these.  Unfortunately I am getting an error of not
>> insalled.  Not sure if I am missing some kind of init function or if
>> this is even possible.
>>
>> K
>>
>> On Thu, 08 Feb 2001 08:42:09 GMT, Urban Lindberg 
>> wrote:
>>
>> >Hi!
>> >
>> >I saw in another post that you were trying to compile C++ support
>into
>> >the bootrom, I am not sure if this is possible. If I remember
>correctly
>> >vxworks resets almost the entire memory and therefor no constructors
>can
>> >be called before vxworks has started. There are probably a couple of
>> >more reasons why it will not work. I am not an expert on C++, but
>trying
>> >to use C++ before the OS has started is maybe not a good idea. If you
>> >get it to work, please let me know how. Is it possible to load
>vxworks
>> >if you remove the C++ parts? If so, then at least you know it was not
>> >the aout support that was missing. Btw. INCLUDE_AOUT, thats Tornado
>1.x
>> >configuration? It's always a good idea to say what Tornado version
>and
>> >which arch. (BSP) you have, when posting.
>> >
>> >Regards
>> >
>> >Urban
>> >
>> >In article <3a81779b.6423375@192.168.1.214>,
>> >  kimp@avtecinc.com (Kim Putnam) wrote:
>> >> I ran objdump386 -a and agot a.out....
>> >>
>> >> I believe INCLUDE_AOUT is defined.
>> >>
>> >> K
>> >> On Wed, 07 Feb 2001 13:33:48 GMT, Urban Lindberg
>
>> >> wrote:
>> >>
>> >> >Hi!
>> >> >
>> >> >The error you get is because you have not included support for the
>> >> >object format you are trying to load. Try with including ELF
>support
>> >> >when you build the bootrom. objdump -a filename tells you
>what
>> >> >format the file is.
>> >> >
>> >> >HTH
>> >> >
>> >> >Urban
>> >> >
>> >> >In article <3a804305.21210562@192.168.1.214>,
>> >> >  kimp@avtecinc.com (Kim Putnam) wrote:
>> >> >> I need to locate any information on bootLoadModule(). I am
>getting
>> >the
>> >> >> following error "S_bootLoadLib_ROUTINE_NOT_INSTALLED"
>> >> >>
>> >> >> kim
>> >> >>
>> >> >
>> >> >
>> >> >Sent via Deja.com
>> >> >http://www.deja.com/
>> >>
>> >>
>> >
>> >
>> >Sent via Deja.com
>> >http://www.deja.com/
>>
>>
>
>
>Sent via Deja.com
>http://www.deja.com/


---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: loadModule from bootConfig.c
Date: Thu, 08 Feb 2001 18:11:38 GMT
From: adrianmichaud@my-deja.com
Organization: Deja.com
Message-ID: <95ungf$4d5$1@nnrp1.deja.com>
References: <3a82d11a.16939390@192.168.1.214>



LoadModule() will return a NULL if you have not setup a load routine.
Basically, you have to call one of the following setup routines before
you call LoadModule():

loadElfInit()  -- Setup the loader code to deal with ELF images.
loadEcoffInit() - Setup the load code to deal with COFF images.
loadAoutInit() .....

etc.


If you are already doing this (Calling the appropriate loadxxxxInit()
function), then there is something wrong with the Image you're loading
(Or a mismatch between the type of image your loading).


Adrian


In article <3a82d11a.16939390@192.168.1.214>,
  kimp@avtecinc.com (Kim Putnam) wrote:
> I am running on a VMICPCI-7751 board, with Tornado 2.0.  We are using
> NDDS multicast for communications. We are currenty creating a
> bootrom_uncmp, using tftp to load vxworks.  What we would like to do
> is use the NDDS facitilites to load vxworks. (Not even sure this is
> possible).
>
> I have loaded the NDDS shared libs required onto flash. I have create
> a  filesrv.out and loaded this onto flash as well. This contains code
> to init NDDS, and get the vxworks file.
>
> I modified bootConfig.c to not call netLoad, but instead call a
> function that opens a file on flash, and then perform  the following:
>
> errnoSet(0);
>
> if((id = loadModule(fd,LOAD_ALL_SYMBOLS)) == 0)
> {
>         err = errnoGet();
>         printErr("Error loading module errno= %x\n",err);
>
>         return(ERROR);
> }
>
> The bootrom_uncmp loads, and I get the following error from loadModule
> 0xe0001.
>
> I figure that I am missing some kind of config, or init function, or
> cannot execute this from bootConfig.c
>
> Any ideas?
>
> Kim
>
>


Sent via Deja.com
http://www.deja.com/

---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: loadModule from bootConfig.c
Date: Thu, 08 Feb 2001 18:21:14 GMT
From: kimp@avtecinc.com (Kim Putnam)
Organization: Info Avenue Internet Services
Message-ID: <3a82e2da.21483375@192.168.1.214>
References: <3a82d11a.16939390@192.168.1.214> <95ungf$4d5$1@nnrp1.deja.com>

Tried that and got messages from the loadAoutInit() that it could not
find things like fopen in the symbol Table.  So....as this is in
bootrom_uncmp, my next question was is there even a symbol table in
this image?

I am quickly becoming of the opinion that I just can't do this kind of
thing in the bootrom_uncmp, but only in the vxworks image.

K
On Thu, 08 Feb 2001 18:11:38 GMT, adrianmichaud@my-deja.com wrote:

>
>
>LoadModule() will return a NULL if you have not setup a load routine.
>Basically, you have to call one of the following setup routines before
>you call LoadModule():
>
>loadElfInit()  -- Setup the loader code to deal with ELF images.
>loadEcoffInit() - Setup the load code to deal with COFF images.
>loadAoutInit() .....
>
>etc.
>
>
>If you are already doing this (Calling the appropriate loadxxxxInit()
>function), then there is something wrong with the Image you're loading
>(Or a mismatch between the type of image your loading).
>
>
>Adrian
>
>
>In article <3a82d11a.16939390@192.168.1.214>,
>  kimp@avtecinc.com (Kim Putnam) wrote:
>> I am running on a VMICPCI-7751 board, with Tornado 2.0.  We are using
>> NDDS multicast for communications. We are currenty creating a
>> bootrom_uncmp, using tftp to load vxworks.  What we would like to do
>> is use the NDDS facitilites to load vxworks. (Not even sure this is
>> possible).
>>
>> I have loaded the NDDS shared libs required onto flash. I have create
>> a  filesrv.out and loaded this onto flash as well. This contains code
>> to init NDDS, and get the vxworks file.
>>
>> I modified bootConfig.c to not call netLoad, but instead call a
>> function that opens a file on flash, and then perform  the following:
>>
>> errnoSet(0);
>>
>> if((id = loadModule(fd,LOAD_ALL_SYMBOLS)) == 0)
>> {
>>         err = errnoGet();
>>         printErr("Error loading module errno= %x\n",err);
>>
>>         return(ERROR);
>> }
>>
>> The bootrom_uncmp loads, and I get the following error from loadModule
>> 0xe0001.
>>
>> I figure that I am missing some kind of config, or init function, or
>> cannot execute this from bootConfig.c
>>
>> Any ideas?
>>
>> Kim
>>
>>
>
>
>Sent via Deja.com
>http://www.deja.com/


---------------------------

Newsgroups: comp.os.vxworks
Subject: Problems with the Intel eeV driver
Date: Thu, 8 Feb 2001 18:34:36 +0000
From: "Wade T. Oram" 
Organization: Wade and Helen
Message-ID: 
Reply-To: "Wade T. Oram" 

Hello,

I am trying to get networking operational on a Jumptec ETX-P1 board 
(Pentium 266) using the standard Tornado II Pentium BSP as the starting 
point.

I have downloaded and installed the Intel eeV driver (for Intel 82559 
based network interfaces) but I am having some problems.

The instructions with the Tornado II driver say to use the Pentium Pro 
BSP but since I do not have a pentium pro I have installed the driver in 
the Pentium BSP following the instructions given for the Pentium Pro 
Tornado II installation.

The VxWorks build works fine.


On startup get messages:

Attached TCP/IP interface to eeE unit 0
Attaching lo0...done

     ....


WDB: Ready

Hence I know my network card has been found. However, I then 
periodically (every 2 seconds) get :

0x1f9e294 (tNetTask): Panic: where are those rfd buffers


It appears that something in the initialisation is not right. I have 
re-checked that I have not missed anything when doing the installation 
and can find nothing obviously wrong.

Not being familiar with VxWorks networking, I do not know what might be 
causing this problem and at present, I am not sure how to proceed.

Any help offered will be much appreciated.

- -- 
Wade Oram

---------------------------

Newsgroups: comp.os.vxworks
Subject: the format of .out file
Date: Thu, 8 Feb 2001 14:02:03 -0500
From: "Jason Jiang" 
Organization: Verio
Message-ID: <13Cg6.15702$Sl.685666@iad-read.news.verio.net>

I built a downloadable project under Tonado and got the output file aaa.out.
But when I used 'ld' command under shell to load it into the processor
(8260), I got the errors as below:

    Relocation value does not fit in 24 bits.
    Relocation value does not fit in 24 bits.
    .......
    Relocation value does not fit in 24 bits.
    ld error: error reading file (errno = 0x3d0001).

But I can use Tonado to download it. Who can tell me what's happening?




---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: MVME162-20 == MVME162LX ????
Date: Thu, 08 Feb 2001 12:00:34 -0700
From: Pete Willemsen 
Organization: School of Computing, University of Utah
Message-ID: <3A82ECD2.7CC4D9B3@cs.utah.edu>
References: <3A81CF00.3DBA5783@cs.utah.edu> <3A821ACB.C666CDE3@home.com>

Bill Dennen wrote:
> 
> Pete Willemsen wrote:
> >
> > Hi.
> >
> > We've got a MVME 162-20 board and a MVME 162-020A board that we're using
> > for an application.  We also have an updated ROM for each of these
> > boards, but the names printed on the ROM packages are MVME 162LX and
> > MVME 162/LC/FX.
> >
> > What I have been unable to determine is which of these ROMs go with our
> > boards.  I've found some information on the Motorola site that would
> > lead me to believe that the following equivalencies are true:
> >
> >    MVME 162-20   ==  MVME 162LX
> >    MVME 162-020A ==  MVME 162/LC/FX
> >
> > but I'm not for sure.  Can anyone verify this for me?  Or point me to
> > some documentation?  At first I thought that the LC designated the
> > 68040LC processor, but the processor on the 020A is not a 68040LC
> > processor (at least it isn't printed on the chip as such).
> >
> > Thanks for your help.
> >
> > Pete Willemsen
> > School of Computing
> > University of Utah
> I believe you are correct in your research but since we marketed a wide
> variety of these boards over time I cannot be certain.  A quick rule of
> thumb is that if it has 2 IP sites and all front panel IO it's an LX, if
> there are 4 IP sites its an FX.  If you can supply the engineering part
> numbers of the board I can identify it precisely.  That number can be
> found
> on the solder side of the board and begins 01-W.  Be sure to include the
> contents of the sticker found nearby which identifies the revision.
> 
> Regards
> --
> >@<
> Bill Dennen             william.dennen@motorola.com
> Motorola Computer Group, TOPAZ

Bill,

Thanks for your help on this.  Here's the information from the boards.

For the board labeled "MVME 162-20", the numbers are

   01-W38884B 08B

For the board labeled "MVME 162-020A", the numbers are

   01-W38884B 65B

It looks like they're the same board, but different revisions based on
the part numbers.

Pete

---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: the format of .out file
Date: Thu, 08 Feb 2001 19:06:25 GMT
From: kimp@avtecinc.com (Kim Putnam)
Organization: Info Avenue Internet Services
Message-ID: <3a82edf6.24327468@192.168.1.214>
References: <13Cg6.15702$Sl.685666@iad-read.news.verio.net>

There is a lot of info on this in the archives. I just went thru these
myself.

I searched on loadModule.

K
On Thu, 8 Feb 2001 14:02:03 -0500, "Jason Jiang" 
wrote:

>I built a downloadable project under Tonado and got the output file aaa.out.
>But when I used 'ld' command under shell to load it into the processor
>(8260), I got the errors as below:
>
>    Relocation value does not fit in 24 bits.
>    Relocation value does not fit in 24 bits.
>    .......
>    Relocation value does not fit in 24 bits.
>    ld error: error reading file (errno = 0x3d0001).
>
>But I can use Tonado to download it. Who can tell me what's happening?
>
>
>


---------------------------

Newsgroups: comp.os.vxworks
Subject: re: dialog with WindRiver - reasonable response
Date: Thu, 8 Feb 2001 10:13:26 -0800
From: Jim Way 
Organization: Lawrence Berkeley National Laboratory
Message-ID: <95ur9c$93t$1@overload.lbl.gov>

Bill wrote:

> I should note that some people at WRS are really helpful.  Some of
> them could even be developers for WRS who post helpful messages here
> on occasion.  At any rate, I don't think it is bad to `flame' their
> technical support.  It is pretty atrocious.  I had asked if anyone
> here had positive comments on it.  I didn't hear any, but maybe that
> is apathy.

I just want to amplify what you've said about support. The phone-in support
does seem to be less than what I would expect, considering the cost of the
software, licenses, etc. HOWEVER, my experiences with my local sales team
including the FAE (Field Application Engineer) have been positive and
helpful. But my best resource for support has been this user group. Up until
a couple months ago, I was the only person in my department with any vxWorks
experience. This group has become my lifeline.

You are also correct that sometimes WRS developers do respond to posts in
this group.

FWIW, I've been on the receiving side of customer support calls. I
understand how difficult it can be to extract useful information from a
caller. But WRS doesn't make the process any easier by requiring you to
first validate your license, then get a call-back telling you you're
validated, then wait hours/days for a callback from someone to try to help
with your problem. I suspect that they have a two-tiered system where one
group is exclusively phone handlers and the other group is exclusively
development engineering. Perhaps they could benefit from rotating their
development staff through the customer support function (in very short
shifts, of course, of no more than one week at a time). I also don't have
much sense that they divide the incoming calls up between different host
platforms or different target processors, although that *could* be happening
and I just don't see it.

WRS could also help their customers by developing user-focused
documentation. What they have at the moment I describe as "menu
documentation". Here's a list of everything you can do. But what I want is
documentation that answers the question, "How do I ...?". That's the
question I bring to the manuals, and I'm always frustrated in my quest for
knowledge. I know (again from personal experience) that this documentation
takes a long time to develop. I also know that once the customers find it
and learn to trust it, it can dramatically reduce the amount of tech support
that needs to be provided in real-time.

Repeating a complaint (suggestion?) I've made here before, a great start
would be to create a Master Index that spans all the documentation. Too
often I see people chastising to "RTFM". Without a Master Index, how do you
know *which* FM?

The bottom line is, we are smart, motivated, hardworking people who are
trying to use vxWorks as a tool to do our jobs. We don't need to be
spoonfed. We don't need to be belittled or intimidated either.

My $0.02,
Jim
- -----------------------------------------
Jim Way, Software Engineer
Datum Austin (Austron Inc.)
voice: 512.721.4170
fax  : 512.990.9712
email: jwayATdatumDOTcom (no spam please)
- -----------------------------------------
Tornado vxWorks

---------------------------

Newsgroups: comp.os.vxworks
Subject: re: C++ in BootRom
Date: Thu, 8 Feb 2001 13:45:52 -0500
From: rnaismit@squared.com
Organization: Lawrence Berkeley National Laboratory
Message-ID: <95ur9d$93u$1@overload.lbl.gov>


Original Question:

- ---------------

Newsgroups: comp.os.vxworks
Subject: Building a bootrom with C++ support
Date: Wed, 07 Feb 2001 13:04:38 GMT
From: [RMRL]BSR 
Organization: Deja.com
Message-ID: <95rh53$99p$1@nnrp1.deja.com>

Hello,

I'm having problems building a bootrom for one of our processors (a
Motorola Scout card).  The original BSP builds OK, but I want to add
more functionality to it.  This additional functionality has been coded
in C++ but when I try to link the bootrom I get unresolved external
references (see below).

The additional functionality has been added successfully to a bootable
VxWorks image.

Does anyone know whether it is possible to link C++ code with the
bootrom and if so what I need to do?

Cheers,

Paul Crowley

- ------


I have C++ in my BootRom.  I had to modify the rule in rules.bsp, as follows:

You need to add the step to create the ctdt.o and link it in.  The ctdt.o has
the constructors/destructors.

Good Luck,
Ron

bootrom_uncmp : depend.$(BSP_NAME) bootInit_uncmp.o romInit.o \
               bootConfig.o $(MACH_DEP) $(LDDEPS) $(LIBS)
               - @ $(RM) $@
               - @ $(RM) $(IMI)
               - @ $(RM) $(IBR)
               - @ $(RM) version.o
               - @ $(RM) ctmp.o ctdt.c ctdt.o
               $(CC) -c $(CFLAGS) -o version.o $(CONFIG_ALL)/version.c
               $(LD) -o ctmp.o $(LDFLAGS) $(LD_PARTIAL_FLAGS) bootConfig.o \
               $(MACH_DEP) version.o $(LIBS)
# Here is one added line to create the ctdt.o
          $(NM) ctmp.o | $(MUNCH) >ctdt.c
     $(MAKE) CC_COMPILER="-traditional" ctdt.o
# Add the ctdt.o in the link in the following line

     $(LD) $(LDFLAGS) $(ROM_LDFLAGS) -e $(ROM_ENTRY) $(LD_HIGH_FLAGS) \
     -o $@ romInit.o bootInit_uncmp.o  ctdt.o version.o bootConfig.o \
     $(MACH_DEP) $(LIBS)

     $(BINHEX) $(HEX_FLAGS) $(SECT_SPEC)bootrom_uncmp$(TXT_OFFSET) \
     $(IMI_SPEC)$(IMI)$(IMI_OFFSET) > $@.hex
     -@ $(MV_BIN)
     $(CONVERT) $@.hex -a -l f0000100 -u f03fffff

     $(CONVERT) $@ -c -m arm
     $(NM) -A -n -r -t x $@ > map
     $(ROMSIZEPROG) -b $(ROM_SIZE) $@
     $(LDOUT_CONV) $@











---------------------------

Newsgroups: comp.os.vxworks
Subject: Can't download big file.out to Vxsim
Date: Thu, 08 Feb 2001 19:27:35 GMT
From: "Eugene Zlobin" 
Organization: EarthLink Inc. -- http://www.EarthLink.net
Message-ID: 

Hello,

I have the following problem. I can't download
big xxx.out file (~ 20 Mb) to VxWorks simulator.
I am able to load about 3-4 Mb and then downloading process is
frozen.

I use tornado 2 and simpc (NT simulator).

Has anyone solve this problem ?

Thanks,
Eugene.





---------------------------

Newsgroups: comp.os.vxworks
Subject: Can't download big XXX.out file to VxWorks simulator
Date: Thu, 08 Feb 2001 19:27:40 GMT
From: "Eugene Zlobin" 
Organization: EarthLink Inc. -- http://www.EarthLink.net
Message-ID: 

 Hello,

I have the following problem. I can't download big XXX.out file (~20 MB) to
VxWorks simulator.   Simulator memory  is growing at 3 MB  and then process
is frozen.

I use Tornado 2 (VxWorks 5.4) and NT simulator.

Has anyone solve this problem?

Thanks,
Eugene.




---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: Building a bootrom with C++ support
Date: Thu, 08 Feb 2001 19:29:28 GMT
From: kimp@avtecinc.com (Kim Putnam)
Organization: Info Avenue Internet Services
Message-ID: <3a82f32b.25660843@192.168.1.214>
References: <95rh53$99p$1@nnrp1.deja.com>  <95uakg$n9i$1@nnrp1.deja.com>

I am not much of a C++ person ( seems like I am going to have to be).

I had to add cplus support to my bootrom code. cplusLibInit() to name
but one.  Don't know if this helps, also there is a call
cplusCtorsLink()...I found these in the reference and users guide. I
have not gotten that far....I am still trying to get loadModule to
work in the bootrom.
K

On Thu, 08 Feb 2001 14:31:44 GMT, pcrowley 
wrote:

>Thanks Dave and Kim.
>
>I modified the BSP bootrom makefile to munch the C++ files and this
>eliminated the unresolved external references (phew).
>
>However, the static objects do not seem to be created by the bootrom.
>I think I probably need to call some function which will iterate
>through the list of static objects (created by the munching stage).
>Can anyone confirm that this is the case and if so, which function I
>need to call?
>
>Cheers,
>
>Paul
>
>
>Sent via Deja.com
>http://www.deja.com/


---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: Please help with C++ examples for VxWorks
Date: Thu, 08 Feb 2001 19:41:51 +0000
From: molochai 
Message-ID: 
References: <95tkt5$5rt$1@nnrp1.deja.com>

In message <95tkt5$5rt$1@nnrp1.deja.com>
          talrostoker@my-deja.com wrote:

> Hi Guys,
> 
> I need your help with C++ examples for VxWorks.
> Things like spawning task that exacute a class function (I would like
> this class to be active).

I, (and several other people for that matter) have posted this sort of thing
before.. have a trawl of dejanews for it.


---------------------------

Newsgroups: comp.os.vxworks
Subject: Impact of using C++ Multiple Inheritance
Date: Thu, 8 Feb 2001 16:06:28 -0500
From: "C Bethune" 
Organization: Bell Northern Research
Message-ID: <95v1ft$1pp$1@bcrkh13.ca.nortel.com>

Hi,

I'm looking at using multiple inheritance as part of a design I'm working
on, and wanted to get a feel for what sort of impact it has on code
performance.  I've gone through some old posts here, and noticed that
someone said that while a virtual function call under single inheritance has
minimal overhead, multiple inheritance imposes a more significant
performance penalty.  There was no explanation as to the severity of the
performance hit, or why one should expect it.  Can anyone provide any info,
or possibly point me to a resource that would explain how G++ goes about
implementing single and multiple inheritance?  The code I'm writing is on
the critical path in our application, so I'm trying to avoid imposing
unecessary overhead (beyond that typically associated with an OO design).

Cheers,

Chris Bethune



---------------------------

Newsgroups: comp.os.vxworks
Subject: Vx task stack protect
Date: Thu, 8 Feb 2001 11:17:59 +0800
From: sardine 
Organization: utstarcomsz
Message-ID: 

hi all,
  i want protect Vx task stack,to do it like as follow:
  
   create task use taskInit instead of taskSpawn,and add a gap on every task stack,
   and the gap is readonly at any time!

 how can i do it!


sardine_sz@163.net

---------------------------

Newsgroups: comp.os.vxworks
Subject: Sporadic reset
Date: Thu, 08 Feb 2001 21:26:12 GMT
From: nogap@my-deja.com
Organization: Deja.com
Message-ID: <95v2tc$epu$1@nnrp1.deja.com>

Hi guys!
Maybe you could help me with this problem.

I am using VxWorks version 5.4. on a Motorola MCP750 card in a compact PCI
cage. A major feature was added to the software and ever since the system
card reboots itself together with the non system card. Reading the boards
last reset register shows that the reason for the reset on the system card
is... pushing the front panel reset button and the reason for reset of the
non-system card is a compact PCI reset (probably triggered by the system
card). I registered a function to be called on reboot by using rebootHookAdd(
), but it hasn't been called on the spoken reboots.

Any ideas about how I can find out the cause for those resets?

Thank a lot!



Sent via Deja.com
http://www.deja.com/

---------------------------

Newsgroups: comp.os.vxworks
Subject: Sporadic reboots
Date: Thu, 08 Feb 2001 21:47:51 GMT
From: nogap@my-deja.com
Organization: Deja.com
Message-ID: <95v464$fq4$1@nnrp1.deja.com>

Hi guys!
Maybe you could help me with this problem.

I am using VxWorks version 5.4. on a Motorola MCP750 card in a compact PCI
cage. A major feature was added to the software and ever since the system
card reboots itself together with the non system card. Reading the boards
last reset register shows that the reason for the reset on the system card
is... pushing the front panel reset button and the reason for reset of the
non-system card is a compact PCI reset (probably triggered by the system
card). I registered a function to be called on reboot by using rebootHookAdd(
), but it hasn't been called on the spoken reboots.

Any ideas about how I can find out the cause for those resets?

Thank a lot!


Sent via Deja.com
http://www.deja.com/

---------------------------

Newsgroups: comp.os.vxworks
Subject: Sporadic reboots
Date: Thu, 08 Feb 2001 21:59:26 GMT
From: eitan6158@my-deja.com
Organization: Deja.com
Message-ID: <95v4rr$gb0$1@nnrp1.deja.com>

Hi guys!
Maybe you could help me with this problem.

I am using VxWorks version 5.4. on a Motorola MCP750 card in a compact
PCI cage.
A major feature was added to the software and ever since the system card
reboots itself
together with the non system card. Reading the boards last reset
register shows that the
reason for the reset on the system card is... pushing the front panel
reset button and
the reason for reset of the non-system card is a compact PCI reset
(probably triggered
by the system card).
I registered a function to be called on reboot by using rebootHookAdd(
), but it hasn't
been called on the spoken reboots.

Any ideas about how I can find out the cause for those resets?

Thank a lot!


Sent via Deja.com
http://www.deja.com/

---------------------------

Newsgroups: comp.os.vxworks
Subject: Sporadic rebbots
Date: Thu, 08 Feb 2001 22:33:10 GMT
From: nogap@my-deja.com
Organization: Deja.com
Message-ID: <95v6r5$i76$1@nnrp1.deja.com>

Hi guys!
Maybe you could help me with this problem.

I am using VxWorks version 5.4. on a Motorola MCP750 card in a compact PCI
cage. A major feature was added to the software and ever since the system
card reboots itself together with the non system card. Reading the boards
last reset register shows that the reason for the reset on the system card
is... pushing the front panel reset button and the reason for reset of the
non-system card is a compact PCI reset (probably triggered by the system
card). I registered a function to be called on reboot by using rebootHookAdd(
), but it hasn't been called on the spoken reboots.

Any ideas about how I can find out the cause for those resets?

Thank a lot!


Sent via Deja.com
http://www.deja.com/

---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: Impact of using C++ Multiple Inheritance
Date: Thu, 08 Feb 2001 18:00:20 -0600
From: Kevin Cline 
Organization: Mayan Networks
Message-ID: <3A833314.35852990@mayannetworks.com>
References: <95v1ft$1pp$1@bcrkh13.ca.nortel.com>

C Bethune wrote:

> Hi,
>
> I'm looking at using multiple inheritance as part of a design I'm working
> on, and wanted to get a feel for what sort of impact it has on code
> performance.  I've gone through some old posts here, and noticed that
> someone said that while a virtual function call under single inheritance has
> minimal overhead, multiple inheritance imposes a more significant
> performance penalty.  There was no explanation as to the severity of the
> performance hit, or why one should expect it.  Can anyone provide any info,
> or possibly point me to a resource that would explain how G++ goes about
> implementing single and multiple inheritance?  The code I'm writing is on
> the critical path in our application, so I'm trying to avoid imposing
> unecessary overhead (beyond that typically associated with an OO design).
>
> Cheers,
>
> Chris Bethune

There is no additional overhead for a virtual function
call in the presence of multiple inheritance.  There is
some cost in memory because multiply derived classes
contain multiple vtbl pointers, but it's hard to imagine
that would be significant.

Most of the arguments I've heard against multiple inheritance were
merely folklore, or were based on a bad experience
introducing multiple inheritance into system full
of clever hacks that assumed a single virtual table pointer
at the beginning of an object.  Keep your base classes abstract,
your code conforming, and you'll be fine.  But if you or
your team tries to outsmart the compiler by casting pointers
to void* and back, or by using memcpy on objects, then
use of multiple inheritance will quickly show you the error
of your ways.

Kevin Cline


---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: LXT971 MDIO Problem
Date: Fri, 09 Feb 2001 00:12:46 GMT
From: Shahid Mahmood 
Organization: Deja.com
Message-ID: <95vclp$n2s$1@nnrp1.deja.com>
References: <95slv6$dhd$1@nnrp1.deja.com>

Nevermind folks. Its fixed.


In article <95slv6$dhd$1@nnrp1.deja.com>,
  Shahid Mahmood  wrote:
>
>
> This is what I have:
> vxWorks 5.4/T2
> MPC860P
> BSP for Fads860T (with lxt970a)
> Custom board: 860P with lxt971
> Driver source for motFecEnd
>
> Problem:
> Cannot access lxt registers.
> Details:
> During initialization the driver code writes a read command
> (0x60020000) to the MPC860 MII_DATA register (IMMR+0x0e80). the speed
> register (IMMR + 0xe84) is 0x14 (2.5 MHz @ 50MHz system clk). Then it
> waits for semaphore, which comes. Now the contents of MII_DATA (IMMR +
> 0x0e80) should be 0x600x. Instead It reads 0xFFFFFFFF
> Any attempt to write anything to MII_DATA register has no effect. ie.
> the contents remain 0xFFFFFFFF and MDIO pind show no activity.
However,
> MDC gives appropriate clock and I can change it with MII_SPEED.
>
> The contents of MII_DATA can be changed if I write 0 to the MII_SPEED
> register. But the moment a valid number is written to the mii_speed
> register, the mii_data becomes 0xffffffff.
>
> While on FADS board, I can write data to MII_DATA anytime and it
> intantly appears on the MDIO pins. Reading the MII_DATA also shows
> meaningful numbers.
>
> any help ?
>
> Sent via Deja.com
> http://www.deja.com/
>


Sent via Deja.com
http://www.deja.com/

---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: Timers
Date: Thu, 08 Feb 2001 19:17:52 -0500
From: "Macleod, Randy [CAR:CF81:EXCH]" 
Organization: Nortel Networks
Message-ID: <3A833730.816FF371@americasm01.nt.com>
References: <3A82C277.7C6A626B@lucent.com>


  I'll give it a go. 

viv wrote:
> 
> Can someone explain the relation between 
> the hardware timer, 

  Some crystal oscillator + ASIC that can interrupt the CPU.

> the sysClkRateSet(and the ticks) 

If your hardware timers supports it, you can change the
rate of timer interrupts. This determines the system tick time.
After each tick, vxWorks interrupts the running task and 
may reschedule, etc.

BTW, on my PPC hardware I'm okay at 2000 Hz but 4000Hz 
locks the board!
Caution! The higher the rate the less CPU time your tasks get. 

> the vxWorks timer, 

  Do you mean:

Watchdog timers:

myWdTmrId = wdCreate();
wdStart(myWdTmrId, nTicks, pointerMyFunc, args)
  Adds a watchdog timer to the system tick queue.
  Calls MyFunc at interrupt level after nTicks.
  Call wdStart again in MyFunc to get a periodic timer.

or

timer_create()
  Less efficient Posix API that uses watchdog timers internally.
  Sends a signal to a task.
  You can setup your own signal handler using timer_connect().
  Time is dealt with in units of sec,nsec but ticks are
  used under the covers.

>and how does all of this work together?
> 

  Pretty well depending on what you want to do ;-)

If this brief explaination doesn't help, 
perhaps you could explain what you want to do.

You might also be look at:
1- vxWorks manuals
2- http://www.eecis.udel.edu/~ntp/ntpfaq/NTP-s-sw-clocks.htm
3-
http://www.smartmodulartech.com/systems/library/docs/vxworks/vxintro/vx0003.htm#watchdog_sec

// Randy 

- -- 
Randy W. Macleod ()  Phone:  (613) 765-7953
Lab 6.2-C13 Nortel Networks                  Fax:    (613) 765-6613
3500 Carling Ave, P.O. Box 3511, Station C   Dept. #: CF81, GMS:   
043/62/C13
Ottawa, Ontario, K2H 8E9, CANADA            
mailto:macleodr@NortelNetworks.com

---------------------------

Newsgroups: comp.os.vxworks
Subject: WindView and portMapper
Date: Thu, 8 Feb 2001 17:41:40 -0700
From: "CICON,JAMES (HP-Vancouver,ex1)" 
Organization: Lawrence Berkeley National Laboratory
Message-ID: <95vgcc$imi$1@overload.lbl.gov>

When ever I launch WindView the vxWorks system task 'tPortmapd' stack
overflows.  

WindView continues to run and returns trace data that looks fine at first
glance until you notice that it is not recording any context switches.

The trace shows that the tWdbTask is always executing, that the tNetTask is
ready and that all other tasks are either pended or delayed.  This is false
because other tasks are running, for example we know that Tornado is
attached and passing network traffic to the target.  Also LEDs are flashing
and the console continues to work.

My questions are:
1) Why is WindView causing tPortmapd to overflow its stack space.
2) Is there away to increase tPortmapd stack size
3) Why is Windview not reporting any context switches?

Jim Cicon
Inkjet Products Division
Hewlett Packard Company




VxWorks Tornado

---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: Impact of using C++ Multiple Inheritance
Date: Fri, 09 Feb 2001 01:16:10 GMT
From: Bill Pringlemeir 
Organization: Factory of the mind
Message-ID: 
References: <95v1ft$1pp$1@bcrkh13.ca.nortel.com> <3A833314.35852990@mayannetworks.com>
Sender: bpringlemeir@DeadDuck

>>>>> "Kevin" == Kevin Cline  writes:

 Kevin> There is no additional overhead for a virtual function call in
 Kevin> the presence of multiple inheritance.  There is some cost in
 Kevin> memory because multiply derived classes contain multiple vtbl
 Kevin> pointers, but it's hard to imagine that would be significant.

 Kevin> Most of the arguments I've heard against multiple inheritance
 Kevin> were merely folklore, or were based on a bad experience

For our reference, what type of hardware do you code for?  A 16 Mhz
Arm thumb, perhaps an 8 Mhz 68302, or something like a 400 Mhz PowerPC
or 800 Mhz PIII?  What memory footprint; 128k or 64Mb?  And which
compiler?  Not to be facetious, I just wonder when this doesn't seem
to make a difference.

Where you just using multiple inheritance without exceptions, RTTI,
etc.  I have seen some systems grind to a halt when base objects were
declared virtual and no other features of C++ were in use.  This was
with GCC (shipped with vxWorks), 4Mb on a 18Mhz Arm chip.  The system
was not using memcpy, etc in a kludgey way.  Sometimes abstract classes
seem to force the compiler to use a vtable when it is not always
needed.  This system had 100 of objects being created.  Perhaps I
missed something, but removing abstraction solved the problem; I don't
see how multiple inheritance would have made things better.

I like many features of C++, it just seems that once you enable it
programmers _MUST_ use every possible feature.  Many of them don't
seem to understand how this works at the machine level... Of course
you do seem to understand what your compiler is doing, so I wonder
what kind of hardware it was ;-) At any rate, I do believe that the
compiler will have a major influence on these things.  They all do not
behave the same way.

best regards,
Bill

- -- 
Maybe we could paint GOLDIE HAWN a rich PRUSSIAN BLUE--

---------------------------

Newsgroups: comp.os.vxworks
Subject: Multicast Data Reception App.
Date: Fri, 09 Feb 2001 01:33:49 GMT
From: Mike McNaughton 
Organization: Sympatico
Message-ID: <3A8348CB.F726293E@sympatico.ca>

Hello vxworks using  people:

I have a reflective memory /ethernet adapter card sitting in a VME rack.
It transmits  data to a  not so useful, proprietory data acquisition
system (daq pc).

It is supposedly a multicast, udp type of transmission according to the
very sparse info that came with the system.

Data packets are as follows as viewed with  network Sniffer:


Sending port= 2048
sending ip = 172.17.66.63
receiving port=604
receiving  ip= 224.0.32.32

(224.0.32.32  is  not the domain IP address assigned to the daq pc even
tho the daq pc is happily reading data which seems to be directed to
224.0.32.32. The ip address of the daq pc is 10.32.7.226 )

I believe that the 224.0.32.32 is a multicast address.

The reflective meory and the daq pc are connected, cat 5 thru a
mini-hub.
I've tried all manner of winsock connections with VB6 but can't get a
connection.

I would like to write a piece of software that would read the data into
a third PC as it flows from the VME rack to the Black Box PC DAQ
package.

I would welcome some advice as to how to get a connection either in VC6
or VB6.

Thanks
Mike M.




---------------------------

Newsgroups: comp.os.vxworks
Subject: tgtsvr : segmentation fault
Date: Fri, 09 Feb 2001 01:57:43 GMT
From: ppatnaik@my-deja.com
Organization: Deja.com
Message-ID: <95viql$ru8$1@nnrp1.deja.com>

Hi,

I have a problem with the tgtsvr core dumping
due to a segmentation fault

The error is as follows

 Connecting to target agent... succeeded.
    Attaching C++ interface... succeeded.
    Attaching coff OMF reader for I960 CPU
family... succeeded.

[1]    Segmentation fault            tgtsvr swd184
- -A -V -s -n pranab (core dumped)

Has anyone encountered such a problem??

What is the workaround??

thanks

Pranab


Sent via Deja.com
http://www.deja.com/

---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: the format of .out file
Date: Fri, 09 Feb 2001 03:11:35 GMT
From: "GuruDev" 
Organization: Road Runner
Message-ID: 
References: <13Cg6.15702$Sl.685666@iad-read.news.verio.net>

use "-mlongcall " compiler flag during compilation.

- -Chacha
"Jason Jiang"  wrote in message
news:13Cg6.15702$Sl.685666@iad-read.news.verio.net...
> I built a downloadable project under Tonado and got the output file
aaa.out.
> But when I used 'ld' command under shell to load it into the processor
> (8260), I got the errors as below:
>
>     Relocation value does not fit in 24 bits.
>     Relocation value does not fit in 24 bits.
>     .......
>     Relocation value does not fit in 24 bits.
>     ld error: error reading file (errno = 0x3d0001).
>
> But I can use Tonado to download it. Who can tell me what's happening?
>
>
>



---------------------------

Newsgroups: comp.os.vxworks
Subject: PowerPC 7400 Altivec and VxWorks
Date: Thu, 08 Feb 2001 22:28:19 -0500
From: Rick Cellucci 
Message-ID: 

Is there anyone using a PowerPC 7400 based board with Tornado II (VxWorks
5.4.x) to generate AltiVec code?

Are you using gcc (which version?) Supported by Wind River or a hack?

How about the Green Hills Multi compiler? They claim to have optimized
AltiVec code generation for VxWorks. Anyone?

I don't want to buy an expensive board and Tornado and find out I can't use
AltiVec without hacking VxWorks to pieces ( more so than normal :-)  ).

Judging by the past threads on the topic, there seems to be an issue using a
modern gcc with VxWorks to produce AltiVec code. There also seems to be a
problem with WindRiver properly supporting that processor in the kernel (I
assume with VRSAVE instructions).

I am about to upgrade our license from Tornado I to Tornado II, and before
we shell out the bucks, I want to know that Wind River properly supports the
7400. I'll be talking to the FAE next week, but would like to be prepared.

Thanks so much for any advice you can give me.

- ---------------------------------------------------
 Rick Cellucci
 Hartwood Software Engineering, Inc.
 Email: h a r t w o o d @ c r o s s l i n k . n e t
- ---------------------------------------------------



---------------------------

Newsgroups: comp.os.vxworks
Subject: How can I check for Ethernet connection is OK.
Date: Fri, 09 Feb 2001 03:38:50 GMT
From: "GuruDev" 
Organization: Road Runner
Message-ID: 

I want to check direct connectivity between two boards, to make sure
Ethernet connection between those two is OK. As there are more than one
connection (direct and indirect) between the boards and both are running RIP
so I can not use PING. It may reach other board through some other path. It
would not fail is only DIRECT connection is broken.

I have tried etherOutput( ) to check if this call fails, but even if I
remove the cable from the port this call return OK. (Documentation says It
can fail only if system is running out of mbufs). So I can not test status
by using this. (Status NOK means cable is broken or other end is down [in
this case PING will fail]).

1. There is one option in ping  like DO_NOT_ROUTE in VxWorks. Will it help?
If I ping the other end with this option then will it try to ping remote end
using direct connection only?

2. Can I use TTL field of ping (TTL is time value in VxWorks, but I think in
some places it is hop count, and some article on internet says TTL is
implementation dependent)
Can I make TTL in VxWorks to hop count? so after getting reply I can see how
many hop it traversed before coming back? and if hop count is more than two
means it reached other end through some other route.

3. Does RIP take care of congestion? If direct path is congested can RIP use
other path? In this case solution 2 would not work.

4. Can I read routing table information to know what is the next hop for
remote IP address. If next hop is same as IP address of remote end means
direct connection is OK.  Will it work? or my understanding of RIP (and of
routing table) is not good?

Or you can suggest some other technique which I am missing (or some API like
etherOutput( ) which can do it straight forward)

Thanks
- -chacha



---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: Vx task stack protect
Date: Fri, 09 Feb 2001 03:45:41 GMT
From: Pierre Frigon 
Organization: Excite@Home - The Leader in Broadband http://home.com/faster
Message-ID: <3A8367B5.DAD16106@yahoo.com>
References: 

If I remeber correctly you might be able to set virtual page states using vmBaseLib.
For debugging purposes if you have a board that supports hardware breakpoints I would
try the bh command from the shell (setting the brakpoint near the end of the stack).
Then you use the debugger to see where the error came from and dubug from there.

You can get the stack info by calling ti(taskName) command from the shell.

There are also more sophosticated solutions using vxVMI or the new vxWorks AE which
supports stack overflow detection.

sardine wrote:

> hi all,
>   i want protect Vx task stack,to do it like as follow:
>
>    create task use taskInit instead of taskSpawn,and add a gap on every task stack,
>    and the gap is readonly at any time!
>
>  how can i do it!
>
> sardine_sz@163.net


---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: Impact of using C++ Multiple Inheritance
Date: 08 Feb 2001 19:54:09 -0800
From: Matt Wette 
Organization: Jet Propulsion Laboratory
Message-ID: <7k66ik340u.fsf@mr-ed.jpl.nasa.gov>
References: <95v1ft$1pp$1@bcrkh13.ca.nortel.com> <3A833314.35852990@mayannetworks.com>

Kevin Cline  writes:

> C Bethune wrote:
> 
> > Hi,
> >
> > I'm looking at using multiple inheritance as part of a design I'm working
> > on, and wanted to get a feel for what sort of impact it has on code
> > performance.  I've gone through some old posts here, and noticed that
> > someone said that while a virtual function call under single inheritance has
> > minimal overhead, multiple inheritance imposes a more significant
> > performance penalty.  There was no explanation as to the severity of the
> > performance hit, or why one should expect it.  Can anyone provide any info,
> > or possibly point me to a resource that would explain how G++ goes about
> > implementing single and multiple inheritance?  The code I'm writing is on
> > the critical path in our application, so I'm trying to avoid imposing
> > unecessary overhead (beyond that typically associated with an OO design).

> There is no additional overhead for a virtual function
> call in the presence of multiple inheritance.  There is
> some cost in memory because multiply derived classes
> contain multiple vtbl pointers, but it's hard to imagine
> that would be significant.
> 
> Most of the arguments I've heard against multiple inheritance were
> merely folklore, or were based on a bad experience
> introducing multiple inheritance into system full
> of clever hacks that assumed a single virtual table pointer
> at the beginning of an object.  Keep your base classes abstract,
> your code conforming, and you'll be fine.  But if you or
> your team tries to outsmart the compiler by casting pointers
> to void* and back, or by using memcpy on objects, then
> use of multiple inheritance will quickly show you the error
> of your ways.

Can you explain this?  I don't believe your claim that there is
no overhead for multiple inheritance over single inheritance.

Here is my understanding: If you have single inheritance then virtual
function calls boil down to dereferencing a function pointer in a
known offset in a vtable.  For multiple inheritance, certain
adjustments need to be made so that 
  Class Derived: public BaseA, BaseB { ... }
  Derived *b = new B();
  BaseB *a = b;
  a->foo();
works correctly.  See, "Inside the C++ Object Model," by Stanley
Lippman, near p 131, "Virtual Functions under MI".  If I understand
correctly, the statement above "a = b" will incur a operation to 
adjust the "this" pointer.

Matt

- -- 
Matthew.R.Wette at jpl.nasa.gov -- I speak for myself, not for JPL.

---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: I Need Your Help Please!!!
Date: Fri, 09 Feb 2001 04:00:34 GMT
From: Jim Don 
Organization: Sympatico
Message-ID: <3A83699D.E5497983@student.math.uwaterloo.ca>
References: <8mgg6.820$f5.158250@news>   <95ttu5$d27$1@nnrp1.deja.com>



Thanks ... we were having ethernet problems on a new board ... turn out
the hub was the problem ... i was kinda grasping for anything that would
fix the problem ... oh well ... just a little mud in my face ... ;-)

Jim


---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: Please help with C++ examples for VxWorks - xxx.zip (0/1)
Date: Fri, 09 Feb 2001 05:33:37 GMT
From: tangoNO_SPAM@pulp-fiction.com (Max)
Organization: Posted via Supernews, http://www.supernews.com
Message-ID: <3a8380b2.51016015@news.pacificnet.net>
References: <95tkt5$5rt$1@nnrp1.deja.com>

On Thu, 08 Feb 2001 08:20:54 GMT, talrostoker@my-deja.com wrote:

>Hi Guys,
>
>I need your help with C++ examples for VxWorks.
>Things like spawning task that exacute a class function (I would like
>this class to be active).
>I would be very happy with any C++ code for VxWorks.
>
>TIA,
>
>Tal Rostoker
>
>
>Sent via Deja.com
>http://www.deja.com/


Here is an attachment that might be of some use.

Enjoy.

Max.

Cambury Technologies, Inc
embedded wireless systems
http://www.cambury.com

---------------------------

Newsgroups: comp.os.vxworks
Subject: Antwort: re: dialog with WindRiver - reasonable response
Date: Fri, 9 Feb 2001 06:18:45 +0100
From: Fiederling.Ulf@dornier.dasa.de
Organization: Lawrence Berkeley National Laboratory
Message-ID: <9601us$qt2$1@overload.lbl.gov>


Hi VxWorks users

I tried different times to get support from Wind River.
The result was always very very poor.

They were not able to help me getting code for using the RTC on a X86
platform, (they even coplained they couldnt)  i had to use a unix code,
adapted by myself.
They dont inform us about the GNU Compiler uptade with Pentium support,
although we have support contract, and they know about our performance
problems with pentium processors.
They dont react to now about the problem of getting disalignment,
downloading software with double variables, (-Malign-double switch in
makefile set, natually).

Every question for support is going into "nirvana"

Very good support we get by vxwexplo


Many thanks about that

Ulf


---------------------------

Newsgroups: comp.os.vxworks
Subject: RE: Please help with C++ examples for VxWorks
Date: Fri, 9 Feb 2001 11:40:41 +0530
From: Rashmi_Topno 
Organization: Lawrence Berkeley National Laboratory
Message-ID: <9601ut$qt3$1@overload.lbl.gov>

Hi Tal,

A related question was also bothering me a lot.

As far as spawning a class member function, the following code works well
(on vxSim).

class cOne{
	public:
		void  some_func();
};

then u can spawn it by

	taskSpawn ("func", 100, 0, 3000, (FUNCPTR)one_obj.some_finc,
.......);  // where     cOne    one_obj

This will solve what you want.

Here, my question is, if you see the defintion of FUNCPTR, it goes like
this....

typedef int (*FUNCPTR) (...);

SO in my view, it takes the address to a "C" like function, i.e. the offset
of the function in object layout. Then how is it able to spawn a "C++"
function. Somewhere I have read that we should not pass a pinter to a C++
function to a place where compiler is expecting a C function.

Rashmi

Tal Rostoker wrote --->

Hi Guys,

I need your help with C++ examples for VxWorks.
Things like spawning task that exacute a class function (I would like
this class to be active).
I would be very happy with any C++ code for VxWorks.

TIA,

Tal Rostoker



---------------------------

Newsgroups: comp.os.vxworks
Subject: Test ignore it please
Date: 9 Feb 2001 02:46:27 -0600
From: ge_lin@hotmail.com (Lin Ge)
Organization: usenet.com http://www.usenet.com 80,000+ UNCENSORED Newsgroups. The #1 Usenet Service on the Planet!
Message-ID: <3a83ae63@post.usenet.com>

**** Post for FREE via your newsreader at post.usenet.com ****

Hello there 

- -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
 *** Usenet.com - The #1 Usenet Newsgroup Service on The Planet! ***
                      http://www.usenet.com
- -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=

---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: Problems with the Intel eeV driver
Date: Fri, 09 Feb 2001 10:55:46 +0000
From: James Marshall 
Organization: Agilent Technologies UK Ltd
Message-ID: <3A83CCB2.70B7A005@agilent.com>
References: 

Check the BIOS settings and make sure that "PNP OS Installed" or similar is set
to "No".

We stopped using the eeV driver because of instabilities like this and went back
to using WindRiver's standard fei END driver which can handle all three of the
82557/8/9 controllers. If you add another line to sys557PciInit() in sysNetif.c
so it will match the 82559ER device id (0x1209) as well as the 557/8/9 device id
(0x1229), then you have a driver that does all that eeV does without its
problems.

James Marshall.


"Wade T. Oram" wrote:

> Hello,
>
> I am trying to get networking operational on a Jumptec ETX-P1 board
> (Pentium 266) using the standard Tornado II Pentium BSP as the starting
> point.
>
> I have downloaded and installed the Intel eeV driver (for Intel 82559
> based network interfaces) but I am having some problems.
>
> The instructions with the Tornado II driver say to use the Pentium Pro
> BSP but since I do not have a pentium pro I have installed the driver in
> the Pentium BSP following the instructions given for the Pentium Pro
> Tornado II installation.
>
> The VxWorks build works fine.
>
> On startup get messages:
>
> Attached TCP/IP interface to eeE unit 0
> Attaching lo0...done
>
>      ....
>
> WDB: Ready
>
> Hence I know my network card has been found. However, I then
> periodically (every 2 seconds) get :
>
> 0x1f9e294 (tNetTask): Panic: where are those rfd buffers
>
> It appears that something in the initialisation is not right. I have
> re-checked that I have not missed anything when doing the installation
> and can find nothing obviously wrong.
>
> Not being familiar with VxWorks networking, I do not know what might be
> causing this problem and at present, I am not sure how to proceed.
>
> Any help offered will be much appreciated.
>
> --
> Wade Oram


---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: Multicast Data Reception App.
Date: Fri, 09 Feb 2001 11:01:24 +0000
From: James Marshall 
Organization: Agilent Technologies UK Ltd
Message-ID: <3A83CE04.DDAE819A@agilent.com>
References: <3A8348CB.F726293E@sympatico.ca>

The IP Multicast Initiative has a white paper called Writing IP Multicast
Applications:

http://www.ipmulticast.com/community/whitepapers/ipmcapps.html

This paper includes code samples for BSD sockets and Winsock 2. It's a useful
site for
getting to know multicast concepts and has other white papers at:

http://www.ipmulticast.com/techcentral/whitepapers.htm

James.


Mike McNaughton wrote:

> Hello vxworks using  people:
>
> I have a reflective memory /ethernet adapter card sitting in a VME rack.
> It transmits  data to a  not so useful, proprietory data acquisition
> system (daq pc).
>
> It is supposedly a multicast, udp type of transmission according to the
> very sparse info that came with the system.
>
> Data packets are as follows as viewed with  network Sniffer:
>
> Sending port= 2048
> sending ip = 172.17.66.63
> receiving port=604
> receiving  ip= 224.0.32.32
>
> (224.0.32.32  is  not the domain IP address assigned to the daq pc even
> tho the daq pc is happily reading data which seems to be directed to
> 224.0.32.32. The ip address of the daq pc is 10.32.7.226 )
>
> I believe that the 224.0.32.32 is a multicast address.
>
> The reflective meory and the daq pc are connected, cat 5 thru a
> mini-hub.
> I've tried all manner of winsock connections with VB6 but can't get a
> connection.
>
> I would like to write a piece of software that would read the data into
> a third PC as it flows from the VME rack to the Black Box PC DAQ
> package.
>
> I would welcome some advice as to how to get a connection either in VC6
> or VB6.
>
> Thanks
> Mike M.


---------------------------

End of New-News digest
**********************


From vxwexplo-errs@csg.lbl.gov  Fri Feb  9 07:11:21 2001
From: "Michael Lawnick" 
Date: Fri Feb  9 07:11:23 PST 2001
Subject: Alignment on net buffers

Hi VxWorks gurus,

may be a FAQ, but I couldn't find the correct entry point:

Because of my particular H/W, I need my receive and send buffers for 
ethernet not only 4 byte aligned, but 32 bytes (cache line size).
I get the buffers with
            pNewMblk = netTupleGet (pDrvCtrl->end.pNetPool, DP_BUF_SIZE,
                                    M_DONTWAIT, MT_DATA, FALSE);

            rcvBuff = pNeMblk->mBlkHdr.mData, len;

The problem is: by increasing of clDescTbl[0].clSize I can get the buffers in 
an address distance of 32 bytes, but the offset is wrong.

Does anybody know, which buffer is managed in netPool in front of the 
clusters so I can change its size and move the clusters this way ? Or are 
there other helpful recommendations ?

BTW: Sure, I could use fixed local buffers, but the copy overhead is to big.

Regards,

Michael Lawnick, SOFTEC GmbH

==============================================
SOFTEC GmbH               Tel +49-731-96600-0
Promenade 17              Fax +49-731-96600-23
D-89073 Ulm               Michael Lawnick
Germany                   lawnick@softec.de
==============================================



From vxwexplo-errs@csg.lbl.gov  Fri Feb  9 10:01:08 2001
From: Atkinson Andrew 
Date: Fri Feb  9 10:01:10 PST 2001
Subject: Memory Mapping a VME device - 

Hi All

I need to map a hardware device located in A24 address space on the VME
bus. I have only worked with PCI in the past and I am looking for some
pointers in the right direction - any takers. Perhaps some example code
could be passed on to me?

My platform is a Radstone PPC (PowerPC 750) and I am using T2/vxworks
5.3.1.

Any help greatly appreciated.

Regards

Andy Atkinson


-- 
The Information contained in this E-Mail and any subsequent correspondence
is private and is intended solely for the intended recipient(s).
For those other than the recipient any disclosure, copying, distribution, 
or any action taken or omitted to be taken in reliance on such information is
prohibited and may be unlawful.


From vxwexplo-errs@csg.lbl.gov  Fri Feb  9 12:21:47 2001
From: Christopher Slominski 
Date: Fri Feb  9 12:21:50 PST 2001
Subject: Re: Memory Mapping a VME device -

the vxWorks Users Group Exploder wrote:

> Submitted-by vxwexplo-errs@csg.lbl.gov  Fri Feb  9 10:01:08 2001
> Submitted-by: Atkinson Andrew 
>
> Hi All
>
> I need to map a hardware device located in A24 address space on the VME
> bus. I have only worked with PCI in the past and I am looking for some
> pointers in the right direction - any takers. Perhaps some example code
> could be passed on to me?
>
> My platform is a Radstone PPC (PowerPC 750) and I am using T2/vxworks
> 5.3.1.
>
> Any help greatly appreciated.
>
> Regards
>
> Andy Atkinson
>

Andy,
     Is this what you are looking for?

          Chris

------------------
#define VME_BASE 0xDF0000                         /* Base (A24) VME address */
         ...

  char testChar, *base;

  /* Map the A24 bus address into local space.
   */
  if (sysBusToLocalAdrs(VME_AM_STD_SUP_DATA, (char *) VME_BASE, &base) != OK)
  { printf("**Failure**: Mapping error\n");
    return ERROR;
  }

  /* Probe the address space by reading one character.  vxMemPrope traps any
     segmentation fault generated by a bus timeout.
   */
  if (vxMemProbe(base, VX_READ, 1, &testChar) != OK)
  { printf("**Failure**: No hardware response\n");
    return ERROR;
  }

        ...




From vxwexplo-errs@csg.lbl.gov  Fri Feb  9 15:54:43 2001
From: sachithanandam karthikselvan 
Date: Fri Feb  9 15:54:45 PST 2001
Subject: snmp agent problem

Hi all,

1) I removed all MIB2 support from the vxworks image. If I do a get operation
over sysuptime it returns correct sysuptime value. Is it a bug in vxworks?

2) Using mibcomp I compiled and loaded object files which supports system
group in MIB 2 . (I dynamically added the system group like this 
snmpdTreeAdd((char *)"1.3.6.1.2.1.1",(MIBNODE_T *)n_system)).
3) When I do get operation over any of the system group, it is geneating
following exception.

Exception number 14: Task: 0x3e65d00 (tSnmpd)

Page Fault
Program Counter: 0x0018ba9b
Status Register: 0x00010202
Error Code     : 0x00000000
13b627   _snmpdInit     +b7 : _snmpIoMain ([0, 12ea34, 0, 0, 0])
16b6dc   _snmpIoMain    +24 : 16b718 ([3e65cd0, 13b62c, 0, 12ea34, 0])
16b7c4   _snmpIoMain    +10c: _snmpdPktProcess (29, 3e654a0, 3e65cb0, 3e65ca0,
&_snmpSocke 
t)
13b6f8   _snmpdPktProcess+28 : _Process_Rcvd_SNMP_Packet_Async ([29, 3e654a0,
3e65cb0, 3e6 
5ca0, 2000])
16c049   _Process_Rcvd_SNMP_Packet_Async+a1 : 16bf08 ([3efe2ec, 3e654a0, 29,
3e654a0, 3e65 
444])
16bf99   _process_packet_two+131: _process_packet_two ([3efe2ec, 3efe2ec,
3e6543c, 16c04e, 
 3efe2ec])
16bea9   _process_packet_two+41 : _SNMP_Process_Get_PDU (3efe2ec)
18470b   _SNMP_Process_Get_PDU+11b: _find_object_node (3efe1e4, 3efe2ec, 1)

4) After that no operation is sucessful. Everything seems OK. i.e. object
hanlers, header files, mib.c option files)

5) Can anyone through any light on this?.

Regards
karthik

____________________________________________________________________
Get free email and a permanent address at http://www.netaddress.com/?N=1


From vxwexplo-errs@csg.lbl.gov  Sat Feb 10 04:03:23 2001
From: Vxworks Exploder 
Date: Sat Feb 10 04:03:26 PST 2001
Subject: comp.os.vxworks newsdigest

Comp.Os.Vxworks Daily Digest    Sat Feb 10 04:03:05 PST 2001

        Subject: Re: Impact of using C++ Multiple Inheritance
        Subject: TFFS Question
        Subject: image transfer using tftp
        Subject: Alignment on net buffers
        Subject: Tornado debugger watch windows union problem
        Subject: Re: Impact of using C++ Multiple Inheritance
        Subject: Re: *UPDATED* Gcc 2.95.3 for VxWorks-PPC available....
        Subject: Re: PowerPC 7400 Altivec and VxWorks
        Subject: Memory Mapping a VME device -
        Subject: Re: TFFS Question
        Subject: I2C controller
        Subject: emacs for vxworks
        Subject: Re: Impact of using C++ Multiple Inheritance
        Subject: Re: emacs for vxworks
        Subject: Re: Memory Mapping a VME device -
        Subject: Re: Antwort: re: dialog with WindRiver - reasonable response
        Subject: Job Change   Electro-Optical Sensor Software Engineers  - Ottawa
        Subject: Job Change   Real-time Embedded Electro-Optics Software Engineers - Ottawa
        Subject: Job Change   Electro-Optics Software Project Leader   - Ottawa
        Subject: Job Change   EMBEDDED VERIFICATION MANAGER   - Ottawa Telecom
        Subject: re: Memory Mapping a VME device -
        Subject: Re: I2C controller
        Subject: Re: *UPDATED* Gcc 2.95.3 for VxWorks-PPC available....
        Subject: Re: Impact of using C++ Multiple Inheritance
        Subject: Linker error: undefined reference to `__udivdi3
        Subject: Re: Impact of using C++ Multiple Inheritance
        Subject: NTP source code for VxWorks
        Subject: snmp agent problem
        Subject: Re: dialog with WindRiver - reasonable response
        Subject: best way to flush a TCP port without disconnecting it
        Subject: .emacs for vxworks
        Subject: What's the meaning of CACHE_PIPE_FLUSH()?
        Subject: Re: Linker error: undefined reference to `__udivdi3
        Subject: Re: PowerPC 7400 Altivec and VxWorks
        Subject: Linker sections
        Subject: Re: Job Change   Electro-Optical Sensor Software Engineers  - Ottawa

-------------------------------------------------------

Newsgroups: comp.os.vxworks
Subject: Re: Impact of using C++ Multiple Inheritance
Date: Fri, 09 Feb 2001 12:15:35 +0000
From: Paul Black 
Organization: Oxford Semiconductor Ltd
Message-ID: <3A83DF67.D97880D5@oxsemi.com>
References: <95v1ft$1pp$1@bcrkh13.ca.nortel.com> <3A833314.35852990@mayannetworks.com> <7k66ik340u.fsf@mr-ed.jpl.nasa.gov>

Matt Wette  wrote:
> Here is my understanding: If you have single inheritance then virtual
> function calls boil down to dereferencing a function pointer in a
> known offset in a vtable.  For multiple inheritance, certain
> adjustments need to be made so that
>   Class Derived: public BaseA, BaseB { ... }
>   Derived *b = new B();
>   BaseB *a = b;
>   a->foo();
> works correctly.  See, "Inside the C++ Object Model," by Stanley
> Lippman, near p 131, "Virtual Functions under MI".  If I understand
> correctly, the statement above "a = b" will incur a operation to
> adjust the "this" pointer.

Indeed. Also, "a->foo()" may incur the same cost to adjust the this
pointer back.

Paul

---------------------------

Newsgroups: comp.os.vxworks
Subject: TFFS Question
Date: Fri, 9 Feb 2001 09:52:03 -0500
From: "James Robinson" 
Message-ID: <3a84045b_1@news3.buffnet.net>

Does TFFS support the INTEL 28F160 Advance-Boot-Block chips? If so which
driver?
Thanks,
James Robinson



---------------------------

Newsgroups: comp.os.vxworks
Subject: image transfer using tftp
Date: Fri, 09 Feb 2001 15:03:16 GMT
From: Gururaj patkar 
Organization: Optimum Online
Message-ID: <3A8406EA.B262E6F4@optonline.net>

greeting vxworkers,

I am currently a using vxworks 5..4/Tornado2 on windows platform. I am
trying to download a large image file to a local ramdisk on the
evaluation board using tftp. I am currently runnng a freeware tftp
server on the PC and I have included a tftp client in the vxworks image.
The tftp transfer fails with the following message displayed in the log
file " Unknown Error from Peer" , the peer in this case is the target
board. Do I have to do anything special on the target board?
Also for scenarios where I want to transfer such a large image file from
one target board to another (one has tftp server and another tftp
client) should I do anything special which is not documented in the
vxworks manual?

bhavesh


---------------------------

Newsgroups: comp.os.vxworks
Subject: Alignment on net buffers
Date: Fri, 9 Feb 2001 15:44:57 +0100
From: "Michael Lawnick" 
Organization: Lawrence Berkeley National Laboratory
Message-ID: <200102091444.PAA10131@net.softec.de>

Hi VxWorks gurus,

may be a FAQ, but I couldn't find the correct entry point:

Because of my particular H/W, I need my receive and send buffers for 
ethernet not only 4 byte aligned, but 32 bytes (cache line size).
I get the buffers with
            pNewMblk = netTupleGet (pDrvCtrl->end.pNetPool, DP_BUF_SIZE,
                                    M_DONTWAIT, MT_DATA, FALSE);

            rcvBuff = pNeMblk->mBlkHdr.mData, len;

The problem is: by increasing of clDescTbl[0].clSize I can get the buffers in 
an address distance of 32 bytes, but the offset is wrong.

Does anybody know, which buffer is managed in netPool in front of the 
clusters so I can change its size and move the clusters this way ? Or are 
there other helpful recommendations ?

BTW: Sure, I could use fixed local buffers, but the copy overhead is to big.

Regards,

Michael Lawnick, SOFTEC GmbH

==============================================
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: Tornado debugger watch windows union problem
Date: Fri, 9 Feb 2001 16:22:54 +0100
From: "Henrik Ravn" 
Organization: Posted Courtesy of Tele Danmark or one of its European Subsidiaries
Message-ID: <961229$oib$1@news.inet.tele.dk>

Hi,

I'm using Tornado 2.0.1 and I can't always get the debugger to show unions
correct.

typedef  union {
  ...some other types
}   UnionType;
is shown correct and I can wiew each type in the union on a seperate line

but the following isn't shown correct

typedef struct {
  ...some basic types
  union {
    ...some basic types
  } UnionType
} StructWithUnionType

here the basic types are shown correct but the union is shown something
like:

/s {{D0{"\000\000....}..

where DO is the name of the first variable declared in the union. It looks
like it is the form that GDB shows data if You use the commandline debugger.

Has anyone had the same problem, and does a solution exist?

Henrik Ravn



---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: Impact of using C++ Multiple Inheritance
Date: Fri, 9 Feb 2001 14:07:23 -0000
From: "Dave Korn" 
Organization: Lumber Cartel (tinlc) Members #2234-2237 (owing to browser refresh)
Message-ID: <36Wg6.15$yi4.12156@newsr1.u-net.net>
References: <95v1ft$1pp$1@bcrkh13.ca.nortel.com> <3A833314.35852990@mayannetworks.com> <7k66ik340u.fsf@mr-ed.jpl.nasa.gov> <3A83DF67.D97880D5@oxsemi.com>

Paul Black wrote in message <3A83DF67.D97880D5@oxsemi.com>...
>Matt Wette  wrote:
>> Here is my understanding: If you have single inheritance then virtual
>> function calls boil down to dereferencing a function pointer in a
>> known offset in a vtable.  For multiple inheritance, certain
>> adjustments need to be made so that
>>   Class Derived: public BaseA, BaseB { ... }
>>   Derived *b = new B();
>>   BaseB *a = b;
>>   a->foo();
>> works correctly.  See, "Inside the C++ Object Model," by Stanley
>> Lippman, near p 131, "Virtual Functions under MI".  If I understand
>> correctly, the statement above "a = b" will incur a operation to
>> adjust the "this" pointer.
>
>Indeed. Also, "a->foo()" may incur the same cost to adjust the this
>pointer back.

  One add and one subtract per function call isn't an unbearable amount of
overhead in most circumstances...

        DaveK
- --
They laughed at Galileo.  They laughed at Copernicus.  They laughed at
Columbus. But remember, they also laughed at Bozo the Clown.



---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: *UPDATED* Gcc 2.95.3 for VxWorks-PPC available....
Date: Fri, 9 Feb 2001 14:34:42 -0000
From: "Dave Korn" 
Organization: Lumber Cartel (tinlc) Members #2234-2237 (owing to browser refresh)
Message-ID: <46Wg6.17$yi4.12215@newsr1.u-net.net>
References:  <3A8260DC.875C3C84@here.i.am>

Roger wrote in message <3A8260DC.875C3C84@here.i.am>...
>
>I'm new here. I just got a VxWorks box as part of an image processing
system.
>Where can I download this version of gcc?

  It's at http://www.newgcc4vxworks4ppc.cjb.net/

  It currently has 'issues' in working with the sourcelevel debugger; I'm
investigating, but it may take a little while.  What kind of image
processing system is it, out of curiosity?

> Also, where is the FAQ (if any) for this group.

  Check out Johan Borkhuis' excellent VxWorks FAQ at
http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html

> Finally, I would be quite happe for a few pointers to a source of
>vxWorks software. I tried a web search and did not find much. So, where
should
>I have looked?

  There isn't a lot of it around, particularly since the PSO website
disappeared, but there's an archive of bits and pieces at

ftp://ftp.atd.ucar.edu/pub/vxworks/vx

>I'm looking forward to getting a handle on VxWorks. Can anyone suggest a
good
>book? I have been told that vxWorks does not lend itself to a user's guide,
but
>I think that must be odd if true.

  Sorry, I don't know of any.

        DaveK
- --
They laughed at Galileo.  They laughed at Copernicus.  They laughed at
Columbus. But remember, they also laughed at Bozo the Clown.



---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: PowerPC 7400 Altivec and VxWorks
Date: Fri, 9 Feb 2001 14:13:31 -0000
From: "Dave Korn" 
Organization: Lumber Cartel (tinlc) Members #2234-2237 (owing to browser refresh)
Message-ID: <36Wg6.16$yi4.12223@newsr1.u-net.net>
References: 

Rick Cellucci wrote in message ...
>Is there anyone using a PowerPC 7400 based board with Tornado II (VxWorks
>5.4.x) to generate AltiVec code?

  I'm using a G4 board (not 7400) and we currently just ignore the presence
of the AltiVec unit, because...

>Are you using gcc (which version?) Supported by Wind River or a hack?

 ..we're using gcc, which doesn't support Altivec in WRS' 2.7.2 version.

>How about the Green Hills Multi compiler? They claim to have optimized
>AltiVec code generation for VxWorks. Anyone?

  Dunno about this one...

>I don't want to buy an expensive board and Tornado and find out I can't use
>AltiVec without hacking VxWorks to pieces ( more so than normal :-)  ).

  You'd have to use Diab or Green Hills, IIUIC.

>Judging by the past threads on the topic, there seems to be an issue using
a
>modern gcc with VxWorks to produce AltiVec code. There also seems to be a
>problem with WindRiver properly supporting that processor in the kernel (I
>assume with VRSAVE instructions).

  There are indeed issues with using modern gcc with VxWorks in general.
Gcc does not support Altivec, but there is a patched version out there that
does; I was planning to have a go at crosspatching it into 2.95.3 in a
little while.

>I am about to upgrade our license from Tornado I to Tornado II, and before
>we shell out the bucks, I want to know that Wind River properly supports
the
>7400. I'll be talking to the FAE next week, but would like to be prepared.

  You'll need some input from someone other than me on the quality of the
Diab or GH compilers.  Gcc isn't a consideration ATM unless you're willing
to hack on it yourself.

     DaveK
- --
They laughed at Galileo.  They laughed at Copernicus.  They laughed at
Columbus. But remember, they also laughed at Bozo the Clown.



---------------------------

Newsgroups: comp.os.vxworks
Subject: Memory Mapping a VME device -
Date: Fri, 9 Feb 2001 18:00:27 -0000
From: Atkinson Andrew 
Organization: Lawrence Berkeley National Laboratory
Message-ID: <961c4t$irv$1@overload.lbl.gov>

Hi All

I need to map a hardware device located in A24 address space on the VME
bus. I have only worked with PCI in the past and I am looking for some
pointers in the right direction - any takers. Perhaps some example code
could be passed on to me?

My platform is a Radstone PPC (PowerPC 750) and I am using T2/vxworks
5.3.1.

Any help greatly appreciated.

Regards

Andy Atkinson


- -- 
The Information contained in this E-Mail and any subsequent correspondence
is private and is intended solely for the intended recipient(s).
For those other than the recipient any disclosure, copying, distribution, 
or any action taken or omitted to be taken in reliance on such information is
prohibited and may be unlawful.

---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: TFFS Question
Date: Fri, 09 Feb 2001 18:08:47 GMT
From: parreg@my-deja.com
Organization: Deja.com
Message-ID: <961bn6$b07$1@nnrp1.deja.com>
References: <3a84045b_1@news3.buffnet.net>

In article <3a84045b_1@news3.buffnet.net>,
  "James Robinson"  wrote:
> Does TFFS support the INTEL 28F160 Advance-Boot-Block chips? If so
which
> driver?
> Thanks,
> James Robinson
>
> It does if that flash is CFI compatible.

parreg


Sent via Deja.com
http://www.deja.com/

---------------------------

Newsgroups: comp.os.vxworks
Subject: I2C controller
Date: Fri, 09 Feb 2001 18:11:09 GMT
From: parreg@my-deja.com
Organization: Deja.com
Message-ID: <961brj$bb3$1@nnrp1.deja.com>

Hi all,

What the earliest an I2C controller can be initialed using the
MPC8260.  For example can it be initialized in sysHwInit() or is there
some sort of settling down time that must be established?

Thanks.

parreg


Sent via Deja.com
http://www.deja.com/

---------------------------

Newsgroups: comp.os.vxworks
Subject: emacs for vxworks
Date: 9 Feb 2001 12:22:43 -0600
From: ge_lin@hotmail.com (Lin Ge)
Organization: usenet.com http://www.usenet.com 80,000+ UNCENSORED Newsgroups. The #1 Usenet Service on the Planet!
Message-ID: <3a843573$1@post.usenet.com>

**** Post for FREE via your newsreader at post.usenet.com ****

Hello, there 
Is there anybody using emacs to editor vxworks source code. If so, do you 
configure your emacs to be compatible with vxworks code conventions. Could 
you share your .emacs with me 
nice weekend 
Lin Ge 

- -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
 *** Usenet.com - The #1 Usenet Newsgroup Service on The Planet! ***
                      http://www.usenet.com
- -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=

---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: Impact of using C++ Multiple Inheritance
Date: Fri, 09 Feb 2001 13:10:01 -0600
From: Andrew Johnson 
Organization: APS, Argonne National Laboratory
Message-ID: <3A844089.1605C5E4@aps.anl.gov>
References: <95v1ft$1pp$1@bcrkh13.ca.nortel.com>

I'm not going to say very much on your performance question, but just add
that you should be careful that your resulting code doesn't become very
hard to understand.  A programmer trying to understand a member function
of a heavily-derived class may not have any idea which base class a
particular call is coming from - it's easy to create object spaghetti
unless you're careful.  I'd recommend careful design, and try to use
interface classes for all but the first base class (ie such classes only
contain virtual functions and no data).

Virtual base classes can add a significant performance overhead - the
Lippman book that Paul Black recommended describes why, and is worth
reading.

- - Andrew
- -- 
Every great idea appears crazy to start with.

---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: emacs for vxworks
Date: Fri, 09 Feb 2001 19:12:51 GMT
From: Bill Pringlemeir 
Organization: Factory of the mind
Message-ID: 
References: <3a843573$1@post.usenet.com>
Sender: bpringlemeir@DeadDuck

- --=-=-=

>>>>> "Lin" == Lin Ge  writes:

 Lin> Hello, there Is there anybody using emacs to editor vxworks
 Lin> source code. If so, do you configure your emacs to be compatible
 Lin> with vxworks code conventions. Could you share your .emacs with
 Lin> me nice weekend Lin Ge

What do you wish to do?  There are several ways to use `Emacs' with
vxWorks.  I think that the defaults are compatible with the way that
vxWorks `C' source files are formatted.  I have this as my
auto-mode-alist.

(autoload 'vx-asm-mode "vx-asm-mode" "VxWorks assembler mode." t)

;; set auto modes
(setq auto-mode-alist
      (append
       '(("\\.C$"        . c-mode)
         ("\\.H$"        . c-mode)
         ("\\.cc$"       . c++-mode)
         ("\\.hh$"       . c++-mode)
         ("\\.c$"        . c-mode)
         ("\\.h$"        . c-mode)
         ("\\.s\\'"      . vx-asm-mode)
         ("\\.S\\'"      . vx-asm-mode)
         ;; ("\\.java$" . jde-mode)
         ("\\.txt$"      . indented-text-mode)
         ("\\.asm$"      . asm-mode)
         ("\\.ASM$"      . asm-mode)
         ("\\.vhdl*$"    . vhdl-mode)
         ("\\.VHDL*$"    . vhdl-mode)
         ("\\.s?htm?\\'" . html-mode)
         ("\\.stm$"      . html-mode)
         ("\\.m$"        . matlab-mode)
         ("\\.gnus$"     . lisp-mode)
         ("\\.GNUS$"     . lisp-mode)
         ) auto-mode-alist))

You can also run the `gdb' in gud mode and step through source on your
target.  Do you wish to just edit with Emacs or do other things.  It is
much different than using the `GUI tools'.

Here is my `vx assembler mode', which is mostly ripped off from the asm-mode
that ships with Emacs.  However this one understand `C' include, defines and
comments.


- --=-=-=
Content-Type: application/emacs-lisp
Content-Disposition: attachment; filename=vx-asm-mode.el
Content-Description: Assembler mode for emacs.

;;; vx-asm-mode.el --- mode for editing assembler code

;; Copyright (C) 1991 Free Software Foundation, Inc.

;; Author: Eric S. Raymond 
;; Maintainer: FSF
;; Keywords: tools, languages

;; This file is part of GNU Emacs.

;; GNU Emacs is free software; you can redistribute it and/or modify
;; it under the terms of the GNU General Public License as published by
;; the Free Software Foundation; either version 2, or (at your option)
;; any later version.

;; GNU Emacs is distributed in the hope that it will be useful,
;; but WITHOUT ANY WARRANTY; without even the implied warranty of
;; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
;; GNU General Public License for more details.

;; You should have received a copy of the GNU General Public License
;; along with GNU Emacs; see the file COPYING.  If not, write to the
;; Free Software Foundation, Inc., 59 Temple Place - Suite 330,
;; Boston, MA 02111-1307, USA.

;;; Commentary:

;; This mode was written by Eric S. Raymond ,
;; inspired by an earlier vx-asm-mode by Martin Neitzel.

;; This minor mode is based on text mode.  It defines a private abbrev table
;; that can be used to save abbrevs for assembler mnemonics.  It binds just
;; four keys:
;;
;;	TAB		tab to next tab stop
;;	:		outdent preceding label, tab to tab stop
;;	C-j, C-m	newline and tab to tab stop
;;
;; Code is indented to the first tab stop level.

;; This mode runs a hook:
;;   1) an vx-asm-mode-hook at the end of initialization.

;;; Code:

(defgroup asm nil
  "Mode for editing assembler code."
  :group 'languages)

(defvar vx-asm-mode-syntax-table nil
  "Syntax table used while in Asm mode.")

(defvar vx-asm-mode-abbrev-table nil
  "Abbrev table used while in Asm mode.")
(define-abbrev-table 'vx-asm-mode-abbrev-table ())

(defvar vx-asm-mode-map nil
  "Keymap for Asm mode.")

(if vx-asm-mode-map
    nil
  (setq vx-asm-mode-map (make-sparse-keymap))
  (define-key vx-asm-mode-map "\C-c;"   'comment-region)
  (define-key vx-asm-mode-map "\C-i"	'tab-to-tab-stop)
  (define-key vx-asm-mode-map "\C-j"	'vx-asm-newline)
  (define-key vx-asm-mode-map "\C-m"	'vx-asm-newline)
  )

(defconst vx-asm-font-lock-keywords
  '(("^\\(\\(\\sw\\|\\s_\\)+\\)\\>:?[ \t]*\\(\\sw+\\)?"
     (1 font-lock-function-name-face) (3 font-lock-keyword-face nil t))
    ("^\\s +\\(\\(\\sw\\|\\s_\\)+\\)" 1 font-lock-keyword-face)
    ;; Fontify error directives.
    ("^[ \t]*#[ \t]*error[ \t]+\\(.+\\)" 1 font-lock-warning-face prepend)
    ;; Fontify filenames in #include <...> preprocessor directives as strings.
    ("^[ \t]*#[ \t]*\\(import\\|include\\)[ \t]*\\(<[^>\"\n]*>?\\)"
     2 font-lock-string-face)
    ;; Fontify function macro names.
    ("^[ \t]*#[ \t]*define[ \t]+\\(\\sw+\\)(" 1 font-lock-function-name-face)
    ;; Fontify symbol names in #elif or #if .. defined preprocessor directives.
    ("^[ \t]*#[ \t]*\\(elif\\|if\\)\\>"
     ("\\<\\(defined\\)\\>[ \t]*(?\\(\\sw+\\)?" nil nil
      (1 font-lock-builtin-face) (2 font-lock-variable-name-face nil t)))
    ;; All `.directives' 
    ("^[ \t]+\.\\([A-Za-z][A-Za-z0-9]*\\)" 1 font-lock-type-face)
    ;; Fontify otherwise as symbol names, and the preprocessor directive names.
    ("^[ \t]*#[ \t]*\\(\\sw+\\)\\>[ \t!]*\\(\\sw+\\)?"
     (1 font-lock-builtin-face) (2 font-lock-variable-name-face nil t)))
  "Additional expressions to highlight in Assembler mode.")

;;;###autoload
(defun vx-asm-mode ()
  "Major mode for editing vxWorks assembler code.
Features the following bindings:

\\[tab-to-tab-stop]\ttab to next tab stop.
\\[vx-asm-newline]\tnewline, then tab to next tab stop.

Turning on Asm mode runs the hook `vx-asm-mode-hook' at the end of initialization.

Special commands:
\\{vx-asm-mode-map}
"
  (interactive)
  (kill-all-local-variables)
  (setq mode-name "vxAsm")
  (setq major-mode 'vx-asm-mode)
  (setq local-abbrev-table vx-asm-mode-abbrev-table)
  (make-local-variable 'font-lock-defaults)
  (setq font-lock-defaults '(vx-asm-font-lock-keywords))
  (make-local-variable 'vx-asm-mode-syntax-table)
  (setq vx-asm-mode-syntax-table (make-syntax-table))
  (set-syntax-table vx-asm-mode-syntax-table)

  (make-local-variable 'comment-start)
  (setq comment-start "/* ")
  (make-local-variable 'comment-end)
  (setq comment-end " */")
  (make-local-variable 'comment-column)
  (setq comment-column 32)
  (make-local-variable 'comment-start-skip)
  (setq comment-start-skip "/\\*+ *")
  (make-local-variable 'comment-multi-line)
  (setq comment-multi-line t)
  (make-local-variable 'parse-sexp-ignore-comments)
  (setq parse-sexp-ignore-comments t)

  ;; Make our own local child of vx-asm-mode-map
  ;; so we can define our own comment character.
  (use-local-map (nconc (make-sparse-keymap) vx-asm-mode-map))

  (modify-syntax-entry ?/ ". 14" vx-asm-mode-syntax-table)
  (modify-syntax-entry ?* ". 23" vx-asm-mode-syntax-table)

  (setq fill-prefix "    ")
  (run-hooks 'vx-asm-mode-hook))


(defun vx-asm-newline ()
  "Insert LFD + fill-prefix, to bring us back to code-indent level."
  (interactive)
  (if (eolp) (delete-horizontal-space))
  (insert "\n")
  (tab-to-tab-stop)
  )

(provide 'vx-asm-mode)

;;; vx-asm-mode.el ends here








- --=-=-=


hth,
Bill

- -- 
How  can you  say that  all black  dudes should  get ten-inchers  ?  I
firmly  believe that  just  because you  read  it in  the Pravda  that
doesn't mean it's true.

- --=-=-=--

---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: Memory Mapping a VME device -
Date: Fri, 09 Feb 2001 13:22:39 -0600
From: Andrew Johnson 
Organization: APS, Argonne National Laboratory
Message-ID: <3A84437F.40C18220@aps.anl.gov>
References: <961c4t$irv$1@overload.lbl.gov>

Atkinson Andrew wrote:
> 
> I need to map a hardware device located in A24 address space on the VME
> bus. I have only worked with PCI in the past and I am looking for some
> pointers in the right direction - any takers. Perhaps some example code
> could be passed on to me?
> 
> My platform is a Radstone PPC (PowerPC 750) and I am using T2/vxworks
> 5.3.1.

I don't know the Radstone board or BSP, but it's 95% likely that you won't
need to do anything more than pass the A24 base address of your hardware
to sysBusToLocalAdrs(VME_AM_STD_SUP_DATA, ...) to get a pointer to it in
memory.  VMEbus is much simpler to handle than PCI as it doesn't support
dynamic addressing, and most BSPs map the whole of the A24 space into
memory by default.  A32 can be a bit harder though.

- - Andrew
- -- 
Every great idea appears crazy to start with.

---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: Antwort: re: dialog with WindRiver - reasonable response
Date: Fri, 9 Feb 2001 22:01:01 +0100
From: "Martin Raabe" 
Organization: T-Online
Message-ID: <961lqo$vim$02$1@news.t-online.com>
References: <9601us$qt2$1@overload.lbl.gov>

Hallo Ulf,
allow me one question:
did you subscribe the WindSurf proactive email alert?

Speaking for my own, not for my employer!

Ciao

- --
Martin Raabe
 schrieb in im Newsbeitrag:
9601us$qt2$1@overload.lbl.gov...
>
> Hi VxWorks users
>
> I tried different times to get support from Wind River.
> The result was always very very poor.
>
> They were not able to help me getting code for using the RTC on a X86
> platform, (they even coplained they couldnt)  i had to use a unix code,
> adapted by myself.
> They dont inform us about the GNU Compiler uptade with Pentium support,
> although we have support contract, and they know about our performance
> problems with pentium processors.
> They dont react to now about the problem of getting disalignment,
> downloading software with double variables, (-Malign-double switch in
> makefile set, natually).
>
> Every question for support is going into "nirvana"
>
> Very good support we get by vxwexplo
>
>
> Many thanks about that
>
> Ulf
>



---------------------------

Newsgroups: comp.os.vxworks,comp.realtime
Subject: Job Change   Electro-Optical Sensor Software Engineers  - Ottawa
Date: Fri, 9 Feb 2001 16:15:20 -0500
From: "PROCOM" 
Organization: IGS - Information Gateway Services
Message-ID: <961ma6$blr$1@news.igs.net>

Our client is currently seeking Electro-Optical Sensor Systems Software
Engineers to join there development program. Those selected will work with
other professionals performing software development, programming and systems
analysis. Applicants should have a formal education in computer science or
electrical engineering, with at least 3 years of experience in real-time
software development. In addition to possessing initiative and drive,
successful candidates must have significant experience developing software
in C and/or C++ on a real-time operating system, and the ability to read and
understand a complex SRS.

Knowledge of the infrared application domain (environments, models and
targets)and experience working on a military software development program
(MIL-STD-2167A or MIL-STD-498) are highly desired. Also desired are
experience in object-oriented design, C++ and VxWorks, plus team and
leadership experience on software development teams. Dutch or Canadian
citizenship is required. Candidates with an active Canadian security
clearance also will be considered. Applicants must indicate that they meet
this requirement on the subject line of cover letters in order to be
considered for this position.

Compensation:    Open ended Contract and Full time

PROCOM: Established in 1978, Professional Computer Consultants Group Ltd.
(Procom) is a national leader in the provision of Computer personnel on a
contract
and full-time basis. Our clients are comprised of the largest national and
international corporations that utilize technical resources extensively
across a wide range of disciplines. In the Financial Post (March 1999)
Procom was ranked as the 6th largest professional Services Company in
Canada. In November of 1999, Procom was named a Regional finalist in Canada'
s 50 Best Managed Private companies. Our track record is proven with more
than 180 consultants servicing Ottawa's high tech community and more than
1600 Procom consultants currently on assignment throughout. North America.
For further information on this and other opportunities please visit our web
site at www.procom.ca.


Derek Weber
PROCOM

300 March Rd Suite 600
Kanata, Ontario
K2K-2E2
613-270-9339 x231
613-270-9449 (FAX)

derekw@procom.ca
www.procom.ca





---------------------------

Newsgroups: comp.os.vxworks,comp.realtime
Subject: Job Change   Real-time Embedded Electro-Optics Software Engineers - Ottawa
Date: Fri, 9 Feb 2001 16:15:43 -0500
From: "PROCOM" 
Organization: IGS - Information Gateway Services
Message-ID: <961mas$blt$1@news.igs.net>


Our client is searching for entry, mid-level and senior Software Engineers
to develop applications to operate on real-time embedded CPU hardware
platforms for military display and radar systems. The successful applicants
will be responsible for all phases of software development.

Knowledge of C, C++ and UNIX is required. A B.S.C.S. or B.S.E.E. degree also
is required, and applicants must be eligible for a government Secret-level
clearance.

Our client utilizes advanced commercial technology for cost-effective
solutions in the design and manufacture of a variety of air-, land- and
sea-based advanced defense electronics systems, including combat display
systems, portable workstations, trainer and emulation systems, coastal
surveillance systems and radar systems.


Compensation:    Open ended Contract and Full time

PROCOM: Established in 1978, Professional Computer Consultants Group Ltd.
(Procom) is a national leader in the provision of Computer personnel on a
contract
and full-time basis. Our clients are comprised of the largest national and
international corporations that utilize technical resources extensively
across a wide range of disciplines. In the Financial Post (March 1999)
Procom was ranked as the 6th largest professional Services Company in
Canada. In November of 1999, Procom was named a Regional finalist in Canada'
s 50 Best Managed Private companies. Our track record is proven with more
than 180 consultants servicing Ottawa's high tech community and more than
1600 Procom consultants currently on assignment throughout. North America.
For further information on this and other opportunities please visit our web
site at www.procom.ca.



Derek Weber
PROCOM

300 March Rd Suite 600
Kanata, Ontario
K2K-2E2
613-270-9339 x231
613-270-9449 (FAX)

derekw@procom.ca
www.procom.ca

















---------------------------

Newsgroups: comp.os.vxworks,comp.realtime
Subject: Job Change   Electro-Optics Software Project Leader   - Ottawa
Date: Fri, 9 Feb 2001 16:15:32 -0500
From: "PROCOM" 
Organization: IGS - Information Gateway Services
Message-ID: <961mai$bls$1@news.igs.net>

Our client is currently seeking people who wish to join a team of other
professionals providing software development, programming and systems
analysis for our Infrared Search and Track (IRST) development program.

Applicants must have a formal education in computer science or electrical
engineering with at least 8 years of experience developing software in a
structured development environment. In addition to possessing initiative and
drive, the successful candidate is required to have had experience
integrating and testing complex software, as well as the ability for working
independently and collaborating in teams.

Highly desired are programming skills and significant experience in a broad
range of languages (C, C++, Fortran, IDL) and operating systems (Windows NT,
Unix, VxWorks). Also preferred is experience working on structured military
software development programs (Mil-Std-2167A, Mil-Std-498), knowledge of the
infrared application domain (environments, models, targets), real-time
development experience, team and leadership experience on software
development teams, exposure to object-oriented design methodologies,
preferably Booch and UML, and knowledge of or exposure to DSP. Dutch or
Canadian citizenship is a requirement, due to the requirements of the
program. Candidates with an active Canadian security clearance also will be
considered. Please indicate that you meet this requirement on the subject
line of your cover letter in order to be considered for this opportunity.


Compensation:    Open ended Contract and Full time

PROCOM: Established in 1978, Professional Computer Consultants Group Ltd.
(Procom) is a national leader in the provision of Computer personnel on a
contract
and full-time basis. Our clients are comprised of the largest national and
international corporations that utilize technical resources extensively
across a wide range of disciplines. In the Financial Post (March 1999)
Procom was ranked as the 6th largest professional Services Company in
Canada. In November of 1999, Procom was named a Regional finalist in Canada'
s 50 Best Managed Private companies. Our track record is proven with more
than 180 consultants servicing Ottawa's high tech community and more than
1600 Procom consultants currently on assignment throughout. North America.
For further information on this and other opportunities please visit our web
site at www.procom.ca.



Derek Weber
PROCOM

300 March Rd Suite 600
Kanata, Ontario
K2K-2E2
613-270-9339 x231
613-270-9449 (FAX)

derekw@procom.ca
www.procom.ca

















---------------------------

Newsgroups: comp.os.vxworks,comp.realtime
Subject: Job Change   EMBEDDED VERIFICATION MANAGER   - Ottawa Telecom
Date: Fri, 9 Feb 2001 16:15:59 -0500
From: "PROCOM" 
Organization: IGS - Information Gateway Services
Message-ID: <961mbc$bm2$1@news.igs.net>

Job Description

The successful candidate will ensure the functionality of Java Wireless
hardware and software. The selected candidate will be responsible for
multiple products. He/She will be a "driver" and team leader, the Manager
will quickly form relationships throughout the company to focus efforts,
coordinate activities, bridge communications, proactively resolve issues,
and effectively meet milestones, goals and objectives from concept through
launch.

In addition, the Manager will be a strategic thinker, able to funnel market,
technology, corporate and resource demands into a strategic and tactical
plan, achieve corporate buy-in, and execute.

Required Skills

Mandatory

Specifications and requirements analysis.
Embedded design, testbench and test development, and regression management.
C/C++, assembly, processor architecture and datacom experience.


Education

This position requires a B.Sc. In Electrical Engineering, Computer Science,
or equivalent.

General

Excellent verbal and written communication skills are essential.
Effective time management skills.


PROCOM: Established in 1978, Professional Computer Consultants Group Ltd.
(Procom) is a national leader in the provision of Computer personnel on a
contract
and full-time basis. Our clients are comprised of the largest national and
international corporations that utilize technical resources extensively
across a wide range of disciplines. In the Financial Post (March 1999)
Procom was ranked as the 6th largest professional Services Company in
Canada. In November of 1999, Procom was named a Regional finalist in Canada'
s 50 Best Managed Private companies. Our track record is proven with more
than 180 consultants servicing Ottawa's high tech community and more than
1600 Procom consultants currently on assignment throughout. North America.
For further information on this and other opportunities please visit our web
site at www.procom.ca.


- --------------------------------------------------------------
Derek Weber
PROCOM

300 March Rd Suite 600
Kanata, Ontario
K2K-2E2
613-270-9339 x231
613-270-9449 (FAX)

derekw@procom.ca
www.procom.ca















---------------------------

Newsgroups: comp.os.vxworks
Subject: re: Memory Mapping a VME device -
Date: Fri, 09 Feb 2001 15:21:14 -0500
From: Christopher Slominski 
Organization: Lawrence Berkeley National Laboratory
Message-ID: <961mmd$mjt$1@overload.lbl.gov>

the vxWorks Users Group Exploder wrote:

> Submitted-by vxwexplo-errs@csg.lbl.gov  Fri Feb  9 10:01:08 2001
> Submitted-by: Atkinson Andrew 
>
> Hi All
>
> I need to map a hardware device located in A24 address space on the VME
> bus. I have only worked with PCI in the past and I am looking for some
> pointers in the right direction - any takers. Perhaps some example code
> could be passed on to me?
>
> My platform is a Radstone PPC (PowerPC 750) and I am using T2/vxworks
> 5.3.1.
>
> Any help greatly appreciated.
>
> Regards
>
> Andy Atkinson
>

Andy,
     Is this what you are looking for?

          Chris

- ------------------
#define VME_BASE 0xDF0000                         /* Base (A24) VME address */
         ...

  char testChar, *base;

  /* Map the A24 bus address into local space.
   */
  if (sysBusToLocalAdrs(VME_AM_STD_SUP_DATA, (char *) VME_BASE, &base) != OK)
  { printf("**Failure**: Mapping error\n");
    return ERROR;
  }

  /* Probe the address space by reading one character.  vxMemPrope traps any
     segmentation fault generated by a bus timeout.
   */
  if (vxMemProbe(base, VX_READ, 1, &testChar) != OK)
  { printf("**Failure**: No hardware response\n");
    return ERROR;
  }

        ...



---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: I2C controller
Date: Fri, 9 Feb 2001 13:25:14 -0800
From: 
Organization: SBC Internet Services
Message-ID: 
References: <961brj$bb3$1@nnrp1.deja.com>

Dear Parreq:
    Your master can transmit anytime after SDA (the data line) and SCL (the
clock line) have settled high after reset. The slaves you are communicating
with may have their individual response times before they can respond after
reset, but it should be much quicker then the system can get to sysHwInit().
Of course, it is recommended by Windriver that you initialize a minimum in
sysHwInit and hold as much as possible until sysHwInit2. I believe the
reason for this is so the more complicated chip subsystems you talk with can
settle (PCI bridge, timers, memory controllers, etc.). IIC is a very simple
protocol and requires no complicated gyrations like PCI (for instance) does.

 wrote in message news:961brj$bb3$1@nnrp1.deja.com...
> Hi all,
>
> What the earliest an I2C controller can be initialed using the
> MPC8260.  For example can it be initialized in sysHwInit() or is there
> some sort of settling down time that must be established?
>
> Thanks.
>
> parreg
>
>
> Sent via Deja.com
> http://www.deja.com/
>



---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: *UPDATED* Gcc 2.95.3 for VxWorks-PPC available....
Date: Fri, 09 Feb 2001 15:36:19 -0600
From: Kevin Cline 
Organization: Mayan Networks
Message-ID: <3A8462D3.1AADFB7C@mayannetworks.com>
References: 

Dave Korn wrote:

>  Finally I've recompiled the whole thing for i386, so it should run on any
> CPU!

Can you document the process so that others can repeat your work for different
targets?  Thanks.

Kevin Cline



---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: Impact of using C++ Multiple Inheritance
Date: Fri, 09 Feb 2001 15:43:21 -0600
From: Kevin Cline 
Organization: Mayan Networks
Message-ID: <3A846479.CB0857D@mayannetworks.com>
References: <95v1ft$1pp$1@bcrkh13.ca.nortel.com> <3A833314.35852990@mayannetworks.com> <7k66ik340u.fsf@mr-ed.jpl.nasa.gov>

Matt Wette wrote:

> Here is my understanding: If you have single inheritance then virtual
> function calls boil down to dereferencing a function pointer in a
> known offset in a vtable.  For multiple inheritance, certain
> adjustments need to be made so that
>   Class Derived: public BaseA, BaseB { ... }
>   Derived *b = new B();
>   BaseB *a = b;
>   a->foo();
> works correctly.  See, "Inside the C++ Object Model," by Stanley
> Lippman, near p 131, "Virtual Functions under MI".  If I understand
> correctly, the statement above "a = b" will incur a operation to
> adjust the "this" pointer.

Yes, it will.  The pointer must be adjusted
by a compile-time constant offset whenever a
derived class pointer is converted to a base class
pointer.  But the virtual function call a->foo()
will be no slower.

If some portion of your system is sensitive to the extra
instruction needed for that offset, then perhaps
that portion should not be coded in C++.

Kevin Cline


---------------------------

Newsgroups: comp.os.vxworks
Subject: Linker error: undefined reference to `__udivdi3
Date: Fri, 09 Feb 2001 22:11:31 GMT
From: "Eugene Zlobin" 
Organization: EarthLink Inc. -- http://www.EarthLink.net
Message-ID: 

Hello,

I got two linker errors :
undefined reference to `__udivdi3
undefined reference to `__divdi3

How i can guess it is related with float point support but I don't know
how to solve this problem.

I use Tornado 2 (VxWorks 5.4) and Windows 2000 simulator.
Did anyone face same problem and what could be the solution for it?

Thanks,
Eugene.





---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: Impact of using C++ Multiple Inheritance
Date: Fri, 09 Feb 2001 16:08:38 -0600
From: Kevin Cline 
Organization: Mayan Networks
Message-ID: <3A846A65.458F766D@mayannetworks.com>
References: <95v1ft$1pp$1@bcrkh13.ca.nortel.com> <3A833314.35852990@mayannetworks.com> 

Bill Pringlemeir wrote:

> >>>>> "Kevin" == Kevin Cline  writes:
>
>  Kevin> There is no additional overhead for a virtual function call in
>  Kevin> the presence of multiple inheritance.  There is some cost in
>  Kevin> memory because multiply derived classes contain multiple vtbl
>  Kevin> pointers, but it's hard to imagine that would be significant.
>
>  Kevin> Most of the arguments I've heard against multiple inheritance
>  Kevin> were merely folklore, or were based on a bad experience
>
> For our reference, what type of hardware do you code for?  A 16 Mhz
> Arm thumb, perhaps an 8 Mhz 68302, or something like a 400 Mhz PowerPC
> or 800 Mhz PIII?  What memory footprint; 128k or 64Mb?  And which
> compiler?  Not to be facetious, I just wonder when this doesn't seem
> to make a difference.

I admit that I have a lot of memory to play with,
and that I had forgotten about the pointer offset
that may be required for a virtual function call
with multiple inheritance.  Bit the addition
of a single instruction to a function call is
not something I would worry about for very long.

I have seldom found that trying to bum
one word from an already sizable object is worthwhile,
and the cost to produce a work-around to the canonical
solution usually exceeds any savings realized.
Similarly, programming in C++ while worrying about every
instruction emitted is counterproductive.
The problems that I have seen as a result of such premature
optimization have been enormously expensive, and the
savings almost always insignificant.  Indeed, these optimizations
often result in an entangling of interface and implementation,
with the consequence that it becomes impossible to
optimize the implementation when the true performance
problems are discovered.

Of course this does not mean that one should freely
create arbitrarily complex inheritance hierarchies,
just that one should not strive to avoid multiple
inheritance when it is the appropriate solution to the problem.

Generally I will inherit only from abstract base classes,
preferably containing no implementation or data.  I've
never used virtual inheritance and would look carefully
at any design that did.  I just wanted to help dispel
the often-held notion that multiple inheritance should
be scrupulously avoided.  The implementations that result
from that sort of thinking are often horrific.

I gained some real insight on this subject from an HP engineer
who said "compared to memory access, instructions are free".

Kevin Cline


---------------------------

Newsgroups: comp.os.vxworks
Subject: NTP source code for VxWorks
Date: Fri, 09 Feb 2001 23:09:24 GMT
From: mat_lfc@my-deja.com
Organization: Deja.com
Message-ID: <961tav$rr4$1@nnrp1.deja.com>

Hi,

I'm looking for the source code for NTP(network time protocol)
for VxWorks and PowerPC.
Even linux implementations would do ok.

Any help would be appreciated.

Thanks,
Mathew


Sent via Deja.com
http://www.deja.com/

---------------------------

Newsgroups: comp.os.vxworks
Subject: snmp agent problem
Date: 9 Feb 2001 16:54:36 MST
From: sachithanandam karthikselvan 
Organization: Lawrence Berkeley National Laboratory
Message-ID: <96217s$r17$1@overload.lbl.gov>

Hi all,

1) I removed all MIB2 support from the vxworks image. If I do a get operation
over sysuptime it returns correct sysuptime value. Is it a bug in vxworks?

2) Using mibcomp I compiled and loaded object files which supports system
group in MIB 2 . (I dynamically added the system group like this 
snmpdTreeAdd((char *)"1.3.6.1.2.1.1",(MIBNODE_T *)n_system)).
3) When I do get operation over any of the system group, it is geneating
following exception.

Exception number 14: Task: 0x3e65d00 (tSnmpd)

Page Fault
Program Counter: 0x0018ba9b
Status Register: 0x00010202
Error Code     : 0x00000000
13b627   _snmpdInit     +b7 : _snmpIoMain ([0, 12ea34, 0, 0, 0])
16b6dc   _snmpIoMain    +24 : 16b718 ([3e65cd0, 13b62c, 0, 12ea34, 0])
16b7c4   _snmpIoMain    +10c: _snmpdPktProcess (29, 3e654a0, 3e65cb0, 3e65ca0,
&_snmpSocke 
t)
13b6f8   _snmpdPktProcess+28 : _Process_Rcvd_SNMP_Packet_Async ([29, 3e654a0,
3e65cb0, 3e6 
5ca0, 2000])
16c049   _Process_Rcvd_SNMP_Packet_Async+a1 : 16bf08 ([3efe2ec, 3e654a0, 29,
3e654a0, 3e65 
444])
16bf99   _process_packet_two+131: _process_packet_two ([3efe2ec, 3efe2ec,
3e6543c, 16c04e, 
 3efe2ec])
16bea9   _process_packet_two+41 : _SNMP_Process_Get_PDU (3efe2ec)
18470b   _SNMP_Process_Get_PDU+11b: _find_object_node (3efe1e4, 3efe2ec, 1)

4) After that no operation is sucessful. Everything seems OK. i.e. object
hanlers, header files, mib.c option files)

5) Can anyone through any light on this?.

Regards
karthik

____________________________________________________________________
Get free email and a permanent address at http://www.netaddress.com/?N=1

---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: dialog with WindRiver - reasonable response
Date: 09 Feb 2001 13:28:45 -1000
From: Jim Thomas 
Organization: Canada France Hawai`i Telescope
Message-ID: 
References: <95ur9c$93t$1@overload.lbl.gov>

>>>>> "Jim" == Jim Way  writes:

 Jim> I just want to amplify what you've said about support. The phone-in
 Jim> support does seem to be less than what I would expect, considering
 Jim> the cost of the software, licenses, etc.

I never got that far.  They never would answer my question about their
requirement that I have a modem hooked to my system for them to use.

 Jim> ... HOWEVER, my experiences with my local sales team including the
 Jim> FAE (Field Application Engineer) have been positive and helpful.

Presumably that must depend on volume?  For those of us who have a couple
of seats and ~10 BSP's it's hard to get a sales droid to answer calls.  I
understand economics, but there has to be some response.  They still send
me flyers for new products, but I guess they haven't noticed that our
support has lapsed.

 Jim> ... FWIW, I've been on the receiving side of customer support calls.

So have I, though it was years ago and the user volume was not huge
(PDP-10).  But I answered the phone directly.

 Jim> ... But WRS doesn't make the process any easier by requiring you to
 Jim> first validate your license, then get a call-back telling you you're
 Jim> validated, then wait hours/days for a callback from someone to try to
 Jim> help with your problem. I suspect that they have a two-tiered system
 Jim> where one group is exclusively phone handlers and the other group is
 Jim> exclusively development engineering.

That sounds like three levels.  Phone interference, response, and
development?

 Jim> ... Tornado vxWorks

---------------------------

Newsgroups: comp.os.vxworks
Subject: best way to flush a TCP port without disconnecting it
Date: Fri, 9 Feb 2001 15:26:30 -0800
From: "GTO" 
Organization: SSO-IT, Hewlett-Packard Co.
Message-ID: <961ub8$ip8$1@web1.cup.hp.com>

Assume you had a socket connection open to your target between a TCP server
and client. Further, assume you had an ojbectID that is always coming in
first (before the rest of the object's data).

Once the client verifies the objectID, it continues to receive the rest of
the object. What is the best way to react on the fact that the objectID does
not match? Maybe the server is sending the wrong object down the pipe or
something went wrong with TCP/IP and all the data is byte shifted.

if (objectID != DESIRED_OBJECT)
    portflush(s);

with

void portflush(SOCKET s)
{
    int rc = 0;
    char buf[1024];
    for (;;)
    {
        rc = recv(s, buf, sizeof(buf),0);
        if (rc <= 0)
            break;
    }
}

does not work, since recv() is a blocking call and all the program from then
on is doing is sitting inside this endless loop.

I also would like to avoid TIME_WAIT assassination with SO_LINGER. I would
like to keep the connection open.

- -GT



---------------------------

Newsgroups: comp.os.vxworks
Subject: .emacs for vxworks
Date: 9 Feb 2001 17:59:54 -0600
From: ge_lin@hotmail.com (Lin Ge)
Organization: usenet.com http://www.usenet.com 80,000+ UNCENSORED Newsgroups. The #1 Usenet Service on the Planet!
Message-ID: <3a84847a@post.usenet.com>

**** Post for FREE via your newsreader at post.usenet.com ****

Hello, 
Thanks a lot 
Your .emacs is really cool. 
However I just need the .emacs for c code. You know the vxworks/windriver 
has a very special code conventions for their c code. So is there any .emacs 
for that format/style. 
Thanks again 
Lin Ge 


- -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
 *** Usenet.com - The #1 Usenet Newsgroup Service on The Planet! ***
                      http://www.usenet.com
- -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=

---------------------------

Newsgroups: comp.os.vxworks
Subject: What's the meaning of CACHE_PIPE_FLUSH()?
Date: Sat, 10 Feb 2001 01:26:37 GMT
From: zhouxiaochen@my-deja.com
Organization: Deja.com
Message-ID: <9625c7$2jj$1@nnrp1.deja.com>

Hi,

In the source code of SCC UART serial driver of PPC8260
(m8260Sio.c),there use CACHE_PIPLE_FLUSH() in some places. But in the
driver of PPC860, we can not find one.

I have seen the online manuals, it saids

STATUS cachePipeFlush (void)
DESCRIPTION
This routine forces the processor output buffers to write their
contents to RAM. A cache flush may have forced its data into the write
buffers, then the buffers need to be flushed to RAM to maintain
coherency.

I do not understand these words exactly and why it is different between
the drivers of PPC860 with the PPC8260.

I use T2, target are PPC8260 and PPC860. BSP are ads860 and ads8260.

Thanks for any help in advance!

Zhou xiaochen



Sent via Deja.com
http://www.deja.com/

---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: Linker error: undefined reference to `__udivdi3
Date: Sat, 10 Feb 2001 01:58:52 GMT
From: Randy MacLeod 
Organization: MyHome
Message-ID: <3A84A095.6D1D8837@home.com>
References: 


  You're missing software support for 64 bit integer 
operations like division. Support comes with vxWorks but in a .a file. 

Here's my recipe to make a .o file which you can load.

On unix:
mkdir gnumath
cd gnumath
ar x /path/to/Tornado/host/sun4-solaris2/\
 lib/gcc-lib/powerpc-wrs-vxworks/cygnus-2.7.2-960126/libgcc.a
rm _pure.o
ldppc -r -X -N -o /tmp/gnu64bitMath.o *.o

then I booted my standard ppc vxWorks kernel.
- ->
- -> ld ld 
> Hello,
> 
> I got two linker errors :
> undefined reference to `__udivdi3
> undefined reference to `__divdi3
> 
> How i can guess it is related with float point support but I don't know
> how to solve this problem.
> 
> I use Tornado 2 (VxWorks 5.4) and Windows 2000 simulator.
> Did anyone face same problem and what could be the solution for it?
> 
> Thanks,
> Eugene.

---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: PowerPC 7400 Altivec and VxWorks
Date: Fri, 09 Feb 2001 21:02:53 -0500
From: Rick Cellucci 
Message-ID: 
References:  <36Wg6.16$yi4.12223@newsr1.u-net.net>

in article 36Wg6.16$yi4.12223@newsr1.u-net.net, Dave Korn at
no.spam@my.mailbox.invalid wrote on 2/9/01 9:13 AM:

> There are indeed issues with using modern gcc with VxWorks in general.
> Gcc does not support Altivec, but there is a patched version out there that
> does; I was planning to have a go at crosspatching it into 2.95.3 in a
> little while.

Thanks for your work on that Dave.  I'll be playing with 2.95.3 shortly,
AltiVec or no.

> You'll need some input from someone other than me on the quality of the
> Diab or GH compilers.  Gcc isn't a consideration ATM unless you're willing
> to hack on it yourself.

I'm thinking about it...

- ---

FWIW, I believe there are still VxWorks-related issues that a new gcc won't
solve.  Specifically, without changing the kernel source code, multiple
tasks will not be able to use AltiVec. We would need a symbol like
VX_FP_TASK for AltiVec to tell the task scheduler to save and restore the
AltiVec registers via VRSAVE.  I guess I could only use AltiVec in only one
task, but yuck.  And forget ISRs.  I'll ask Wind River next week.

Thanks for your help Dave.

- ---------------------------------------------------
 Rick Cellucci
 Hartwood Software Engineering, Inc.
 Email: h a r t w o o d @ c r o s s l i n k . n e t
- ---------------------------------------------------



---------------------------

Newsgroups: comp.os.vxworks
Subject: Linker sections
Date: Fri, 9 Feb 2001 22:35:26 -0500
From: "Normand Barriere" 
Message-ID: 

Hi All,

Recently I discovered that with Tornado 2, if too many files are compiled
with debugging ON, I can't download the image to the target board. It seems
that the size of the "stabs" overwrites another sections and causes the
download to abort. The test I've performed to confirm this, is that the
VxWorks image is built around 3 software blocks. If I set the debug info
switch ON ( -g ) on all the files, I cannot download. If I set this switch
for a single software block, then I can download correctly.

Environment : ARM 7

Does anybody knwo how we should configure the environment in order to
configure the size of this section ?

I need help on this.

Thanks

Normand



---------------------------

Newsgroups: comp.os.vxworks,comp.realtime
Subject: Re: Job Change   Electro-Optical Sensor Software Engineers  - Ottawa
Date: Sat, 10 Feb 2001 08:49:21 GMT
From: "Alf Katz" 
Organization: @Home Network
Message-ID: 
References: <961ma6$blr$1@news.igs.net>
Reply-To: "Alf Katz" 

Derek, posting these ads twice in 3 days on the comp.realtime newsgroup
could be seen by some as a little uncool.  The last iteration hadn't yet had
time to scroll off my outlook (yuk) screen.
- --
Regards,
Alf Katz,
alfkatz@removethis.ieee.org

"PROCOM"  wrote in message
news:961ma6$blr$1@news.igs.net...
> Our client is currently seeking Electro-Optical Sensor Systems Software
> Engineers to join there development program. Those selected will work with
> other professionals performing software development, programming and
systems
> analysis. Applicants should have a formal education in computer science or
> electrical engineering, with at least 3 years of experience in real-time
> software development. In addition to possessing initiative and drive,
> successful candidates must have significant experience developing software
> in C and/or C++ on a real-time operating system, and the ability to read
and
> understand a complex SRS.
>
> Knowledge of the infrared application domain (environments, models and
> targets)and experience working on a military software development program
> (MIL-STD-2167A or MIL-STD-498) are highly desired. Also desired are
> experience in object-oriented design, C++ and VxWorks, plus team and
> leadership experience on software development teams. Dutch or Canadian
> citizenship is required. Candidates with an active Canadian security
> clearance also will be considered. Applicants must indicate that they meet
> this requirement on the subject line of cover letters in order to be
> considered for this position.
>
> Compensation:    Open ended Contract and Full time
>
> PROCOM: Established in 1978, Professional Computer Consultants Group Ltd.
> (Procom) is a national leader in the provision of Computer personnel on a
> contract
> and full-time basis. Our clients are comprised of the largest national and
> international corporations that utilize technical resources extensively
> across a wide range of disciplines. In the Financial Post (March 1999)
> Procom was ranked as the 6th largest professional Services Company in
> Canada. In November of 1999, Procom was named a Regional finalist in
Canada'
> s 50 Best Managed Private companies. Our track record is proven with more
> than 180 consultants servicing Ottawa's high tech community and more than
> 1600 Procom consultants currently on assignment throughout. North America.
> For further information on this and other opportunities please visit our
web
> site at www.procom.ca.
>
>
> Derek Weber
> PROCOM
>
> 300 March Rd Suite 600
> Kanata, Ontario
> K2K-2E2
> 613-270-9339 x231
> 613-270-9449 (FAX)
>
> derekw@procom.ca
> www.procom.ca
>
>
>
>



---------------------------

End of New-News digest
**********************


From vxwexplo-errs@csg.lbl.gov  Sat Feb 10 21:38:48 2001
From: "Thiagarajan Hariharan" 
Date: Sat Feb 10 21:38:50 PST 2001
Subject: codeTEST vs PerformancePak

Hello,

We have the option of getting codeTEST or
PerformancePak (ProfileScope/MemScope from RTI)
to monitor performance and memory on VxWorks.
I don't expect people to be using both at the same
time (so it won't be possible to compare them), but
I'd very much appreciate it if you could post your
opinions about either product.

Looks like codeTest actually measures code coverage,
and tells you which parts of your code have not been
executed at all. ProfileScope, on the other hand, seems
to be more like the Unix profilers, reporting how much
time was spent in each function (not sure if codeTest
does this).

Memory leak monitoring wise, the products seem to
have similar features.

Also, has anyone used RTI's ScopePak (StethoScope
and TraceScope)? Any opinions?

Thanks,
Hari




From vxwexplo-errs@csg.lbl.gov  Sun Feb 11 04:03:09 2001
From: Vxworks Exploder 
Date: Sun Feb 11 04:03:12 PST 2001
Subject: comp.os.vxworks newsdigest

Comp.Os.Vxworks Daily Digest    Sun Feb 11 04:03:03 PST 2001

        Subject: Re: Job Change   Electro-Optical Sensor Software Engineers  - Ottawa
        Subject: How to strip out the symbols?
        Subject: SAR Drivers
        Subject: DOSFS2 S_dosFsLib_FD_OBSOLETE problem...
        Subject: Re: Job Change   Electro-Optical Sensor Software Engineers  - Ottawa
        Subject: Re: How to strip out the symbols?
        Subject: Re: PowerPC 7400 Altivec and VxWorks
        Subject: Re: Impact of using C++ Multiple Inheritance
        Subject: Re: *UPDATED* Gcc 2.95.3 for VxWorks-PPC available....
        Subject: Re: DOSFS2 S_dosFsLib_FD_OBSOLETE problem...
        Subject: Re: PowerPC 7400 Altivec and VxWorks
        Subject: codeTEST vs PerformancePak

-------------------------------------------------------

Newsgroups: comp.os.vxworks,comp.realtime
Subject: Re: Job Change   Electro-Optical Sensor Software Engineers  - Ottawa
Date: Sat, 10 Feb 01 12:02:54 GMT
From: peb@amleth.demon.co.uk ("Paul E. Bennett")
Organization: HIDECS Consultancy
Message-ID: <981806574snz@amleth.demon.co.uk>
References: <961ma6$blr$1@news.igs.net> 
Reply-To: peb@amleth.demon.co.uk

In article 
           alfkatz@removethis.ieee.org  writes:

> Derek, posting these ads twice in 3 days on the comp.realtime newsgroup
> could be seen by some as a little uncool.  The last iteration hadn't yet had
> time to scroll off my outlook (yuk) screen.

.... and he hadn't had the decency to indicate locale either.

- -- 
********************************************************************
Paul E. Bennett ....................
Forth based HIDECS Consultancy .....
Mob: +44 (0)7811-639972 .........NOW AVAILABLE:- HIDECS COURSE......
Tel: +44 (0)1235-814586 .... see http://www.feabhas.com for details.
Going Forth Safely ..... EBA. www.electric-boat-association.org.uk..
********************************************************************


---------------------------

Newsgroups: comp.os.vxworks
Subject: How to strip out the symbols?
Date: Sat, 10 Feb 2001 21:08:35 GMT
From: "murali swaminathan" 
Organization: Verio
Message-ID: 

Folks,
We are in the process of porting our application to Vxworks. Our Image size
is very large and couple of out interfaces are serial. What is strip utility
that we need to run to remove the symbols (for 860/8260 powerpcs). If we
strip the symbols out how do we create a symbol file to be loaded into
tornado for debugging. Thanks for any help you can provide.

Murali





---------------------------

Newsgroups: comp.os.vxworks
Subject: SAR Drivers
Date: Sat, 10 Feb 2001 21:12:53 GMT
From: "murali swaminathan" 
Organization: Verio
Message-ID: 

any experience with 860/8260 SAR drivers that are commercially available for
vxworks? Netplane has one for 860 and Windriver for 8260.



---------------------------

Newsgroups: comp.os.vxworks
Subject: DOSFS2 S_dosFsLib_FD_OBSOLETE problem...
Date: Sun, 11 Feb 2001 02:01:59 GMT
From: ajay 
Organization: EarthLink Inc. -- http://www.EarthLink.net
Message-ID: <3A85F288.1118D1B5@wellknit.com>
Reply-To: ajay@wellknit.com

Hi All,
   I am trying to access an opened file that is stored in compact flash.
On accessing file (using lseek or ioctll(, 8, &))  I am getting
s_dosFsLib_FD_OBSOLETE error. iosFdValue() does not return -1 and
also iosFdShow shows it as valid fd.

I am lost here. Can some one let me know when does the fd becomes
obsolete? Any example scenarios would be great.  Should I be looking at
some structure dumps?

Thanks,
Ajay




---------------------------

Newsgroups: comp.os.vxworks,comp.realtime
Subject: Re: Job Change   Electro-Optical Sensor Software Engineers  - Ottawa
Date: Sun, 11 Feb 2001 02:21:21 -0000
From: "Dave Korn" 
Organization: Lumber Cartel (tinlc) Members #2234-2237 (owing to browser refresh)
Message-ID: 
References: <961ma6$blr$1@news.igs.net>  <981806574snz@amleth.demon.co.uk>


""Paul E. Bennett""  wrote in message news:981806574snz@amleth.demon.co.uk...
> In article 
>            alfkatz@removethis.ieee.org  writes:
>
> > Derek, posting these ads twice in 3 days on the comp.realtime newsgroup
> > could be seen by some as a little uncool.  The last iteration hadn't yet had
> > time to scroll off my outlook (yuk) screen.
>
> .... and he hadn't had the decency to indicate locale either.
>

   There is only one Ottawa, isn't there?

           DaveK



---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: How to strip out the symbols?
Date: Sat, 10 Feb 2001 19:02:41 -0800
From: DrDiags 
Organization: Posted via Supernews, http://www.supernews.com
Message-ID: <3A8600D1.7F834914@flashcom.net>
References: 
Reply-To: drdiags@flashcom.net

Murali,

    If you are using Tornado 2/VxWorks 5.4 GNU Toolchain, you should use objcopy
(I forgot the exact option). I believe the stripppc has problems (which is
probably why you asked this question).

murali swaminathan wrote:

> Folks,
> We are in the process of porting our application to Vxworks. Our Image size
> is very large and couple of out interfaces are serial. What is strip utility
> that we need to run to remove the symbols (for 860/8260 powerpcs). If we
> strip the symbols out how do we create a symbol file to be loaded into
> tornado for debugging. Thanks for any help you can provide.
>
> Murali


---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: PowerPC 7400 Altivec and VxWorks
Date: Sun, 11 Feb 2001 03:19:53 -0000
From: "Dave Korn" 
Organization: Lumber Cartel (tinlc) Members #2234-2237 (owing to browser refresh)
Message-ID: 
References:  <36Wg6.16$yi4.12223@newsr1.u-net.net> 


"Rick Cellucci"  wrote in message news:B6AA0B7D.107D3%hartwood@crosslink.net...

>
> FWIW, I believe there are still VxWorks-related issues that a new gcc won't
> solve.  Specifically, without changing the kernel source code, multiple
> tasks will not be able to use AltiVec. We would need a symbol like
> VX_FP_TASK for AltiVec to tell the task scheduler to save and restore the
> AltiVec registers via VRSAVE.  I guess I could only use AltiVec in only one
> task, but yuck.  And forget ISRs.  I'll ask Wind River next week.

  I imagine you could use a task context switch hook to save and restore the regs, and as for ISRs, they already need to call a
routine that's something like fppSaveContext, you could add similar for the altivec regs.

          DaveK



---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: Impact of using C++ Multiple Inheritance
Date: Sun, 11 Feb 2001 03:21:35 -0000
From: "Dave Korn" 
Organization: Lumber Cartel (tinlc) Members #2234-2237 (owing to browser refresh)
Message-ID: 
References: <95v1ft$1pp$1@bcrkh13.ca.nortel.com> <3A844089.1605C5E4@aps.anl.gov>


"Andrew Johnson"  wrote in message news:3A844089.1605C5E4@aps.anl.gov...
> I'm not going to say very much on your performance question, but just add
> that you should be careful that your resulting code doesn't become very
> hard to understand.  A programmer trying to understand a member function
> of a heavily-derived class may not have any idea which base class a
> particular call is coming from - it's easy to create object spaghetti

 Mmmmm.... object spaghetti...

          Homer Simpson.




---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: *UPDATED* Gcc 2.95.3 for VxWorks-PPC available....
Date: Sun, 11 Feb 2001 03:23:58 -0000
From: "Dave Korn" 
Organization: Lumber Cartel (tinlc) Members #2234-2237 (owing to browser refresh)
Message-ID: 
References:  <3A8462D3.1AADFB7C@mayannetworks.com>


"Kevin Cline"  wrote in message news:3A8462D3.1AADFB7C@mayannetworks.com...
> Dave Korn wrote:
>
> >  Finally I've recompiled the whole thing for i386, so it should run on any
> > CPU!
>
> Can you document the process so that others can repeat your work for different
> targets?  Thanks.
>

   There's already an outline of the build procedure on the FAQ/Docs page that should work fine for any other VxWorks targets; just
change the --target= parameter in the configure commands for the cross binutils and cross compiler.

         DaveK



---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: DOSFS2 S_dosFsLib_FD_OBSOLETE problem...
Date: Sun, 11 Feb 2001 04:04:39 GMT
From: Pierre Frigon 
Organization: Excite@Home - The Leader in Broadband http://home.com/faster
Message-ID: <3A860F20.D9612C06@yahoo.com>
References: <3A85F288.1118D1B5@wellknit.com>

Sounds like you did something to cause the dosfs to think the drive has been
replaced, forcing the fds to be invalidated. There is a call that will do
this (something like dosFsUnmount). Setting the option ...CHANGE_NO_WARN in
the dosVolConfig structure may have the same effect.

ajay wrote:

> Hi All,
>    I am trying to access an opened file that is stored in compact flash.
> On accessing file (using lseek or ioctll(, 8, &))  I am getting
> s_dosFsLib_FD_OBSOLETE error. iosFdValue() does not return -1 and
> also iosFdShow shows it as valid fd.
>
> I am lost here. Can some one let me know when does the fd becomes
> obsolete? Any example scenarios would be great.  Should I be looking at
> some structure dumps?
>
> Thanks,
> Ajay


---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: PowerPC 7400 Altivec and VxWorks
Date: Sun, 11 Feb 2001 01:00:59 -0500
From: Rick Cellucci 
Message-ID: 
References:  <36Wg6.16$yi4.12223@newsr1.u-net.net>  

in article Lxnh6.7597$zz4.180549@news2-win.server.ntlworld.com, Dave Korn at
no.spam@my.mailbox.invalid wrote on 2/10/01 10:19 PM:

> 
>> FWIW, I believe there are still VxWorks-related issues that a new gcc won't
>> solve.  Specifically, without changing the kernel source code, 
> 
> I imagine you could use a task context switch hook to save and restore the
> regs, and as for ISRs, they already need to call a routine that's something
> like fppSaveContext, you could add similar for the altivec regs.


You know, taskSwitchHookAdd just might work if I store pointers to AltiVec
register save areas in the Task Control Blocks. I incorrectly assumed that
the kernel would have to be changed.

Thanks Dave.

- ---------------------------------------------------
 Rick Cellucci
 Hartwood Software Engineering, Inc.
 Email: h a r t w o o d @ c r o s s l i n k . n e t
- ---------------------------------------------------



---------------------------

Newsgroups: comp.os.vxworks
Subject: codeTEST vs PerformancePak
Date: Sat, 10 Feb 2001 21:34:33 -0800
From: "Thiagarajan Hariharan" 
Organization: Lawrence Berkeley National Laboratory
Message-ID: <965ams$pq$1@overload.lbl.gov>

Hello,

We have the option of getting codeTEST or
PerformancePak (ProfileScope/MemScope from RTI)
to monitor performance and memory on VxWorks.
I don't expect people to be using both at the same
time (so it won't be possible to compare them), but
I'd very much appreciate it if you could post your
opinions about either product.

Looks like codeTest actually measures code coverage,
and tells you which parts of your code have not been
executed at all. ProfileScope, on the other hand, seems
to be more like the Unix profilers, reporting how much
time was spent in each function (not sure if codeTest
does this).

Memory leak monitoring wise, the products seem to
have similar features.

Also, has anyone used RTI's ScopePak (StethoScope
and TraceScope)? Any opinions?

Thanks,
Hari



---------------------------

End of New-News digest
**********************


From vxwexplo-errs@csg.lbl.gov  Mon Feb 12 01:55:41 2001
From: Fiederling.Ulf@dornier.dasa.de
Date: Mon Feb 12 01:55:43 PST 2001
Subject: Antwort: Pentium performance

                                                                                                                   
                    "T.                                                                                            
                    Hariharan"           An:                                       
                                 Thema:  Pentium performance                                               
                                                                                                                   
                    09.02.01                                                                                       
                    17:23                                                                                          
                                                                                                                   
                                                                                                                   









Hello Ulf,

Saw your posting about windriver support on the  newsgroup.
I'm a new vxworks user, so I haven't had to use support so  far.
You must've seen the previous posting i've included  below.
May be you should go through your local windriver field  application
engineer.

We are also using a pentium board. You mentioned  'pentium
performance problems' and 'gnu compiler update with  pentium
support'. Can you give some more details? When did  they
release this update? What kind of performance problems  did
you have?

You can post your answer to the newsgroup.

Thanks,
Hari



Hello Hari

Maybe some people are content with the support of WindRiver, but we made
different experiences.

The Compiler update is dated november 2000 i think and contains the new GNU
ccpentium compiler instead of the cc386 delivered with Tornado II.
We had many problem by porting our software from WINNT to VxWorks, because
the performance in arithmetic operations and trigonometric calculations was
very slow under VxWorks. WINNT was in some applications two times faster in
fact. With the new compiler, performance difference is only 10%, in some
apllications WINNT is faster, on the other side in some Applications
VxWorks is faster.

We have until now absolutly no documentation about compile option and
switches, getting the new compiler was only luck, because we had personal
support from WindRiver at our company. They gave us the software. I have to
mention also, that we have a support contract (you know the prices), and we
didnt get the software automaticly, however we informed WinRiver about our
performance problems almoust a year ago.

One great problem is still remaining: When downloading our application
software to the target, the alignment is not guaranteed. We are handling
large arrays of double variables, and regardless to the compiler switch
-malign-double, the arrays are not aligned "double" in many times.
This problem is not a problem of the compiler, because it happend also with
the cc386. The problem makes our software very slow, because copying these
great arrays is becoming much slower in case of disalignment. We told
WindRiver in December about this, but there was no response until now.

Hope I could help You

Ulf



From vxwexplo-errs@csg.lbl.gov  Mon Feb 12 04:03:11 2001
From: Vxworks Exploder 
Date: Mon Feb 12 04:03:13 PST 2001
Subject: comp.os.vxworks newsdigest

Comp.Os.Vxworks Daily Digest    Mon Feb 12 04:03:04 PST 2001

        Subject: Re: Job Change   Electro-Optical Sensor Software Engineers  - Ottawa
        Subject: Re: How to strip out the symbols?
        Subject: Re: How to strip out the symbols?
        Subject: Re: Job Change   Electro-Optical Sensor Software Engineers  - Ottawa
        Subject: Re: Job Change   Electro-Optical Sensor Software Engineers  - Ottawa
        Subject: Re: PowerPC 7400 Altivec and VxWorks
        Subject: Re: Job Change   Electro-Optical Sensor Software Engineers  - Ottawa
        Subject: Re: Job Change   Electro-Optical Sensor Software Engineers  - Ottawa
        Subject: Re: Job Change   Electro-Optical Sensor Software Engineers  - Ottawa
        Subject: Re: Job Change   Electro-Optical Sensor Software Engineers  - Ottawa
        Subject: Small Embedded VxWorks machine needed immediately
        Subject: Novice's question on JNI problem...
        Subject: Re: Job Change   Electro-Optical Sensor Software Engineers  - Ottawa
        Subject: Antwort: Pentium performance
        Subject: Re: SAMBA and VxWorks
        Subject: Re: How to strip out the symbols?

-------------------------------------------------------

Newsgroups: comp.os.vxworks,comp.realtime
Subject: Re: Job Change   Electro-Optical Sensor Software Engineers  - Ottawa
Date: Sun, 11 Feb 01 11:36:24 GMT
From: peb@amleth.demon.co.uk ("Paul E. Bennett")
Organization: HIDECS Consultancy
Message-ID: <981891384snz@amleth.demon.co.uk>
References: <961ma6$blr$1@news.igs.net>  <981806574snz@amleth.demon.co.uk> 
Reply-To: peb@amleth.demon.co.uk

In article 
           no.spam@my.mailbox.invalid "Dave Korn" writes:

> 
> ""Paul E. Bennett""  wrote in message
>  news:981806574snz@amleth.demon.co.uk...
> > In article 
> >            alfkatz@removethis.ieee.org  writes:
> >
> > > Derek, posting these ads twice in 3 days on the comp.realtime newsgroup
> > > could be seen by some as a little uncool.  The last iteration hadn't yet had> > > time to scroll off my outlook (yuk) screen.
> >
> > .... and he hadn't had the decency to indicate locale either.
> >
> 
>    There is only one Ottawa, isn't there?

As far as I know -- but he didn't indicate that in the header.

- -- 
********************************************************************
Paul E. Bennett ....................
Forth based HIDECS Consultancy .....
Mob: +44 (0)7811-639972 .........NOW AVAILABLE:- HIDECS COURSE......
Tel: +44 (0)1235-814586 .... see http://www.feabhas.com for details.
Going Forth Safely ..... EBA. www.electric-boat-association.org.uk..
********************************************************************


---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: How to strip out the symbols?
Date: Sun, 11 Feb 2001 16:53:06 GMT
From: "murali swaminathan" 
Organization: Verio
Message-ID: 
References:  <3A8600D1.7F834914@flashcom.net>

Thanks. Objcopy worked. It took the same time for me to download as with the
symbols. any thoughts, why? Further, how can I get the symbol into tornado
env (on the host side).

"DrDiags"  wrote in message
news:3A8600D1.7F834914@flashcom.net...
> Murali,
>
>     If you are using Tornado 2/VxWorks 5.4 GNU Toolchain, you should use
objcopy
> (I forgot the exact option). I believe the stripppc has problems (which is
> probably why you asked this question).
>
> murali swaminathan wrote:
>
> > Folks,
> > We are in the process of porting our application to Vxworks. Our Image
size
> > is very large and couple of out interfaces are serial. What is strip
utility
> > that we need to run to remove the symbols (for 860/8260 powerpcs). If we
> > strip the symbols out how do we create a symbol file to be loaded into
> > tornado for debugging. Thanks for any help you can provide.
> >
> > Murali
>



---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: How to strip out the symbols?
Date: Sun, 11 Feb 2001 09:22:39 -0800
From: DrDiags 
Organization: Posted via Supernews, http://www.supernews.com
Message-ID: <3A86CA5F.E9790774@flashcom.net>
References:  <3A8600D1.7F834914@flashcom.net> 
Reply-To: drdiags@flashcom.net

Murali,

    I don't have an answer for you download question. As far as making the
symbol available for Tornado, have you tried using/saving the pre-stripped image
and pointing the target server to use this core? I may be butchering your
questions, so if this thread dies, try reposting your new questions. You could
make two, one would speak of your network configuration, BSP, image size, etc..
for the download question and the other would be the symbol question. HTH.

murali swaminathan wrote:

> Thanks. Objcopy worked. It took the same time for me to download as with the
> symbols. any thoughts, why? Further, how can I get the symbol into tornado
> env (on the host side).
>
> "DrDiags"  wrote in message
> news:3A8600D1.7F834914@flashcom.net...
> > Murali,
> >
> >     If you are using Tornado 2/VxWorks 5.4 GNU Toolchain, you should use
> objcopy
> > (I forgot the exact option). I believe the stripppc has problems (which is
> > probably why you asked this question).
> >
> > murali swaminathan wrote:
> >
> > > Folks,
> > > We are in the process of porting our application to Vxworks. Our Image
> size
> > > is very large and couple of out interfaces are serial. What is strip
> utility
> > > that we need to run to remove the symbols (for 860/8260 powerpcs). If we
> > > strip the symbols out how do we create a symbol file to be loaded into
> > > tornado for debugging. Thanks for any help you can provide.
> > >
> > > Murali
> >


---------------------------

Newsgroups: comp.os.vxworks,comp.realtime
Subject: Re: Job Change   Electro-Optical Sensor Software Engineers  - Ottawa
Date: Sun, 11 Feb 2001 17:39:56 GMT
From: "Spehro Pefhany" 
Organization: Rather
Message-ID: 
References: <961ma6$blr$1@news.igs.net>  <981806574snz@amleth.demon.co.uk> 
Sender:  

In comp.realtime Dave Korn  wrote:

>    There is only one Ottawa, isn't there?

Would that be the one in Kansas or the one in Illinois? Or Ottawa county
in Ohio?  Of course I don't think any of those has a Kanata nearby. 

Best regards, 
- -- 
=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
Spehro Pefhany --"it's the network..."            "The Journey is the reward"
speff@interlog.com             Info for manufacturers: http://www.trexon.com
Embedded software/hardware/analog  Info for designers:  http://www.speff.com
Contributions invited->The AVR-gcc FAQ is at: http://www.BlueCollarLinux.com
=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=

---------------------------

Newsgroups: comp.os.vxworks,comp.realtime
Subject: Re: Job Change   Electro-Optical Sensor Software Engineers  - Ottawa
Date: Sun, 11 Feb 2001 15:33:07 -0000
From: "Dave Korn" 
Organization: Lumber Cartel (tinlc) Members #2234-2237 (owing to browser refresh)
Message-ID: 
References: <961ma6$blr$1@news.igs.net>  <981806574snz@amleth.demon.co.uk>  <981891384snz@amleth.demon.co.uk>


""Paul E. Bennett""  wrote in message news:981891384snz@amleth.demon.co.uk...
> In article 
>            no.spam@my.mailbox.invalid "Dave Korn" writes:
>
> >
> > ""Paul E. Bennett""  wrote in message
> >  news:981806574snz@amleth.demon.co.uk...
> > > In article 
> > >            alfkatz@removethis.ieee.org  writes:
> > >
> > > > Derek, posting these ads twice in 3 days on the comp.realtime newsgroup
> > > > could be seen by some as a little uncool.  The last iteration hadn't yet had> > > time to scroll off my outlook (yuk)
screen.
> > >
> > > .... and he hadn't had the decency to indicate locale either.
> > >
> >
> >    There is only one Ottawa, isn't there?
>
> As far as I know -- but he didn't indicate that in the header.

  Have you noticed the subject line of this thread?

         DaveK





---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: PowerPC 7400 Altivec and VxWorks
Date: Sun, 11 Feb 2001 15:34:15 -0000
From: "Dave Korn" 
Organization: Lumber Cartel (tinlc) Members #2234-2237 (owing to browser refresh)
Message-ID: 
References:  <36Wg6.16$yi4.12223@newsr1.u-net.net>   


"Rick Cellucci"  wrote in message news:B6AB94CB.11142%hartwood@crosslink.net...
> in article Lxnh6.7597$zz4.180549@news2-win.server.ntlworld.com, Dave Korn at
> no.spam@my.mailbox.invalid wrote on 2/10/01 10:19 PM:
>
> >
> >> FWIW, I believe there are still VxWorks-related issues that a new gcc won't
> >> solve.  Specifically, without changing the kernel source code, 
> >
> > I imagine you could use a task context switch hook to save and restore the
> > regs, and as for ISRs, they already need to call a routine that's something
> > like fppSaveContext, you could add similar for the altivec regs.
>
>
> You know, taskSwitchHookAdd just might work if I store pointers to AltiVec
> register save areas in the Task Control Blocks. I incorrectly assumed that
> the kernel would have to be changed.
>

  If you give it a go, do let us know how it works.

          DaveK



---------------------------

Newsgroups: comp.os.vxworks,comp.realtime
Subject: Re: Job Change   Electro-Optical Sensor Software Engineers  - Ottawa
Date: Sun, 11 Feb 01 21:50:53 GMT
From: peb@amleth.demon.co.uk ("Paul E. Bennett")
Organization: HIDECS Consultancy
Message-ID: <981928253snz@amleth.demon.co.uk>
References: <961ma6$blr$1@news.igs.net>  <981806574snz@amleth.demon.co.uk>  <981891384snz@amleth.demon.co.uk> 
Reply-To: peb@amleth.demon.co.uk

In article 
           no.spam@my.mailbox.invalid "Dave Korn" writes:

>   Have you noticed the subject line of this thread?

Yes - "Job Change   Electro-Optical Sensor Software Engineer"

- -- 
********************************************************************
Paul E. Bennett ....................
Forth based HIDECS Consultancy .....
Mob: +44 (0)7811-639972 .........NOW AVAILABLE:- HIDECS COURSE......
Tel: +44 (0)1235-814586 .... see http://www.feabhas.com for details.
Going Forth Safely ..... EBA. www.electric-boat-association.org.uk..
********************************************************************


---------------------------

Newsgroups: comp.os.vxworks,comp.realtime
Subject: Re: Job Change   Electro-Optical Sensor Software Engineers  - Ottawa
Date: Sun, 11 Feb 2001 15:21:17 -0800
From: 
Organization: SBC Internet Services
Message-ID: 
References: <961ma6$blr$1@news.igs.net>  <981806574snz@amleth.demon.co.uk>  <981891384snz@amleth.demon.co.uk>  <981928253snz@amleth.demon.co.uk>

This is a multi-part message in MIME format.

- ------=_NextPart_000_0090_01C0943E.47746EF0
Content-Type: text/plain;
	charset="iso-8859-1"
Content-Transfer-Encoding: quoted-printable

Gee now that's the first interesting thing said in this thread. My =
subject line for the message you just posted says:

"Re: Job Change Electro-Optical Sensor Software Engineers - Ottawa"

What about your setup is different do you suppose. Have you got a =
character limitation of some sort on your subject line?

""Paul E. Bennett""  wrote in message =
news:981928253snz@amleth.demon.co.uk...
> In article 
>            no.spam@my.mailbox.invalid "Dave Korn" writes:
>=20
> >   Have you noticed the subject line of this thread?
>=20
> Yes - "Job Change   Electro-Optical Sensor Software Engineer"
>=20
> --=20
> ********************************************************************
> Paul E. Bennett ....................
> Forth based HIDECS Consultancy .....
> Mob: +44 (0)7811-639972 .........NOW AVAILABLE:- HIDECS COURSE......
> Tel: +44 (0)1235-814586 .... see http://www.feabhas.com for details.
> Going Forth Safely ..... EBA. www.electric-boat-association.org.uk..
> ********************************************************************
>=20
>=20

- ------=_NextPart_000_0090_01C0943E.47746EF0
Content-Type: text/html;
	charset="iso-8859-1"
Content-Transfer-Encoding: quoted-printable









Gee now that's the first interesting thing said =
in this=20
thread. My subject line for the message you just posted =
says:


 


"Re: Job Change Electro-Optical Sensor Software =
Engineers=20
- - Ottawa"


 


What about your setup is different do you =
suppose. Have=20
you got a character limitation of some sort on your subject=20
line?


 


""Paul E. Bennett"" <peb@amleth.demon.co.uk> =
wrote in=20
message news:981928253snz@amleth.demon.co.uk...
> In article =
<IeBh6.11436$BQ3.198444@news6-win.server.ntlworld.com<=
/A>>
>=20
           no.spam@my.mailbox.invalid =
"Dave Korn"=20
writes:
> 
> >   Have you noticed the subject =
line of=20
this thread?
> 
> Yes - "Job Change   =
Electro-Optical=20
Sensor Software Engineer"
> 
> -- 
>=20
********************************************************************
&=
gt;=20
Paul E. Bennett ....................<email://peb@amleth.demon.co.uk>
> Forth based HIDECS Consultancy =
.....<http://www.amleth.demon.co.uk/>
> Mob: +44 (0)7811-639972 .........NOW =
AVAILABLE:- HIDECS=20
COURSE......
> Tel: +44 (0)1235-814586 .... see http://www.feabhas.com for=20
details.
> Going Forth Safely ..... EBA. www.electric-boat-association.org.uk..
>=20
********************************************************************
&=
gt;=20

> 

- ------=_NextPart_000_0090_01C0943E.47746EF0--


---------------------------

Newsgroups: comp.os.vxworks,comp.realtime
Subject: Re: Job Change   Electro-Optical Sensor Software Engineers  - Ottawa
Date: Mon, 12 Feb 2001 00:07:11 GMT
From: "Spehro Pefhany" 
Organization: Rather
Message-ID: 
References: <961ma6$blr$1@news.igs.net>  <981806574snz@amleth.demon.co.uk>  <981891384snz@amleth.demon.co.uk>  <981928253snz@amleth.demon.co.uk> 
Sender:  

In comp.realtime cfk@pacbell.net wrote:
> Gee now that's the first interesting thing said in this thread. My subject line for the message you just posted says:

> "Re: Job Change Electro-Optical Sensor Software Engineers - Ottawa"

> What about your setup is different do you suppose. Have you got a character limitation of some sort on your subject line?

Mine says "Re: Job Change Electro-Optical Sensor Soft" until I hit 
^H to view the full headers on tin (unix newsreader). 

If it happened to  truncate between words, it wouldn't be obvious there
was any more stuff to the right. So, the OP is guilty, guilty, guilty of
using an overly-long subject line with vital information on the far
right-hand side of it. 

Best regards, 
- -- 
=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
Spehro Pefhany --"it's the network..."            "The Journey is the reward"
speff@interlog.com             Info for manufacturers: http://www.trexon.com
Embedded software/hardware/analog  Info for designers:  http://www.speff.com
Contributions invited->The AVR-gcc FAQ is at: http://www.BlueCollarLinux.com
=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=

---------------------------

Newsgroups: comp.os.vxworks,comp.realtime
Subject: Re: Job Change   Electro-Optical Sensor Software Engineers  - Ottawa
Date: Mon, 12 Feb 2001 00:56:15 -0000
From: "Dave Korn" 
Organization: Lumber Cartel (tinlc) Members #2234-2237 (owing to browser refresh)
Message-ID: 
References: <961ma6$blr$1@news.igs.net>  <981806574snz@amleth.demon.co.uk>  <981891384snz@amleth.demon.co.uk>  <981928253snz@amleth.demon.co.uk>  


"Spehro Pefhany"  wrote in message news:POFh6.110029$KP3.30592011@news3.rdc1.on.home.com...
> In comp.realtime cfk@pacbell.net wrote:
> > Gee now that's the first interesting thing said in this thread. My subject line for the message you just posted says:
>
> > "Re: Job Change Electro-Optical Sensor Software Engineers - Ottawa"
>
> > What about your setup is different do you suppose. Have you got a character limitation of some sort on your subject line?
>
> Mine says "Re: Job Change Electro-Optical Sensor Soft" until I hit
> ^H to view the full headers on tin (unix newsreader).
>
> If it happened to  truncate between words, it wouldn't be obvious there
> was any more stuff to the right. So, the OP is guilty, guilty, guilty of
> using an overly-long subject line with vital information on the far
> right-hand side of it.

  Hnmm.... the canonical response to this situation is to say "Get A Real Newsreader(tm)".

  Unfortunately, that usually means "ditch your windoze and get *nix".  tin and trn are two that are generally considered "Real"
Newsreaders.  So I guess the wisdom of the ancients has no relevance in this situation.

         DaveK




---------------------------

Newsgroups: comp.os.vxworks,comp.soft-sys.matlab
Subject: Small Embedded VxWorks machine needed immediately
Date: Mon, 12 Feb 2001 05:43:38 GMT
From: ix@fnord.io.com (IX Corp)
Organization: Illuminati Online
Message-ID: 

I'm looking for a small VxWorks target machine to develop 
Mathworks/Simulink code generated by Real Time Workshop on.
The device requirements are: P6-200 class CPU (don't care whether it is
power PC or i960 based, but 68040 class machines will not be enough
juice), 4 serial ports (1 terminal & 3 free). I would prefer it be
physically small, and equipped with something like a 6 channel A/D
converter though these are not necessary.

I come from a VME-board/crate VxWorks "gcc hosted on unix" perspective,
and have had some difficulty finding information on more limited gizmoes
supporting VxWorks. I don't need or want the whole enchelada; I really
just need serial ports (and would like the A/D).

I already have the RTW generated code. It will need minor modifications to
make the multitasking work right on VxWorks.
Bearing this in mind, I'd prefer something "gcc hosted on Unix"-like to
develop my application on.

If someone could point me towards a vendor I would appreciate it. 
thanks,

- -Lupo
"I pine for the sheer stupidity of the old macho days, when men would 
brandish hammers and build huge, bulky cars that sucked up gas and tore
open the ozone layer and crushed small animals beneath totally useless but
totally cool-looking tail fins." -Dennis Leary    

---------------------------

Newsgroups: comp.os.vxworks
Subject: Novice's question on JNI problem...
Date: Mon, 12 Feb 2001 06:13:07 GMT
From: kwonyh@samsung.co.kr
Organization: Deja.com
Message-ID: <967utg$cd1$1@nnrp1.deja.com>

Hello,

i'm working on a project with VxWorks 5.4.1 and pJWorks 3.0.2,
and have a basic problem for the JNI programming.

According to JNI-related books and/or internet resources,
the sample code below must work, i think.

void SendToJava( char *arr )
{
	JavaVM		*jvm;
	jsize		no_jvm;
	JNIEnv		*env;  /* = g_env; */
	jobject		obj;   /* = g_obj; */
	JDK1_1AttachArgs	attach_args;
	jclass		cls;
	jmethodID	mid;
	jbyteArray	arrbuf;
	jint		version = 0x00010001;

	int i;

	if ( JNI_GetCreatedJavaVMs( &jvm, 1, &no_jvm ) != 0 )
		printf("JNI_GetCreatedJavaVMs Error !! \n");


	if ( (*jvm)->GetEnv( jvm, &env, version ) != 0 )
		printf( "GerEnv Error!" );


/*	if ( (*jvm)->AttachCurrentThread( jvm, (void**)&env,
&attach_args ) != 0 ) */
	if ( (*jvm)->AttachCurrentThread( jvm, &env, NULL ) != 0 )
		printf("Error : AttachCurrentThread Error !! \n");

	...
}

When i compile and run it on my target,
the JNI_GetCreatedJavaVMs() returns successfully with valid values
for jvm, (*jvm), and no_jvm; i mean they are not null.

But in the second jni function call, GetEnv(),
it returns the error with '0xFFFFFFFE'.

Could you explain why it happened?
Or, is there any difference in jni functionality
between JVM in desktop/pc and WRS/pJWorks 3.0.2?

Please advise to help a novice.

Thanks much in advance.
YH Kwon


Sent via Deja.com
http://www.deja.com/

---------------------------

Newsgroups: comp.os.vxworks,comp.realtime
Subject: Re: Job Change   Electro-Optical Sensor Software Engineers  - Ottawa
Date: Mon, 12 Feb 01 09:46:38 GMT
From: peb@amleth.demon.co.uk ("Paul E. Bennett")
Organization: HIDECS Consultancy
Message-ID: <981971198snz@amleth.demon.co.uk>
References: 
Reply-To: peb@amleth.demon.co.uk

In article 
           speff@interlog.com "Spehro Pefhany" writes:

> In comp.realtime cfk@pacbell.net wrote:
> > Gee now that's the first interesting thing said in this thread. My subject
>  line for the message you just posted says:
> 
> > "Re: Job Change Electro-Optical Sensor Software Engineers - Ottawa"
> 
> > What about your setup is different do you suppose. Have you got a character
>  limitation of some sort on your subject line?

My (still DOS based) newsreader (although that will changes soon when I 
get the FreeBSD software settled on my new laptop) seemingly truncates
the subject line. I had to save the message to a file and use the text 
editor to find out.
 
> Mine says "Re: Job Change Electro-Optical Sensor Soft" until I hit 
> ^H to view the full headers on tin (unix newsreader). 
> 
> If it happened to  truncate between words, it wouldn't be obvious there
> was any more stuff to the right. So, the OP is guilty, guilty, guilty of
> using an overly-long subject line with vital information on the far
> right-hand side of it. 

It wasn't obvious within the newsreader. I agree with Spehro about
the guilt of the originator in posting an overly long subject line.

Something like:-

  JOB: - Ottowa - Electro-Optical Engineers

would have been just as informative and would probably only have been
read by those near Ottowa (even if there happens to be more than one)

- -- 
********************************************************************
Paul E. Bennett ....................
Forth based HIDECS Consultancy .....
Mob: +44 (0)7811-639972 .........NOW AVAILABLE:- HIDECS COURSE......
Tel: +44 (0)1235-814586 .... see http://www.feabhas.com for details.
Going Forth Safely ..... EBA. www.electric-boat-association.org.uk..
********************************************************************


---------------------------

Newsgroups: comp.os.vxworks
Subject: Antwort: Pentium performance
Date: Mon, 12 Feb 2001 10:55:42 +0100
From: Fiederling.Ulf@dornier.dasa.de
Organization: Lawrence Berkeley National Laboratory
Message-ID: <968d4s$r9j$1@overload.lbl.gov>

                                                                                                                   
                    "T.                                                                                            
                    Hariharan"           An:                                       
                                 Thema:  Pentium performance                                               
                                                                                                                   
                    09.02.01                                                                                       
                    17:23                                                                                          
                                                                                                                   
                                                                                                                   









Hello Ulf,

Saw your posting about windriver support on the  newsgroup.
I'm a new vxworks user, so I haven't had to use support so  far.
You must've seen the previous posting i've included  below.
May be you should go through your local windriver field  application
engineer.

We are also using a pentium board. You mentioned  'pentium
performance problems' and 'gnu compiler update with  pentium
support'. Can you give some more details? When did  they
release this update? What kind of performance problems  did
you have?

You can post your answer to the newsgroup.

Thanks,
Hari



Hello Hari

Maybe some people are content with the support of WindRiver, but we made
different experiences.

The Compiler update is dated november 2000 i think and contains the new GNU
ccpentium compiler instead of the cc386 delivered with Tornado II.
We had many problem by porting our software from WINNT to VxWorks, because
the performance in arithmetic operations and trigonometric calculations was
very slow under VxWorks. WINNT was in some applications two times faster in
fact. With the new compiler, performance difference is only 10%, in some
apllications WINNT is faster, on the other side in some Applications
VxWorks is faster.

We have until now absolutly no documentation about compile option and
switches, getting the new compiler was only luck, because we had personal
support from WindRiver at our company. They gave us the software. I have to
mention also, that we have a support contract (you know the prices), and we
didnt get the software automaticly, however we informed WinRiver about our
performance problems almoust a year ago.

One great problem is still remaining: When downloading our application
software to the target, the alignment is not guaranteed. We are handling
large arrays of double variables, and regardless to the compiler switch
- -malign-double, the arrays are not aligned "double" in many times.
This problem is not a problem of the compiler, because it happend also with
the cc386. The problem makes our software very slow, because copying these
great arrays is becoming much slower in case of disalignment. We told
WindRiver in December about this, but there was no response until now.

Hope I could help You

Ulf


---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: SAMBA and VxWorks
Date: Mon, 12 Feb 2001 12:21:05 +0100
From: "Jorgen Jakobsen" 
Organization: Posted Courtesy of Tele Danmark or one of its European Subsidiaries
Message-ID: <968h0u$log$1@news.inet.tele.dk>
References: <95u52t$qv3$1@news.inet.tele.dk>

Found a link on WRS themselves :

http://www.windriver.com/cgi-bin/corporate/tradeWinds/viewProd.cgi?id=381


/Jørgen

"Jorgen Jakobsen"  wrote in message
news:95u52t$qv3$1@news.inet.tele.dk...
> Dear VxWork'ers
>
> Anybody know of a VxWorks port of SAMBA (GNU SMB) ?
>
> /Jørgen
>
>



---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: How to strip out the symbols?
Date: Mon, 12 Feb 2001 12:47:34 +0100
From: alain 
Organization: Nortel
Message-ID: 
References: 

There is an utility elfXsyms which keep only symbols from an image
file.
Alain.


---------------------------

End of New-News digest
**********************


From vxwexplo-errs@csg.lbl.gov  Mon Feb 12 05:36:09 2001
From: james bond 
Date: Mon Feb 12 05:36:11 PST 2001
Subject: I am new to vxworks, can any one help me out.

Hi,
This is Sumeet from Bangalore.
I am new to VxWorks, Can anyone help in this regard.
Regards
Sumeet



__________________________________________________
Do You Yahoo!?
Get personalized email addresses from Yahoo! Mail - only $35 
a year!  http://personal.mail.yahoo.com/


From vxwexplo-errs@csg.lbl.gov  Mon Feb 12 11:25:52 2001
From: Joseph Sebastian 
Date: Mon Feb 12 11:25:55 PST 2001
Subject: MIPS Evaluation Board.

Hi all,
I am looking for a MIPS evaluation board with PCI interface.
Is there a source from where I can get an evaluation board with VxWorks
BSP.
Joseph Sebastian.




From vxwexplo-errs@csg.lbl.gov  Mon Feb 12 23:50:16 2001
From: Bob Kibrick 
Date: Mon Feb 12 23:50:18 PST 2001
Subject: MVME2304 w/ VxWorks 5.3.1 is slow reading VMEbus

We are currently conducting tests on a Motorola MVME2304-0133 board that is
running VxWorks version 5.3.1.  We are rather surprised to find that on
certain tests of VMEbus throughput, the MVME2304 (which has a 333 MHz.
PowerPC 604e) delivers significantly lower performance than an obsolete
Force 5CE board (which has an 85 MHz. SPARC processor running VxWorks 5.2).

The particular test in question involves transferring approximately 145 MB
of data between a bulk VME memory board (Chrislin CI-VME64 256 MB) and the
local DRAM on the CPU board.  This test was performed in both directions,
and using both D16 and D32 transfer widths.  The identical C program was
used on both processors, compiled (or cross-compiled) using gcc (2.9.5 for
Sparc, cygnus-2.7.2 for PowerPC) on a Sparc Solaris 2.6 host.  The same
VME memory board was used in all tests.  The tests were conducted on an
otherwise idle system with no other competing processes (the tests were 
run in the context of the VxWorks shell) nor any concurrent activity on
the VMEbus.

The transfers were coded as a tight loop of the form:

	while( vmeptr < end_address )
		localmem = *vmeptr++;

for transferring from VME memory to local DRAM, and

	while( vmeptr < end_address )
		*vmeptr++ = localmem;

for the reverse direction.

On the PowerPC, the read loop for the 32-bit case (int32 *vmeptr) assembles to:

.L4:
        lwz 0,0(31)
        addi 31,31,4
        cmplw 1,31,29
        bc 12,4,.L4

On the SPARC, the corresponding read loop assembles to:

L4:
        cmp %l0,%l1
        blu L6
        nop
        b L5
        nop
L6:
        ld [%l0],%o0
        st %o0,[%fp-24]
        add %l0,4,%l0
        b L4
        nop
L5:

For single cycle accesses, the Chrislin VME board has a 90ns access time
for writing, and 140ns access time for reading.

The results (elapsed time required to transfer 145 MB) are as follows:

Writing from CPU DRAM into VMEbus          Reading from VMEbus into CPU DRAM
----------------------------------         ----------------------------------
MVME2304,  32-bit transfers: 15.1s	   MVME2304,  32-bit transfers: 39.6s
Force 5CE, 32-bit transfers: 21.1s	   Force 5CE, 32-bit transfers: 29.4s

MVME2304,  16-bit transfers: 30.2s	   MVME2304,  16-bit transfers: 79.3s
Force 5CE, 32-bit transfers: 42.3s	   Force 5CE, 16-bit transfers: 58.9s

Thus, while the Force 5CE is about 40% slower than the MVME2304 when writing
to the VMEbus, the MVME2304 is about 35% slower than the Force 5CE when reading
from the VMEbus.  

Furthermore, the MVME2304 takes 2.6 times longer to read from the VMEbus than
it does to to write to it.  But the corresponding read cycle time on the
Chrislin board is only 1.6 times the write cycle time.

We are trying to understand why the MVME2304 performs so slowly when reading
from the VMEbus memory, and are wondering what options, if any, exist for
improving this performance.  For example, are there any alternate settings
of the control registers of the Tundra Universe chip (on the MVME2304) that
might allow for higher throughput on VMEbus reads (We are currently using the 
default values set by VxWorks 5.3.1)?  Alternatively, is there any way from
software (running under VxWorks 5.3.1 on the MVME2304) to initiate either
A32/D32 block transfer read cycles on the VMEbus (i.e., the type of bus
cycles implied by VME address modifier codes 0x0F or 0x0B) or to initiate a
DMA-like transfer from VMEbus memory to local DRAM on the CPU board that
would generate such block transfer read cycles?

Any pointers or suggestions would be greatly appreciated.

Bob Kibrick, University of California Observatories / Lick Observatory
kibrick@ucolick.org	831-459-2262	831-459-2298 (FAX)


From vxwexplo-errs@csg.lbl.gov  Tue Feb 13 04:03:14 2001
From: Vxworks Exploder 
Date: Tue Feb 13 04:03:17 PST 2001
Subject: comp.os.vxworks newsdigest

Comp.Os.Vxworks Daily Digest    Tue Feb 13 04:03:05 PST 2001

        Subject: What causes memPartFree to call taskSuspend?
        Subject: Re: *UPDATED* Gcc 2.95.3 for VxWorks-PPC available....
        Subject: loadLib, loadModule....
        Subject: Re: Tornado Compiler
        Subject: I am new to vxworks, can any one help me out.
        Subject: Problem booting vxworks from DosFs
        Subject: Re: integrating network drivers
        Subject: Re: integrating network drivers
        Subject: Re: integrating network drivers
        Subject: SM polling on PPC
        Subject: Re: dynamically adding PCI space for vxworks
        Subject: Re: *UPDATED* Gcc 2.95.3 for VxWorks-PPC available....
        Subject: MIPS Evaluation Board.
        Subject: Re: Problem booting vxworks from DosFs
        Subject: Re: building with TrueFFS
        Subject: Re: RouteLookup function
        Subject: CppUnit port to VxWorks
        Subject: documenation for configAll.h?
        Subject: Problem with NFS
        Subject: compiler differences between T1 and T2?
        Subject: steve.konish@windriver.com
        Subject: Tornado on Solaris 7?
        Subject: Re: Error during using JavaCodeCompact
        Subject: Problem with Vxworks Timers/Signals
        Subject: Re: Linker error: undefined reference to `__udivdi3
        Subject: Re: VxWorks for Linux emulation
        Subject: How can I get the etheraddr from end driver send?
        Subject: Re: What causes memPartFree to call taskSuspend?
        Subject: Re: loadLib, loadModule....
        Subject: Re: How to calculate Task stack space?
        Subject: Re: Problem with Vxworks Timers/Signals
        Subject: Re: Shared memory network problems
        Subject: GDI rendering for VxWorks
        Subject: MVME2304 w/ VxWorks 5.3.1 is slow reading VMEbus
        Subject: Re: documenation for configAll.h?
        Subject: Re: compiler differences between T1 and T2?
        Subject: Re: *UPDATED* Gcc 2.95.3 for VxWorks-PPC available....
        Subject: Learning a new OS [was Re: dialog with WindRiver - reasonable response]
        Subject: Re: unknown error message
        Subject: Re: Problem booting vxworks from DosFs

-------------------------------------------------------

Newsgroups: comp.os.vxworks
Subject: What causes memPartFree to call taskSuspend?
Date: Mon, 12 Feb 2001 12:53:03 +0000
From: Iain Gourlay 
Organization: Agilent Technologies Ltd.
Message-ID: <3A87DCAF.4E187C6@agilent.com>

The scenario appears simple enough - C++ code creating object instances during initialisation of an
application running in the main thread on Pentium architecture. Observations so far,

- - This is not a case of a double delete.
- - The heap is pretty clean and not fragmented.
- - memShow shows plenty of space and subsequent heap based mallocs and frees are ok.
- - None of the other tasks appear to be hogging the memory allocation and freeing library.
- - The task stack high water mark is good.

The hang-up appears to be dependent on the size of the call stack.

Thoughts?

Iain Gourlay
Agilent Technologies






- -> tt "Main"
23fcca vxTaskEntry    +a  : 3c871ef (1, 3e9fba4, 0, 0, 0, 0, 0, 0, 0, 0)
3c87258 main           +2c8: MeasSysState::Initialize(bool) (3ffd254, 3c80000)
3db8893 MeasSysState::Initialize(bool)+1b : MeasSubSystem::InitNorm(void) ([3ffd244, 38a14e0,
3c8725d, 3ffd254, 3c80000])
3db69d3 MeasSubSystem::InitNorm(void)+23 : MeasSubSystem::Initialize(void) (3ffd244)
3db6b63 MeasSubSystem::Initialize(void)+5b : MeasThread::MeasThread(char const *,
IHardwareAssemblies *) (3ffd1c4, 3db6ac7, 3e792d4)
3db9291 MeasThread::MeasThread(char const *, IHardwareAssemblies *)+149:
MeasurementSystem::MeasurementSystem(IHardwareAssemblies *, MutexBool *) (3ffd0b8, 3e792d4, 3ffd1f8)
3dbc70e MeasurementSystem::MeasurementSystem(IHardwareAssemblies *, MutexBool *)+5a6:
MeasMeasurements::MeasMeasurements(void) (3ffcdbc)
3db6613 MeasMeasurements::MeasMeasurements(void)+6b : __builtin_delete (3ffcda4)
24e886 __builtin_delete+42 : free ([3ffcda4, 38a0e44, 38a0d98, 3e86168, 38a0e0c])
1e95e8 free           +c  : memPartFree ([2a2f98, 3ffcda4, 38a0e24, 24e88b, 3ffcda4])
1e9204 memPartFree    +f8 : taskSuspend (0)
value = 0 = 0x0

---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: *UPDATED* Gcc 2.95.3 for VxWorks-PPC available....
Date: Mon, 12 Feb 2001 14:01:39 +0100
From: Roger 
Message-ID: <3A87DEB3.725A6DDB@opq.se>
References:  <3A8462D3.1AADFB7C@mayannetworks.com> 

Dave Korn wrote:
> 
> "Kevin Cline"  wrote in message news:3A8462D3.1AADFB7C@mayannetworks.com...
> > Dave Korn wrote:
> >
> > >  Finally I've recompiled the whole thing for i386, so it should run on any
> > > CPU!
> >
> > Can you document the process so that others can repeat your work for different
> > targets?  Thanks.
> >
> 
>    There's already an outline of the build procedure on the FAQ/Docs page that should work fine for any other VxWorks targets; just
> change the --target= parameter in the configure commands for the cross binutils and cross compiler

I am new to this. Where are the FAQ pages? 

 .
> 
>          DaveK

---------------------------

Newsgroups: comp.os.vxworks,comp.arch.embedded
Subject: loadLib, loadModule....
Date: Mon, 12 Feb 2001 13:18:11 +0000
From: Owain Phillips 
Organization: Siemens Inc.
Message-ID: <3A87E293.2DB2BF2@siemenscomms.co.uk>

This is a multi-part message in MIME format.
- --------------611BC98CD11AA8E7FA047CE1
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit

Hi comp.os.vxworks,

I am working on a new product that has numerous varients and each of
those variants may have  many functional areas that are common to
other variants.

Could we use a modular approach not only in development, but also in
the end product where a load-list specifies lists of loadable modules
that
are loaded  from a filesystem before starting the configured system.

How does/could this hang together as regards cross references?
Must we ensure that each module is built to ensure that after its load
we have no undefined references?

I could see that such an approach could work, when we develop using
windshell for incremental development as windriver describes; but what
about those awkward bought in modules where we don't have finger in
the pie?

If any of you have experience of using the loadLib and such a modular
load list I would be very interested to hear your views.

Kind regards,
Owain Phillips

- --------------611BC98CD11AA8E7FA047CE1
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

- --------------611BC98CD11AA8E7FA047CE1--


---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: Tornado Compiler
Date: Mon, 12 Feb 2001 15:04:01 +0100
From: "RedFenris" 
Organization: Mess
Message-ID: <968qea$2h5$1@wanadoo.fr>
References: <8vv0t6$f7h$1@overload.lbl.gov> <6D%W5.251$4T1.189523@news.pacbell.net>
Reply-To: "RedFenris" 

Nope.....

remove -ansi from compiler invokation parameters


Jianfei XU  a écrit dans le message :
6D%W5.251$4T1.189523@news.pacbell.net...
> with gcc, you can use -x to specify the language.
>
> "Thach Vu"  wrote in message
> news:8vv0t6$f7h$1@overload.lbl.gov...
> > 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_01C058C5.050C2EE0
> > Content-Type: text/plain;
> > charset="iso-8859-1"
> >
> > Hi,
> >
> > I have a question regarding to Tornado 2.0 Compiler.
> > I tried to compile a header file that contains C++ comment (//) but it
> > didn't accept this comment style.
> > I had to change it to C comment style (/* */) in order to make the
> compiler
> > work.
> > I found it strange that if the file is cpp files then it will accept the
> C++
> > comment style.
> > Is there any way that the compiler will read the C++ comment style
inside
> my
> > header file.
> > The reason I ask is that the header file contains an awful lot of C++
> > comments and I don't want to convert them into C comment.
> > Thank you for any of your help.
> >
> > Thach Vu
> >
> >
> > ------_=_NextPart_001_01C058C5.050C2EE0
> > Content-Type: text/html;
> > charset="iso-8859-1"
> > Content-Transfer-Encoding: quoted-printable
> >
> > 
> > 
> > 
> >  > charset=3Diso-8859-1">
> >  > 5.5.2652.35">
> > Tornado Compiler
> > 
> > 
> >
> > 
Hi,
> > 

> >
> > 
I have a question regarding to Tornado
=
> > 2.0 Compiler.
> > 
I tried to compile a header file that
=
> > contains C++ comment (//) but it didn't accept this comment =
> > style.
> > 
I had to change it to C comment style
=
> > (/* */) in order to make the compiler work.
> > 
I found it strange that if the file =
> > is cpp files then it will accept the C++ comment style.
> > 
Is there any way that the compiler =
> > will read the C++ comment style inside my header file.
> > 
The reason I ask is that the header =
> > file contains an awful lot of C++ comments and I don't want to convert =
> > them into C comment.

> >
> > 
Thank you for any of your help.
> > 

> >
> > 
Thach Vu
> > 

> >
> > 
> > 
> > ------_=_NextPart_001_01C058C5.050C2EE0--
> >
>
>



---------------------------

Newsgroups: comp.os.vxworks
Subject: I am new to vxworks, can any one help me out.
Date: Mon, 12 Feb 2001 05:36:07 -0800 (PST)
From: james bond 
Organization: Lawrence Berkeley National Laboratory
Message-ID: <968r6r$50t$1@overload.lbl.gov>

Hi,
This is Sumeet from Bangalore.
I am new to VxWorks, Can anyone help in this regard.
Regards
Sumeet



__________________________________________________
Do You Yahoo!?
Get personalized email addresses from Yahoo! Mail - only $35 
a year!  http://personal.mail.yahoo.com/

---------------------------

Newsgroups: comp.os.vxworks
Subject: Problem booting vxworks from DosFs
Date: Mon, 12 Feb 2001 16:12:21 +0100
From: Torsten Levin 
Organization: Colt Telecom GmbH
Message-ID: <3A87FD55.AD34FD82@kayser-threde.de>

Hello to all,

i have a problem booting from a PCMCIA flash card on a PowerPC target.
Host is Tornado 2.0 on a Windows NT 4.0 Pentium PC.

The code inside the boot loader does the following:

1. Initialize PCMCIA hardware (third party driver function)
   Return: OK
2. generating PCMCIA block device (third party driver function)
   Return: OK
3. initializing DOS FS (dosFsInit)
   Return: OK
4. Attaching DOS file system to pcmcia block device (dosFsDevInit)
   Return: ERROR
   errno = 0x3d0001 = VOLUME_NOT_AVAILABLE

The absolute identical code packed inside a "test" function and running
on the shell after booting the system via Ethernet works without any
problem. The PCMCIA card can be fully accessed.

Any ideas?
Thank you in advance!

Torsten

- -- 
Torsten Levin             Tel.: ++49-(0)89-72495-451
Kayser-Threde GmbH        Fax:  ++49-(0)89-72495-291
Perchtinger Str. 3        mailto:TL@kayser-threde.de
D-81379 München           http://www.kayser-threde.de

---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: integrating network drivers
Date: Mon, 12 Feb 2001 16:16:01 GMT
From: Bill Pringlemeir 
Organization: Factory of the mind
Message-ID: 
References: <90eniu$cms$1@news.netmar.com> <3A2AF70B.A74CA02@flashcom.net> 
Sender: bpringlemeir@DeadDuck

[bcc'ed to `abuse@rr.com']

Did anyone else complain to `abuse@rr.com'?  I have several new posts
in comp.os.vxworks that seem to have happened long ago.  Are the
people from `RoadRunner - TampaBay' going to delete all of the
duplicate posts or is that impossible?

Here is the header of my stored post,

[original header]
X-From-Line: nobody Mon Dec 04 09:26:30 2000
Sender: bpringlemeir@DeadDuck
Newsgroups: comp.os.vxworks
Subject: Re: integrating network drivers
References: <90eniu$cms$1@news.netmar.com> <3A2AF70B.A74CA02@flashcom.net>
From: Bill Pringlemeir 
Mail-Copies-To: nobody
Organization: Factory of the mind
X-Microsoft-die: Ministry - Everyday is Halloween.
X-die-Microsoft: "http://www.opensource.org/halloween.html"
Date: 04 Dec 2000 09:26:30 -0500
Message-ID: 
User-Agent: Gnus/5.0803 (Gnus v5.8.3) Emacs/20.7
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Lines: 18
Xref: DeadDuck misc-news:12
X-Gnus-Article-Number: 12   Mon Dec 04 09:26:30 2000
[end original header]

This is the new one...

[start header]
Path: news20.bellglobal.com!news3.bellglobal.com!sunqbc.risq.qc.ca!newsfeed.cwix.com!europa.netcrusader.net!208.184.7.66!newsfeed.skycache.com!Cidera!cyclone.tampabay.rr.com!typhoon.tampabay.rr.com.POSTED!not-for-mail
Sender: bpringlemeir@DeadDuck
Newsgroups: comp.os.vxworks
Subject: Re: integrating network drivers
References: <90eniu$cms$1@news.netmar.com> <3A2AF70B.A74CA02@flashcom.net>
From: Bill Pringlemeir 
Mail-Copies-To: nobody
X-Microsoft-die: Ministry - Everyday is Halloween.
X-die-Microsoft: "http://www.opensource.org/halloween.html"
Message-ID: 
Lines: 18
User-Agent: Gnus/5.0803 (Gnus v5.8.3) Emacs/20.7
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Date: Mon, 12 Feb 2001 13:30:37 GMT
NNTP-Posting-Host: 65.32.75.52
X-Complaints-To: abuse@rr.com
X-Trace: typhoon.tampabay.rr.com 981984637 65.32.75.52 (Mon, 12 Feb 2001 08:30:37 EST)
NNTP-Posting-Date: Mon, 12 Feb 2001 08:30:37 EST
Organization: RoadRunner - TampaBay
Xref: news3.bellglobal.com comp.os.vxworks:47236
[end header]

At least the make the exploder look smart...

regards,
Bill Pringlemeir

- -- 
BOMBED BY IRC SERVER 12W 23N. SINKING. U-950.

---------------------------

Newsgroups: comp.os.vxworks
Subject: SM polling on PPC
Date: Mon, 12 Feb 2001 12:58:53 -0500
From: "Tom Dropka" 
Organization: TEK Microsystems
Message-ID: <5uVh6.27625$OD.10740775@typhoon.snet.net>

We have implemented the shared memory network on a dual-PPC board, but
unfortunately we don't have VME interrupts on the secondary CPU.  We've
gotten SM polling working, but we've found that the tsmPollTask is hitting
the master board's memory every 5 microseconds or so.  Since the VME
controller is on the PCI bus, the PCI performance is getting killed.

Has anybody seen this or know of a fix or workaround?

Thanks,

Tom Dropka
TEK Microsystems




---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: dynamically adding PCI space for vxworks
Date: Mon, 12 Feb 2001 19:03:01 +0000
From: James Marshall 
Organization: Agilent Technologies UK Ltd
Message-ID: <3A883365.384968AE@agilent.com>
References: <9094b5$ah3$1@overload.lbl.gov> <3A2A2327.105A99A8@club-internet.fr>

Try vmBaseStateSet(NULL, address, len, stateMask, state).


 must be page aligned and  must be a multiple of the page size,
which limits the usefulness of mapping small PCI address ranges.

You'll need to unmap (state will be VM_STATE_VALID_NOT) the address range for the
removed card as well.

sysMmuMapAdd() is only meant to be used before usrMmuInit() is called.

James.


Sebastien Fauris wrote:

> Hi Raj,
>
> try using
>
> STATUS sysMmuMapAdd(void *address, UINT len, UINT initialStateMask,
>                     UINT initialState);
>
> defined in sysLib.c. Let me know if it works for you.
> I meet some problems trying to use it (I still have page faults while
> accessing my PCI registers after I called sysMmuMapAdd).
>
> Regards,
>
> Sebastien.
>
> rajat.bandhu@conexant.com wrote:
>
> > dynamically adding PCI space for vxworks
> >
> > We are using a pentium based system using Vxworks where we need to
> > dynamically add PCI space as we detect PCI cards. I am aware for static
> > addition of PCI space we need to  simplify modify the sysPhysMemDesc
> > structure in sysLib.c. This table tells the MMU in the chip which pages are
> > accessible.And then usrRoot which gets called in the context of kernel does
> > the memory initisations(usrMmuInit et al).
> >
> > But how do we implement addition of PCI space on a dynamic basis i.e. from
> > hotswap standpoint where the system has booted and vxworks kernel has been
> > launched and the application code started and now we plug in a PCI card
> > into the chasis/system and want to update the MMU with respect to this
> > newly plugged in card.
> >
> > Any advice/pointers is appreciated.
> >
> > regards,
> > raj


---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: *UPDATED* Gcc 2.95.3 for VxWorks-PPC available....
Date: Mon, 12 Feb 2001 13:48:13 -0600
From: Kevin Cline 
Organization: Mayan Networks
Message-ID: <3A883DFD.26386D7D@mayannetworks.com>
References:  <3A8462D3.1AADFB7C@mayannetworks.com> 

Dave Korn wrote:

> "Kevin Cline"  wrote in message news:3A8462D3.1AADFB7C@mayannetworks.com...
> > Dave Korn wrote:
> >
> > >  Finally I've recompiled the whole thing for i386, so it should run on any
> > > CPU!
> >
> > Can you document the process so that others can repeat your work for different
> > targets?  Thanks.
> >
>
>    There's already an outline of the build procedure on the FAQ/Docs page that should work fine for any other VxWorks targets; just
> change the --target= parameter in the configure commands for the cross binutils and cross compiler.
>
>          DaveK

Thanks.  I'll give it a shot.

Kevin Cline


---------------------------

Newsgroups: comp.os.vxworks
Subject: MIPS Evaluation Board.
Date: Mon, 12 Feb 2001 11:23:23 -0800
From: Joseph Sebastian 
Organization: Lawrence Berkeley National Laboratory
Message-ID: <969g9r$lk0$1@overload.lbl.gov>

Hi all,
I am looking for a MIPS evaluation board with PCI interface.
Is there a source from where I can get an evaluation board with VxWorks
BSP.
Joseph Sebastian.



---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: Problem booting vxworks from DosFs
Date: Mon, 12 Feb 2001 21:07:19 GMT
From: andreyk@home.com (Andray Kaganovsky)
Organization: @Home
Message-ID: 
References: <3A87FD55.AD34FD82@kayser-threde.de>

Torsten Levin  wrote in <3A87FD55.AD34FD82@kayser-
threde.de>:

>Hello to all,
>
>i have a problem booting from a PCMCIA flash card on a PowerPC target.
>Host is Tornado 2.0 on a Windows NT 4.0 Pentium PC.
>
>The code inside the boot loader does the following:
>
>1. Initialize PCMCIA hardware (third party driver function)
>   Return: OK
>2. generating PCMCIA block device (third party driver function)
>   Return: OK
>3. initializing DOS FS (dosFsInit)
>   Return: OK
>4. Attaching DOS file system to pcmcia block device (dosFsDevInit)
>   Return: ERROR
>   errno = 0x3d0001 = VOLUME_NOT_AVAILABLE
>
>The absolute identical code packed inside a "test" function and running
>on the shell after booting the system via Ethernet works without any
>problem. The PCMCIA card can be fully accessed.
>
>Any ideas?
>Thank you in advance!
>

[AndrayK]. Are you sure you have enough RAM available to your application
at that point ? 

    regards, Andray 


---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: building with TrueFFS
Date: Mon, 12 Feb 2001 21:39:36 +0000
From: molochai 
Message-ID: <7fd3db4b4a%molochai@vapour-trail.demon.co.uk>
References: <90jmkq$pmm$1@nnrp1.deja.com> <3A2DC050.1030004@flashcom.net>

In message <3A2DC050.1030004@flashcom.net>
          DrDiags  wrote:

> The file tfllite.o which should have been part of your 
> lib80486(TOOL)(ENV) , contains these two functions. You could look on 
> your system for it. The fact that it is missing could mean that you have 
> the -00 TDK release of your architecture. Ask your account manager for 
> the -01 release of the Intel x86 CD. Here is a nm of the file in my 
> 40486 obj directory:
> 
[snip]

I've just encountered this one too. [waiting for support to get back to me..
at least I know what it is now ;-)].

Does pose the question: Why isn't a problem which is evidently widespread on
Windsurf? (Or did I just not find it?)


---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: RouteLookup function
Date: Mon, 12 Feb 2001 21:54:13 +0000
From: molochai 
Message-ID: <252add4b4a%molochai@vapour-trail.demon.co.uk>
References: <90gmtj$p26$1@overload.lbl.gov>

In message <90gmtj$p26$1@overload.lbl.gov>
          "Komuraiah Poodari"  wrote:

> Hi,
> 
> Is there any function in VxWorks SENS stack to perform routeLookup,  i.e.
> to find out the interface corresponding to given  destination address ?
> 
Might be worth getting hold of the BSD net/3 source code as the VxWorks SENS
stack is apparently a derivative. Odds are the function names haven't changed
(much). 

TCP/IP Illustrated Vol. II (W.Stevens) provides a thorough description of
net/3 and is probably essential reading for this sort of stuff.

---------------------------

Newsgroups: comp.os.vxworks
Subject: CppUnit port to VxWorks
Date: Mon, 12 Feb 2001 17:52:40 -0500
From: Fake 
Organization: UUNET Canada News Reader Service
Message-ID: <3A886938.E0BA12C8@address.com>

Hi All,

Is anyone out there managed to port it (version 1.5) successfully to
this OS? CppUnit is a testing framework made popular by
http://www.xprogramming.com/software.htm and was created by Michael
Feathers. I couldn't even get it to compile.

TIA.

Cheers


---------------------------

Newsgroups: comp.os.vxworks
Subject: documenation for configAll.h?
Date: Mon, 12 Feb 2001 14:49:37 -0800
From: "Ryan Tanner" 
Organization: Posted via Supernews, http://www.supernews.com
Message-ID: 

What is the best source of information about the VxWorks configuration
macros in configAll.h and config.h?

For instance, but not limited to:

#define INCLUDE_ENV_VARS /* unix compatable environment variables */


Thanks



---------------------------

Newsgroups: comp.os.vxworks
Subject: Problem with NFS
Date: Mon, 12 Feb 2001 23:16:26 GMT
From: "Vasilis Papadakis" 
Organization: EMS Technologies Canada Ltd.
Message-ID: 

Hello,

We are experiencing a problem while trying to use the NFS library. In the
following lines I will try to give a picture of our setup, and also
described the problem that we are facing.

1. We have configured an NFS Server (using Hummingbird NFS Server version
7.00) running on a Windows NT Workstation.
2. From a PowerPC 604 board, running vxWorks we run a process which tries to
generate a file in the NFS server.
  i) If the NFS server is configured correctly, the process of creating and
saving
     the file is successful.
  ii) During our tests, we change NFS servers (changing IP Address). From a
      valid one, in the sense that we were able to create and save a file,
to
      another one not correctly configured. As expected the process of
creating
      and saving a file failed. When we try to switch back to the previously
      working server, the process of creating and saving a file failed. The
only
      way to get out of that situation was by rebooting the board.

Is there a way to avoid the reboot of the board? What remains locked after a
file creation failure?

Thanks.

- - Vasilis



---------------------------

Newsgroups: comp.os.vxworks
Subject: compiler differences between T1 and T2?
Date: Mon, 12 Feb 2001 15:57:25 -0800
From: "Ryan Tanner" 
Organization: Posted via Supernews, http://www.supernews.com
Message-ID: 

I'm using the same flags I used for x86 with the new compiler, and I'm
getting a bigger object file than I used to. Have other people experienced
the same trouble?

Thanks



---------------------------

Newsgroups: comp.os.vxworks
Subject: steve.konish@windriver.com
Date: Tue, 13 Feb 2001 00:20:51 GMT
From: Craig Vanderborgh 
Organization: Excite@Home - The Leader in Broadband http://home.com/faster
Message-ID: <3A887DE3.DC67C42F@voxware.com>

Hi steve:

Could you please tell me if Tornado for ARM v2.01 will function
on Solaris 7? 

Thanks!
craig

---------------------------

Newsgroups: comp.os.vxworks
Subject: Tornado on Solaris 7?
Date: Tue, 13 Feb 2001 00:24:41 GMT
From: Craig Vanderborgh 
Organization: Excite@Home - The Leader in Broadband http://home.com/faster
Message-ID: <3A887EC9.2F2773E5@voxware.com>

Hi VxFolks:

Is it possible to use Tornado 2.01 on Solaris 7/SPARC?

Please advise.

Thanks in advance!
craig vanderborgh
voxware incorporated

---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: Error during using JavaCodeCompact
Date: Tue, 13 Feb 2001 00:32:42 GMT
From: Craig Vanderborgh 
Organization: Excite@Home - The Leader in Broadband http://home.com/faster
Message-ID: <3A8880A9.B500B0B8@voxware.com>
References: <90fosm$ip5$1@nnrp1.deja.com>

Hi Toby:

I can help you with this one.  I have been trying to get JavaCodeCompact
(JCC) to work on Tornado/ARM for several *years* now.  Many TSR's have been
filed, and Wind River knows all about the problems.  However, they have
not responded to the JCC bugs and are most likely not going to.  The
failure I was getting the last time I tried JCC on our ARM platform 
(about 2 months ago) looks just like yours.  JavaCodeCompact DOES NOT WORK.
It is another WRS "virtual feature", designed to get you to buy JWorks.  
The trick worked on us, and perhaps on you too...

regards,
craig vanderborgh
voxware incorporated.

tobyhage@my-deja.com wrote:
> 
> We want to use JavaCode compact. When we configure the Java classes
> with "complete ROMmized configuration" and try to start java -version
> or a java application we receive the following error messages in the
> shell:
> 
> -> javaConfigShow()
> ------------------------
> Native libraries         file zip net awt jpeg math
> ------------------------
> java.class.path          /java/lib/classes.zip:/java/lib/wrs/jar
> java.home                /java
> java.user.name           target
> java.user.home           /java
> java.user.dir            /java
> path.separator           :
> file.separator           /
> ------------------------
> Native stack size        32768 bytes
> Max Java heap size       33554432 bytes
> Stack overflow checks    FAST (minimal)
> Java thread semaphores   priority inversion safe
> Native library loading   DISABLED
> ------------------------
> Idle thread frequency    10 Hz
> Async GC period          1 seconds
> Async GC threshold       99 %
> ------------------------
> On VxWorks exception     ABORT VM
> Java VM restart          ENABLED
> JNI arg init hook        NULL
> Instrumentation hook     _javaInstrumentHook (DISABLED)
> ------------------------
> awt.toolkit
> windriver.awt.chameleon.ChameleonToolkit
> sun.graphicssystem               sun.awt.gfU.GraphicsSystem
> sun.graphicssystem.width         800
> sun.graphicssystem.height        600
> sun.awt.gfU.pixelType            color:8
> value = 0 = 0x0
> 
> -> java -version
> 
> Exception number 16: Task: 0x3f85bb0 (t27)
> 
> Data abort
> Program Counter:   0x17399c
> Status Register:   0xa0000013
> d67bc    _vxTaskEntry   +28 : 325b60 ()
> 325c04   _wdbFuncCallLibInit+128: _java ()
> 173bbc   _java          +14 : _javaWait ()
> 173b18   _javaWait      +1c : 173950 ()
> value = 0 = 0x0
> -> java -version
> 
> Exception number 16: Task: 0x3f80a04 (t28)
> 
> Data abort
> Program Counter:   0x17399c
> Status Register:   0xa0000013
> d67bc    _vxTaskEntry   +28 : 325b60 ()
> 325c04   _wdbFuncCallLibInit+128: _java ()
> 173bbc   _java          +14 : _javaWait ()
> 173b18   _javaWait      +1c : 173950 ()
> value = 0 = 0x0
> 
> -> java("-verbose HelloWorld")
> value = 0 = 0x0
> ->
> Exception number 16: Task: 0x3ee7fd4 (tJmain)
> 
> Data abort
> Program Counter:   0xf3ccc
> Status Register:   0x60000013
> 0                           : ???? ()
> 0                           : _ROMPureMethods ()
> 243044   _ROMPureMethods+310e4: .text2 ()
> 302e94   .text2         +893c: _ExecuteJava ()
> 101190   _ExecuteJava   +37f0: _excEnterUndef ()
> 352f14   _excEnterUndef +9c : _excExcContinue ()
> 323214   _excExcContinue+94 : _excExcHandle ()
> 322f94   _excExcHandle  +174: 32faa8 ()
> 32fad4   _sigPendDestroy+2a8: 32fa30 ()
> 32fa88   _sigPendDestroy+25c: _sigPendKill ()
> 32f7e4   _sigPendKill   +10c: 32f380 ()
> 32f6a0   _sigqueue      +938: 32efc8 ()
> 32f0f4   _sigqueue      +38c: 177994 ()
> 177a08   __jhpiThreadUnlock+3a8: _jhpiAbort ()
> 1794c8   _jhpiAbort     +9c : _javaDumpThreads ()
> fd47c    _javaDumpThreads+64 : _jhpiThreadEnumerateOver ()
> 178c54   _jhpiThreadEnumerateOver+60 : _DumpThreadsHelper ()
> fd34c    _DumpThreadsHelper+20 : _threadDumpInfo ()
> fd1d4    _threadDumpInfo+38 : _jio_fprintf ()
> 120dfc   _jio_fprintf   +18 : _jio_vfprintf ()
> 120e20   _jio_vfprintf  +1c : _vfprintf ()
> 33e580   _vfprintf      +148: _fflush ()
> 335c64   _fflush        +70 : ___sflush ()
> 335d34   ___sflush      +c4 : ___swrite ()
> 35b070   ___swrite      +4c : _write ()
> e9e74    _write         +c  : _iosWrite ()
> 3307a4   _iosWrite      +b0 : _tyWrite ()
> 350d18   _tyWrite       +a4 : 351670 ()
> 3516b0   _tyIRd         +668: 3260c8 ()
> 3260d8   _wdbVioDrv     +200: _wdbEventPost ()
> 323ff0   _wdbEventPost  +80 : _wdbNotifyHost ()
> 31cc20   _wdbNotifyHost +34 : 322a60 ()
> 322a80   _udpRcv        +26c: 466c8 ()
> 
> and the following on the output screen:
> 
> Usage: (null) [-options] class
> 
> where options include:
>     -help             print out this message
>     -version          print out the build version
>     -v -verbose       turn on verbose mode
>     -debug            enable remote JAVA debugging
>     -asyncgc          allow asynchronous garbage collection
>     -verbosegc        print a message when garbage collection occurs
>     -noclassgc        disable class garbage collection
>     -ss       set the maximum native stack size for any thread
>     -oss      set the maximum Java stack size for any thread
>     -ms       set the initial Java heap size
>     -mx       set the maximum Java heap size
>     -mr       set the red heap reserve size
>     -my       set the yellow heap reserve size
>     -D=  set a system property
>     -classpath 
>                       list directories in which to look for classes
>     -verify           verify all classes when read in
>     -verifyremote     verify classes read in over the network [default]
>     -noverify         do not verify any class
> 
> Data abort
> Exception address: 0x0017399c
> Current Processor Status Register: 0xa0000013
> Task: 0x3f85bb0 "t27"
> 
> Data abort
> Exception address: 0x0017399c
> Current Processor Status Register: 0xa0000013
> Task: 0x3f80a04 "t28"
> SIGSEGV occured.
> 
> Full thread dump:
>     "main" (TID:0x1edffe0, sys_thread_t:0x3ef4288, taskId:0x3ee7fd4
> state:runnable name:tJmain, stack_base:0x3ee7fd4, stack_high:0xee0
> stack_size:32092) prio=5 *current thread*
>         java.lang.System.initializeSystemClass()
>     "Finalizer thread" (TID:0x1ee0010, sys_thread_t:0x3eeb898,
> taskId:0x3ef38f0 state:runnable name:Finalizer threa,
> stack_base:0x3ef38e8, stack_high:0x138 stack_size:32784) prio=1
> 
> Monitor Cache Dump:
> Registered Monitor Dump:
>     Thread queue lock: 0x3ef42f0 0x0 n 00 m 0x3f7ad10 b
> 0x3f7acec #e 0
>     Name and type hash table lock: 0x3ef434c 0x0 n 00 m
> 0x3f7ad58 b 0x3f7ad34 #e 0
>     String intern lock: 0x3ef439c 0x0 n 00 m 0x3f7ada0 b
> 0x3f7ad7c #e 0
>     JNI pinning lock: 0x3ef4bd8 0x0 n 00 m 0x3f7ade8 b
> 0x3f7adc4 #e 0
>     JNI global reference lock: 0x3ef4c30 0x0 n 00 m 0x3f7ae30
> b 0x3f7ae0c #e 0
>     BinClass lock: 0x3ef4c7c 0x0 n 00 m 0x3f7ae78 b 0x3f7ae54
> #e 0
>     Class loading lock: 0x3ef4ccc 0x0 n 00 m 0x3f7aec0 b
> 0x3f7ae9c #e 0
>     Java stack lock: 0x3ef4d18 0x0 n 00 m 0x3f7af08 b
> 0x3f7aee4 #e 0
>     Code rewrite lock: 0x3ef4d68 0x0 n 00 m 0x3f7af50 b
> 0x3f7af2c #e 0
>     Heap lock: 0x3ef4db0 0x0 n 00 m 0x3f7af98 b 0x3f7af74 #e 0
>     Has finalization queue lock: 0x3ef4e08 0x0 n 00 m
> 0x3f7afe0 b 0x3f7afbc #e 0
>     Finalize me queue lock: 0x3ef4e5c 0x0 n 00 m 0x3f7b028 b
> 0x3f7b004 #e 0
>     Monitor registry: 0x3ef8100 0x3ee7fd4 n 01 m 0x3f7b998 b
> 0x3f7b974 #e 0
> 
> Data abort
> Exception address: 0x000f3ccc
> Current Processor Status Register: 0x60000013
> Task: 0x3ee7fd4 "tJmain"
> 
> what could this be??
> 
> with kind regards,
> 
> Toby Hage
> 
> Sent via Deja.com http://www.deja.com/
> Before you buy.

---------------------------

Newsgroups: comp.os.vxworks
Subject: Problem with Vxworks Timers/Signals
Date: 13 Feb 2001 01:29:02 GMT
From: mrmcsnoots@aol.com (Mrmcsnoots)
Organization: AOL http://www.aol.com
Message-ID: <20010212202902.12937.00001506@ng-cg1.aol.com>

All,

I need to create a 1 second periodic timer using Vxworks timers which are
delivered through the signal mechanism.  I have written the follow test code. 
I never see my callback function, timer(), executed.  Any ideas?

Thanks,
Paul

static void timer(int arg)
{
    static int i=0;

    printf("\nTimer expired %d", ++i);
}

void doTimers(void)
{
    timer_t timer_id;
    sigset_t sigset, osigset;
    struct itimerspec itv;

    if ((timer_create (CLOCK_REALTIME, NULL, &timer_id) < 0) ||
        (timer_connect(timer_id, (VOIDFUNCPTR) timer, 0) < 0))
    {
        notifyExit();
    }
    itv.it_interval.tv_sec = 1;
    itv.it_value.tv_sec = 1;
    itv.it_interval.tv_nsec = 0;
    itv.it_value.tv_nsec = 0;
    timer_settime(timer_id, CLOCK_REALTIME, &itv, NULL);    
    // Setup signals
    sigemptyset(&sigset);
    sigaddset(&sigset, SIGALRM);
    sigprocmask(SIG_SETMASK, &sigset, NULL);
}

---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: Linker error: undefined reference to `__udivdi3
Date: Tue, 13 Feb 2001 01:59:03 GMT
From: "Eugene Zlobin" 
Organization: EarthLink Inc. -- http://www.EarthLink.net
Message-ID: 
References:  <3A84A095.6D1D8837@home.com>

Hello,
It is really good advice. Thanks a lot !!!
But there are some modification in my case.
I couldn't use libgcc.a from cygnus directory (loader can't recognize object
format) and instead of use path
\Tornado\host\x86-win32\lib\gcc-lib\i386-pc-mingw32\egcs-2.90.29\libgcc.a
 and second one
I used macros called EXTRA_MODULES. It works fine.

Best regards,
Eugene.


"Randy MacLeod"  wrote in message
news:3A84A095.6D1D8837@home.com...
>
>   You're missing software support for 64 bit integer
> operations like division. Support comes with vxWorks but in a .a file.
>
> Here's my recipe to make a .o file which you can load.
>
> On unix:
> mkdir gnumath
> cd gnumath
> ar x /path/to/Tornado/host/sun4-solaris2/\
>  lib/gcc-lib/powerpc-wrs-vxworks/cygnus-2.7.2-960126/libgcc.a
> rm _pure.o
> ldppc -r -X -N -o /tmp/gnu64bitMath.o *.o
>
> then I booted my standard ppc vxWorks kernel.
> ->
> -> ld -> ld no unresolved symbols and my 64bit division test code works!
>
> N.B. I got rid of _pure.o because  it contained a function
> _exit that the loader complained about and _exit was not
> called by any of the other .o files. I think its some Unix thing.
>
> There's an EXTRA_OBJECTS option in the kernel compile procedure
> that would link gnu64bitMath.o  into the kernel but I haven't
> done that yet.
>
> // Randy
>
> Eugene Zlobin wrote:
> >
> > Hello,
> >
> > I got two linker errors :
> > undefined reference to `__udivdi3
> > undefined reference to `__divdi3
> >
> > How i can guess it is related with float point support but I don't know
> > how to solve this problem.
> >
> > I use Tornado 2 (VxWorks 5.4) and Windows 2000 simulator.
> > Did anyone face same problem and what could be the solution for it?
> >
> > Thanks,
> > Eugene.
>



---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: VxWorks for Linux emulation
Date: Mon, 12 Feb 2001 19:53:05 -0600
From: Andrew Johnson 
Organization: APS, Argonne National Laboratory
Message-ID: <3A889381.9FECD2C1@aps.anl.gov>
References: <90hvri$cs6$1@nnrp1.deja.com>

bsjung8472@my-deja.com wrote:
> 
> Anyway, I found the pSOS2Linux project
> for pSOS emulation in Linux environment...
> 
> Do anybody know similar project for VxWorks?

Yes, and it may have been with the same distributor as psos2linux.  If not
try MonteVista or the Hard Hat Linux guys.  I'm sure you could find it
with a Google search.

- - Andrew
- -- 
Every great idea appears crazy to start with.

---------------------------

Newsgroups: comp.os.vxworks
Subject: How can I get the etheraddr from end driver send?
Date: Tue, 13 Feb 2001 09:59:07 +0800
From: "tanghaoyu" 
Organization: Bentium Ltd.
Message-ID: <96a4fh$psg$1@mail.cn99.com>

Hello All,

    I am writing an end driver for CompactPCI,but I don't know how to get
the physical address from xxEndSend(). Who can help me?

Tanghy



---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: What causes memPartFree to call taskSuspend?
Date: Tue, 13 Feb 2001 02:27:38 -0000
From: "Dave Korn" 
Organization: Lumber Cartel (tinlc) Members #2234-2237 (owing to browser refresh)
Message-ID: 
References: <3A87DCAF.4E187C6@agilent.com>


"Iain Gourlay"  wrote in message news:3A87DCAF.4E187C6@agilent.com...
> The scenario appears simple enough - C++ code creating object instances during initialisation of an
> application running in the main thread on Pentium architecture. Observations so far,
>
> - This is not a case of a double delete.
> - The heap is pretty clean and not fragmented.
> - memShow shows plenty of space and subsequent heap based mallocs and frees are ok.
> - None of the other tasks appear to be hogging the memory allocation and freeing library.
> - The task stack high water mark is good.
>
> The hang-up appears to be dependent on the size of the call stack.
>
> 24e886 __builtin_delete+42 : free ([3ffcda4, 38a0e44, 38a0d98, 3e86168, 38a0e0c])
> 1e95e8 free           +c  : memPartFree ([2a2f98, 3ffcda4, 38a0e24, 24e88b, 3ffcda4])
> 1e9204 memPartFree    +f8 : taskSuspend (0)

  If it aint a double delete, something else is trashing the heap.

         DaveK



---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: loadLib, loadModule....
Date: Tue, 13 Feb 2001 03:54:00 GMT
From: Pierre Frigon 
Organization: Excite@Home - The Leader in Broadband http://home.com/faster
Message-ID: <3A8890B3.B42C3264@yahoo.com>
References: <3A87E293.2DB2BF2@siemenscomms.co.uk>

Look at vxWorks AE. It can handle this issue just the way you want.

Owain Phillips wrote:

> Hi comp.os.vxworks,
>
> I am working on a new product that has numerous varients and each of
> those variants may have  many functional areas that are common to
> other variants.
>
> Could we use a modular approach not only in development, but also in
> the end product where a load-list specifies lists of loadable modules
> that
> are loaded  from a filesystem before starting the configured system.
>
> How does/could this hang together as regards cross references?
> Must we ensure that each module is built to ensure that after its load
> we have no undefined references?
>
> I could see that such an approach could work, when we develop using
> windshell for incremental development as windriver describes; but what
> about those awkward bought in modules where we don't have finger in
> the pie?
>
> If any of you have experience of using the loadLib and such a modular
> load list I would be very interested to hear your views.
>
> Kind regards,
> Owain Phillips


---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: How to calculate Task stack space?
Date: Tue, 13 Feb 2001 04:07:08 GMT
From: Pierre Frigon 
Organization: Excite@Home - The Leader in Broadband http://home.com/faster
Message-ID: <3A8893C7.19233322@yahoo.com>
References: <200011301401.GAA03645@mail22.bigmailbox.com> <90gtk7$nfm$1@neptunium.btinternet.com> <83rX5.611$4o3.510590@newsr1.u-net.net>

That would imply the greenhill utility knows the worst case stack usage of scanf
and printf which it does not. It would also imply that the greenhill would be
smart enough to see that the call to printf is actually calling the write method
of the serial port device driver which it cannot.

Dave Korn wrote:

> David Laight wrote in message <90gtk7$nfm$1@neptunium.btinternet.com>...
>
> >Years ago I wrote a program that parsed the object file listing (from a
> >PASCAL compiler) which tracked the stack depth in each routine at which it
> >called other routines.  This let me calculate a maximum stack usage for the
> >high level routines.  Maximum stack use was often in obscure error
> >paths......
>
>   Hey, could you do me a favour, and run it over a PASCAL
> conversion of this code please?
>
> int do_test(int, int);
>
> int main(void)
> {
> int x, n;
>     n = scanf(" %d", x);
>     if (n)
>         printf("The answer is %d\n", do_test(x, 0));
>     return EXIT_SUCCESS;
> }
>
> int do_test(int n, int d)
> {
>     if (n == 1) return d;
>     n = (n & 1) ? 3 * n + 1 : n / 2;
>     return do_test(n, d + 1);
> }
>
>     ;-)
>         DaveK
> --
> They laughed at Galileo.  They laughed at Copernicus.  They laughed at
> Columbus. But remember, they also laughed at Bozo the Clown.


---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: Problem with Vxworks Timers/Signals
Date: Tue, 13 Feb 2001 04:20:59 GMT
From: Pierre Frigon 
Organization: Excite@Home - The Leader in Broadband http://home.com/faster
Message-ID: <3A889706.663D1F5C@yahoo.com>
References: <20010212202902.12937.00001506@ng-cg1.aol.com>

The reason that you timer is not executing is probably because the task that
created the timer no longer exists. If you must use signals then you will have to
look at this section in the vxWorks Programmers guide and read it VERY carfully.

You might also want to give this a shot:

void myTaskCode()
{
FOREVER
    {
    taskDelay(sysClkRateGet()); /* Block for 1 second */
    do_my_thing();
    }
}

void mainLoop
{
taskSpawn(...myTaskCode...);
}

Mrmcsnoots wrote:

> All,
>
> I need to create a 1 second periodic timer using Vxworks timers which are
> delivered through the signal mechanism.  I have written the follow test code.
> I never see my callback function, timer(), executed.  Any ideas?
>
> Thanks,
> Paul
>
> static void timer(int arg)
> {
>     static int i=0;
>
>     printf("\nTimer expired %d", ++i);
> }
>
> void doTimers(void)
> {
>     timer_t timer_id;
>     sigset_t sigset, osigset;
>     struct itimerspec itv;
>
>     if ((timer_create (CLOCK_REALTIME, NULL, &timer_id) < 0) ||
>         (timer_connect(timer_id, (VOIDFUNCPTR) timer, 0) < 0))
>     {
>         notifyExit();
>     }
>     itv.it_interval.tv_sec = 1;
>     itv.it_value.tv_sec = 1;
>     itv.it_interval.tv_nsec = 0;
>     itv.it_value.tv_nsec = 0;
>     timer_settime(timer_id, CLOCK_REALTIME, &itv, NULL);
>     // Setup signals
>     sigemptyset(&sigset);
>     sigaddset(&sigset, SIGALRM);
>     sigprocmask(SIG_SETMASK, &sigset, NULL);
> }


---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: Shared memory network problems
Date: Tue, 13 Feb 2001 08:37:43 +0100
From: Marc Waber 
Organization: STS Switching Test Solutions
Message-ID: <3A88E447.C5B7C9EA@stest.ch>
References: <3a2c0d56.3444883@news.ic24.net>

Mark Lamb wrote:
> 
> Hi,
> 
> I'm currently having problems using TCP sockets over the shared memory
> network. I get panics from (usually) tNetTask and eventually mbufShow
> shows corruption of its mbuf statistics (the totals don't agree). This
> is often accompanied by a fail light coming on on my slot 0
> controller, a National Instruments VXIcpu-030 (68030-based). I'm
> currently running vxWorks 5.1.1 (can't change at the moment).
> 
> I have a couple of questions:
> 
> 1. Has anyone else had problems using the shared memory network?
> 
> 2. Has anyone developed or know of an alternative implementation of
> TCP sockets which doesn't use the shared memory?
> 
> Thanks,
> 
> Mark Lamb
We 've had tons of problems with shared memory.
Have you tried to increase your network stack?
There is a product soultion document concerning this from WIndRiver:
SENS, T2:vxxWorks Network tack Configuration/Performance

Hope thos helps,
Marc

---------------------------

Newsgroups: comp.os.vxworks
Subject: GDI rendering for VxWorks
Date: Tue, 13 Feb 2001 07:38:51 GMT
From: lourens@mecalc.co.za (Lourens Geldenhuys)
Organization: Mecalc (Pty) Ltd
Message-ID: <3a88e2ee.349109833@modem.mecalc>
Reply-To: lourens@mecalc.co.za

HI,

Does anybody know if there exists (from WRS or 3rd party) any GDI
rendering capability for VxWorks. The specific application calls for
graphic images to be printed to a GDI printer (PENTAX PocketJet II).

Lourens

================================================================
Lourens Geldenhuys                Tel:        +27 (0)12 665 1480
Project Engineer                  Fax:        +27 (0)12 665 1495
Mecalc (Pty) Ltd                  e-mail:   lourens@mecalc.co.za
86 Oak Avenue, Highveld Technopark, Centurion, South Africa
================================================================

---------------------------

Newsgroups: comp.os.vxworks
Subject: MVME2304 w/ VxWorks 5.3.1 is slow reading VMEbus
Date: Mon, 12 Feb 2001 23:49:42 -0800 (PST)
From: Bob Kibrick 
Organization: Lawrence Berkeley National Laboratory
Message-ID: <200102130749.XAA25653@celeste.ucolick.org>

We are currently conducting tests on a Motorola MVME2304-0133 board that is
running VxWorks version 5.3.1.  We are rather surprised to find that on
certain tests of VMEbus throughput, the MVME2304 (which has a 333 MHz.
PowerPC 604e) delivers significantly lower performance than an obsolete
Force 5CE board (which has an 85 MHz. SPARC processor running VxWorks 5.2).

The particular test in question involves transferring approximately 145 MB
of data between a bulk VME memory board (Chrislin CI-VME64 256 MB) and the
local DRAM on the CPU board.  This test was performed in both directions,
and using both D16 and D32 transfer widths.  The identical C program was
used on both processors, compiled (or cross-compiled) using gcc (2.9.5 for
Sparc, cygnus-2.7.2 for PowerPC) on a Sparc Solaris 2.6 host.  The same
VME memory board was used in all tests.  The tests were conducted on an
otherwise idle system with no other competing processes (the tests were 
run in the context of the VxWorks shell) nor any concurrent activity on
the VMEbus.

The transfers were coded as a tight loop of the form:

	while( vmeptr < end_address )
		localmem = *vmeptr++;

for transferring from VME memory to local DRAM, and

	while( vmeptr < end_address )
		*vmeptr++ = localmem;

for the reverse direction.

On the PowerPC, the read loop for the 32-bit case (int32 *vmeptr) assembles to:

.L4:
        lwz 0,0(31)
        addi 31,31,4
        cmplw 1,31,29
        bc 12,4,.L4

On the SPARC, the corresponding read loop assembles to:

L4:
        cmp %l0,%l1
        blu L6
        nop
        b L5
        nop
L6:
        ld [%l0],%o0
        st %o0,[%fp-24]
        add %l0,4,%l0
        b L4
        nop
L5:

For single cycle accesses, the Chrislin VME board has a 90ns access time
for writing, and 140ns access time for reading.

The results (elapsed time required to transfer 145 MB) are as follows:

Writing from CPU DRAM into VMEbus          Reading from VMEbus into CPU DRAM
- ----------------------------------         ----------------------------------
MVME2304,  32-bit transfers: 15.1s	   MVME2304,  32-bit transfers: 39.6s
Force 5CE, 32-bit transfers: 21.1s	   Force 5CE, 32-bit transfers: 29.4s

MVME2304,  16-bit transfers: 30.2s	   MVME2304,  16-bit transfers: 79.3s
Force 5CE, 32-bit transfers: 42.3s	   Force 5CE, 16-bit transfers: 58.9s

Thus, while the Force 5CE is about 40% slower than the MVME2304 when writing
to the VMEbus, the MVME2304 is about 35% slower than the Force 5CE when reading
from the VMEbus.  

Furthermore, the MVME2304 takes 2.6 times longer to read from the VMEbus than
it does to to write to it.  But the corresponding read cycle time on the
Chrislin board is only 1.6 times the write cycle time.

We are trying to understand why the MVME2304 performs so slowly when reading
from the VMEbus memory, and are wondering what options, if any, exist for
improving this performance.  For example, are there any alternate settings
of the control registers of the Tundra Universe chip (on the MVME2304) that
might allow for higher throughput on VMEbus reads (We are currently using the 
default values set by VxWorks 5.3.1)?  Alternatively, is there any way from
software (running under VxWorks 5.3.1 on the MVME2304) to initiate either
A32/D32 block transfer read cycles on the VMEbus (i.e., the type of bus
cycles implied by VME address modifier codes 0x0F or 0x0B) or to initiate a
DMA-like transfer from VMEbus memory to local DRAM on the CPU board that
would generate such block transfer read cycles?

Any pointers or suggestions would be greatly appreciated.

Bob Kibrick, University of California Observatories / Lick Observatory
kibrick@ucolick.org	831-459-2262	831-459-2298 (FAX)

---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: documenation for configAll.h?
Date: 13 Feb 2001 08:19:00 GMT
From: borkhuis@agere.com (Johan Borkhuis)
Organization: Lucent Technologies
Message-ID: 
References: 

"Ryan Tanner"  wrote:

>What is the best source of information about the VxWorks configuration
>macros in configAll.h and config.h?
>
>For instance, but not limited to:
>
>#define INCLUDE_ENV_VARS /* unix compatable environment variables */

The file configAll.h only provides a default configuration. Most of the 
defines you have there can also be defined from within Tornado. If you want 
some more info (for example about INCLUDE_ENV_VARS) you open a bootable 
project and go to the menu Project->Find Object. Here you select 
INCLUDE_ENV_VARS and you find were you can include/exclude it and the 
properties of this item. In this case it will point you to envLib.

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: compiler differences between T1 and T2?
Date: 13 Feb 2001 08:20:42 GMT
From: borkhuis@agere.com (Johan Borkhuis)
Organization: Lucent Technologies
Message-ID: 
References: 

"Ryan Tanner"  wrote:

>I'm using the same flags I used for x86 with the new compiler, and I'm
>getting a bigger object file than I used to. Have other people experienced
>the same trouble?

Is the compiler the only difference, or are you using other libraries as 
well?

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: *UPDATED* Gcc 2.95.3 for VxWorks-PPC available....
Date: Mon, 12 Feb 2001 16:23:26 -0000
From: "Dave Korn" 
Organization: Lumber Cartel (tinlc) Members #2234-2237 (owing to browser refresh)
Message-ID: <2e7i6.146$yi4.164057@newsr1.u-net.net>
References:  <3A8462D3.1AADFB7C@mayannetworks.com>  <3A87DEB3.725A6DDB@opq.se>

Roger wrote in message <3A87DEB3.725A6DDB@opq.se>...
>Dave Korn wrote:
>>
>> "Kevin Cline"  wrote in message
news:3A8462D3.1AADFB7C@mayannetworks.com...
>> > Dave Korn wrote:
>> >
>> > >  Finally I've recompiled the whole thing for i386, so it should run
on any
>> > > CPU!
>> >
>> > Can you document the process so that others can repeat your work for
different
>> > targets?  Thanks.
>> >
>>
>>    There's already an outline of the build procedure on the FAQ/Docs page
that should work fine for any other VxWorks targets; just
>> change the --target= parameter in the configure commands for the cross
binutils and cross compiler
>
>I am new to this. Where are the FAQ pages?

  On my website as referred to in that other post I replied to you in.  I
don't want to keep quoting the URL, it makes me feel like a spammer, so
here's a pointer to that post instead:

news:<46Wg6.17$yi4.12215@newsr1.u-net.net>

      DaveK
- --
They laughed at Galileo.  They laughed at Copernicus.  They laughed at
Columbus. But remember, they also laughed at Bozo the Clown.



---------------------------

Newsgroups: comp.os.vxworks
Subject: Learning a new OS [was Re: dialog with WindRiver - reasonable response]
Date: Mon, 12 Feb 2001 16:41:17 -0000
From: "Dave Korn" 
Organization: Lumber Cartel (tinlc) Members #2234-2237 (owing to browser refresh)
Message-ID: <2e7i6.147$yi4.163786@newsr1.u-net.net>
References: <95ur9c$93t$1@overload.lbl.gov>

Jim Way wrote in message <95ur9c$93t$1@overload.lbl.gov>...
>Repeating a complaint (suggestion?) I've made here before, a great start
>would be to create a Master Index that spans all the documentation. Too
>often I see people chastising to "RTFM". Without a Master Index, how do you
>know *which* FM?

  By reading *all* the FMs and familiarising yourself with their contents.
When you first start working under a new OS, you should read, browse and
skim through all the available manuals, to get at least a working view of
what's where.  Borrow them from the office overnight, read them on the bus
home, browse through them on the toilet, flip through the pages in your
lunch hour; steep yourself in the knowledge you are trying to acquire.  I
suppose you could call it 'method' programming.  During the course of all
this you will get a good subconscious knowledge of what's where.  You can
skim through or skip over the bits that are most irrelevant to your job;
you'll still know where to find them if you need them.  And there are things
that you know in advance you'll need to pay attention to, because they're
the same things that are tricky in all OS's and architectures: how do
interrupts work, and what restrictions do they place on system calls?  How
is the scheduling handled?  How do you lock task changes and interrupts out?
Where are the clocks/timers and what are they used for?  Then there are the
things that are pretty much the same between all OS's: semaphores, file
handles, malloc and free; you can assume you already know how these work and
skip fairly quickly through those bits of the manual.  There's no shortcut
for doing your homework.

  All this is of course IMO and YMMV, but it's worked for me across a whole
series of OS's and architectures.  There's a steep learning curve when
beginning work under a new architecture, but if you take it slow for the
first couple of weeks and take the time to deeply familiarise yourself with
the new system, it will more than repay your effort in terms of better effic
iency later on.  Of course, you might be in the situation when your
management throws something new at you and tells you you've got to get it
done in a couple of weeks.  Then you're shafted, and your only real option
is to tell them to choose between having it done well-but-slower, or
on-time-but-shit.

  I'd be interested to hear any other suggestions people have for adapting
to a new and unknown OS and finding out the pitfalls and gotchas.

      DaveK
- --
They laughed at Galileo.  They laughed at Copernicus.  They laughed at
Columbus. But remember, they also laughed at Bozo the Clown.



---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: unknown error message
Date: Mon, 12 Feb 2001 16:50:46 -0000
From: "Dave Korn" 
Organization: Lumber Cartel (tinlc) Members #2234-2237 (owing to browser refresh)
Message-ID: <2e7i6.148$yi4.164049@newsr1.u-net.net>
References: <3a804305.21210562@192.168.1.214> <95rirs$ams$1@nnrp1.deja.com> <3a81779b.6423375@192.168.1.214> <95tm51$7i5$1@nnrp1.deja.com> <3a828ff5.262984@192.168.1.214>

Kim Putnam wrote in message <3a828ff5.262984@192.168.1.214>...
>I am using Tornado 2.0, and VMICPCI-7715.
>
>I've taken out the c++ code from the bootrom. I have put the runtime
>libs on flash, as well as the c++ code. I am now trying to use
>loadModule to load these.  Unfortunately I am getting an error of not
>insalled.  Not sure if I am missing some kind of init function or if
>this is even possible.
>
>K

  Kim, can you be more precise about exactly what you've put in your flash?
Are you referring to .a files?

      DaveK
- --
They laughed at Galileo.  They laughed at Copernicus.  They laughed at
Columbus. But remember, they also laughed at Bozo the Clown.



---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: Problem booting vxworks from DosFs
Date: Tue, 13 Feb 2001 12:14:36 +0100
From: Torsten Levin 
Organization: Colt Telecom GmbH
Message-ID: <3A89171C.E1D66D27@kayser-threde.de>
References: <3A87FD55.AD34FD82@kayser-threde.de> 

Andray Kaganovsky schrieb:
> Are you sure you have enough RAM available to your application
> at that point ?
> 
>     regards, Andray

This should not be the problem, when I use "dosFsMkfs" instead of
"dosFsDevInit", the file system is successfully installed, but then, of
course, all files are lost. So it seems not to be a ressource problem.
There must be any differences in a bootloader and a normal application.

Anybody else out there any ideas?

---------------------------

End of New-News digest
**********************


From vxwexplo-errs@csg.lbl.gov  Tue Feb 13 04:37:41 2001
From: james bond 
Date: Tue Feb 13 04:37:43 PST 2001
Subject: Re: Vx Exploder Digest

Hi all,

This is James from India.
I am new to vxworks. I am interested to get more
information on the same.
If someone has some useful documents relating to
vxworks architecture and fundamentals,kindly send me
the same.
You can also send me the links on the web for various
documents and information on vxworks.

regards
James

__________________________________________________
Do You Yahoo!?
Get personalized email addresses from Yahoo! Mail - only $35 
a year!  http://personal.mail.yahoo.com/


From vxwexplo-errs@csg.lbl.gov  Tue Feb 13 04:37:50 2001
From: james bond 
Date: Tue Feb 13 04:37:52 PST 2001
Subject: Re: Vx Exploder Digest

Hi all,

This is James from India.
I am new to vxworks. I am interested to get more
information on the same.
If someone has some useful documents relating to
vxworks architecture and fundamentals,kindly send me
the same.
You can also send me the links on the web for various
documents and information on vxworks.

regards
James

__________________________________________________
Do You Yahoo!?
Get personalized email addresses from Yahoo! Mail - only $35 
a year!  http://personal.mail.yahoo.com/


From vxwexplo-errs@csg.lbl.gov  Tue Feb 13 06:42:51 2001
From: David Abbott 
Date: Tue Feb 13 06:42:53 PST 2001
Subject: Re: MVME2304 w/ VxWorks 5.3.1 is slow reading VMEbus

On Mon, 12 Feb 2001, the vxWorks Users Group Exploder wrote:
> Submitted-by vxwexplo-errs@csg.lbl.gov  Mon Feb 12 23:50:16 2001
> Submitted-by: Bob Kibrick 
> 
> We are currently conducting tests on a Motorola MVME2304-0133 board that is
> running VxWorks version 5.3.1.  We are rather surprised to find that on
> certain tests of VMEbus throughput, the MVME2304 (which has a 333 MHz.
> PowerPC 604e) delivers significantly lower performance than an obsolete
> Force 5CE board (which has an 85 MHz. SPARC processor running VxWorks 5.2).
> 
> The particular test in question involves transferring approximately 145 MB
> of data between a bulk VME memory board (Chrislin CI-VME64 256 MB) and the
> local DRAM on the CPU board.  This test was performed in both directions,
> and using both D16 and D32 transfer widths.  The identical C program was
> used on both processors, compiled (or cross-compiled) using gcc (2.9.5 for
> Sparc, cygnus-2.7.2 for PowerPC) on a Sparc Solaris 2.6 host.  The same
> VME memory board was used in all tests.  The tests were conducted on an
> otherwise idle system with no other competing processes (the tests were 
> run in the context of the VxWorks shell) nor any concurrent activity on
> the VMEbus.
> 
>
>  ... Parts of original message cut here ...
>
> 
> The results (elapsed time required to transfer 145 MB) are as follows:
> 
> Writing from CPU DRAM into VMEbus          Reading from VMEbus into CPU DRAM
> ----------------------------------         ----------------------------------
> MVME2304,  32-bit transfers: 15.1s	   MVME2304,  32-bit transfers: 39.6s
> Force 5CE, 32-bit transfers: 21.1s	   Force 5CE, 32-bit transfers: 29.4s
> 
> MVME2304,  16-bit transfers: 30.2s	   MVME2304,  16-bit transfers: 79.3s
> Force 5CE, 32-bit transfers: 42.3s	   Force 5CE, 16-bit transfers: 58.9s
> 
> Thus, while the Force 5CE is about 40% slower than the MVME2304 when writing
> to the VMEbus, the MVME2304 is about 35% slower than the Force 5CE when reading
> from the VMEbus.  
> 
> Furthermore, the MVME2304 takes 2.6 times longer to read from the VMEbus than
> it does to to write to it.  But the corresponding read cycle time on the
> Chrislin board is only 1.6 times the write cycle time.
> 
> We are trying to understand why the MVME2304 performs so slowly when reading
> from the VMEbus memory, and are wondering what options, if any, exist for
> improving this performance.  For example, are there any alternate settings
> of the control registers of the Tundra Universe chip (on the MVME2304) that
> might allow for higher throughput on VMEbus reads (We are currently using the 
> default values set by VxWorks 5.3.1)?  Alternatively, is there any way from
> software (running under VxWorks 5.3.1 on the MVME2304) to initiate either
> A32/D32 block transfer read cycles on the VMEbus (i.e., the type of bus
> cycles implied by VME address modifier codes 0x0F or 0x0B) or to initiate a
> DMA-like transfer from VMEbus memory to local DRAM on the CPU board that
> would generate such block transfer read cycles?
> 
> Any pointers or suggestions would be greatly appreciated.
> 
> Bob Kibrick, University of California Observatories / Lick Observatory
> kibrick@ucolick.org	831-459-2262	831-459-2298 (FAX)
> 

  Bob,
    While I cannot speak for the Force 5CE board, I am somewhat familiar
with the MVME2304 board. Your numbers for the 32 bit transfers
reading/writing are right on target with what I have measured with various
VME slave boards that we use.
 
    For 145MB transfer:
         Read :  3.6MBytes/sec  or 1.09 microsec/32bit word
         Write:  9.6Mbytes/sec  or 400  nsec/32bit word         

   (the 16bit reads/writes logically take twice as many cycles)

   The general problem with programmed (CPU Controlled) I/O to and from
the VME bus on the MVME PowerPC Arch boards is that you are actually
having to transfer data over two different buses. To get to the CPU's DRAM
the data must be transfered over VME to the Universe chip, and from the
Universe chip over the PCI bus to the Raven PCI/PowerPC bridge and
finally to DRAM. 
    For the Reads from VME the two buses are coupled and everything is
handshaked until the data is recieved in DRAM. You are helped in the
Writes because (based on the above numbers) you have the Universe chip
programmed to support Posted Writes which means that the CPU is not
waiting for a successful write over the VME bus. This part of the reason
why writes to invalid VME addresses do not generate Bus error exceptions
on this PowerPC Arch.
    In all likelyhood the Force Board is using a different VME interface
bridge that is a direct bridge with the SPARC/memory bus. This is why the
reads are somewhat faster. 

    However, programmed I/O of 145MBytes of data over the VME bus is
crazy. You should definitely use the DMA engine which is integrated in
with the Universe chip. I can send you some example code for implementing
transfers using the DMA engine if you like. But as you suggest with the
block tansfer modes B32 (or in your case with the Chrislin memory boards
they most certainly support MBLK 64 bit transfers) you ought to be able
read data into DRAM at around 30Mbytes/sec or perhaps slightly better. 

             Regards,
              David

---------------------------------------------------------
David Abbott                      Jefferson Lab
Data Acquisition Group            MS 12H
EMAIL: abbottd@jlab.org           12000 Jefferson Ave.
Tel: (757) 269-7190               Newport News, VA  23606
FAX: (757) 269-5800
---------------------------------------------------------



From vxwexplo-errs@csg.lbl.gov  Tue Feb 13 07:14:00 2001
From: Jim Way 
Date: Tue Feb 13 07:14:03 PST 2001
Subject: RE: comp.os.vxworks newsdigest

DaveK wrote:

>             ...Of course, you might be in the situation when your
> management throws something new at you and tells you you've got to get
> it done in a couple of weeks.  Then you're shafted, ...

Dave,

You pretty much pegged the situation. There's never time to do things the
right way. And I agree that reading all the FMs is the best way to go. If
you have the time.

Perhaps I could have made my point more clearly. When a Subject Matter
Expert (SME) is attempting to explain the Subject Matter (SM), the burden of
communication is not on the listener but on the speaker. In this case, the
SM is vxWorks and the SME is WindRiver. The burden is [should be] on them to
provide a clearly defined entry point into the manuals that takes you
through a Quick Start and a survey of basic concepts. From there, it should
point you to the various other manuals with suggestions on how to explore
them. One of the other ways a SME can improve communication is by providing
a Master Index. This is especially important when the SM is as huge as
vxWorks and many of the components can be configured in or out.

One bit of evidence that there is a problem can be found here in the User
Group. Let's take a simple example. How many postings do we see where
someone is asking, "What does this error message mean?" To me, this is an
obvious question, but I'm not sure the manuals are organized to allow me to
answer it. (For comparison, I know that the manual for my product has a
"messages" section where *all* the messages are located along with further
explanation of what they mean. It took time to develop that list, but the
results are worth it.)

Dave, as always your comments are insightful and enlightening. One day I
hope to have the familiarity with vxWorks that you've earned.

Jim
-----------------------------------------
Jim Way, Software Engineer
Datum Austin (Austron Inc.)
voice: 512.721.4170
fax  : 512.990.9712
email: jwayATdatumDOTcom (no spam please)
-----------------------------------------


From vxwexplo-errs@csg.lbl.gov  Tue Feb 13 09:31:11 2001
From: Richard Dickson 
Date: Tue Feb 13 09:31:13 PST 2001
Subject: Re: MVME2304 w/ VxWorks 5.3.1 is slow reading VMEbus

This is a multi-part message in MIME format.
--------------6BA9C2153D603E27871A595A
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit

Bob,
     The attached function will do DMA from A24 space to RAM using D16.  Adjust
the parameters of FB_UNIV_DCTL to suit your needs.  Hope this helps.

Richard

> Alternatively, is there any way from
> software (running under VxWorks 5.3.1 on the MVME2304) to initiate either
> A32/D32 block transfer read cycles on the VMEbus (i.e., the type of bus
> cycles implied by VME address modifier codes 0x0F or 0x0B) or to initiate a
> DMA-like transfer from VMEbus memory to local DRAM on the CPU board that
> would generate such block transfer read cycles?
> 
> Any pointers or suggestions would be greatly appreciated.
> 

-- 
Richard Dickson
Thomas Jefferson National Accelerator Facility
dickson@jlab.org
(757)269-5082
--------------6BA9C2153D603E27871A595A
Content-Type: text/plain; charset=us-ascii;
 name="dma_driver_mv2300.c"
Content-Transfer-Encoding: 7bit
Content-Disposition: inline;
 filename="dma_driver_mv2300.c"

/* Richard Dickson, Jefferson Lab - Code to perform 16 bit DMA transfer (VME_A24->RAM) for mv2306 */

#include 
#include 
#include 
#include 	/* This works for my mv2306 */
#include 
#include 

/* Most of these defs may be found in $VW_BASE/target/config/mv2300/universe.h */
/* Long swap must be done because PCI is little endian */
#define DGCS_RESET_STAT         LONGSWAP(DGCS_STOP|DGCS_HALT|DGCS_DONE|DGCS_LERR|DGCS_VERR|DGCS_P_ERR)
#define FB_UNIV_DGCS            DGCS_RESET_STAT|LONGSWAP(DGCS_GO)

#ifdef DO_BLOCK_XFER
/* Set 16 bit VME, 64 bit PCI, direction VME->PCI, block transfer enable  */
#define FB_UNIV_DCTL            LONGSWAP(DCTL_VDW_16 | DCTL_VAS_A24 | DCTL_LD64EN | DCTL_VCT_EN & ~DCTL_L2V)
#else				/* not block transfer */
/* Set 16 bit VME, 64 bit PCI, direction VME->PCI */
#define FB_UNIV_DCTL            LONGSWAP(DCTL_VDW_16 | DCTL_VAS_A24 | DCTL_LD64EN & ~DCTL_L2V)
#endif

extern unsigned long sysMemTop;


int dma_xfer(unsigned long VME_address,
             unsigned long RAM_address,
             unsigned long byte_count)
 {
   unsigned dma_vme_addr, dma_pci_ram_addr, dma_byte_cnt;

   if (VME_address < sysMemTop || RAM_address >= sysMemTop || byte_count == 0)
     return ERROR;

   dma_vme_addr = LONGSWAP(VME_address);
   dma_pci_ram_addr = LONGSWAP((unsigned)RAM_address + PCI2DRAM_BASE_ADRS);
   dma_byte_cnt = LONGSWAP(byte_count);

   /* set up DMA controller */
   *UNIVERSE_DCTL = FB_UNIV_DCTL;
   *UNIVERSE_DTBC = dma_byte_cnt;
   *UNIVERSE_DVA  = dma_vme_addr;
   *UNIVERSE_DLA  = dma_pci_ram_addr;
   *UNIVERSE_DGCS = FB_UNIV_DGCS;		/* start DMA */

   while (!(LONGSWAP(*UNIVERSE_DGCS) & (DGCS_DONE|DGCS_LERR|DGCS_VERR|DGCS_P_ERR)))
    {				/* wait for complete or error */
    }

   return OK;
 }

--------------6BA9C2153D603E27871A595A--



From vxwexplo-errs@csg.lbl.gov  Tue Feb 13 11:43:11 2001
From: David Abbott 
Date: Tue Feb 13 11:43:13 PST 2001
Subject: Re: MVME2304 w/ VxWorks 5.3.1 is slow reading VMEbus

On Tue, 13 Feb 2001, the vxWorks Users Group Exploder wrote:
> Submitted-by vxwexplo-errs@csg.lbl.gov  Tue Feb 13 09:31:11 2001
> Submitted-by: Richard Dickson 
> 
> Bob,
>      The attached function will do DMA from A24 space to RAM using D16.  Adjust
> the parameters of FB_UNIV_DCTL to suit your needs.  Hope this helps.
> 
> Richard
> 
> > Alternatively, is there any way from
> > software (running under VxWorks 5.3.1 on the MVME2304) to initiate either
> > A32/D32 block transfer read cycles on the VMEbus (i.e., the type of bus
> > cycles implied by VME address modifier codes 0x0F or 0x0B) or to initiate a
> > DMA-like transfer from VMEbus memory to local DRAM on the CPU board that
> > would generate such block transfer read cycles?
> > 
> > Any pointers or suggestions would be greatly appreciated.
> > 
> 
> -- 
> Richard Dickson
> Thomas Jefferson National Accelerator Facility
> dickson@jlab.org
> (757)269-5082
> --------------6BA9C2153D603E27871A595A
> Content-Type: text/plain; charset=us-ascii;
>  name="dma_driver_mv2300.c"
> Content-Transfer-Encoding: 7bit
> Content-Disposition: inline;
>  filename="dma_driver_mv2300.c"
> 
> /* Richard Dickson, Jefferson Lab - Code to perform 16 bit DMA transfer (VME_A24->RAM) for mv2306 */
> 
> #include 
> #include 
> #include 
> #include 	/* This works for my mv2306 */
> #include 
> #include 
> 
> /* Most of these defs may be found in $VW_BASE/target/config/mv2300/universe.h */
> /* Long swap must be done because PCI is little endian */
> #define DGCS_RESET_STAT         LONGSWAP(DGCS_STOP|DGCS_HALT|DGCS_DONE|DGCS_LERR|DGCS_VERR|DGCS_P_ERR)
> #define FB_UNIV_DGCS            DGCS_RESET_STAT|LONGSWAP(DGCS_GO)
> 
> #ifdef DO_BLOCK_XFER
> /* Set 16 bit VME, 64 bit PCI, direction VME->PCI, block transfer enable  */
> #define FB_UNIV_DCTL            LONGSWAP(DCTL_VDW_16 | DCTL_VAS_A24 | DCTL_LD64EN | DCTL_VCT_EN & ~DCTL_L2V)
> #else				/* not block transfer */
> /* Set 16 bit VME, 64 bit PCI, direction VME->PCI */
> #define FB_UNIV_DCTL            LONGSWAP(DCTL_VDW_16 | DCTL_VAS_A24 | DCTL_LD64EN & ~DCTL_L2V)
> #endif
> 
> extern unsigned long sysMemTop;
> 
> 
> int dma_xfer(unsigned long VME_address,
>              unsigned long RAM_address,
>              unsigned long byte_count)
>  {
>    unsigned dma_vme_addr, dma_pci_ram_addr, dma_byte_cnt;
> 
>    if (VME_address < sysMemTop || RAM_address >= sysMemTop || byte_count == 0)
>      return ERROR;
> 
>    dma_vme_addr = LONGSWAP(VME_address);
>    dma_pci_ram_addr = LONGSWAP((unsigned)RAM_address + PCI2DRAM_BASE_ADRS);
>    dma_byte_cnt = LONGSWAP(byte_count);
> 
>    /* set up DMA controller */
>    *UNIVERSE_DCTL = FB_UNIV_DCTL;
>    *UNIVERSE_DTBC = dma_byte_cnt;
>    *UNIVERSE_DVA  = dma_vme_addr;
>    *UNIVERSE_DLA  = dma_pci_ram_addr;
>    *UNIVERSE_DGCS = FB_UNIV_DGCS;		/* start DMA */
> 
>    while (!(LONGSWAP(*UNIVERSE_DGCS) & (DGCS_DONE|DGCS_LERR|DGCS_VERR|DGCS_P_ERR)))
>     {				/* wait for complete or error */
>     }
> 
>    return OK;
>  }
> 

  Hi Richard,
     Just thought I would throw my 2 cents worth in here.

  Generally waiting for the DMA to finish is OK (more time efficient) if
the transfers are "small" (i.e. below a few hundred microseconds or less).
  However,
  In Bob's case he seems to want to transfer a large amount of data over
the VMEBus at one time. First of all unless his VME Slave board requires
it (or the VME crate only has a P1 connector) there is no reason to force
16bit transfers over the VMEBus. Both the Chrislin board and the MVME2304
support the VME64 standard. Secondly, It may be unwise to sit in a
(potentially infinite) while loop waiting for 100+ MBytes to be
transfered. Even with the DMA this can take seconds. Inside an ISR the
above code would create a sure Crash. At the least he will be sucking up a
lot of CPU cycles that may be blocking other lower priority tasks from
running. Third, by constantly checking the Universe chip for the status
one actually is interfering with the DMA by utilizing the PCI bus during
the DMA. This can effectively slow down the whole transfer.
   As I see it one has a couple of options. First if this is not in an ISR 
and you know how much data is being moved one could simply taskDelay() for
the "appropriate" amount of time. A more elegant solution would be to 
use the DMA Done interrupt (see Universe DGCS Register)
          sysUnivIntEnable(UNIVERSE_DMA_INT) 
and attached ISR 
          sysUnivIntConnect(UNIVERSE_DMA_INT, (VOIDFUNCPTR) myDMAIsr, 0)
that checks the status and gives a semaphore. So then when your code
starts the DMA it can then just sleep on a semaphore until myDMAIsr() 
runs.

          Regards,
           David

---------------------------------------------------------
David Abbott                      Jefferson Lab
Data Acquisition Group            MS 12H
EMAIL: abbottd@jlab.org           12000 Jefferson Ave.
Tel: (757) 269-7190               Newport News, VA  23606
FAX: (757) 269-5800
---------------------------------------------------------



From vxwexplo-errs@csg.lbl.gov  Tue Feb 13 14:18:01 2001
From: "Tan, Tat Kin" 
Date: Tue Feb 13 14:18:03 PST 2001
Subject: Optic

Vxworks/Tornado

Hi All,

Am new to the VxWorks world and hoping to get some info from you guys.

Basically i need to integrate an optical fibre PHY into a board. The board
has already been running a BSP and what do i need to do to this BSP to make
this PHY works? I mean, which part of this BSP i need to modify? For sure
the Device Driver needs modification. Perhaps add the optical device driver
? What more? Appreciated helps. millions of thanks.

   with warm regards,
   Tat Kin, Tan




From vxwexplo-errs@csg.lbl.gov  Tue Feb 13 14:39:38 2001
From: "Baron, Ken" 
Date: Tue Feb 13 14:39:40 PST 2001
Subject: host/target symbol sync

Hi,

I am using Tornado2/VxWorks5.4. I recently tried using the target shell and
synchronizing the target server symbol table with the target shell symbol
table. Whenever I start the target server I get the following printout on my
console:
-> Fatal WTX error (0x10136), synchronization stopped
Fatal WTX error (0x1012f), synchronization stopped
Otherwise, both target and host shell are working fine. I am using Win95 and
a Radstone PPC board (these are the only 2 items on the ethernet).

Can anyone provide some help?

Ken Baron
BAE SYSTEMS
Reconnaissance & Surveillance Systems
300 Robbins Lane, Syosset, NY 11791
Voice: (516) 349-2132
Fax: (516) 349-2211
Email: ken.baron@lmco.com




From vxwexplo-errs@csg.lbl.gov  Tue Feb 13 21:11:11 2001
From: =?ks_c_5601-1987?B?sejFwrHU?= 
Date: Tue Feb 13 21:11:14 PST 2001
Subject: UDP Socket Performance

Hi,

I am using Tornado2/VxWorks5.4 on two MCP750 targets. I recently tried to 
test udp socket performance. The test program has 64 UDP connections which send
and receive 50 packets per second between the two targets. All connections are appear to
operate normaly, but more the half of the packets are not received.  

Can anyone provide some help?

Taegue Kim
Contela, Inc.
9-1, Sunae-dong, Bundang-gu, Sungnam-si,
Kyunggi-do, 463-784, Korea
Phone: +82-31-710-4834
Email: tgkim@contela.com


From vxwexplo-errs@csg.lbl.gov  Tue Feb 13 21:23:40 2001
From: "Vijayaraghavan Rajasekaran" 
Date: Tue Feb 13 21:23:42 PST 2001
Subject: Interrupt:dc0 no carrier error

This is a multi-part message in MIME format.
--------------3DE79F3D5F5BCF58A013FCED
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit

VxWorks/Tornado

Hi all,
        we are using Vxworks with Mcp750 Bsp. During and after boot up
continuously "Interrupt:dc0 no carrier error" is displayed . This error
appears whenever host tools or any node on the lan tries to contact
target over LAN.We even Installed "Tornado-Comp-Drv-December2000 release
v070".
After that error stops appearing , but we cud not connect TELNET/FTP
servers.
Pls suggest us the  remedy , ur immediate response will be highly
appreciated .

Thanks & Regards,
R.vijayaraghavan



--------------3DE79F3D5F5BCF58A013FCED
Content-Type: text/x-vcard; charset=us-ascii;
 name="vijayaraghavan.raja.vcf"
Content-Transfer-Encoding: 7bit
Content-Description: Card for vijay
Content-Disposition: attachment;
 filename="vijayaraghavan.raja.vcf"

begin:vcard 
n:Rajasekaran;Vijayaraghavan
x-mozilla-html:TRUE
url:www.wipro.com
org:WiproTechnologies;Embedded & Internet Access Solution
version:2.1
email;internet:vijayaraghavan.raja@wipro.com
adr;quoted-printable:;;Wipro Ltd. No. 111, Mount Road=0D=0A  Guindy=0D=0A=0D=0A=0D=0A=0D=0A=0D=0A=0D=0A=0D=0A;Chennai;TamilNadu;600032;India
fn:R.Vijayaraghavan
end:vcard

--------------3DE79F3D5F5BCF58A013FCED--



From vxwexplo-errs@csg.lbl.gov  Wed Feb 14 00:11:30 2001
From: "Michael Lawnick" 
Date: Wed Feb 14 00:11:32 PST 2001
Subject: Re: host/target symbol sync

Hi Ken,

this messages indicate that there occured problems on reaching the host 
symbol table by the target. AFAIK here are three major problems (assuming 
you are using WinNt as HostOS):
First: you might have specified localRegistry - change it to remoteRegistry 
with the ipAddress/targetName of your host.
Scnd: you might use 2 NIC or IP-addresses on your host. Connect your 
target to the first one.
Third: your target resides in a different subnet - haven't ever found a solution 
for this situation.

Are you sure you need the synchronisation ? Building-in the symbol table 
into vxWorks will consume the same amount of target memory. You only will 
lose the automatic update of symbols created after boot, but this will only be 
needed when working with target shell and host shell at the same time.

HTH
regards,
Michael

> Submitted-by: "Baron, Ken" 
> 
> Hi,
> 
> I am using Tornado2/VxWorks5.4. I recently tried using the target shell and
> synchronizing the target server symbol table with the target shell symbol
> table. Whenever I start the target server I get the following printout on my
> console:
> -> Fatal WTX error (0x10136), synchronization stopped
> Fatal WTX error (0x1012f), synchronization stopped
> Otherwise, both target and host shell are working fine. I am using Win95 and
> a Radstone PPC board (these are the only 2 items on the ethernet).
> 
> Can anyone provide some help?
> 
> Ken Baron
> BAE SYSTEMS
> Reconnaissance & Surveillance Systems
> 300 Robbins Lane, Syosset, NY 11791
> Voice: (516) 349-2132
> Fax: (516) 349-2211
> Email: ken.baron@lmco.com


Michael Lawnick, SOFTEC GmbH

==============================================
SOFTEC GmbH               Tel +49-731-96600-0
Promenade 17              Fax +49-731-96600-23
D-89073 Ulm               Michael Lawnick
Germany                   lawnick@softec.de
==============================================



From vxwexplo-errs@csg.lbl.gov  Wed Feb 14 00:15:30 2001
From: "Gokul Gopalasamy" 
Date: Wed Feb 14 00:15:32 PST 2001
Subject: VxWorks 

Hi all,

I'm a beginner in VxWorks. Can anyone suggest me some links or whitepapers regarding
VxWorks. Suggestions will be sincerely appreciated.

Thanks and Regards
Gokul

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
"Have the courage to face the Truth"
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Gokul Gopalasamy,
Global R&D,
Wipro Technologies,
Tel: 91-44-4500200-209   Ext: 2287
Fax: 91-44-4500250
Email: gokul.gopalasamy@wipro.com
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~




From vxwexplo-errs@csg.lbl.gov  Wed Feb 14 04:02:10 2001
From: "Bosis, Ilia" 
Date: Wed Feb 14 04:02:13 PST 2001
Subject: RE: Optic



> -----Original Message-----
> From: vxwexplo@lbl.gov [mailto:vxwexplo@lbl.gov]
> Sent: Wednesday, February 14, 2001 12:32 AM
> To: vxworks_users@csg.lbl.gov
> Subject: Optic
> 
> 
> Submitted-by vxwexplo-errs@csg.lbl.gov  Tue Feb 13 14:18:01 2001
> Submitted-by: "Tan, Tat Kin" 
> 
> Vxworks/Tornado
> 
> Hi All,
> 
> Am new to the VxWorks world and hoping to get some info from you guys.
> 
> Basically i need to integrate an optical fibre PHY into a 
> board. The board
> has already been running a BSP and what do i need to do to 
> this BSP to make
> this PHY works? I mean, which part of this BSP i need to 
> modify? For sure
> the Device Driver needs modification. Perhaps add the optical 
> device driver
> ? What more? Appreciated helps. millions of thanks.

Device driver , for sure.
B.T.W. If You need an example of how to write it,
go to target/unsupported/drv in the Tornado tree
and look for some driver source. 
Ethernet one is probably the closest similar.

Before this, You need to map the device into the CPU address space 
to make it accessible.
For PCI or VME devices this is rather a non-trivial task.
You may need to configure chip select(s), bus bridge(s)
and in case of PCI device also the device' PCI interface -
all depends on Your specific HW & BSP.
 
The simplest way (from my experience) 
is to find an unused space of the appropriate mapping
in the default/currently used configuration of Your BSP 
and map the device there.
Thus You save a lot of hassle of configuring complex 
and too often poorly documented hardware.
Not to mention that incorrect configuration can easely make 
the board I/O unaccessible so that the whole board looks dead
(no debugger connection, no console, nothing :(  ).

Hopefully, that's all.


> 
>    with warm regards,
>    Tat Kin, Tan
>
 
Success for You.
		Ilia

iliab at engineer dot com
----------------------
Complexity comes for free, Simplicity you have to work for.


From vxwexplo-errs@csg.lbl.gov  Wed Feb 14 04:03:10 2001
From: Vxworks Exploder 
Date: Wed Feb 14 04:03:13 PST 2001
Subject: comp.os.vxworks newsdigest

Comp.Os.Vxworks Daily Digest    Wed Feb 14 04:03:06 PST 2001

        Subject: Re: How to calculate Task stack space?
        Subject: Re: integrating network drivers
        Subject: re: Vx Exploder Digest
        Subject: re: Vx Exploder Digest
        Subject: NFS  server daemon on est8260
        Subject: Standalone applications
        Subject: Re: What happens when using zero for tftpdInit()
        Subject: Problem booting vxworks from DosFs
        Subject: re: MVME2304 w/ VxWorks 5.3.1 is slow reading VMEbus
        Subject: RE: comp.os.vxworks newsdigest
        Subject: Compiler Question: gprof ???
        Subject: va_list ???
        Subject: power pc -- 64bit operation
        Subject: Re: Vx Exploder Digest
        Subject: Re: Tornado on Solaris 7?
        Subject: re: MVME2304 w/ VxWorks 5.3.1 is slow reading VMEbus
        Subject: vxWorks 5.4/mv2700: networks goes north (down) - help
        Subject: re: MVME2304 w/ VxWorks 5.3.1 is slow reading VMEbus
        Subject: Re: Multicast Data Reception App.
        Subject: Adding new DNS resource types
        Subject: Debugger problems
        Subject: Re: vxWorks 5.4/mv2700: networks goes north (down) - help
        Subject: Re: Debugger problems
        Subject: Compile Errors
        Subject: appcompat for Win2k - Tornado 2.0
        Subject: Optic
        Subject: host/target symbol sync
        Subject: How about that RapidControl?
        Subject: Re: appcompat for Win2k - Tornado 2.0
        Subject: Re: [Question] usrAppInit() function in Vxworks Project
        Subject: UDP Socket Performance
        Subject: Re: How about that RapidControl?
        Subject: Interrupt:dc0 no carrier error
        Subject: Re: MIPS Evaluation Board.
        Subject: File system on VxWorks
        Subject: re: host/target symbol sync
        Subject: VxWorks
        Subject: Re: Compile Errors
        Subject: [Q] MUX interface between network and link layers
        Subject: (beginer) Problem "catching" a cPCI interrupt
        Subject: Re: *UPDATED* Gcc 2.95.3 for VxWorks-PPC available....
        Subject: Re: UDP Socket Performance
        Subject: Increase shared memory size
        Subject: Re: [Question] usrAppInit() function in Vxworks Project
        Subject: Re: VxWorks

-------------------------------------------------------

Newsgroups: comp.os.vxworks
Subject: Re: How to calculate Task stack space?
Date: Tue, 13 Feb 2001 11:20:16 -0000
From: "Dave Korn" 
Organization: Lumber Cartel (tinlc) Members #2234-2237 (owing to browser refresh)
Message-ID: 
References: <200011301401.GAA03645@mail22.bigmailbox.com> <90gtk7$nfm$1@neptunium.btinternet.com> <83rX5.611$4o3.510590@newsr1.u-net.net> <3A8893C7.19233322@yahoo.com>

Pierre Frigon wrote in message <3A8893C7.19233322@yahoo.com>...
>That would imply the greenhill utility knows the worst case stack usage of
scanf
>and printf which it does not. It would also imply that the greenhill would
be
>smart enough to see that the call to printf is actually calling the write
method
>of the serial port device driver which it cannot.

  Actually it was David Laight's own custom utility, and if it gave an
answer to the maximum stack depth of that program the stopping problem would
be solved!

     DaveK
- --
They laughed at Galileo.  They laughed at Copernicus.  They laughed at
Columbus. But remember, they also laughed at Bozo the Clown.




---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: integrating network drivers
Date: Tue, 13 Feb 2001 11:27:52 -0000
From: "Dave Korn" 
Organization: Lumber Cartel (tinlc) Members #2234-2237 (owing to browser refresh)
Message-ID: 
References: <90eniu$cms$1@news.netmar.com> <3A2AF70B.A74CA02@flashcom.net>  

Bill Pringlemeir wrote in message ...
>[bcc'ed to `abuse@rr.com']
>
>Did anyone else complain to `abuse@rr.com'?  I have several new posts
>in comp.os.vxworks that seem to have happened long ago.  Are the
>people from `RoadRunner - TampaBay' going to delete all of the
>duplicate posts or is that impossible?

  It's The Return of the Living Dead Threads from Beyond the Grave!  Zombie
Posts ate my Usenet!

  It seems to have stopped now, fortunately.  Anyone using a *decent*
newsreader could just filter on the Organization: RoadRunner - TampaBay
header, or the combination of Org. and date.  (Sorry to any RR users out
there!)

       DaveK
- --
They laughed at Galileo.  They laughed at Copernicus.  They laughed at
Columbus. But remember, they also laughed at Bozo the Clown.



---------------------------

Newsgroups: comp.os.vxworks
Subject: re: Vx Exploder Digest
Date: Tue, 13 Feb 2001 04:37:38 -0800 (PST)
From: james bond 
Organization: Lawrence Berkeley National Laboratory
Message-ID: <96bc2b$ssc$1@overload.lbl.gov>

Hi all,

This is James from India.
I am new to vxworks. I am interested to get more
information on the same.
If someone has some useful documents relating to
vxworks architecture and fundamentals,kindly send me
the same.
You can also send me the links on the web for various
documents and information on vxworks.

regards
James

__________________________________________________
Do You Yahoo!?
Get personalized email addresses from Yahoo! Mail - only $35 
a year!  http://personal.mail.yahoo.com/

---------------------------

Newsgroups: comp.os.vxworks
Subject: re: Vx Exploder Digest
Date: Tue, 13 Feb 2001 04:37:47 -0800 (PST)
From: james bond 
Organization: Lawrence Berkeley National Laboratory
Message-ID: <96bc2c$ssd$1@overload.lbl.gov>

Hi all,

This is James from India.
I am new to vxworks. I am interested to get more
information on the same.
If someone has some useful documents relating to
vxworks architecture and fundamentals,kindly send me
the same.
You can also send me the links on the web for various
documents and information on vxworks.

regards
James

__________________________________________________
Do You Yahoo!?
Get personalized email addresses from Yahoo! Mail - only $35 
a year!  http://personal.mail.yahoo.com/

---------------------------

Newsgroups: comp.os.vxworks
Subject: NFS  server daemon on est8260
Date: Tue, 13 Feb 2001 13:40:21 GMT
From: Gururaj patkar 
Organization: Optimum Online
Message-ID: <3A893988.DC0C4E01@optonline.net>

Hi vxworkers,

I have built a custom bootable vxworks image with "INCLUDE_NFS_SERVER".
Is there a limit on how many NFS client can talk to the NFS server
running on the target board. I am contemplating the use of NFS mounted
filesystem for remote software download of vxworks image from master
system controller to the slave controller?

Also, for a flash file system, are there any benchmark regarding the
write access time/erase time for PCMCIA based ATA flash cards (Sandisk,
siliconTech...). Can a enlightened guru tell  me what software
implementation/design should be considered for a PCMCAIA based  ata
flash cared as opposed to one for regular flash memory from Intel's
flash memory (Strataflash).
Any recommendations/suggestions is highly appreciated.

Gururaj


---------------------------

Newsgroups: comp.os.vxworks
Subject: Standalone applications
Date: Tue, 13 Feb 2001 14:01:56 GMT
From: "Scott Smail" 
Organization: Time Warner Road Runner - Rochester NY
Message-ID: 

Hello All,

I have the following system:

Tornado 2.0 with Zinc 5.0
host: AMD 586, 128MB RAM
target: PC486 Dos File Sys 32 MB RAM

Could someone tell me how to make my system standalone (automatically start
my application on bootup)?
I have modified the makefile and usrAppInit( ) and then run make vxWorks.st.
This compiled, but now I'm not sure how to proceed.

What do I do with vxWorks.st?

How do I get the image onto my target's ata?

Thanks,
- -Scott





---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: What happens when using zero for tftpdInit()
Date: Tue, 13 Feb 2001 14:07:15 GMT
From: Gururaj patkar 
Organization: Optimum Online
Message-ID: <3A893FD5.15CAFFFD@optonline.net>
References: <3a2ce730_2@nnrp1.news.uk.psi.net>

The default is 10.

hth
gururaj

Stephen Hill wrote:

> When making the function call tftpdInit (0, 0, 0, FALSE, 0), is the maximum
> number of TFTP clients unlimited or 4 (like the FTP server)?


---------------------------

Newsgroups: comp.os.vxworks
Subject: Problem booting vxworks from DosFs
Date: Tue, 13 Feb 2001 15:14:34 GMT
From: andreyk@home.com (Andray Kaganovsky)
Organization: @Home
Message-ID: 
References: <3A87FD55.AD34FD82@kayser-threde.de>  <3A89171C.E1D66D27@kayser-threde.de>

Torsten Levin  wrote in <3A89171C.E1D66D27@kayser-
threde.de>:

>Andray Kaganovsky schrieb:
>> Are you sure you have enough RAM available to your application
>> at that point ?
>> 
>>     regards, Andray
>
>This should not be the problem, when I use "dosFsMkfs" instead of
>"dosFsDevInit", the file system is successfully installed, but then, of
>course, all files are lost. So it seems not to be a ressource problem.
>There must be any differences in a bootloader and a normal application.
>
>Anybody else out there any ideas?

[AndrayK]. If you happen to have the source code of your driver, 
check out which sectors dosFs reads from the "disk", and what are
their contents, both in cases of dosFsMkfs() and dosFsDevInit(). 
This might give you a clue where the problem is. If you don't have 
the source, write you own block device driver which is stacked on 
top of this 3rd party driver, and simply passes all read/write 
calls to the it.  

    regards, Andray

---------------------------

Newsgroups: comp.os.vxworks
Subject: re: MVME2304 w/ VxWorks 5.3.1 is slow reading VMEbus
Date: Tue, 13 Feb 2001 09:42:08 -0500 (EST)
From: David Abbott 
Organization: Lawrence Berkeley National Laboratory
Message-ID: <96bj3b$2ng$1@overload.lbl.gov>

On Mon, 12 Feb 2001, the vxWorks Users Group Exploder wrote:
> Submitted-by vxwexplo-errs@csg.lbl.gov  Mon Feb 12 23:50:16 2001
> Submitted-by: Bob Kibrick 
> 
> We are currently conducting tests on a Motorola MVME2304-0133 board that is
> running VxWorks version 5.3.1.  We are rather surprised to find that on
> certain tests of VMEbus throughput, the MVME2304 (which has a 333 MHz.
> PowerPC 604e) delivers significantly lower performance than an obsolete
> Force 5CE board (which has an 85 MHz. SPARC processor running VxWorks 5.2).
> 
> The particular test in question involves transferring approximately 145 MB
> of data between a bulk VME memory board (Chrislin CI-VME64 256 MB) and the
> local DRAM on the CPU board.  This test was performed in both directions,
> and using both D16 and D32 transfer widths.  The identical C program was
> used on both processors, compiled (or cross-compiled) using gcc (2.9.5 for
> Sparc, cygnus-2.7.2 for PowerPC) on a Sparc Solaris 2.6 host.  The same
> VME memory board was used in all tests.  The tests were conducted on an
> otherwise idle system with no other competing processes (the tests were 
> run in the context of the VxWorks shell) nor any concurrent activity on
> the VMEbus.
> 
>
>  ... Parts of original message cut here ...
>
> 
> The results (elapsed time required to transfer 145 MB) are as follows:
> 
> Writing from CPU DRAM into VMEbus          Reading from VMEbus into CPU DRAM
> ----------------------------------         ----------------------------------
> MVME2304,  32-bit transfers: 15.1s	   MVME2304,  32-bit transfers: 39.6s
> Force 5CE, 32-bit transfers: 21.1s	   Force 5CE, 32-bit transfers: 29.4s
> 
> MVME2304,  16-bit transfers: 30.2s	   MVME2304,  16-bit transfers: 79.3s
> Force 5CE, 32-bit transfers: 42.3s	   Force 5CE, 16-bit transfers: 58.9s
> 
> Thus, while the Force 5CE is about 40% slower than the MVME2304 when writing
> to the VMEbus, the MVME2304 is about 35% slower than the Force 5CE when reading
> from the VMEbus.  
> 
> Furthermore, the MVME2304 takes 2.6 times longer to read from the VMEbus than
> it does to to write to it.  But the corresponding read cycle time on the
> Chrislin board is only 1.6 times the write cycle time.
> 
> We are trying to understand why the MVME2304 performs so slowly when reading
> from the VMEbus memory, and are wondering what options, if any, exist for
> improving this performance.  For example, are there any alternate settings
> of the control registers of the Tundra Universe chip (on the MVME2304) that
> might allow for higher throughput on VMEbus reads (We are currently using the 
> default values set by VxWorks 5.3.1)?  Alternatively, is there any way from
> software (running under VxWorks 5.3.1 on the MVME2304) to initiate either
> A32/D32 block transfer read cycles on the VMEbus (i.e., the type of bus
> cycles implied by VME address modifier codes 0x0F or 0x0B) or to initiate a
> DMA-like transfer from VMEbus memory to local DRAM on the CPU board that
> would generate such block transfer read cycles?
> 
> Any pointers or suggestions would be greatly appreciated.
> 
> Bob Kibrick, University of California Observatories / Lick Observatory
> kibrick@ucolick.org	831-459-2262	831-459-2298 (FAX)
> 

  Bob,
    While I cannot speak for the Force 5CE board, I am somewhat familiar
with the MVME2304 board. Your numbers for the 32 bit transfers
reading/writing are right on target with what I have measured with various
VME slave boards that we use.
 
    For 145MB transfer:
         Read :  3.6MBytes/sec  or 1.09 microsec/32bit word
         Write:  9.6Mbytes/sec  or 400  nsec/32bit word         

   (the 16bit reads/writes logically take twice as many cycles)

   The general problem with programmed (CPU Controlled) I/O to and from
the VME bus on the MVME PowerPC Arch boards is that you are actually
having to transfer data over two different buses. To get to the CPU's DRAM
the data must be transfered over VME to the Universe chip, and from the
Universe chip over the PCI bus to the Raven PCI/PowerPC bridge and
finally to DRAM. 
    For the Reads from VME the two buses are coupled and everything is
handshaked until the data is recieved in DRAM. You are helped in the
Writes because (based on the above numbers) you have the Universe chip
programmed to support Posted Writes which means that the CPU is not
waiting for a successful write over the VME bus. This part of the reason
why writes to invalid VME addresses do not generate Bus error exceptions
on this PowerPC Arch.
    In all likelyhood the Force Board is using a different VME interface
bridge that is a direct bridge with the SPARC/memory bus. This is why the
reads are somewhat faster. 

    However, programmed I/O of 145MBytes of data over the VME bus is
crazy. You should definitely use the DMA engine which is integrated in
with the Universe chip. I can send you some example code for implementing
transfers using the DMA engine if you like. But as you suggest with the
block tansfer modes B32 (or in your case with the Chrislin memory boards
they most certainly support MBLK 64 bit transfers) you ought to be able
read data into DRAM at around 30Mbytes/sec or perhaps slightly better. 

             Regards,
              David

- ---------------------------------------------------------
David Abbott                      Jefferson Lab
Data Acquisition Group            MS 12H
EMAIL: abbottd@jlab.org           12000 Jefferson Ave.
Tel: (757) 269-7190               Newport News, VA  23606
FAX: (757) 269-5800
- ---------------------------------------------------------


---------------------------

Newsgroups: comp.os.vxworks
Subject: RE: comp.os.vxworks newsdigest
Date: Tue, 13 Feb 2001 07:07:38 -0800
From: Jim Way 
Organization: Lawrence Berkeley National Laboratory
Message-ID: <96bj3c$2nh$1@overload.lbl.gov>

DaveK wrote:

>             ...Of course, you might be in the situation when your
> management throws something new at you and tells you you've got to get
> it done in a couple of weeks.  Then you're shafted, ...

Dave,

You pretty much pegged the situation. There's never time to do things the
right way. And I agree that reading all the FMs is the best way to go. If
you have the time.

Perhaps I could have made my point more clearly. When a Subject Matter
Expert (SME) is attempting to explain the Subject Matter (SM), the burden of
communication is not on the listener but on the speaker. In this case, the
SM is vxWorks and the SME is WindRiver. The burden is [should be] on them to
provide a clearly defined entry point into the manuals that takes you
through a Quick Start and a survey of basic concepts. From there, it should
point you to the various other manuals with suggestions on how to explore
them. One of the other ways a SME can improve communication is by providing
a Master Index. This is especially important when the SM is as huge as
vxWorks and many of the components can be configured in or out.

One bit of evidence that there is a problem can be found here in the User
Group. Let's take a simple example. How many postings do we see where
someone is asking, "What does this error message mean?" To me, this is an
obvious question, but I'm not sure the manuals are organized to allow me to
answer it. (For comparison, I know that the manual for my product has a
"messages" section where *all* the messages are located along with further
explanation of what they mean. It took time to develop that list, but the
results are worth it.)

Dave, as always your comments are insightful and enlightening. One day I
hope to have the familiarity with vxWorks that you've earned.

Jim
- -----------------------------------------
Jim Way, Software Engineer
Datum Austin (Austron Inc.)
voice: 512.721.4170
fax  : 512.990.9712
email: jwayATdatumDOTcom (no spam please)
- -----------------------------------------

---------------------------

Newsgroups: comp.os.vxworks
Subject: Compiler Question: gprof ???
Date: Tue, 13 Feb 2001 15:25:38 GMT
From: "Jim" 
Organization: Magma Communications Ltd.
Message-ID: 

Hello all,
This question may be a little unreasonable ... but i figured i'd ask anyways
... i found a utility called gprof last night which can do profiling of your
code ... ie how much time was spent in what function etc ...

I haven't used gprof yet ... but apparently you only need to compile with
gcc -gp ... to get it to run on unix ... is it possible to have gprof run
with vxWorks ...

Keep in mind i don't know much about gprof yet i am just curious if it is
supported in vxWorks version of gcc ... or even if the gprof application can
be used ...

Jim



---------------------------

Newsgroups: comp.os.vxworks
Subject: va_list ???
Date: Tue, 13 Feb 2001 15:40:54 GMT
From: "Jim" 
Organization: Magma Communications Ltd.
Message-ID: 

A quick question ... is there any known problems with va_list ... i am
hoping to write my own printf type function for a serial device ... but i
have heard there are some problems with "va_list args;" in vxWorks anyone
know for sure ...

Jim

ps this is what i am kinda what I am hoping to do ...
        va_list args;
        va_start(args, fmt);
        vprintf(fmt, args);
        va_end(args);





---------------------------

Newsgroups: comp.os.vxworks
Subject: power pc -- 64bit operation
Date: Tue, 13 Feb 2001 16:53:42 GMT
From: "Lou Kale" 
Organization: EarthLink Inc. -- http://www.EarthLink.net
Message-ID: 

I'm using a ppc604 (750) with Tornado2.

How do I know if it is running code in 64 bit mode?
Is there a gnu compiler switch to turn 64 bit on?


Thanx,
Lou Kale
KeNETec Inc.
550 Spring Street
Naugatuck CT 06770
Tel (203)-723-4242 x330
Fax (203)-723-4187
lou.kale@kenetec.com




---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: Vx Exploder Digest
Date: Tue, 13 Feb 2001 14:12:55 -0000
From: "Dave Korn" 
Organization: Lumber Cartel (tinlc) Members #2234-2237 (owing to browser refresh)
Message-ID: 
References: <96bc2c$ssd$1@overload.lbl.gov>

james bond wrote in message <96bc2c$ssd$1@overload.lbl.gov>...
>Hi all,
>
>This is James from India.

  Yeah?  You were called Sumeet yesterday!

     DaveK
- --
They laughed at Galileo.  They laughed at Copernicus.  They laughed at
Columbus. But remember, they also laughed at Bozo the Clown.



---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: Tornado on Solaris 7?
Date: Tue, 13 Feb 2001 11:04:14 -0700
From: Russ Johnson 
Organization: RMI.NET
Message-ID: <3A89771E.AEA5A6C2@bigfoot.com>
References: <3A887EC9.2F2773E5@voxware.com>

Craig Vanderborgh wrote:
> Is it possible to use Tornado 2.01 on Solaris 7/SPARC?

We do, and it works fine for us.

Russ

---------------------------

Newsgroups: comp.os.vxworks
Subject: re: MVME2304 w/ VxWorks 5.3.1 is slow reading VMEbus
Date: Tue, 13 Feb 2001 12:30:37 -0500
From: Richard Dickson 
Organization: Lawrence Berkeley National Laboratory
Message-ID: <96btkq$bu3$1@overload.lbl.gov>

This is a multi-part message in MIME format.
- --------------6BA9C2153D603E27871A595A
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit

Bob,
     The attached function will do DMA from A24 space to RAM using D16.  Adjust
the parameters of FB_UNIV_DCTL to suit your needs.  Hope this helps.

Richard

> Alternatively, is there any way from
> software (running under VxWorks 5.3.1 on the MVME2304) to initiate either
> A32/D32 block transfer read cycles on the VMEbus (i.e., the type of bus
> cycles implied by VME address modifier codes 0x0F or 0x0B) or to initiate a
> DMA-like transfer from VMEbus memory to local DRAM on the CPU board that
> would generate such block transfer read cycles?
> 
> Any pointers or suggestions would be greatly appreciated.
> 

- -- 
Richard Dickson
Thomas Jefferson National Accelerator Facility
dickson@jlab.org
(757)269-5082
- --------------6BA9C2153D603E27871A595A
Content-Type: text/plain; charset=us-ascii;
 name="dma_driver_mv2300.c"
Content-Transfer-Encoding: 7bit
Content-Disposition: inline;
 filename="dma_driver_mv2300.c"

/* Richard Dickson, Jefferson Lab - Code to perform 16 bit DMA transfer (VME_A24->RAM) for mv2306 */

#include 
#include 
#include 
#include 	/* This works for my mv2306 */
#include 
#include 

/* Most of these defs may be found in $VW_BASE/target/config/mv2300/universe.h */
/* Long swap must be done because PCI is little endian */
#define DGCS_RESET_STAT         LONGSWAP(DGCS_STOP|DGCS_HALT|DGCS_DONE|DGCS_LERR|DGCS_VERR|DGCS_P_ERR)
#define FB_UNIV_DGCS            DGCS_RESET_STAT|LONGSWAP(DGCS_GO)

#ifdef DO_BLOCK_XFER
/* Set 16 bit VME, 64 bit PCI, direction VME->PCI, block transfer enable  */
#define FB_UNIV_DCTL            LONGSWAP(DCTL_VDW_16 | DCTL_VAS_A24 | DCTL_LD64EN | DCTL_VCT_EN & ~DCTL_L2V)
#else				/* not block transfer */
/* Set 16 bit VME, 64 bit PCI, direction VME->PCI */
#define FB_UNIV_DCTL            LONGSWAP(DCTL_VDW_16 | DCTL_VAS_A24 | DCTL_LD64EN & ~DCTL_L2V)
#endif

extern unsigned long sysMemTop;


int dma_xfer(unsigned long VME_address,
             unsigned long RAM_address,
             unsigned long byte_count)
 {
   unsigned dma_vme_addr, dma_pci_ram_addr, dma_byte_cnt;

   if (VME_address < sysMemTop || RAM_address >= sysMemTop || byte_count == 0)
     return ERROR;

   dma_vme_addr = LONGSWAP(VME_address);
   dma_pci_ram_addr = LONGSWAP((unsigned)RAM_address + PCI2DRAM_BASE_ADRS);
   dma_byte_cnt = LONGSWAP(byte_count);

   /* set up DMA controller */
   *UNIVERSE_DCTL = FB_UNIV_DCTL;
   *UNIVERSE_DTBC = dma_byte_cnt;
   *UNIVERSE_DVA  = dma_vme_addr;
   *UNIVERSE_DLA  = dma_pci_ram_addr;
   *UNIVERSE_DGCS = FB_UNIV_DGCS;		/* start DMA */

   while (!(LONGSWAP(*UNIVERSE_DGCS) & (DGCS_DONE|DGCS_LERR|DGCS_VERR|DGCS_P_ERR)))
    {				/* wait for complete or error */
    }

   return OK;
 }

- --------------6BA9C2153D603E27871A595A--


---------------------------

Newsgroups: comp.os.vxworks
Subject: vxWorks 5.4/mv2700: networks goes north (down) - help
Date: Tue, 13 Feb 2001 15:23:39 -0500
From: Al Johnston 
Organization: MindSpring Enterprises
Message-ID: <3A8997CB.6411EE51@mindspring.com>

At some point in running our code we start getting error messages
that indicate the networking part of vxWorks has gone down.

A ping from the target and a ping to the target confirm this
(destination unreachable and or time-out).

Can someone give me some hints on how to debug this?  Things
like ifShow() seem to say everything is okay.

The network is a 10bT port on the card;  connection is
to solaris boxes.

ping of "localhost", of "sn6-n2-1", and of "xxx.yyy.zzz.aaa" 
(where the last two are the name and ip address of the target
card in question) all work at that point, but a ping to
any other ip address times out.


help!

---------------------------

Newsgroups: comp.os.vxworks
Subject: re: MVME2304 w/ VxWorks 5.3.1 is slow reading VMEbus
Date: Tue, 13 Feb 2001 14:42:37 -0500 (EST)
From: David Abbott 
Organization: Lawrence Berkeley National Laboratory
Message-ID: <96c4lr$een$1@overload.lbl.gov>

On Tue, 13 Feb 2001, the vxWorks Users Group Exploder wrote:
> Submitted-by vxwexplo-errs@csg.lbl.gov  Tue Feb 13 09:31:11 2001
> Submitted-by: Richard Dickson 
> 
> Bob,
>      The attached function will do DMA from A24 space to RAM using D16.  Adjust
> the parameters of FB_UNIV_DCTL to suit your needs.  Hope this helps.
> 
> Richard
> 
> > Alternatively, is there any way from
> > software (running under VxWorks 5.3.1 on the MVME2304) to initiate either
> > A32/D32 block transfer read cycles on the VMEbus (i.e., the type of bus
> > cycles implied by VME address modifier codes 0x0F or 0x0B) or to initiate a
> > DMA-like transfer from VMEbus memory to local DRAM on the CPU board that
> > would generate such block transfer read cycles?
> > 
> > Any pointers or suggestions would be greatly appreciated.
> > 
> 
> -- 
> Richard Dickson
> Thomas Jefferson National Accelerator Facility
> dickson@jlab.org
> (757)269-5082
> --------------6BA9C2153D603E27871A595A
> Content-Type: text/plain; charset=us-ascii;
>  name="dma_driver_mv2300.c"
> Content-Transfer-Encoding: 7bit
> Content-Disposition: inline;
>  filename="dma_driver_mv2300.c"
> 
> /* Richard Dickson, Jefferson Lab - Code to perform 16 bit DMA transfer (VME_A24->RAM) for mv2306 */
> 
> #include 
> #include 
> #include 
> #include 	/* This works for my mv2306 */
> #include 
> #include 
> 
> /* Most of these defs may be found in $VW_BASE/target/config/mv2300/universe.h */
> /* Long swap must be done because PCI is little endian */
> #define DGCS_RESET_STAT         LONGSWAP(DGCS_STOP|DGCS_HALT|DGCS_DONE|DGCS_LERR|DGCS_VERR|DGCS_P_ERR)
> #define FB_UNIV_DGCS            DGCS_RESET_STAT|LONGSWAP(DGCS_GO)
> 
> #ifdef DO_BLOCK_XFER
> /* Set 16 bit VME, 64 bit PCI, direction VME->PCI, block transfer enable  */
> #define FB_UNIV_DCTL            LONGSWAP(DCTL_VDW_16 | DCTL_VAS_A24 | DCTL_LD64EN | DCTL_VCT_EN & ~DCTL_L2V)
> #else				/* not block transfer */
> /* Set 16 bit VME, 64 bit PCI, direction VME->PCI */
> #define FB_UNIV_DCTL            LONGSWAP(DCTL_VDW_16 | DCTL_VAS_A24 | DCTL_LD64EN & ~DCTL_L2V)
> #endif
> 
> extern unsigned long sysMemTop;
> 
> 
> int dma_xfer(unsigned long VME_address,
>              unsigned long RAM_address,
>              unsigned long byte_count)
>  {
>    unsigned dma_vme_addr, dma_pci_ram_addr, dma_byte_cnt;
> 
>    if (VME_address < sysMemTop || RAM_address >= sysMemTop || byte_count == 0)
>      return ERROR;
> 
>    dma_vme_addr = LONGSWAP(VME_address);
>    dma_pci_ram_addr = LONGSWAP((unsigned)RAM_address + PCI2DRAM_BASE_ADRS);
>    dma_byte_cnt = LONGSWAP(byte_count);
> 
>    /* set up DMA controller */
>    *UNIVERSE_DCTL = FB_UNIV_DCTL;
>    *UNIVERSE_DTBC = dma_byte_cnt;
>    *UNIVERSE_DVA  = dma_vme_addr;
>    *UNIVERSE_DLA  = dma_pci_ram_addr;
>    *UNIVERSE_DGCS = FB_UNIV_DGCS;		/* start DMA */
> 
>    while (!(LONGSWAP(*UNIVERSE_DGCS) & (DGCS_DONE|DGCS_LERR|DGCS_VERR|DGCS_P_ERR)))
>     {				/* wait for complete or error */
>     }
> 
>    return OK;
>  }
> 

  Hi Richard,
     Just thought I would throw my 2 cents worth in here.

  Generally waiting for the DMA to finish is OK (more time efficient) if
the transfers are "small" (i.e. below a few hundred microseconds or less).
  However,
  In Bob's case he seems to want to transfer a large amount of data over
the VMEBus at one time. First of all unless his VME Slave board requires
it (or the VME crate only has a P1 connector) there is no reason to force
16bit transfers over the VMEBus. Both the Chrislin board and the MVME2304
support the VME64 standard. Secondly, It may be unwise to sit in a
(potentially infinite) while loop waiting for 100+ MBytes to be
transfered. Even with the DMA this can take seconds. Inside an ISR the
above code would create a sure Crash. At the least he will be sucking up a
lot of CPU cycles that may be blocking other lower priority tasks from
running. Third, by constantly checking the Universe chip for the status
one actually is interfering with the DMA by utilizing the PCI bus during
the DMA. This can effectively slow down the whole transfer.
   As I see it one has a couple of options. First if this is not in an ISR 
and you know how much data is being moved one could simply taskDelay() for
the "appropriate" amount of time. A more elegant solution would be to 
use the DMA Done interrupt (see Universe DGCS Register)
          sysUnivIntEnable(UNIVERSE_DMA_INT) 
and attached ISR 
          sysUnivIntConnect(UNIVERSE_DMA_INT, (VOIDFUNCPTR) myDMAIsr, 0)
that checks the status and gives a semaphore. So then when your code
starts the DMA it can then just sleep on a semaphore until myDMAIsr() 
runs.

          Regards,
           David

- ---------------------------------------------------------
David Abbott                      Jefferson Lab
Data Acquisition Group            MS 12H
EMAIL: abbottd@jlab.org           12000 Jefferson Ave.
Tel: (757) 269-7190               Newport News, VA  23606
FAX: (757) 269-5800
- ---------------------------------------------------------


---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: Multicast Data Reception App.
Date: Tue, 13 Feb 2001 12:28:34 -0800
From: "D. H. Everett" 
Organization: Posted via Supernews, http://www.supernews.com
Message-ID: 
References: <3A8348CB.F726293E@sympatico.ca>

Seems to me you need Winsock2 to use multicast.   I think the aforementioned
Multicast Initiative has a sample multicast application using Winsock2.

Dave

Mike McNaughton  wrote in message
news:3A8348CB.F726293E@sympatico.ca...
> Hello vxworks using  people:
>
> I have a reflective memory /ethernet adapter card sitting in a VME rack.
> It transmits  data to a  not so useful, proprietory data acquisition
> system (daq pc).
>
> It is supposedly a multicast, udp type of transmission according to the
> very sparse info that came with the system.
>
> Data packets are as follows as viewed with  network Sniffer:
>
>
> Sending port= 2048
> sending ip = 172.17.66.63
> receiving port=604
> receiving  ip= 224.0.32.32
>
> (224.0.32.32  is  not the domain IP address assigned to the daq pc even
> tho the daq pc is happily reading data which seems to be directed to
> 224.0.32.32. The ip address of the daq pc is 10.32.7.226 )
>
> I believe that the 224.0.32.32 is a multicast address.
>
> The reflective meory and the daq pc are connected, cat 5 thru a
> mini-hub.
> I've tried all manner of winsock connections with VB6 but can't get a
> connection.
>
> I would like to write a piece of software that would read the data into
> a third PC as it flows from the VME rack to the Black Box PC DAQ
> package.
>
> I would welcome some advice as to how to get a connection either in VC6
> or VB6.
>
> Thanks
> Mike M.
>
>
>



---------------------------

Newsgroups: comp.os.vxworks
Subject: Adding new DNS resource types
Date: Tue, 13 Feb 2001 20:30:43 GMT
From: "Daniel Roy" 
Organization: Magma Communications Ltd.
Message-ID: 

Hello,

I'm trying to introduce DNS resource types not currently supported by
resolvLib into my VxWorks platform.  The only solution I've been able to
come up with at this point is to port portions of the Internet Software
Consortium's BIND client to VxWorks and to bypass resolvLib all together.
This smacks of overkill, especially since all I need is resolvLib with a few
extra typedefs and extra entries in a CASE statement.  Can anyone suggest an
alternate solution/approach ?

Regards,
Daniel



---------------------------

Newsgroups: comp.os.vxworks
Subject: Debugger problems
Date: Tue, 13 Feb 2001 15:45:19 -0500
From: "Shaila Kambli" 
Organization: SNET Internet
Message-ID: 

Hello,
 I am working with Tornado 2.0. I have recently been experiencing debugger
problems. Some of the errors are listed here :
    - "Unable to retrieve task list. WTX error 0x100de
AGENT_COMMUNICATION_ERROR)
   - "Target connection has been lost! Stopping the debugger now"
   - "Serious internal error occured while executing TORNADO.EXE .."

A log file is attached for your reference. Your input will be greately
appreciated.

Shaila Kambli


begin 666 TORNADO-Problem-20010213-153055.log
M17AC97!T:6]N(#!X0S P,# P,#4Z($580T505$E/3E]!0T-%4U-?5DE/3$%4
M24].#0H-"DQO9R!F:6QE(&AA# P-CA$1#0T72P@,"P@,C Q,S,P.#DX
M,"!;,'@W.# P03@S-%TL(#8X-S,Q,S8@6S!X,# V.$4P,S!=("XN+BD-"G-T
M# R-T5=( D@*" V.## P-CA$1#0T72P@-C@W,S$S
M-B!;,'@P,#8X13 S,%TL(#8X-S(Q.#0@6S!X,# V.$1#-SA=+" Q-S(Y,3@Q
M,3@T(%LP>#8W,3$S-C P72 N+BXI#0IS=')L96Y;,'@P,31%72 )("@@-C@W
M,C,X."!;,'@P,#8X1$0T-%TL(#8X-S,Q,S8@6S!X,# V.$4P,S!=+" V.## P-CA$0S# P-CA%,#,P72P@-C@W,C$X-"!;,'@P
M,#8X1$,W.%TL(#8X-S(Q-#0@6S!X,# V.$1#-3!=("XN+BD-"E]C# U.3!=( D@*" V.## P-CA$1#0T72P@
M-C@W,S$S-B!;,'@P,#8X13 S,%TL(#8X-S(Q.#0@6S!X,# V.$1#-SA=+" V
M.## P-CA$0S4P72 N+BXI#0IS=')L96Y;,'@P-D0P72 )("@@
M-C@W,C,X."!;,'@P,#8X1$0T-%TL(#8X-S,Q,S8@6S!X,# V.$4P,S!=+" V
M.## P-CA$0S# P-CA%,#,P72P@-C@W,C$X-"!;,'@P,#8X1$,W.%TL
M(#8X-S(Q-#0@6S!X,# V.$1#-3!=("XN+BD-"E9I45X6S!X
M,#)$-#9=( D@*" V.## P-CA$0S#!&.5T@
M"2 H(#8X-S(U,#0@6S!X,# V.$1$0CA=+" V.## P-CA$13 X
M72P@-C@W,C$)&1C$4P-D0W,S8S72P@,2 N+BXI#0I&:6YD0VQO# V,3-=( D@*" V.## P-CA$
M1D0P72P@,3# Q
M,##!!-SA=( D@*" P+" T-S0R,# T(%LP># P-#@U0C#8V0S%#,C# P-#@U0C##8U-C@W-#(P72P@,3@Q.#,R-C4V,"!;,'@V0S8Q
M-S8R,%TL(#$X-C0S.30Q,#$@6S!X-D8R,#8U-S5=("XN+BD-"@T*4F5G:7-T
M97)S.@T*96%X/3,P,S@W.#,P(&5B>#TV-S$P-D1&.2!E8W@],S S.# T*,# V.$1$-#0@,# V.$4P
M,S @-CVTN9RC?:"XN+BXN#0HP,# P,# P," P
M,# P,# P," P,#8X1$$W," P,#8X1#@U." @+BXN+BXN+BYPVF@N6-AH+@T*
M,# V.$1"0S @-S@P,$5&,#,@-S@P,D5$.# @1D9&1D9&1D8@(,#;:"XN[RYX
M@.TN>/____\-"C P-CA$044X(#$3=:"XPX&@N#0HP,#8X1$,W." V-S$Q,S8P," V-S$P-D1&
M.2 P,# P,# P," @>-QH+BXV+F?Y;2YG+BXN+@T*,# P,# Q,# @,# V.$4P
M,S @,# V.$1#-3 @-C
Organization: Magma Communications Ltd.
Message-ID: 
References: <3A8997CB.6411EE51@mindspring.com>


This may sound stupid ... but i had a problem abou a week ago ... my board
would die after about 250 or so pings ... problem was we were connecting to
a cheap hub ... after we switched (and much time wasted) the board was
steady as a rock ...

Hope this helps ... otherwise goodluck ...

Jim

"Al Johnston"  wrote in message
news:3A8997CB.6411EE51@mindspring.com...
> At some point in running our code we start getting error messages
> that indicate the networking part of vxWorks has gone down.
>
> A ping from the target and a ping to the target confirm this
> (destination unreachable and or time-out).
>
> Can someone give me some hints on how to debug this?  Things
> like ifShow() seem to say everything is okay.
>
> The network is a 10bT port on the card;  connection is
> to solaris boxes.
>
> ping of "localhost", of "sn6-n2-1", and of "xxx.yyy.zzz.aaa"
> (where the last two are the name and ip address of the target
> card in question) all work at that point, but a ping to
> any other ip address times out.
>
>
> help!



---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: Debugger problems
Date: Tue, 13 Feb 2001 20:59:53 GMT
From: "Daniel Roy" 
Organization: Magma Communications Ltd.
Message-ID: 
References: 

Perhaps if you logged WTX during your debugging session, you might be able
to pin-point the problem.  To turn WTX logging on, modify your target-server
settings:
1) Tools-> Target Server -> Configure
2) Select the target server in question
3) Use the "Target Server Properties" drop-down list to select Logging

"Backend Log File" records messages between the target server running on
your host and the WDB agent on the target (probably not what you want, but
who knows...)

"WTX Log File" records messages between the target server and which ever
tools your happen to be running at the time, presumably the Debugger.

Good luck,
Daniel

"Shaila Kambli"  wrote in message
news:m1hi6.27740$OD.10819188@typhoon.snet.net...
> Hello,
>  I am working with Tornado 2.0. I have recently been experiencing debugger
> problems. Some of the errors are listed here :
>     - "Unable to retrieve task list. WTX error 0x100de
> AGENT_COMMUNICATION_ERROR)
>    - "Target connection has been lost! Stopping the debugger now"
>    - "Serious internal error occured while executing TORNADO.EXE .."



---------------------------

Newsgroups: comp.os.vxworks
Subject: Compile Errors
Date: Tue, 13 Feb 2001 16:33:03 -0500
From: Fake 
Organization: UUNET Canada News Reader Service
Message-ID: <3A89A80F.CFFB026E@address.com>

Hi All,

Since the response to my last question on CppUnit is deafening (:)), so
let's me ask the question in another format. I have the class defined
as:

- --------------------------------------------------
#include "Test.h"
#include 
using std::string;

class TestSuite: public Test
{
  public:
   TestSuite(const char* name);

  void AddTest(Test);
 template void Add(void (Fixture_::* fp)(),string name)
 {
  AddTest(Test(new TestCaller(fp,name)));
 }
  bool Run(TestResult&);
  const char* const Name() const;
};

#define ADD_TEST(test) (suite.Add(test,#test));
- -------------------------------------------------------
I invoked it as follow:

Test ComplexTestSuite()
{
  TestSuite suite("ComplexTestSuite");

  ADD_TEST(ComplexTester::TestAdd);
  return suite;
}
- --------------------------------------------------
But when I compiled it I got the following errors (adding -ansi produced
the same result):
Note: I did not include the test.h as I don't think it caused the error.

ccsimpc -U__WINNT -UWIN32 -U__WINNT__ -UWINNT -U__MINGW32__ -U_WIN32
- -U__WIN32 -U__WIN32__
 -U__WIN32 -mpentium  -nostdinc -g -nostdlib -fno-builtin -fno-defer-pop
- -Wall -DRW_MULTI_
THREAD -D_REENTRANT -I. -ID:\Tornado\target\h -DCPU=SIMNT
- -cD:\Tornado\target\proj\Cpp\TestComplex.cpp
D:\Tornado\target\proj\Cpp\TestComplex.cpp: In function `class Test
ComplexTestSuite()':
D:\Tornado\target\proj\Cpp\TestComplex.cpp:35: no matching function for
call to `TestSuite::Add (void (ComplexTester::)(), char[23])'
D:\Tornado\target\proj\Cpp\TestSuite.h:34: candidates are:
TestSuite::Add(void (ComplexTester::*)(),
basic_string,__d
efault_alloc_template >)
make: *** [TestComplex.o] Error 0x1

================================================
I wish someone here could help me and I would really appreciate.

Cheers.



---------------------------

Newsgroups: comp.os.vxworks
Subject: appcompat for Win2k - Tornado 2.0
Date: Tue, 13 Feb 2001 15:03:06 -0800
From: "Christian Bengtson" 
Organization: CTSnet Internet Services
Message-ID: <96cee6$282p$1@thoth.cts.com>

I have installed Tornado 2.0 for MIPS on a new Win2K machine and I get the
following faults:

RPCExit: extra call error a couple of times

and

a message saying couldn't read file for "...\WindView.win32.tcl" : no such
file or directory.

I was told by another engineer that a program called appcompat could solve
this so I downloaded a program called appupd.exe from microsoft homepage,
but it still doesn't work. Does anyone know anything about this?

Thanks,

Christian



---------------------------

Newsgroups: comp.os.vxworks
Subject: Optic
Date: Tue, 13 Feb 2001 14:03:43 -0800
From: "Tan, Tat Kin" 
Organization: Lawrence Berkeley National Laboratory
Message-ID: <96cf7b$k96$1@overload.lbl.gov>

Vxworks/Tornado

Hi All,

Am new to the VxWorks world and hoping to get some info from you guys.

Basically i need to integrate an optical fibre PHY into a board. The board
has already been running a BSP and what do i need to do to this BSP to make
this PHY works? I mean, which part of this BSP i need to modify? For sure
the Device Driver needs modification. Perhaps add the optical device driver
? What more? Appreciated helps. millions of thanks.

   with warm regards,
   Tat Kin, Tan



---------------------------

Newsgroups: comp.os.vxworks
Subject: host/target symbol sync
Date: Tue, 13 Feb 2001 17:30:26 -0500
From: "Baron, Ken" 
Organization: Lawrence Berkeley National Laboratory
Message-ID: <96cf7c$k97$1@overload.lbl.gov>

Hi,

I am using Tornado2/VxWorks5.4. I recently tried using the target shell and
synchronizing the target server symbol table with the target shell symbol
table. Whenever I start the target server I get the following printout on my
console:
- -> Fatal WTX error (0x10136), synchronization stopped
Fatal WTX error (0x1012f), synchronization stopped
Otherwise, both target and host shell are working fine. I am using Win95 and
a Radstone PPC board (these are the only 2 items on the ethernet).

Can anyone provide some help?

Ken Baron
BAE SYSTEMS
Reconnaissance & Surveillance Systems
300 Robbins Lane, Syosset, NY 11791
Voice: (516) 349-2132
Fax: (516) 349-2211
Email: ken.baron@lmco.com



---------------------------

Newsgroups: comp.os.vxworks,comp.dcom.net-management,comp.protocols.snmp
Subject: How about that RapidControl?
Date: Tue, 13 Feb 2001 23:56:12 +0000
From: Kerry Kimbrough 
Message-ID: <3A89C99C.ADFA79@netpliance.net>

Have you used the RapidControl toolset for unified access to SNMP MIBs
via HTTP/CLI/etc.? (Formerly from RapidLogic, now offered by Wind
River). Does it work well? What are its pros and cons? Any tips much
appreciated.

- -- 

Regards,

Kerry

- ------------------------------------------------------------------------------
mailto:kerryk@netpliance.net 
- ------------------------------------------------------------------------------

---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: appcompat for Win2k - Tornado 2.0
Date: Wed, 14 Feb 2001 01:22:53 GMT
From: "Mark Fanara" 
Organization: Excite@Home - The Leader in Broadband http://home.com/faster
Message-ID: 
References: <96cee6$282p$1@thoth.cts.com>

Check wrs.com patches - there is a fix for this problem - see SPR 62517





---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: [Question] usrAppInit() function in Vxworks Project
Date: Wed, 14 Feb 2001 01:56:16 GMT
From: "Eddy Raineri" 
Organization: Excite@Home - The Leader in Broadband http://home.com/faster
Message-ID: <4Bli6.72893$bb.1084753@news1.rdc1.tx.home.com>
References: <90fj7h$e4o$1@nnrp1.deja.com>

The problem is that the usrApplInit.c file does not include config.h or
prjconfig.h.  Because of this INCLUDE_USER_APPL is not defined.  Modify the
usrApplInit.c file to include config.h and it will work

Good luck
E
 wrote in message news:90fj7h$e4o$1@nnrp1.deja.com...
>
>
>  Hi, VxWorks Developers.
>
>  I have one question about usrAppInit() in VxWorks Project file.
>
>  I added following lines to my BSP directory's config.h
>
> #define INCLUDE_USER_APPL
> #define USER_APPL_INIT taskSpawn ("sample_app", 20, 0,
> 2000, (FUNCPTR) sample_app, 0,0,0,0,0,0,0,0,0,0)
>
>  of course, I defined sample_app function also....
>  and I create new bootable project based on my BSP directory.
>  So, I got some project files and I found that usrAppInit()
>
>  void usrAppInit (void)
>     {
> #ifdef USER_APPL_INIT
> USER_APPL_INIT; /* for backwards compatibility */
> #endif
>
>     /* add application specific code here */
>     }
>
>  When I made ROM based Vxworks (ie. Vxworks_rom.hex) and succeeded in
> booting. but usrAppInit() function never run....
>  so, I checked whether checked USER_APPL_INIT is defined.
>
>  unfortunately, in my usrAppInit() fucntion .
>  this function said that ......
>
>  void usrAppInit (void)
>     {
> #ifdef USER_APPL_INIT
> USER_APPL_INIT; /* for backwards compatibility */
>          printf("defined.......");
> #else
>          printf("undedined.......");
> #endif
>
>     /* add application specific code here */
>     }
>
>  the result is "undefined......."
>
>  Is something that I forgot ??
>
> I defined
> #define INCLUDE_USER_APPL
> #define USER_APPL_INIT
> in my config.h file in my BSP directory...
>
>  I don't know what else should I change or add....
>
>  please post your comments......
>  Thank you..........
>
>  kingseft.
>
>
>
>
> Sent via Deja.com http://www.deja.com/
> Before you buy.



---------------------------

Newsgroups: comp.os.vxworks
Subject: UDP Socket Performance
Date: Wed, 14 Feb 2001 14:02:23 +0900
From: =?ks_c_5601-1987?B?sejFwrHU?= 
Organization: Lawrence Berkeley National Laboratory
Message-ID: <96d4aa$sg3$1@overload.lbl.gov>

Hi,

I am using Tornado2/VxWorks5.4 on two MCP750 targets. I recently tried to 
test udp socket performance. The test program has 64 UDP connections which send
and receive 50 packets per second between the two targets. All connections are appear to
operate normaly, but more the half of the packets are not received.  

Can anyone provide some help?

Taegue Kim
Contela, Inc.
9-1, Sunae-dong, Bundang-gu, Sungnam-si,
Kyunggi-do, 463-784, Korea
Phone: +82-31-710-4834
Email: tgkim@contela.com

---------------------------

Newsgroups: comp.os.vxworks,comp.dcom.net-management,comp.protocols.snmp
Subject: Re: How about that RapidControl?
Date: Tue, 13 Feb 2001 23:17:45 -0600
From: John Bundy 
Organization: CoreComm LTD - Chicago, IL
Message-ID: <3a8a1716$0$70806$1dc6e903@news.corecomm.net>
References: <3A89C99C.ADFA79@netpliance.net>

Rapid Control seemed expensive, both up front software cost and run time royalties
considered.

The person who came out to demo the product had never actually run the web server
on anything other than a PC.  We didn't get a detailed answer about how the html
files get to the embedded environment.

It looks like a good idea if you have a MIB and are also doing a web interface and
a CLI.

John Bundy


Kerry Kimbrough wrote:

> Have you used the RapidControl toolset for unified access to SNMP MIBs
> via HTTP/CLI/etc.? (Formerly from RapidLogic, now offered by Wind
> River). Does it work well? What are its pros and cons? Any tips much
> appreciated.
>
> --
>
> Regards,
>
> Kerry
>
> ------------------------------------------------------------------------------
> mailto:kerryk@netpliance.net
> ------------------------------------------------------------------------------


---------------------------

Newsgroups: comp.os.vxworks
Subject: Interrupt:dc0 no carrier error
Date: Wed, 14 Feb 2001 10:44:31 +0530
From: "Vijayaraghavan Rajasekaran" 
Organization: Lawrence Berkeley National Laboratory
Message-ID: <96d7qr$d2$1@overload.lbl.gov>

This is a multi-part message in MIME format.
- --------------3DE79F3D5F5BCF58A013FCED
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit

VxWorks/Tornado

Hi all,
        we are using Vxworks with Mcp750 Bsp. During and after boot up
continuously "Interrupt:dc0 no carrier error" is displayed . This error
appears whenever host tools or any node on the lan tries to contact
target over LAN.We even Installed "Tornado-Comp-Drv-December2000 release
v070".
After that error stops appearing , but we cud not connect TELNET/FTP
servers.
Pls suggest us the  remedy , ur immediate response will be highly
appreciated .

Thanks & Regards,
R.vijayaraghavan



- --------------3DE79F3D5F5BCF58A013FCED
Content-Type: text/x-vcard; charset=us-ascii;
 name="vijayaraghavan.raja.vcf"
Content-Transfer-Encoding: 7bit
Content-Description: Card for vijay
Content-Disposition: attachment;
 filename="vijayaraghavan.raja.vcf"

begin:vcard 
n:Rajasekaran;Vijayaraghavan
x-mozilla-html:TRUE
url:www.wipro.com
org:WiproTechnologies;Embedded & Internet Access Solution
version:2.1
email;internet:vijayaraghavan.raja@wipro.com
adr;quoted-printable:;;Wipro Ltd. No. 111, Mount Road=0D=0A  Guindy=0D=0A=0D=0A=0D=0A=0D=0A=0D=0A=0D=0A=0D=0A;Chennai;TamilNadu;600032;India
fn:R.Vijayaraghavan
end:vcard

- --------------3DE79F3D5F5BCF58A013FCED--


---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: MIPS Evaluation Board.
Date: Wed, 14 Feb 2001 16:37:05 +0900
From: "Sung Won Ho" 
Organization: ELIMnet Internet News
Message-ID: <96dcha$gg6$1@news.elim.net>
References: <969g9r$lk0$1@overload.lbl.gov>

Hi

Why don't you try to contact V3 semiconductor.
Visit www.vcubed.com for more information.

Regards,
Sung.


"Joseph Sebastian"  wrote in message
news:969g9r$lk0$1@overload.lbl.gov...
> Hi all,
> I am looking for a MIPS evaluation board with PCI interface.
> Is there a source from where I can get an evaluation board with VxWorks
> BSP.
> Joseph Sebastian.
>
>



---------------------------

Newsgroups: comp.os.vxworks
Subject: File system on VxWorks
Date: Wed, 14 Feb 2001 16:47:03 +0900
From: "Sung Won Ho" 
Organization: ELIMnet Internet News
Message-ID: <96dd40$ggu$1@news.elim.net>

Hi all,

Is it possible for me to write and bind my own file system on VxWorks?
Does VxWorks offer any APIs or something for customized file system?
Is there any resources ralating this subject?

Thanks in advance.
Sung.






---------------------------

Newsgroups: comp.os.vxworks
Subject: re: host/target symbol sync
Date: Wed, 14 Feb 2001 08:42:26 +0100
From: "Michael Lawnick" 
Organization: Lawrence Berkeley National Laboratory
Message-ID: <200102140742.IAA31230@net.softec.de>

Hi Ken,

this messages indicate that there occured problems on reaching the host 
symbol table by the target. AFAIK here are three major problems (assuming 
you are using WinNt as HostOS):
First: you might have specified localRegistry - change it to remoteRegistry 
with the ipAddress/targetName of your host.
Scnd: you might use 2 NIC or IP-addresses on your host. Connect your 
target to the first one.
Third: your target resides in a different subnet - haven't ever found a solution 
for this situation.

Are you sure you need the synchronisation ? Building-in the symbol table 
into vxWorks will consume the same amount of target memory. You only will 
lose the automatic update of symbols created after boot, but this will only be 
needed when working with target shell and host shell at the same time.

HTH
regards,
Michael

> Submitted-by: "Baron, Ken" 
> 
> Hi,
> 
> I am using Tornado2/VxWorks5.4. I recently tried using the target shell and
> synchronizing the target server symbol table with the target shell symbol
> table. Whenever I start the target server I get the following printout on my
> console:
> -> Fatal WTX error (0x10136), synchronization stopped
> Fatal WTX error (0x1012f), synchronization stopped
> Otherwise, both target and host shell are working fine. I am using Win95 and
> a Radstone PPC board (these are the only 2 items on the ethernet).
> 
> Can anyone provide some help?
> 
> Ken Baron
> BAE SYSTEMS
> Reconnaissance & Surveillance Systems
> 300 Robbins Lane, Syosset, NY 11791
> Voice: (516) 349-2132
> Fax: (516) 349-2211
> Email: ken.baron@lmco.com


Michael Lawnick, SOFTEC GmbH

==============================================
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: VxWorks
Date: Wed, 14 Feb 2001 13:45:44 +0530
From: "Gokul Gopalasamy" 
Organization: Lawrence Berkeley National Laboratory
Message-ID: <96ders$9gd$1@overload.lbl.gov>

Hi all,

I'm a beginner in VxWorks. Can anyone suggest me some links or whitepapers regarding
VxWorks. Suggestions will be sincerely appreciated.

Thanks and Regards
Gokul

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
"Have the courage to face the Truth"
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Gokul Gopalasamy,
Global R&D,
Wipro Technologies,
Tel: 91-44-4500200-209   Ext: 2287
Fax: 91-44-4500250
Email: gokul.gopalasamy@wipro.com
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~



---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: Compile Errors
Date: Wed, 14 Feb 2001 09:13:38 +0100
From: "Geurt Vos" 
Organization: Rohill Technologies BV
Message-ID: 
References: <3A89A80F.CFFB026E@address.com>
Sender: rohill2@rohill.nl


It should be either:
    ADD_TEST(&ComplexTester::TestAdd);
or:
    #define ADD_TEST(test) (suite.Add(&test,#test));

If you compile it with a newer GCC compiler (2.95.2),
you'll get a (decent) warning:
- ---
test.cpp: In function `void X()':
test.cpp:54: assuming & on `ComplexTester::TestAdd'
- ---

HTH,
Geurt






> Hi All,
>
> Since the response to my last question on CppUnit is deafening (:)), so
> let's me ask the question in another format. I have the class defined
> as:
>
> --------------------------------------------------
> #include "Test.h"
> #include 
> using std::string;
>
> class TestSuite: public Test
> {
>   public:
>    TestSuite(const char* name);
>
>   void AddTest(Test);
>  template void Add(void (Fixture_::* fp)(),string name)
>  {
>   AddTest(Test(new TestCaller(fp,name)));
>  }
>   bool Run(TestResult&);
>   const char* const Name() const;
> };
>
> #define ADD_TEST(test) (suite.Add(test,#test));
> -------------------------------------------------------
> I invoked it as follow:
>
> Test ComplexTestSuite()
> {
>   TestSuite suite("ComplexTestSuite");
>
>   ADD_TEST(ComplexTester::TestAdd);
>   return suite;
> }
> --------------------------------------------------
> But when I compiled it I got the following errors (adding -ansi produced
> the same result):
> Note: I did not include the test.h as I don't think it caused the error.
>
> ccsimpc -U__WINNT -UWIN32 -U__WINNT__ -UWINNT -U__MINGW32__ -U_WIN32
> -U__WIN32 -U__WIN32__
>  -U__WIN32 -mpentium  -nostdinc -g -nostdlib -fno-builtin -fno-defer-pop
> -Wall -DRW_MULTI_
> THREAD -D_REENTRANT -I. -ID:\Tornado\target\h -DCPU=SIMNT
> -cD:\Tornado\target\proj\Cpp\TestComplex.cpp
> D:\Tornado\target\proj\Cpp\TestComplex.cpp: In function `class Test
> ComplexTestSuite()':
> D:\Tornado\target\proj\Cpp\TestComplex.cpp:35: no matching function for
> call to `TestSuite::Add (void (ComplexTester::)(), char[23])'
> D:\Tornado\target\proj\Cpp\TestSuite.h:34: candidates are:
> TestSuite::Add(void (ComplexTester::*)(),
> basic_string,__d
> efault_alloc_template >)
> make: *** [TestComplex.o] Error 0x1
>
> ================================================
> I wish someone here could help me and I would really appreciate.
>
> Cheers.
>
>



---------------------------

Newsgroups: comp.os.vxworks
Subject: [Q] MUX interface between network and link layers
Date: Wed, 14 Feb 2001 17:48:46 +0900
From: "SungJin Lee" 
Organization: Korea Telecom
Message-ID: <96dgnc$ekv$1@news2-2.kornet.net>
Reply-To: "SungJin Lee" 

I'm just getting start with the VxWORK and studying on network related
stuff.
Includes PPP and L2TP modules

MUX interface provides API between network layer and data link layer.
It is clear with the figures of vxworks and tornado articles.

However, i got come questions for the network structure.
The MUX seems to be located right above the device driver
and right below the network layer.. cuz the device driver directly
register with the MUX and them are depicted on the pictures on the manual.

then, where can i insert network layer modules ?

It said                              -------------------------------
                                      |          network layer ( IP )
|
                                       -------------------------------
                                      |                   MUX
|
                                       _______________________________
                                      |         END,    Device driver
|
                                       -------------------------------


In fact, IP protocol can communicate with PPP via MUX  ?
Or, all nework protocols are located over MUX and have different priorities
to take the packet first.... Is the Device driver is network layer protocol
itself ?

I can include PPP modules in the vxworks image with INCLUDE_PPP options
but can't understand where the PPP works on. ( is the PPP is device driver
itself ? )






---------------------------

Newsgroups: comp.os.vxworks
Subject: (beginer) Problem "catching" a cPCI interrupt
Date: Wed, 14 Feb 2001 11:07:57 +0200
From: "David Ganor" 
Organization: Bezeq International Ltd.
Message-ID: <3a8a4aab@news.bezeqint.net>

Hello,

Some backgroug info, first:
1. VxWorks 5.4
2. Target: Motorola with CPU PPC750
3. BSP version: 1.2/0
4. Tornado II

I have a system board with a cPCI interface, on which a PLX9054 chip is the
PCI bridge (the board has a local bus...).
I have configured the CPU_PCI_MEM_SIZE to be 32MBytes and the board uses
about 10MBytes.
I have wrote an ISR to be activated when the board, througth the PLX chip,
drives an interrupt. There are no problems configuring the PLX and the
interrupt
line is "active".
When running the code the ISR isn't activated, thought the intConnect()
returns
with OK. I used intConnecte() like that:
    lvl = PCI_cfg->DEV_INT_LINE;
    ...
    if (intConnect (INUM_TO_IVEC (lvl), (VOIDFUNCPTR) ISR, (int) this) !=
OK)
    {
        printf("[PLX] ERROR - Failed to connect interrupt to ISR!\n");
    }
Could anyone sujest an insight? What am I doing wrong?
Maybe the PLX9054 isn't configured right? -> What are the main configuration
regarding
the described problem?

Thanks in advance.
David Ganor



---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: *UPDATED* Gcc 2.95.3 for VxWorks-PPC available....
Date: Wed, 14 Feb 2001 10:28:55 +0100
From: Roger 
Message-ID: <3A8A4FD7.DE9A720F@here.there.every.where>
References:  <3A8462D3.1AADFB7C@mayannetworks.com>  <3A87DEB3.725A6DDB@opq.se> <2e7i6.146$yi4.164057@newsr1.u-net.net>

Dave Korn wrote:
> 
> Roger wrote in message <3A87DEB3.725A6DDB@opq.se>...
> >Dave Korn wrote:
> >>
> >> "Kevin Cline"  wrote in message
> news:3A8462D3.1AADFB7C@mayannetworks.com...
> >> > Dave Korn wrote:
> >> >
> >> > >  Finally I've recompiled the whole thing for i386, so it should run
> on any
> >> > > CPU!
> >> >
> >> > Can you document the process so that others can repeat your work for
> different
> >> > targets?  Thanks.
> >> >
> >>
> >>    There's already an outline of the build procedure on the FAQ/Docs page
> that should work fine for any other VxWorks targets; just
> >> change the --target= parameter in the configure commands for the cross
> binutils and cross compiler
> >
> >I am new to this. Where are the FAQ pages?
> 
>   On my website as referred to in that other post I replied to you in.  I
> don't want to keep quoting the URL, it makes me feel like a spammer, so
> here's a pointer to that post instead:
> 
> news:<46Wg6.17$yi4.12215@newsr1.u-net.net>

If you tried to reply directly, that did not work. My reply address in
news postings is not a real one. For various reasons. So, I did not see
it. And here I thought no one wanted to talk to me...

My ISP says that there is no such news article. I think they keep these
things 30 seconds or so. I have never been able to get articles this
way.

Do you think you could send a copy to: roger@opq.se

Thanks in advance.

---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: UDP Socket Performance
Date: Wed, 14 Feb 2001 10:36:12 +0100
From: Roger 
Message-ID: <3A8A518C.65771E0D@here.there.every.where>
References: <96d4aa$sg3$1@overload.lbl.gov>

=B1=E8=C5=C2=B1=D4 wrote:
> =

> Hi,
> =

> I am using Tornado2/VxWorks5.4 on two MCP750 targets. I recently tried =
to
> test udp socket performance. The test program has 64 UDP connections wh=
ich send
> and receive 50 packets per second between the two targets. All connecti=
ons are appear to
> operate normaly, but more the half of the packets are not received.

Some say the 'U' in UDP means unreliable. These packets are not
guaranteed to arrive. You must have your own protocol for this. In a
similiar system with one coax between two computers, I was suprised that
packets would go missing. This happened mainly when the system was in a
vehicle operating on vehicle power. Still, we had to add a protocol
layer. It was simple in our case as the packets are distance related. If
one goes missing, there is no way to know this becaust the vehicle may,
in fact, be stopped. Still, we came up with a nice solution.

Depending on the packet characteristics, you may be happy with TCP. Just
don't use one socket for bi-directional comms. This is a slow down. TCP
makes sure packets arrive. But, it your net is noisy, TCP may slow down
as part of an attempt to let the network regain sanity. So, you may also
need to see if the network is poor.



> =

> Can anyone provide some help?
> =

> Taegue Kim
> Contela, Inc.
> 9-1, Sunae-dong, Bundang-gu, Sungnam-si,
> Kyunggi-do, 463-784, Korea
> Phone: +82-31-710-4834
> Email: tgkim@contela.com

---------------------------

Newsgroups: comp.os.vxworks
Subject: Increase shared memory size
Date: Wed, 14 Feb 2001 11:25:07 +0100
From: Marc Waber 
Organization: STS Switching Test Solutions
Message-ID: <3A8A5D03.51C4FEBB@stest.ch>

Hi,

We are using shared memory with a CPV5350 as master and a MCPN765 as
slave.
I tried to increase the shared memory on the master by changing
SM_MEM_SIZE.
TO accomodate this larger shard memory, I have to move RAM_LOW_ADRS
accordingly (at least that's what I think...)
in config.h and Makefile.
After buildiing vxWorks and bootrom, what happens with these parameters
changed is that the image loads and starts executing, but whenever it
starts to configure the shared memory, the system crashes and reboots.

Any ideas how to properly make those changes?

Thanks,
Marc

---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: [Question] usrAppInit() function in Vxworks Project
Date: Wed, 14 Feb 2001 10:48:15 -0000
From: "Kelvin Lawson" 
Message-ID: <96dnnc$kdd9u$1@ID-56276.news.dfncis.de>
References: <90fj7h$e4o$1@nnrp1.deja.com> <4Bli6.72893$bb.1084753@news1.rdc1.tx.home.com>

The problem with this is that you have to add the #include "config.h" line
every time you generate a new project. Otherwise you could add it to
comps/src/stubAppInit.c, and all new projects would use it. This is also a
problem in that you may not want it included for all of your BSPs. Does
anyone have a neater way round this ?

I'd rather use my own usrConfig.c than mess around with these
project-generated files. However I've tried the following in my Makefile:

USRCONFIG = usrConfig.c

and it still builds prjConfig.o instead of usrConfig.o.

Any ideas why ?

Cheers,
Kelvin.


"Eddy Raineri"  wrote in message
news:4Bli6.72893$bb.1084753@news1.rdc1.tx.home.com...
> The problem is that the usrApplInit.c file does not include config.h or
> prjconfig.h.  Because of this INCLUDE_USER_APPL is not defined.  Modify
the
> usrApplInit.c file to include config.h and it will work
>
> Good luck
> E
>  wrote in message
news:90fj7h$e4o$1@nnrp1.deja.com...
> >
> >
> >  Hi, VxWorks Developers.
> >
> >  I have one question about usrAppInit() in VxWorks Project file.
> >
> >  I added following lines to my BSP directory's config.h
> >
> > #define INCLUDE_USER_APPL
> > #define USER_APPL_INIT taskSpawn ("sample_app", 20, 0,
> > 2000, (FUNCPTR) sample_app, 0,0,0,0,0,0,0,0,0,0)
> >
> >  of course, I defined sample_app function also....
> >  and I create new bootable project based on my BSP directory.
> >  So, I got some project files and I found that usrAppInit()
> >
> >  void usrAppInit (void)
> >     {
> > #ifdef USER_APPL_INIT
> > USER_APPL_INIT; /* for backwards compatibility */
> > #endif
> >
> >     /* add application specific code here */
> >     }
> >
> >  When I made ROM based Vxworks (ie. Vxworks_rom.hex) and succeeded in
> > booting. but usrAppInit() function never run....
> >  so, I checked whether checked USER_APPL_INIT is defined.
> >
> >  unfortunately, in my usrAppInit() fucntion .
> >  this function said that ......
> >
> >  void usrAppInit (void)
> >     {
> > #ifdef USER_APPL_INIT
> > USER_APPL_INIT; /* for backwards compatibility */
> >          printf("defined.......");
> > #else
> >          printf("undedined.......");
> > #endif
> >
> >     /* add application specific code here */
> >     }
> >
> >  the result is "undefined......."
> >
> >  Is something that I forgot ??
> >
> > I defined
> > #define INCLUDE_USER_APPL
> > #define USER_APPL_INIT
> > in my config.h file in my BSP directory...
> >
> >  I don't know what else should I change or add....
> >
> >  please post your comments......
> >  Thank you..........
> >
> >  kingseft.
> >
> >
> >
> >
> > Sent via Deja.com http://www.deja.com/
> > Before you buy.
>
>



---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: VxWorks
Date: Wed, 14 Feb 2001 06:38:21 -0800
From: Joe Durusau 
Organization: Lockheed Martin Corporation
Message-ID: <3A8A985D.650B83B2@lmco.com>
References: <96ders$9gd$1@overload.lbl.gov>

www.wrs.com

Speaking only for myself,

Joe Durusau


Gokul Gopalasamy wrote:
> 
> Hi all,
> 
> I'm a beginner in VxWorks. Can anyone suggest me some links or whitepapers regarding
> VxWorks. Suggestions will be sincerely appreciated.
> 
> Thanks and Regards
> Gokul
> 
> ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
> "Have the courage to face the Truth"
> ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
> Gokul Gopalasamy,
> Global R&D,
> Wipro Technologies,
> Tel: 91-44-4500200-209   Ext: 2287
> Fax: 91-44-4500250
> Email: gokul.gopalasamy@wipro.com
> ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

---------------------------

End of New-News digest
**********************


From vxwexplo-errs@csg.lbl.gov  Wed Feb 14 08:22:34 2001
From: "Michael Baumann" 
Date: Wed Feb 14 08:22:36 PST 2001
Subject: Different Problems with Host/Target sync. Torando 2/Vx5.4

This is not the same as Ken Barons's problem, as
we don't get the "fatal error" message

Vxworks 5.4:
Problem is as follows: A simple load in the windshell
results in 3
"warning: WTX timeout" errors. No symbols are
loaded into the target. No complaint about syncronization failing.

Following symboltable related options configured in:
CONFIGURATION_5_2
NET_SYM_TBL
SYM_TBL
SYM_TBL_SYNC

Sync task shows up as attached to the target.
Tgtsvr started with synchronization flag.
Devel environment: Solaris.
Target MVME167
Used to work just fine.

Any ideas?
--
Michael Baumann  Optivus Technology Inc.|Loma Linda University Medical
Center
San Bernardino, California. (909)799-8308 |Internet: baumann@llumc.edu




From vxwexplo-errs@csg.lbl.gov  Wed Feb 14 16:51:41 2001
From: "Girish V. Gulawani" 
Date: Wed Feb 14 16:51:43 PST 2001
Subject: Re: power pc -- 64bit operation

> ---------------------------
> 
> Newsgroups: comp.os.vxworks
> Subject: power pc -- 64bit operation
> Date: Tue, 13 Feb 2001 16:53:42 GMT
> From: "Lou Kale" 
> Organization: EarthLink Inc. -- http://www.EarthLink.net
> Message-ID: 
> 
> I'm using a ppc604 (750) with Tornado2.
> 
> How do I know if it is running code in 64 bit mode?

> Is there a gnu compiler switch to turn 64 bit on?
I think the current GNU CC of T2 does not have this option. But the next
version (2.95.2+) has -mpowerpc64 option. Do you have 64-bit BSP/kernel?

HTH.
Girish.


From vxwexplo-errs@csg.lbl.gov  Wed Feb 14 18:07:42 2001
From: "Ganesan, M" 
Date: Wed Feb 14 18:07:44 PST 2001
Subject: VxWorks - Minimum Initializations for PCI

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_01C096F6.8909DCA0
Content-Type: text/plain;
	charset="iso-8859-1"

Hi,

I am using MPC 8240 Kahlua based evaluation board. When I am trying to plug
PM/PPC (artesyn board) into the unused PCI slot of MPC 8240 board, I am not
able to boot MPC 8240 board. I have added code in syslib.c to issue
PCI_CMD_MASTER_ENABLE to PM/PPC who waits for that. Other than this, I have
not included any code for memory space mapping as I am not going to access
any memory from MPC 8240 as of now. 

My questions are

1. What are all the minimum initialization which are mandatory to be done
for plugging an PCI device in the bus from bus arbiter point of view.

2. If I need to access memory from both end (from MPC 8240 to PM/PPC and
vice-versa), with MPC8240 as bus arbiter, what are all the steps I need to
follow. What init codes I need to add in both bsp?

Please help me in this regard

bye

ram


------_=_NextPart_001_01C096F6.8909DCA0
Content-Type: text/html;
	charset="iso-8859-1"











Hi,


I am using MPC 8240 Kahlua based evaluation board. When I am trying to plug 
PM/PPC (artesyn board) into the unused PCI slot of MPC 8240 board, I am not able 
to boot MPC 8240 board. I have added code in syslib.c to issue 
PCI_CMD_MASTER_ENABLE to PM/PPC who waits for that. Other than this, I have not 
included any code for memory space mapping as I am not going to access any 
memory from MPC 8240 as of now. 


My questions are


1. What are all the minimum initialization which are mandatory to be done for 
plugging an PCI device in the bus from bus arbiter point of view.


2. If I need to access memory from both end (from MPC 8240 to PM/PPC and 
vice-versa), with MPC8240 as bus arbiter, what are all the steps I need to 
follow. What init codes I need to add in both bsp?


Please help me in this regard


bye


ram


------_=_NextPart_001_01C096F6.8909DCA0--


From vxwexplo-errs@csg.lbl.gov  Wed Feb 14 21:32:43 2001
From: "Girish V. Gulawani" 
Date: Wed Feb 14 21:32:45 PST 2001
Subject: Re: VxWorks - Minimum Initializations for PCI

Hi, Ram.


> 1. What are all the minimum initialization which are mandatory to be done
> for plugging an PCI device in the bus from bus arbiter point of view.

The WindRiver provides the PCI AutoConfiguration driver w/ source.
During bootup the PCI devices should come up initialized - semi plug and
play. 

Anyway, typically to configure the PCI device the Configuration header
is filled which is nothing but the PCI initialization. The steps other
than device specific init are to fill up following headers -
Command Register : Enable IO/Memory, Parity Error. Also see if your
device wants to be bus master.
Cache Line : Typically 32.
Latency    : Full, 0xFF.
Interrupts : Almost devices use INTA. Again the PCI library mentioned
above, provides routines to allocate and manage interrupts.
BARs       : Remember the PCI BAR values are physical addresses as seen
by PCI device and not virtual addresses seen from CPU. So for example IO
add offset 0x80000000 and to access memory you add offset of 0xC0000000
to the physical address to get virtual address. To get physcial from
virtual, I bet you know what to do! The actual memory map depends on
your board configuration. Refer your config.h it should be somewhere
there. 

> 2. If I need to access memory from both end (from MPC 8240 to PM/PPC and
> vice-versa), with MPC8240 as bus arbiter, what are all the steps I need to
> follow. What init codes I need to add in both bsp?
Once you configure the PCI header with appropriate BAR and other values.
The IO and/or Memory access is performed as given in BAR explanation
above.

Also a mundane note but very important, PCI is a Little Endian bus. 

HTH.
Girish.


From vxwexplo-errs@csg.lbl.gov  Thu Feb 15 04:03:12 2001
From: Vxworks Exploder 
Date: Thu Feb 15 04:03:14 PST 2001
Subject: comp.os.vxworks newsdigest

Comp.Os.Vxworks Daily Digest    Thu Feb 15 04:03:05 PST 2001

        Subject: RE: Optic
        Subject: Re: UDP Socket Performance
        Subject: How to repeat a target shell command?
        Subject: Re: How to repeat a target shell command?
        Subject: problem using shared memory network
        Subject: Re: How to repeat a target shell command?
        Subject: Forcing enum size to byte size
        Subject: Different Problems with Host/Target sync. Torando 2/Vx5.4
        Subject: Re: please help me
        Subject: Use std::string
        Subject: Re: Problem with absolute include path Tornado project environment
        Subject: Re: [elfHex] option
        Subject: Re: UDP Socket Performance
        Subject: STDIN problem on target
        Subject: T2 man pages???
        Subject: Re: T2 man pages???
        Subject: re: power pc -- 64bit operation
        Subject: Re: T2 man pages???
        Subject: VxWorks - Minimum Initializations for PCI
        Subject: FEC(10/100) of MPC855T connects to switch
        Subject: re: VxWorks - Minimum Initializations for PCI
        Subject: Re: vxWorks 5.4/mv2700: networks goes north (down) - help
        Subject: Re: CppUnit port to VxWorks
        Subject: Re: UDP Socket Performance
        Subject: Re: UDP Socket Performance
        Subject: pthreads
        Subject: Re: Standalone applications
        Subject: Q: Build erro in VxMP ?

-------------------------------------------------------

Newsgroups: comp.os.vxworks
Subject: RE: Optic
Date: Wed, 14 Feb 2001 14:03:01 +0200
From: "Bosis, Ilia" 
Organization: Lawrence Berkeley National Laboratory
Message-ID: <96dstq$hu2$1@overload.lbl.gov>



> -----Original Message-----
> From: vxwexplo@lbl.gov [mailto:vxwexplo@lbl.gov]
> Sent: Wednesday, February 14, 2001 12:32 AM
> To: vxworks_users@csg.lbl.gov
> Subject: Optic
> 
> 
> Submitted-by vxwexplo-errs@csg.lbl.gov  Tue Feb 13 14:18:01 2001
> Submitted-by: "Tan, Tat Kin" 
> 
> Vxworks/Tornado
> 
> Hi All,
> 
> Am new to the VxWorks world and hoping to get some info from you guys.
> 
> Basically i need to integrate an optical fibre PHY into a 
> board. The board
> has already been running a BSP and what do i need to do to 
> this BSP to make
> this PHY works? I mean, which part of this BSP i need to 
> modify? For sure
> the Device Driver needs modification. Perhaps add the optical 
> device driver
> ? What more? Appreciated helps. millions of thanks.

Device driver , for sure.
B.T.W. If You need an example of how to write it,
go to target/unsupported/drv in the Tornado tree
and look for some driver source. 
Ethernet one is probably the closest similar.

Before this, You need to map the device into the CPU address space 
to make it accessible.
For PCI or VME devices this is rather a non-trivial task.
You may need to configure chip select(s), bus bridge(s)
and in case of PCI device also the device' PCI interface -
all depends on Your specific HW & BSP.
 
The simplest way (from my experience) 
is to find an unused space of the appropriate mapping
in the default/currently used configuration of Your BSP 
and map the device there.
Thus You save a lot of hassle of configuring complex 
and too often poorly documented hardware.
Not to mention that incorrect configuration can easely make 
the board I/O unaccessible so that the whole board looks dead
(no debugger connection, no console, nothing :(  ).

Hopefully, that's all.


> 
>    with warm regards,
>    Tat Kin, Tan
>
 
Success for You.
		Ilia

iliab at engineer dot com
- ----------------------
Complexity comes for free, Simplicity you have to work for.

---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: UDP Socket Performance
Date: Wed, 14 Feb 2001 12:54:45 -0000
From: "Dave Korn" 
Organization: Lumber Cartel (tinlc) Members #2234-2237 (owing to browser refresh)
Message-ID: <78vi6.212$yi4.201420@newsr1.u-net.net>
References: <96d4aa$sg3$1@overload.lbl.gov>

±èű wrote in message <96d4aa$sg3$1@overload.lbl.gov>...
>Hi,
>
>I am using Tornado2/VxWorks5.4 on two MCP750 targets. I recently tried to
>test udp socket performance. The test program has 64 UDP connections which
send
>and receive 50 packets per second between the two targets. All connections
are appear to
>operate normaly, but more the half of the packets are not received.

  How does 64 * 50 * packetsize compare to the bandwidth of your network ?

       DaveK
- --
They laughed at Galileo.  They laughed at Copernicus.  They laughed at
Columbus. But remember, they also laughed at Bozo the Clown.



---------------------------

Newsgroups: comp.os.vxworks
Subject: How to repeat a target shell command?
Date: Wed, 14 Feb 2001 13:31:28 +0100
From: "Eelco Chaudron" 
Message-ID: <982155030.12990.0.pluto.d4ee6f20@news.demon.nl>

Hi All,

I would like to execute a single command from the target shell a couple of
times with an increasing parameter.

For example I would like to call the function dumpBlock  about 100
times starting with 0 going up to 99.

I can simply type:
- ->dumpBlock 0
- ->dumpBlock 1
..
..
..
- ->dumpBlock 99

This is not very easy to do :-( So can I do this another way? I tried the
repeat command, but that won't work, as it does not increase the counter.

example:
- ->x=0
- -> repeat 100,dumpBlock,x++

But this would only increase x once...

Any ideas?

Thanks,

Eelco



---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: How to repeat a target shell command?
Date: Wed, 14 Feb 2001 15:19:46 +0100
From: "Jorgen Jakobsen" 
Organization: Posted Courtesy of Tele Danmark or one of its European Subsidiaries
Message-ID: <96e48d$d6g$1@news.inet.tele.dk>
References: <982155030.12990.0.pluto.d4ee6f20@news.demon.nl>

Hi Eelco

You type "esc k" (not at the same time) to get the last command. Then
"Shift-A (at the same time) to edit the line from the end.

Page 194 in Tornado Users Guide :)

/Jørgen

"Eelco Chaudron"  wrote in message
news:982155030.12990.0.pluto.d4ee6f20@news.demon.nl...
> Hi All,
>
> I would like to execute a single command from the target shell a couple of
> times with an increasing parameter.
>
> For example I would like to call the function dumpBlock  about 100
> times starting with 0 going up to 99.
>
> I can simply type:
> ->dumpBlock 0
> ->dumpBlock 1
> ..
> ..
> ..
> ->dumpBlock 99
>
> This is not very easy to do :-( So can I do this another way? I tried the
> repeat command, but that won't work, as it does not increase the counter.
>
> example:
> ->x=0
> -> repeat 100,dumpBlock,x++
>
> But this would only increase x once...
>
> Any ideas?
>
> Thanks,
>
> Eelco
>
>



---------------------------

Newsgroups: comp.os.vxworks
Subject: problem using shared memory network
Date: 14 Feb 2001 14:51:42 GMT
From: sagi486@deja.com (sagi shain)
Organization: QI
Message-ID: 

I am having the following problem - I want to use shared memory for udp 
between 2 ppc on cpci bus
as long as I am using the standard configuration it all goes well. 
but in our SW we need to call sysPciAutoConfig() after thet when I do 
smNetShow() I am getting uninitialized. 

does any one encountered the same problem and how can I solve this 
any help is welcome

---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: How to repeat a target shell command?
Date: Wed, 14 Feb 2001 08:49:52 -0600
From: "M. Kotiaho" 
Organization: LMMFC - D
Message-ID: <3A8A9B10.819CE6BE@m_a_i_l.com>
References: <982155030.12990.0.pluto.d4ee6f20@news.demon.nl>

Eelco Chaudron wrote:


> This is not very easy to do :-( So can I do this another way? I tried the
> repeat command, but that won't work, as it does not increase the counter.
>
> example:
> ->x=0
> -> repeat 100,dumpBlock,x++
>
> But this would only increase x once...
>
> Any ideas?
>
> Thanks,
>
> Eelco

Hi Eelco,

You can use the built in Tcl interpreter for looping.  From the
shell, enter ? to get into Tcl mode.

- -> ?
tcl> for {set x 0} {$x<100} {incr x} {
tcl>   shParse "dumpBlock $x"
tcl> }
tcl> ?
- -> ...

The "?" toggles between Tcl and C mode.

HTH,
Markku Kotiaho



---------------------------

Newsgroups: comp.os.vxworks
Subject: Forcing enum size to byte size
Date: Wed, 14 Feb 2001 16:43:25 +0100
From: "Henrik Ravn" 
Organization: Posted Courtesy of Tele Danmark or one of its European Subsidiaries
Message-ID: <96e94v$rqd$1@news.inet.tele.dk>

Hi,

A gcc option is -fshort-enums which result in enums being as small as
possible. This options works for me, an other ways should be to specify
"__attribute__(( packed ))" together with the enum declaration, this doesn't
work for me.

Has anybody tried using the last method?

I'm running Tornado 2.0.1 on a Windows platform.

Henrik RAvn



---------------------------

Newsgroups: comp.os.vxworks
Subject: Different Problems with Host/Target sync. Torando 2/Vx5.4
Date: Wed, 14 Feb 2001 08:17:12 -0800
From: "Michael Baumann" 
Organization: Lawrence Berkeley National Laboratory
Message-ID: <96eega$q4h$1@overload.lbl.gov>

This is not the same as Ken Barons's problem, as
we don't get the "fatal error" message

Vxworks 5.4:
Problem is as follows: A simple load in the windshell
results in 3
"warning: WTX timeout" errors. No symbols are
loaded into the target. No complaint about syncronization failing.

Following symboltable related options configured in:
CONFIGURATION_5_2
NET_SYM_TBL
SYM_TBL
SYM_TBL_SYNC

Sync task shows up as attached to the target.
Tgtsvr started with synchronization flag.
Devel environment: Solaris.
Target MVME167
Used to work just fine.

Any ideas?
- --
Michael Baumann  Optivus Technology Inc.|Loma Linda University Medical
Center
San Bernardino, California. (909)799-8308 |Internet: baumann@llumc.edu



---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: please help me
Date: 14 Feb 2001 12:57:00 -0500
From: vanbaren@falcon.si.com (vanbaren_gerald)
Organization: Smiths Industries
Message-ID: <3a8ac6ec@news.si.com>
References:  <3A2C18DB.A32EAA47@hotmail.com>

Ben  writes:

>hi
>    i am  trying to install vxworks image on my target board(ppc), but i

>am getting the following error. can you please tell me where am i going
>wrong. i am also attaching the information which i am getting in ftp.
>i will be grateful to you for any assistance in this regard.
>thank you

>following is the information i am getting in terminal emulator
>[VxWorks Boot]: @

>boot device          : dc
>processor number     : 0
>host name            : U.S.U..usu.edu
>file name            : c:\tornado\target\config\mtx603\vxWorks
>inet on ethernet (e) : 129.123.4.127:ffffff00
>host inet (h)        : 129.123.4.237
>user (u)             : sl149
>ftp password (pw)    : goud
>flags (f)            : 0x0
>target name (tn)     : tornado

>Attaching network interface dc0... done.
>Attaching network interface lo0... done.
>Loading... 670192

>......and it gets stuck here...

>the follwoing information in ftp is
>[L 0010] 11/30/00 15:46:51 Connection accepted from 129.123.4.127
>[C 0010] 11/30/00 15:46:51 Command "USER sl149" received
>[C 0010] 11/30/00 15:46:51 PASSword accepted
>[L 0010] 11/30/00 15:46:51 User sl149 logged in.
>[C 0010] 11/30/00 15:46:51 Command "TYPE I" received
>[C 0010] 11/30/00 15:46:51 TYPE set to I N
>[C 0010] 11/30/00 15:46:51 Command "PORT 129,123,4,127,4,1" received
>[C 0010] 11/30/00 15:46:51 PORT set to 129.123.4.127 - 1025 (4,1)
>[C 0010] 11/30/00 15:46:51 Command "RETR
>c:\tornado\target\config\mtx603\vxworks" received
>[C 0010] 11/30/00 15:46:51 RETRieve started on file
>c:\tornado\target\config\mtx603\vxworks
>[C 0010] 11/30/00 15:46:58 Transfer aborted
>[G 0010] 11/30/00 15:46:58 Got file
>c:\tornado\target\config\mtx603\vxworks unsuccessfully
>[! 0010] 11/30/00 15:46:58 Timeout while sending data
>[C 0010] 11/30/00 15:46:58 QUIT or close - user sl149 logged ou


Where is your vxWorks download image ORGed?  Most likely, your download
image is crashing your BootROM as it is being downloaded.

My first guess is that you are loading the image on top of your BootROM
or BootROM variables.  It is possible that you are trying to load it
into non existing memory and getting a bus fault too.

gvb

- --
+---------------------------------------------------------------------------+
| Jerry Van Baren / vanbaren_gerald@si.com / Grand Rapids Mi / 616-241-7973 |
|   My employer is a company.  Companies are artifacts of a legal system.   |
|________________Artifacts are incapable of having opinions.________________|

---------------------------

Newsgroups: comp.os.vxworks
Subject: Use std::string
Date: Wed, 14 Feb 2001 13:04:18 -0500
From: Fake 
Organization: UUNET Canada News Reader Service
Message-ID: <3A8AC8A2.5C891CEE@address.com>

Hi All,

Is there any special library file needed to be included when using Use
std::string?

I was able to compile the codes but when I tried to load the .out file
into the VxWorks, I got the following unresolved errors:

(I was using the following comiple options:
ccsimpc -U__WINNT -UWIN32 -U__WINNT__ -UWINNT -U__MINGW32__ -U_WIN32
- -U__WIN32 -U__WIN32__  -U__WIN32 -mpentium -ansi -nostdinc -g -nostdlib
- -fno-builtin -fno-defer-pop -Wall -DRW_MULTI_THREAD -D_REENTRANT)

_length__t18string_char_traits1ZcPCc

_clone__Q2t12basic_string3ZcZt18string_char_traits1ZcZt24__default_alloc_template2b1i03Rep

__t12basic_string3ZcZt18string_char_traits1ZcZt24__default_alloc_template2b1i0$nilRep

_compare__Ct12basic_string3ZcZt18string_char_traits1ZcZt24__default_alloc_template2b1i0RCt12basic_string3ZcZt18string_char_traits1ZcZt24__default_alloc_template2b1i0UiUi

___dl__Q2t12basic_string3ZcZt18string_char_traits1ZcZt24__default_alloc_template2b1i03RepPv

_assign__t18string_char_traits1ZcRcRCc

_replace__t12basic_string3ZcZt18string_char_traits1ZcZt24__default_alloc_template2b1i0UiUiPCcUi

_replace__t12basic_string3ZcZt18string_char_traits1ZcZt24__default_alloc_template2b1i0UiUiRCt12basic_string3ZcZt18string_char_traits1ZcZt24__default_alloc_template2b1i0UiUi

_eos__t18string_char_traits1Zc

___ls__H3ZcZt18string_char_traits1ZcZt24__default_alloc_template2b1i0_R7ostreamRCt12basic_string3ZX01ZX11ZX21_R7ostream

Thanks in advance.

Cheers


---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: Problem with absolute include path Tornado project environment
Date: 14 Feb 2001 13:03:59 -0500
From: vanbaren@falcon.si.com (vanbaren_gerald)
Organization: Smiths Industries
Message-ID: <3a8ac88f@news.si.com>
References: <90iqvq$id$1@nnrp1.deja.com> <90jj6f$ehc$1@neptunium.btinternet.com>


...also, the "make" operation takes place in a subdirectory so you may
need an extra ../ to get to where you want to be.

gvb



"Graham Baxter"  writes:

>Toby,

>Make the back slashes '\' forward slashes '/'.

>Backward slash has a special meaning (like in C) in makefiles.

>From your project window select the C/C++ options and add as
>many -I as you want but with forward slashes.


>I hope this helps.

>--

>Regards,

>Graham Baxter
>Freelance Software Engineer
>vxWorks and pSOS BSPs
>EMAIL: gbaxter@NOSPAMbcs.org.uk

> wrote in message news:90iqvq$id$1@nnrp1.deja.com...
>> We want to use relative inlude paths in the Tornado Project
>> environment. But it isn't possible to use relative include paths in the
>> Tornado project environment, eg: -I..\..\h\x.h
>>
>> Does somebody knows a workaround for this or are there more people with
>> the same problem?
>>
>> Thanx in advance.
>>
>> with kind regards,
>>
>> Toby Hage
>>
>>
>> Sent via Deja.com http://www.deja.com/
>> Before you buy.


- --
+---------------------------------------------------------------------------+
| Jerry Van Baren / vanbaren_gerald@si.com / Grand Rapids Mi / 616-241-7973 |
|   My employer is a company.  Companies are artifacts of a legal system.   |
|________________Artifacts are incapable of having opinions.________________|

---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: [elfHex] option
Date: 14 Feb 2001 13:06:11 -0500
From: vanbaren@falcon.si.com (vanbaren_gerald)
Organization: Smiths Industries
Message-ID: <3a8ac913@news.si.com>
References:  

Read the help (hint: in vxWorks, select Help->Manuals Index and then
use the "Index" tab in the help dialog box)

file:///C:/Tornado/docs/tornado/hutils/elfHex.html

gvb


"sukyum"  writes:

>Hi vxWorkers...

>Could you tell me about "elfHex" options please?

>uasge : elfHex [-a ] [-l] [-v] [-p ] [-s ] file





- --
+---------------------------------------------------------------------------+
| Jerry Van Baren / vanbaren_gerald@si.com / Grand Rapids Mi / 616-241-7973 |
|   My employer is a company.  Companies are artifacts of a legal system.   |
|________________Artifacts are incapable of having opinions.________________|

---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: UDP Socket Performance
Date: 14 Feb 2001 13:11:47 -0500
From: vanbaren@falcon.si.com (vanbaren_gerald)
Organization: Smiths Industries
Message-ID: <3a8aca63$1@news.si.com>
References:  <96d4aa$sg3$1@overload.lbl.gov>


Is someone _always_ ready to receive a packet on the receive side?
vxWorks does not appear to buffer UDP packets at all.  If the packet
arrives in the forest and nobody is there listening for it, it gets
silently discarded (oooh, that was BAD).

gvb


=?ks_c_5601-1987?B?sejFwrHU?=  writes:

>Hi,

>I am using Tornado2/VxWorks5.4 on two MCP750 targets. I recently tried to 
>test udp socket performance. The test program has 64 UDP connections which send
>and receive 50 packets per second between the two targets. All connections are appear to
>operate normaly, but more the half of the packets are not received.  

>Can anyone provide some help?

>Taegue Kim
>Contela, Inc.
>9-1, Sunae-dong, Bundang-gu, Sungnam-si,
>Kyunggi-do, 463-784, Korea
>Phone: +82-31-710-4834
>Email: tgkim@contela.com

- --
+---------------------------------------------------------------------------+
| Jerry Van Baren / vanbaren_gerald@si.com / Grand Rapids Mi / 616-241-7973 |
|   My employer is a company.  Companies are artifacts of a legal system.   |
|________________Artifacts are incapable of having opinions.________________|

---------------------------

Newsgroups: comp.os.vxworks
Subject: STDIN problem on target
Date: Wed, 14 Feb 2001 11:39:39 -0800
From: "Gang Wang" 
Organization: None
Message-ID: 

Hi,
I am new to vxWorks and WinSh. I am using Tornado II. I am using the WinSh
to exec the following code. ( i.e. in windsh , after load in the .o file,
type >testStdIO). I wish I could see the characters I typed in there
reflected in the window, just as xterm. However, it seems that the data is
swallowed by the application.

How can I do this.

Gang Wang

#include 

int testStdIO()
{
   FILE *fp;
   char  data[256];

   /*if ( ( fp = fopen(stdin, "r")) == NULL )
     {
       printf("ERROR in open stdin");
       return 0;
      }*/
   while (1)
   {
     fgets(data, 256, stdin);
     if ( data[0] == 'q' )
        return 1;
     else
       printf("I get something\n");
   }
   return 0 ;
}



---------------------------

Newsgroups: comp.os.vxworks
Subject: T2 man pages???
Date: Wed, 14 Feb 2001 14:22:22 -0600
From: Bob Irwin 
Message-ID: <3A8AE8FE.D8334752@sis-dev.com>

Is it just me or are the Unix style man pages missing in T2??

I got so used to finding libs and functions by using vapropos (man -k).
I find the html pages a pain, they take up so much room on the screen.

With the good old unix man pages, I could have a couple of man pages,
a couple of includes and a couple of .c files displayed on my monitor
with all visible, no overlap.

Is there a way to restore the man pages?  Does anyone miss them?

  tnx.
- -- 
Bob Irwin | bobi@sis-dev.com | SIS Development | 281-484-6661 | Houston 
   "Research causes cancer in rats."

---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: T2 man pages???
Date: Wed, 14 Feb 2001 21:56:51 +0100
From: Sebastien Fauris 
Message-ID: <3A8AF113.580E570A@libertysurf.fr>
References: <3A8AE8FE.D8334752@sis-dev.com>

Hi Bob,

I am afraid the man pages have been replaced in T2 by HTML documentation.
Personnally I think also that it was much more quicker and efficient to
find
the information I was looking for using the man pages.

Talking about this, has someone ever tried to configure the man utility
ported
to Windows to run with the man pages provided with Tornado 1 Windows
version ?

Sebastien.

Bob Irwin wrote:

> Is it just me or are the Unix style man pages missing in T2??
>
> I got so used to finding libs and functions by using vapropos (man -k).
> I find the html pages a pain, they take up so much room on the screen.
>
> With the good old unix man pages, I could have a couple of man pages,
> a couple of includes and a couple of .c files displayed on my monitor
> with all visible, no overlap.
>
> Is there a way to restore the man pages?  Does anyone miss them?
>
>   tnx.
> --
> Bob Irwin | bobi@sis-dev.com | SIS Development | 281-484-6661 | Houston
>    "Research causes cancer in rats."


---------------------------

Newsgroups: comp.os.vxworks
Subject: re: power pc -- 64bit operation
Date: Thu, 15 Feb 2001 09:54:35 +0900
From: "Girish V. Gulawani" 
Organization: Lawrence Berkeley National Laboratory
Message-ID: <96faka$dqk$1@overload.lbl.gov>

> ---------------------------
> 
> Newsgroups: comp.os.vxworks
> Subject: power pc -- 64bit operation
> Date: Tue, 13 Feb 2001 16:53:42 GMT
> From: "Lou Kale" 
> Organization: EarthLink Inc. -- http://www.EarthLink.net
> Message-ID: 
> 
> I'm using a ppc604 (750) with Tornado2.
> 
> How do I know if it is running code in 64 bit mode?

> Is there a gnu compiler switch to turn 64 bit on?
I think the current GNU CC of T2 does not have this option. But the next
version (2.95.2+) has -mpowerpc64 option. Do you have 64-bit BSP/kernel?

HTH.
Girish.

---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: T2 man pages???
Date: Wed, 14 Feb 2001 20:26:06 -0500
From: Joe Durusau 
Message-ID: <3A8B302E.8C67A395@bellsouth.net>
References: <3A8AE8FE.D8334752@sis-dev.com>

	Sounds like WRS can't tell the diff between a 
feature and a major deficiency.  Menawhile, does anyone have
the time to write a script to regen the man pages out of 
all that masty html stuff???

Speaking only for myself,

Joe Durusau


Bob Irwin wrote:
> 
> Is it just me or are the Unix style man pages missing in T2??
> 
> I got so used to finding libs and functions by using vapropos (man -k).
> I find the html pages a pain, they take up so much room on the screen.
> 
> With the good old unix man pages, I could have a couple of man pages,
> a couple of includes and a couple of .c files displayed on my monitor
> with all visible, no overlap.
> 
> Is there a way to restore the man pages?  Does anyone miss them?
> 
>   tnx.
> --
> Bob Irwin | bobi@sis-dev.com | SIS Development | 281-484-6661 | Houston
>    "Research causes cancer in rats."

---------------------------

Newsgroups: comp.os.vxworks
Subject: VxWorks - Minimum Initializations for PCI
Date: Wed, 14 Feb 2001 21:25:17 -0500
From: "Ganesan, M" 
Organization: Lawrence Berkeley National Laboratory
Message-ID: <96fe4r$jfc$1@overload.lbl.gov>

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_01C096F6.8909DCA0
Content-Type: text/plain;
	charset="iso-8859-1"

Hi,

I am using MPC 8240 Kahlua based evaluation board. When I am trying to plug
PM/PPC (artesyn board) into the unused PCI slot of MPC 8240 board, I am not
able to boot MPC 8240 board. I have added code in syslib.c to issue
PCI_CMD_MASTER_ENABLE to PM/PPC who waits for that. Other than this, I have
not included any code for memory space mapping as I am not going to access
any memory from MPC 8240 as of now. 

My questions are

1. What are all the minimum initialization which are mandatory to be done
for plugging an PCI device in the bus from bus arbiter point of view.

2. If I need to access memory from both end (from MPC 8240 to PM/PPC and
vice-versa), with MPC8240 as bus arbiter, what are all the steps I need to
follow. What init codes I need to add in both bsp?

Please help me in this regard

bye

ram


- ------_=_NextPart_001_01C096F6.8909DCA0
Content-Type: text/html;
	charset="iso-8859-1"











Hi,


I am using MPC 8240 Kahlua based evaluation board. When I am trying to plug 
PM/PPC (artesyn board) into the unused PCI slot of MPC 8240 board, I am not able 
to boot MPC 8240 board. I have added code in syslib.c to issue 
PCI_CMD_MASTER_ENABLE to PM/PPC who waits for that. Other than this, I have not 
included any code for memory space mapping as I am not going to access any 
memory from MPC 8240 as of now. 


My questions are


1. What are all the minimum initialization which are mandatory to be done for 
plugging an PCI device in the bus from bus arbiter point of view.


2. If I need to access memory from both end (from MPC 8240 to PM/PPC and 
vice-versa), with MPC8240 as bus arbiter, what are all the steps I need to 
follow. What init codes I need to add in both bsp?


Please help me in this regard


bye


ram


- ------_=_NextPart_001_01C096F6.8909DCA0--

---------------------------

Newsgroups: comp.os.vxworks
Subject: FEC(10/100) of MPC855T connects to switch
Date: Thu, 15 Feb 2001 13:32:19 +0900
From: "Jinwoo Kim" 
Organization: Korea Telecom
Message-ID: <96flvi$e4p$1@news2-2.kornet.net>

On my cusstom board, FEC(10/100) of MPC855T connects to
switch(non-manageable switch) but not PHY. In this case, I don't know where
I have to modify the source code of motFecEnd.c.

That is to say,
        The Ethernet channels of our custom board are connected below.
            cpm <---> PHY                                     : works well
            fec   <---> Switch(MII connection)        : ?

I wish your comments...

Jinwoo Kim




---------------------------

Newsgroups: comp.os.vxworks
Subject: re: VxWorks - Minimum Initializations for PCI
Date: Thu, 15 Feb 2001 14:35:46 +0900
From: "Girish V. Gulawani" 
Organization: Lawrence Berkeley National Laboratory
Message-ID: <96fs6r$pnv$1@overload.lbl.gov>

Hi, Ram.


> 1. What are all the minimum initialization which are mandatory to be done
> for plugging an PCI device in the bus from bus arbiter point of view.

The WindRiver provides the PCI AutoConfiguration driver w/ source.
During bootup the PCI devices should come up initialized - semi plug and
play. 

Anyway, typically to configure the PCI device the Configuration header
is filled which is nothing but the PCI initialization. The steps other
than device specific init are to fill up following headers -
Command Register : Enable IO/Memory, Parity Error. Also see if your
device wants to be bus master.
Cache Line : Typically 32.
Latency    : Full, 0xFF.
Interrupts : Almost devices use INTA. Again the PCI library mentioned
above, provides routines to allocate and manage interrupts.
BARs       : Remember the PCI BAR values are physical addresses as seen
by PCI device and not virtual addresses seen from CPU. So for example IO
add offset 0x80000000 and to access memory you add offset of 0xC0000000
to the physical address to get virtual address. To get physcial from
virtual, I bet you know what to do! The actual memory map depends on
your board configuration. Refer your config.h it should be somewhere
there. 

> 2. If I need to access memory from both end (from MPC 8240 to PM/PPC and
> vice-versa), with MPC8240 as bus arbiter, what are all the steps I need to
> follow. What init codes I need to add in both bsp?
Once you configure the PCI header with appropriate BAR and other values.
The IO and/or Memory access is performed as given in BAR explanation
above.

Also a mundane note but very important, PCI is a Little Endian bus. 

HTH.
Girish.

---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: vxWorks 5.4/mv2700: networks goes north (down) - help
Date: Thu, 15 Feb 2001 06:35:18 GMT
From: Al Johnston 
Organization: Excite@Home - The Leader in Broadband http://home.com/faster
Message-ID: <3A8B7B3D.EDA5E6CC@mindspring.com>
References: <3A8997CB.6411EE51@mindspring.com>

We found it.  We had a "one way" udp rpc client call whose server was sending a
reply back.  That filled up sockets/netBufs that were never read.

This defect in our code has been around forever... back to pre vxW 3.02 days.
Strange that it is just showing up now...  Looks like tor2 is a little less
forgiving... 

Not that I really blaming it.  We should have caught this one long ago.  

For anyone that cares, the issue was a clnt_call() with a timeout value
of 0;  The server routine in such a case should not call svc_sendreply().
Ours did.  It will be a pain to track these down... 

Thanks to all who helped.

- -al

---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: CppUnit port to VxWorks
Date: Thu, 15 Feb 2001 01:57:19 -0600
From: "John Poplett" 
Organization: MindSpring Enterprises
Message-ID: <96g25e$pha$1@slb5.atl.mindspring.net>
References: <3A886938.E0BA12C8@address.com>

We have it compiling. The main difficulty, as I recall, has to do with
auto_ptr, which is commented out in Tornado. It was a simple matter to
rewrite CppUnit to eliminate references to auto_ptr.

John Poplett
CommWorks Corp.

"Fake"  wrote in message
news:3A886938.E0BA12C8@address.com...
> Hi All,
>
> Is anyone out there managed to port it (version 1.5) successfully to
> this OS? CppUnit is a testing framework made popular by
> http://www.xprogramming.com/software.htm and was created by Michael
> Feathers. I couldn't even get it to compile.
>
> TIA.
>
> Cheers
>



---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: UDP Socket Performance
Date: Thu, 15 Feb 2001 09:55:58 +0100
From: Roger 
Message-ID: <3A8B999E.A2B21971@here.there.every.where>
References: <96d4aa$sg3$1@overload.lbl.gov> <78vi6.212$yi4.201420@newsr1.u-net.net>

Dave Korn wrote:
> =

> =B1=E8=C5=C2=B1 wrote in message <96d4aa$sg3$1@overload.lbl.gov>...
> >Hi,
> >
> >I am using Tornado2/VxWorks5.4 on two MCP750 targets. I recently tried=
 to
> >test udp socket performance. The test program has 64 UDP connections w=
hich
> send
> >and receive 50 packets per second between the two targets. All connect=
ions
> are appear to
> >operate normaly, but more the half of the packets are not received.
> =

>   How does 64 * 50 * packetsize compare to the bandwidth of your networ=
k?

We move lots of data. Sometimes 2*32000*20 bytes/sec. Some comms are
UDP-based (synchronous events and control commands and the like). Others
are TCP-based (streams of tagged data that arrive asynchronously,
typically as indirect responses to the UDP commands). This mixture
allows the time-critical things to be UDP and quick, both in terms of
speed and error recovery when a packet goes missing. It also allows it
to deal with properly tagged data streams for high volume data with
guaranteed delivery.

Ability to do this depends on the nature of the system control. We had
to ensure, for example, that if a missing command had to be resent, it
was not executed again on the remote system if, in fact, it was only the
ack that went missing.

> =

>        DaveK
> --
> They laughed at Galileo.  They laughed at Copernicus.  They laughed at
> Columbus. But remember, they also laughed at Bozo the Clown.

---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: UDP Socket Performance
Date: Thu, 15 Feb 2001 09:59:49 +0100
From: Roger 
Message-ID: <3A8B9A85.5E0B0C88@here.there.every.where>
References: <96d4aa$sg3$1@overload.lbl.gov> <3a8aca63$1@news.si.com>

vanbaren_gerald wrote:
> 
> Is someone _always_ ready to receive a packet on the receive side?
> vxWorks does not appear to buffer UDP packets at all.  If the packet
> arrives in the forest and nobody is there listening for it, it gets
> silently discarded (oooh, that was BAD).

You must have a listener. And, each sent packet must be read with a
comparable read command. That is, if the sender sends a 128 byte
message, the packet must be read with a single 128 byte read command.
Not 2 consecutive 64 byte reads. Reads and writes are atomic. So, a UDP
read is usually always the max allowed UDP packet size. The return is
always what was actually receicved. You can't ask for more.

(At least this is the way it is everywhere else. Don't know about
VxWorks, but I would be suprised if it was not like this).

> 
> gvb
> 
> =?ks_c_5601-1987?B?sejFwrHU?=  writes:
> 
> >Hi,
> 
> >I am using Tornado2/VxWorks5.4 on two MCP750 targets. I recently tried to
> >test udp socket performance. The test program has 64 UDP connections which send
> >and receive 50 packets per second between the two targets. All connections are appear to
> >operate normaly, but more the half of the packets are not received.
> 
> >Can anyone provide some help?
> 
> >Taegue Kim
> >Contela, Inc.
> >9-1, Sunae-dong, Bundang-gu, Sungnam-si,
> >Kyunggi-do, 463-784, Korea
> >Phone: +82-31-710-4834
> >Email: tgkim@contela.com
> 
> --
> +---------------------------------------------------------------------------+
> | Jerry Van Baren / vanbaren_gerald@si.com / Grand Rapids Mi / 616-241-7973 |
> |   My employer is a company.  Companies are artifacts of a legal system.   |
> |________________Artifacts are incapable of having opinions.________________|

---------------------------

Newsgroups: comp.os.vxworks
Subject: pthreads
Date: Thu, 15 Feb 2001 11:39:31 +0100
From: "Fred Zwarts" 
Organization: Kernfysisch Versneller Instituut
Message-ID: <96gbl3$l95$1@info.service.rug.nl>

I would like to know whether the pthreads header files and API
is supported by VxWorks.

We use VxWorks for AlphaVME and AXPvme processors, which is a
version of VxWorks that has some additions.
This version has a pthreads implementation, which we use to=20
write portable code. Since VxWorks for the Alpha is declared
end-of-life, we are looking for another platform. To ease
application migration, we prefer a platform with pthreads=20
support. Therefore the question: Does VxWorks support pthreads
also on the other hardware platforms, or is (was) it an=20
specific extension for the Alpha platform?


---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: Standalone applications
Date: Thu, 15 Feb 2001 07:03:04 -0500
From: "Serge Sarraillon" 
Message-ID: 
References: 

This oneway, that I prefer, to do it (if you have a file system on your
target)

1) Does your macro INCLUDE_STARTUP_SCRIPT    is defined ?
     it is used in the file ' config/all/usrConfig.h'

#ifdef INCLUDE_STARTUP_SCRIPT   /* run a startup script */
    if (sysBootParams.startupScript [0] != EOS)
 usrStartupScript (sysBootParams.startupScript);
#endif /* INCLUDE_STARTUP_SCRIPT */

2) You have to modified the bootline using the vxWorks 'bootChange' command
    and refer a 'startup script file' which should be on your file system.

3) a script file it is simply contain vxWorks command (as prompt) as

        ld < program.o /* load program */
        taskSpawn ( testFunction, ...)  /* start your main function */

Using startup script allows you to have plenty of program on your target and
you can change
directly in prompt (serial term) without re-compilation

Presently, the makefile, rules,... are besd on the old vxWorks 5.3.1 -
Tornado I.   Notice that vxWorks 5.3.1
has the Standalone feature and Tornado II doesn't.  Wind River remove it
from their makefile but if you put it back
it works perfectly.

Serge S.



"Scott Smail"  wrote in message
news:o7bi6.129989$o91.13373691@typhoon.nyroc.rr.com...
> Hello All,
>
> I have the following system:
>
> Tornado 2.0 with Zinc 5.0
> host: AMD 586, 128MB RAM
> target: PC486 Dos File Sys 32 MB RAM
>
> Could someone tell me how to make my system standalone (automatically
start
> my application on bootup)?
> I have modified the makefile and usrAppInit( ) and then run make
vxWorks.st.
> This compiled, but now I'm not sure how to proceed.
>
> What do I do with vxWorks.st?
>
> How do I get the image onto my target's ata?
>
> Thanks,
> -Scott
>
>
>
>



---------------------------

Newsgroups: comp.os.vxworks
Subject: Q: Build erro in VxMP ?
Date: Thu, 15 Feb 2001 12:02:46 GMT
From: S.Flache@signaal.de (Stefan Flache)
Organization: News from ezri
Message-ID: <3a8bc3f7.12707031@10.1.1.1>
Reply-To: S.Flache@signaal.de

Using shared memory objects as build option for the vxWorks boot image
deliveres the following error:

ldppc -X -N -e _sysInit -Ttext 00010000 \
	dataSegPad.o partialImage.o ctdt.o -o vxWorks
partialImage.o: In function `smUtilIntGen':
partialImage.o(.text+0x10e0f4): undefined reference to `sysBusIntGen'
partialImage.o: In function `smUtilIntConnect':
partialImage.o(.text+0x10e3c4): undefined reference to
`sysMailboxConnect'
partialImage.o(.text+0x10e3d4): undefined reference to
`sysMailboxEnable'
make: *** [vxWorks] Error 0x1

What can I do to prevent this error?


Greetings Stefan

---------------------------

End of New-News digest
**********************


From vxwexplo-errs@csg.lbl.gov  Thu Feb 15 06:24:48 2001
From: "Baron, Ken" 
Date: Thu Feb 15 06:24:50 PST 2001
Subject: RE: STDIN problem on target


	Your code below works fine for me. If you run testStdIO from the
windsh the function receives input from the shell window.
	If you spawn testStdIO as a new task, then the task receives input
from the console. (vxworks sim window or your primary serial port on your
hardware.

> Hi,
> I am new to vxWorks and WinSh. I am using Tornado II. I am using the WinSh
> to exec the following code. ( i.e. in windsh , after load in the .o file,
> type >testStdIO). I wish I could see the characters I typed in there
> reflected in the window, just as xterm. However, it seems that the data is
> swallowed by the application.
> 
> How can I do this.
> 
> Gang Wang
> 
> #include 
> 
> int testStdIO()
> {
>    FILE *fp;
>    char  data[256];
> 
>    /*if ( ( fp = fopen(stdin, "r")) == NULL )
>      {
>        printf("ERROR in open stdin");
>        return 0;
>       }*/
>    while (1)
>    {
>      fgets(data, 256, stdin);
>      if ( data[0] == 'q' )
>         return 1;
>      else
>        printf("I get something\n");
>    }
>    return 0 ;
> }
> 


From vxwexplo-errs@csg.lbl.gov  Thu Feb 15 11:05:17 2001
From: "Sinn, Larry" 
Date: Thu Feb 15 11:05:19 PST 2001
Subject: Re: T2 man pages???

I agree!!!
We upgraded from T1 to T2 and kept the T1 man page directories and created a
Tornado man command:
	alias tman "$WIND_BASE_T1/target/man:$WIND_BASE_T1/host/man"

Larry.

> 	Sounds like WRS can't tell the diff between a 
> feature and a major deficiency.  Menawhile, does anyone have
> the time to write a script to regen the man pages out of 
> all that masty html stuff???
> 
> Speaking only for myself,
> 
> Joe Durusau


> Bob Irwin wrote:
> > 
> > Is it just me or are the Unix style man pages missing in T2??
> > 
> > I got so used to finding libs and functions by using vapropos (man -k).
> > I find the html pages a pain, they take up so much room on the screen.
>>  
> > With the good old unix man pages, I could have a couple of man pages,
> > a couple of includes and a couple of .c files displayed on my monitor
> > with all visible, no overlap.
> > 
> > Is there a way to restore the man pages?  Does anyone miss them?
> > 
> >   tnx.
> > --
> > Bob Irwin | bobi@sis-dev.com | SIS Development | 281-484-6661 | Houston
> >    "Research causes cancer in rats."
> 

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  Thu Feb 15 12:06:41 2001
From: "Balaji  Ramachandran" 
Date: Thu Feb 15 12:06:44 PST 2001
Subject: Can't create vxWorks_rom for mcp750

VxWorks 5.4
BSP:MCP750

Hi All,

I have tried to build "vxworks_rom" image from a bootable application
created from T-II project facility.

But it gives linker error "undefined reference to "SysGetBusSpd" and
"SysGetDramSpd" error in the "rominit.o" file.

Can any one throw light on this, how to solve it?.


Pl.. see below, the build output


Thanks in advance,

Balaji.

--------------------------------------------------------------------------------------------------------------

vxrm *.o *.rpo ctdt.c symTbl.c vxWorks*
vxrm C:\Vx_test\Romable\prjComps.h C:\Vx_test\Romable\prjParams.h
C:\Vx_test\Romable\prjCo
nfig.c C:\Vx_test\Romable\linkSyms.c
vxrm C:\Vx_test\Romable\libs.nm C:\Vx_test\Romable\libs.size
wtxtcl C:\Tornado\host\src\hutils\configGen.tcl
C:\Vx_test\Romable\romable.wpj
ccppc -g -mstrict-align -ansi -nostdinc -fvolatile -fno-builtin
-fno-for-scope -P -x assem
bler-with-cpp -IC:\Vx_test\Romable -IC:\Tornado\target\config\mcp750
-IC:\Tornado\target\h
 -IC:\Tornado\target\config\comps\src -IC:\Tornado\target\src\drv
-DCPU=PPC604 -DPRJ_BUILD
 -DMCP750 -DTARGET_DIR="\"mcp750\"" -c
C:\Tornado\target\config\mcp750\sysALib.s -o sysALi
b.o
C:\TEMP\cca00234.s: Assembler messages:
C:\TEMP\cca00234.s:4764: Warning: operand out of range (-32768 not
between 0 and 65535)
C:\TEMP\cca00234.s:4853: Warning: operand out of range (57343 not
between -32768 and 32767
)
ccppc -O2 -g -mstrict-align -ansi -nostdinc -DRW_MULTI_THREAD
-D_REENTRANT -fvolatile -fno
-builtin -fno-for-scope -IC:\Vx_test\Romable
-IC:\Tornado\target\config\mcp750 -IC:\Tornad
o\target\h -IC:\Tornado\target\config\comps\src
-IC:\Tornado\target\src\drv -DCPU=PPC604 -
DPRJ_BUILD -DMCP750 -DTARGET_DIR="\"mcp750\"" -c
C:\Tornado\target\config\mcp750\sysLib.c
ccppc -O2 -g -mstrict-align -ansi -nostdinc -DRW_MULTI_THREAD
-D_REENTRANT -fvolatile -fno
-builtin -fno-for-scope -IC:\Vx_test\Romable
-IC:\Tornado\target\config\mcp750 -IC:\Tornad
o\target\h -IC:\Tornado\target\config\comps\src
-IC:\Tornado\target\src\drv -DCPU=PPC604 -
DPRJ_BUILD -DMCP750 -DTARGET_DIR="\"mcp750\"" -c
C:\Tornado\target\config\mcp750\sysSpeed.
c
ccppc -O2 -g -mstrict-align -ansi -nostdinc -DRW_MULTI_THREAD
-D_REENTRANT -fvolatile -fno
-builtin -fno-for-scope -IC:\Vx_test\Romable
-IC:\Tornado\target\config\mcp750 -IC:\Tornad
o\target\h -IC:\Tornado\target\config\comps\src
-IC:\Tornado\target\src\drv -DCPU=PPC604 -
DPRJ_BUILD -DMCP750 -DTARGET_DIR="\"mcp750\"" -c
C:\Tornado\target\config\mcp750\usrLib.c
ccppc -O2 -g -mstrict-align -ansi -nostdinc -DRW_MULTI_THREAD
-D_REENTRANT -fvolatile -fno
-builtin -fno-for-scope -IC:\Vx_test\Romable
-IC:\Tornado\target\config\mcp750 -IC:\Tornad
o\target\h -IC:\Tornado\target\config\comps\src
-IC:\Tornado\target\src\drv -DCPU=PPC604 -
DPRJ_BUILD -DMCP750 -DTARGET_DIR="\"mcp750\"" -c
C:\Vx_test\Romable\usrAppInit.c
ccppc -O2 -g -mstrict-align -ansi -nostdinc -DRW_MULTI_THREAD
-D_REENTRANT -fvolatile -fno
-builtin -fno-for-scope -IC:\Vx_test\Romable
-IC:\Tornado\target\config\mcp750 -IC:\Tornad
o\target\h -IC:\Tornado\target\config\comps\src
-IC:\Tornado\target\src\drv -DCPU=PPC604 -
DPRJ_BUILD -DMCP750 -DTARGET_DIR="\"mcp750\"" -c
C:\Vx_test\Romable\prjConfig.c
ccppc -O2 -g -mstrict-align -ansi -nostdinc -DRW_MULTI_THREAD
-D_REENTRANT -fvolatile -fno
-builtin -fno-for-scope -IC:\Vx_test\Romable
-IC:\Tornado\target\config\mcp750 -IC:\Tornad
o\target\h -IC:\Tornado\target\config\comps\src
-IC:\Tornado\target\src\drv -DCPU=PPC604 -
DPRJ_BUILD -DMCP750 -DTARGET_DIR="\"mcp750\"" -c
C:\Vx_test\Romable\linkSyms.c
ccppc -O2 -g -mstrict-align -ansi -nostdinc -DRW_MULTI_THREAD
-D_REENTRANT -fvolatile -fno
-builtin -fno-for-scope -IC:\Vx_test\Romable
-IC:\Tornado\target\config\mcp750 -IC:\Tornad
o\target\h -IC:\Tornado\target\config\comps\src
-IC:\Tornado\target\src\drv -DCPU=PPC604 -
DPRJ_BUILD -DMCP750 -DTARGET_DIR="\"mcp750\"" -c
C:\Tornado\target\config\comps\src\versio
n.c
ccppc -O2 -g -mstrict-align -ansi -nostdinc -DRW_MULTI_THREAD
-D_REENTRANT -fvolatile -fno
-builtin -fno-for-scope -IC:\Vx_test\Romable
-IC:\Tornado\target\config\mcp750 -IC:\Tornad
o\target\h -IC:\Tornado\target\config\comps\src
-IC:\Tornado\target\src\drv -DCPU=PPC604 -
DPRJ_BUILD -DMCP750 -DTARGET_DIR="\"mcp750\"" -c
C:\Tornado\target\config\comps\src\usrEnt
ry.c
ccppc -BC:\Tornado\host\x86-win32\lib\gcc-lib\ -nostdlib -r -Wl,-X \
    usrEntry.o -Wl,@C:\Vx_test\Romable\prjObjs.lst  version.o
C:\Tornado\target\lib\libPP
C604gnuvx.a -o partialImage.o
nmppc partialImage.o @C:\Vx_test\Romable\prjObjs.lst | wtxtcl
C:\Tornado\host\src\hutils\m
unch.tcl -asm ppc > ctdt.c
ccppc -c -fdollars-in-identifiers -O2 -g -mstrict-align -ansi -nostdinc
-DRW_MULTI_THREAD
-D_REENTRANT -fvolatile -fno-builtin -fno-for-scope -IC:\Vx_test\Romable
-IC:\Tornado\targ
et\config\mcp750 -IC:\Tornado\target\h
-IC:\Tornado\target\config\comps\src -IC:\Tornado\t
arget\src\drv -DCPU=PPC604 -DPRJ_BUILD -DMCP750
-DTARGET_DIR="\"mcp750\"" ctdt.c -o ctdt.o
ccppc -O2 -g -mstrict-align -ansi -nostdinc -DRW_MULTI_THREAD
-D_REENTRANT -fvolatile -fno
-builtin -fno-for-scope -IC:\Vx_test\Romable
-IC:\Tornado\target\config\mcp750 -IC:\Tornad
o\target\h -IC:\Tornado\target\config\comps\src
-IC:\Tornado\target\src\drv -DCPU=PPC604 -
DPRJ_BUILD -DMCP750 -DTARGET_DIR="\"mcp750\"" -c
C:\Tornado\target\config\comps\src\dataSe
gPad.c
ldppc -X -N -e _sysInit -Ttext 00100000  \
 dataSegPad.o partialImage.o ctdt.o -o vxWorks
C:\Tornado\host\x86-win32\bin\vxsize PPC -v 00300000  00100000  vxWorks
vxWorks: 799248(t) + 96244(d) +  35796(b) = 931288 (1165864 unused)
ccppc -c -g -mstrict-align -ansi -nostdinc -fvolatile -fno-builtin
-fno-for-scope -P -x as
sembler-with-cpp -IC:\Vx_test\Romable -IC:\Tornado\target\config\mcp750
-IC:\Tornado\targe
t\h -IC:\Tornado\target\config\comps\src -IC:\Tornado\target\src\drv
-DCPU=PPC604 -DPRJ_BU
ILD -DMCP750 -DTARGET_DIR="\"mcp750\"" -DROM_COPY
-DRAM_DATA_ADRS=0x00300000  -DRAM_DST_AD
RS=0x00100000  C:\Tornado\target\config\mcp750\romInit.s -o romInit.o
C:\TEMP\cca00304.s: Assembler messages:
C:\TEMP\cca00304.s:4745: Warning: operand out of range (-32768 not
between 0 and 65535)
C:\TEMP\cca00304.s:4823: Warning: operand out of range (57343 not
between -32768 and 32767
)
ccppc -c -O2 -g -mstrict-align -ansi -nostdinc -DRW_MULTI_THREAD
-D_REENTRANT -fvolatile -
fno-builtin -fno-for-scope -IC:\Vx_test\Romable
-IC:\Tornado\target\config\mcp750 -IC:\Tor
nado\target\h -IC:\Tornado\target\config\comps\src
-IC:\Tornado\target\src\drv -DCPU=PPC60
4 -DPRJ_BUILD -DMCP750 -DTARGET_DIR="\"mcp750\"" -DROM_COPY
-DRAM_DATA_ADRS=0x00300000  -D
RAM_DST_ADRS=0x00100000  C:\Tornado\target\config\comps\src\romStart.c
-o romStart.o
copy vxWorks vxWorks.o
        1 file(s) copied.
C:\Tornado\host\x86-win32\bin\elfToBin < vxWorks.o > tmp.Z
C:\Tornado\host\x86-win32\bin\binToAsm tmp.Z > vxWorks.Z.s
ccppc -c -O2 -g -mstrict-align -ansi -nostdinc -DRW_MULTI_THREAD
-D_REENTRANT -fvolatile -
fno-builtin -fno-for-scope -IC:\Vx_test\Romable
-IC:\Tornado\target\config\mcp750 -IC:\Tor
nado\target\h -IC:\Tornado\target\config\comps\src
-IC:\Tornado\target\src\drv -DCPU=PPC60
4 -DPRJ_BUILD -DMCP750 -DTARGET_DIR="\"mcp750\"" vxWorks.Z.s -o
vxWorks.Z.o
ccppc -O2 -g -mstrict-align -ansi -nostdinc -DRW_MULTI_THREAD
-D_REENTRANT -fvolatile -fno
-builtin -fno-for-scope -IC:\Vx_test\Romable
-IC:\Tornado\target\config\mcp750 -IC:\Tornad
o\target\h -IC:\Tornado\target\config\comps\src
-IC:\Tornado\target\src\drv -DCPU=PPC604 -
DPRJ_BUILD -DMCP750 -DTARGET_DIR="\"mcp750\"" -c
C:\Tornado\target\config\comps\src\versio
n.c
ldppc -X -N -e _romInit  -Ttext fff00100  -Tdata 00300000  \
    romInit.o romStart.o version.o
C:\Tornado\target\lib\libPPC604gnuvx.a vxWorks.Z.o -o
vxWorks_rom
romInit.o(.text+0x40c): undefined reference to `sysGetDramSpd'
romInit.o(.text+0x438): undefined reference to `sysGetBusSpd'
make: *** [vxWorks_rom] Error 0x1

Done.
-------------------------------------------------------------------------------------------------------------------------

VxWorks / Tornado




From vxwexplo-errs@csg.lbl.gov  Thu Feb 15 23:37:43 2001
From: "Joe Zhou" 
Date: Thu Feb 15 23:37:46 PST 2001
Subject: Question on SNMP agent getBulk operation

This is a multi-part message in MIME format.

------=_NextPart_000_0007_01C097A8.0A27BC10
Content-Type: text/plain;
	charset="iso-8859-1"
Content-Transfer-Encoding: quoted-printable

Hi, all

I am new to this, so bear with me.
I am trying to implement GETBULK operation on SNMP agent, using Tornado2 =
WindNet SNMPV1/v2c.
I've finished GET,GETNEXT and SET operation, but somehow I am a little =
bit confused with GETBULK.=20
Using a SNMP manager, I send a GETBUK to the SNMP agent. It seems that =
agent can interpret the packet correctly (showing the pdu type=3D5, =
correct max_reps and non_reps) and seemingly, the result returned to the =
manager is corret. So my question is -- I never customerily implemented =
my getbulk functions, has it been taken care of by Tornado internally?

From the backtrace window, I can see that following procedures have been =
invoked in sequence:
snmpdMain-->snmpIoMain-->snmpIoBody-->snmpdPktProcess-->Process_Rcvd_SNMP=
_Packet_Asyn-->process_packet_one-->process_packet_twon-->snmpdContinue--=
> my_own_getnext_procedure
So does it mean that Tornado treat GETBULK as getnext internally? Do I =
still need to do anything?

I am sure there are people out there are much more experienced than I =
am, please help! Thanks in advance.

Regards

Joe Zhou


------=_NextPart_000_0007_01C097A8.0A27BC10
Content-Type: text/html;
	charset="iso-8859-1"
Content-Transfer-Encoding: quoted-printable









Hi, all


 


I am new to this, so bear with =
me.


I am trying to implement GETBULK =
operation on=20
SNMP agent, using Tornado2 WindNet SNMPV1/v2c.


I've finished GET,GETNEXT and SET =
operation,=20
but somehow I am a little bit confused with GETBULK. 


Using a SNMP manager, I send a GETBUK =
to the SNMP=20
agent. It seems that agent can interpret the packet correctly (showing =
the pdu=20
type=3D5, correct max_reps and non_reps) and seemingly, the result returned to the manager is corret. So my =
question is=20
-- I never customerily implemented my getbulk functions, has it been =
taken care=20
of by Tornado internally?


 


From the backtrace window, I can see =
that following=20
procedures have been invoked in sequence:


snmpdMain-->snmpIoMain-->snmpIoBody-->snmpdPktProcess--=
>Process_Rcvd_SNMP_Packet_Asyn-->process_packet_one-->process_pa=
cket_twon-->snmpdContinue-->=20
my_own_getnext_procedure


So does it mean that Tornado treat =
GETBULK as=20
getnext internally? Do I still need to do anything?


 


I am sure there are people out there =
are much more=20
experienced than I am, please help! Thanks in advance.


 


Regards


 


Joe Zhou


------=_NextPart_000_0007_01C097A8.0A27BC10--



From vxwexplo-errs@csg.lbl.gov  Fri Feb 16 03:47:48 2001
From: prabhala rajendra 
Date: Fri Feb 16 03:47:50 PST 2001
Subject: VxMP on CPCI back plane

vxworks tornado

Hi vxworkers,
I am planing to use VxMP for remote function
invocation from one telcom board to another one over a
cpci backplane using shared message queues.

I have never used VxMP before. I have checked the
manual for some details about this package
but the following things are not clear to me. 

Could somebody please clarify the following regarding
VxMP ? 

1.Do we have to allocate and reserve this
shared-memory partition (dedicated to shared memory
system pool) during the init phase by calling the
routine smMemAddToPool()? 
	
2.Is there any restriction on the type of back plane
(VME / PCI/CPCI) used when we use VxMP ? For non-VME
backplanes (eg.CPCI) are there any extra
settings/issues to be considered ? Manual seems to be
very specific about VME.

3.Has anybody tried using VxMP in typical call
processing applications ? How is the performance ?

Thanks and best regards,
rajendra.

__________________________________________________
Do You Yahoo!?
Get personalized email addresses from Yahoo! Mail - only $35 
a year!  http://personal.mail.yahoo.com/


From vxwexplo-errs@csg.lbl.gov  Fri Feb 16 04:03:12 2001
From: Vxworks Exploder 
Date: Fri Feb 16 04:03:15 PST 2001
Subject: comp.os.vxworks newsdigest

Comp.Os.Vxworks Daily Digest    Fri Feb 16 04:03:08 PST 2001

        Subject: Re: How to strip out the symbols?
        Subject: BSP port MPC750 to 7400 fails
        Subject: Which function can i use in SENS instead on etherHook of version 5.3 ?
        Subject: Re: UDP Socket Performance
        Subject: RE: STDIN problem on target
        Subject: RSVP Stack
        Subject: Rebuilding gcc for ARM
        Subject: Syncing symbol tables between target-resident and host shells
        Subject: Re: CppUnit port to VxWorks
        Subject: Re: pthreads
        Subject: WTX C API Interface to target server on a Windows Machine
        Subject: Re: Rebuilding gcc for ARM
        Subject: re: T2 man pages???
        Subject: Re: WTX C API Interface to target server on a Windows Machine
        Subject: passFs problem
        Subject: Can't create vxWorks_rom for mcp750
        Subject: Re: STDIN problem on target
        Subject: Re: How to spawn two target server for two different boards in one 
        Subject: Re: Can't create vxWorks_rom for mcp750
        Subject: Re: Shared memory network problems
        Subject: file I/O performance
        Subject: File I/O through Serial Link on RAD6000
        Subject: Question on SNMP agent getBulk operation
        Subject: Re: (beginer) Problem "catching" a cPCI interrupt
        Subject: Re: Syncing symbol tables between target-resident and host shells
        Subject: Re: T2 man pages???

-------------------------------------------------------

Newsgroups: comp.os.vxworks
Subject: Re: How to strip out the symbols?
Date: Thu, 15 Feb 2001 07:11:15 -0500
From: "Serge Sarraillon" 
Message-ID: <5DPi6.26$Cx3.1303@wagner.videotron.net>
References:  <3A8600D1.7F834914@flashcom.net> 

For the host, I still refer to the 'unstrip' file which contains synbols and
debug information and it works.

"murali swaminathan"  wrote in message
news:Srzh6.13336$1%2.692711@sjc-read.news.verio.net...
> Thanks. Objcopy worked. It took the same time for me to download as with
the
> symbols. any thoughts, why? Further, how can I get the symbol into tornado
> env (on the host side).
>
> "DrDiags"  wrote in message
> news:3A8600D1.7F834914@flashcom.net...
> > Murali,
> >
> >     If you are using Tornado 2/VxWorks 5.4 GNU Toolchain, you should use
> objcopy
> > (I forgot the exact option). I believe the stripppc has problems (which
is
> > probably why you asked this question).
> >
> > murali swaminathan wrote:
> >
> > > Folks,
> > > We are in the process of porting our application to Vxworks. Our Image
> size
> > > is very large and couple of out interfaces are serial. What is strip
> utility
> > > that we need to run to remove the symbols (for 860/8260 powerpcs). If
we
> > > strip the symbols out how do we create a symbol file to be loaded into
> > > tornado for debugging. Thanks for any help you can provide.
> > >
> > > Murali
> >
>
>



---------------------------

Newsgroups: comp.os.vxworks
Subject: BSP port MPC750 to 7400 fails
Date: Thu, 15 Feb 2001 07:09:17 -0500
From: "Greg Griffes" 
Organization: Smiths Industries
Message-ID: <3a8bc74e$1@news.si.com>

We developed a BSP for our PPC 750 hardware which has been fine until we
upgraded it with a 7400 CPU.  Now the bootrom fails to load the vxWorks
image via Ethernet (gets all the way to "Loading..." and times out).  We are
looking into the Ethernet driver.  Has anyone had similar experiences or
advice?

Thanks in advance,

Greg Griffes
Smiths Industries
griffes_greg@si.com



---------------------------

Newsgroups: comp.os.vxworks
Subject: Which function can i use in SENS instead on etherHook of version 5.3 ?
Date: Thu, 15 Feb 2001 21:16:22 +0900
From: "SungJin Lee" 
Organization: Korea Telecom
Message-ID: <96gh8h$gtb$1@news2-2.kornet.net>
Reply-To: "SungJin Lee" 

Which function can i use of SENSE version to hook packets
instead of etherHook in VxWORKS 5.3 ?





---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: UDP Socket Performance
Date: Thu, 15 Feb 2001 13:26:25 +0000
From: James Marshall 
Organization: Agilent Technologies UK Ltd
Message-ID: <3A8BD901.3B173B2@agilent.com>
References: <96d4aa$sg3$1@overload.lbl.gov>

The first point of packet loss, assuming your LAN works okay, is the receive buffers
available to the LAN controller - when there are none, packets are missed. You'll need to
increase these massively if you receive a lot of data in a short time. The number of receive
and transmit buffers is usually supplied as a parameter to the driver init routine which you
can adjust. You could use Windview to see if you are getting roughly the number of receive
packet interrupts that you expect, though this would depend on how quickly your system can
handle packets.

You might also want to look at the zBuf interface to avoid the cost of copying from network
buffer to application buffer in the socket call. See the Network Programmer's Guide.

James Marshall.


±èÅÂ±Ô wrote:

> Hi,
>
> I am using Tornado2/VxWorks5.4 on two MCP750 targets. I recently tried to
> test udp socket performance. The test program has 64 UDP connections which send
> and receive 50 packets per second between the two targets. All connections are appear to
> operate normaly, but more the half of the packets are not received.
>
> Can anyone provide some help?
>
> Taegue Kim
> Contela, Inc.
> 9-1, Sunae-dong, Bundang-gu, Sungnam-si,
> Kyunggi-do, 463-784, Korea
> Phone: +82-31-710-4834
> Email: tgkim@contela.com


---------------------------

Newsgroups: comp.os.vxworks
Subject: RE: STDIN problem on target
Date: Thu, 15 Feb 2001 09:25:45 -0500
From: "Baron, Ken" 
Organization: Lawrence Berkeley National Laboratory
Message-ID: <96grrb$e8t$1@overload.lbl.gov>


	Your code below works fine for me. If you run testStdIO from the
windsh the function receives input from the shell window.
	If you spawn testStdIO as a new task, then the task receives input
from the console. (vxworks sim window or your primary serial port on your
hardware.

> Hi,
> I am new to vxWorks and WinSh. I am using Tornado II. I am using the WinSh
> to exec the following code. ( i.e. in windsh , after load in the .o file,
> type >testStdIO). I wish I could see the characters I typed in there
> reflected in the window, just as xterm. However, it seems that the data is
> swallowed by the application.
> 
> How can I do this.
> 
> Gang Wang
> 
> #include 
> 
> int testStdIO()
> {
>    FILE *fp;
>    char  data[256];
> 
>    /*if ( ( fp = fopen(stdin, "r")) == NULL )
>      {
>        printf("ERROR in open stdin");
>        return 0;
>       }*/
>    while (1)
>    {
>      fgets(data, 256, stdin);
>      if ( data[0] == 'q' )
>         return 1;
>      else
>        printf("I get something\n");
>    }
>    return 0 ;
> }
> 

---------------------------

Newsgroups: comp.os.vxworks
Subject: RSVP Stack
Date: Thu, 15 Feb 2001 15:57:49 GMT
From: "Christophe Belmont" 
Organization: Verio
Message-ID: <10Ti6.2178$ws2.132897@ord-read.news.verio.net>

Hi,

I'm wondering if one of you have already heard about an implementation (even
though it's imcomplete) of an RSVP stack for VxWorks.

Thank you,
Christophe.



---------------------------

Newsgroups: comp.os.vxworks
Subject: Rebuilding gcc for ARM
Date: 15 Feb 2001 11:51:00 -0600
From: Kevin Cline 
Organization: MAYAN Networks
Message-ID: 
Sender: kcline@vector.mayannetworks.com

I'm trying to build Dave Korn's gcc source for the ARM,
but I'm stumped on a basic question -- what is the
target architecture.  Is there a list?  Is there a way
to find out the target architecture for the WRS-supplied
compilers and use that?
- -- 
Kevin Cline

---------------------------

Newsgroups: comp.os.vxworks
Subject: Syncing symbol tables between target-resident and host shells
Date: Thu, 15 Feb 2001 18:05:30 +0000
From: Tom Ko 
Organization: Schlumberger Omnes
Message-ID: <3A8C1A69.58396313@ferndown.tt.slb.com>

Hi

In the vxWorks, you can use the startup script to load modules. To do
this, you need to use the target-resident shell. Once loaded, the symbol
information are stored in the symbol table of the target-resident shell.
Now if you launch the host shell from NT, these modules/symbols you
loaded during the startup cannot be seen from the host shell. For
example, 'moduleShow' on the host shell will not list modules loaded
during the startup. However, 'moduleShow' in the target-resident shell
i.e. the hyper-terminal window, will list the loaded modules.


Is there a way to sync these tables?


Regards


---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: CppUnit port to VxWorks
Date: Thu, 15 Feb 2001 11:54:37 -0500
From: Newbie 
Organization: UUNET Canada News Reader Service
Message-ID: <3A8C09CD.DFA2F2F8@NoSuchAddress.com>
References: <3A886938.E0BA12C8@address.com> <96g25e$pha$1@slb5.atl.mindspring.net>

Hi John,

What library do you need to include in order to use the std::string?

TIA


John Poplett wrote:

> We have it compiling. The main difficulty, as I recall, has to do with
> auto_ptr, which is commented out in Tornado. It was a simple matter to
> rewrite CppUnit to eliminate references to auto_ptr.
>
> John Poplett
> CommWorks Corp.
>
> "Fake"  wrote in message
> news:3A886938.E0BA12C8@address.com...
> > Hi All,
> >
> > Is anyone out there managed to port it (version 1.5) successfully to
> > this OS? CppUnit is a testing framework made popular by
> > http://www.xprogramming.com/software.htm and was created by Michael
> > Feathers. I couldn't even get it to compile.
> >
> > TIA.
> >
> > Cheers
> >


---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: pthreads
Date: Thu, 15 Feb 2001 17:04:33 GMT
From: Bill Pringlemeir 
Organization: Factory of the mind
Message-ID: 
References: <96gbl3$l95$1@info.service.rug.nl>
Sender: bpringlemeir@DeadDuck

>>>>> "Fred" == Fred Zwarts  writes:

 Fred> I would like to know whether the pthreads header files and API
 Fred> is supported by VxWorks.
[snip]
 Fred> support. Therefore the question: Does VxWorks support pthreads
 Fred> also on the other hardware platforms, or is (was) it an
 Fred> specific extension for the Alpha platform?

I have just looked at pthread library for a port to ucLinux. Anyways,
the pthread library has some modules that *can* use assembler code.
You can also use the generic `C' routines, but performance will suffer
and some features might not be available.  Do you know if your pthread
port you have to the Alpha uses assembler code?  I don't think that it
would be a problem to re-compile the code on another architecture with
the generic `C' routines.

In any case, I believe that `Dot4' supports pthread on vxWorks, NOT wrs.
Check, 

 "http://www.wrs.com/products/html/pthreads.html"

This only list m68k, x86, and PowerPC as ports (not Alpha).

hth,
Bill

- -- 
Mmmmmm-MMMMMM!!  A plate of STEAMING PIECES of a PIG mixed with the
 shreds of SEVERAL CHICKENS!!...  Oh BOY!!  I'm about to swallow a
 TORN-OFF section of a COW'S LEFT LEG soaked in COTTONSEED OIL and
 SUGAR!!  ..  Let's see..  Next, I'll have the GROUND-UP flesh of
 CUTE, BABY LAMBS fried in the MELTED, FATTY TISSUES from a
 warm-blooded animal someone once PETTED!!  ...  YUM!!  That was
 GOOD!!  For DESSERT, I'll have a TOFU BURGER with BEAN SPROUTS on a
 stone-ground, WHOLE WHEAT BUN!!

---------------------------

Newsgroups: comp.os.vxworks
Subject: WTX C API Interface to target server on a Windows Machine
Date: Thu, 15 Feb 2001 10:42:29 -0800
From: Christopher A Leddy 
Organization: Raytheon Company
Message-ID: <3A8C2315.78E0D3D9@west.raytheon.com>

This isn't a request for help, just sharing some info.

I actually decided that I should write the info below for myself in my
notebook for later use, but why not post it share others some grief?

I had write a WTX tool under Tornado 1.0.1 and Windows NT 4.0 and began
by following the detailed example and explicit build/link instructions
found in the Tornado API guide. Not even close to building, let alone
working on my setup. 

To be fair, I am compiling with Visual Studio C++, version 6, but there
are still a lot of things missing, regardless of compiler versions. And
I know I could be much more suave and sophisticated by using environment
variables in the paths, but sometimes you just need to get it done (and
a lot of the errors in paths were due to trying to be smooth with
..\..\i_hope_it_is_two_up_on_your_machine_too nonsense).

So here are the mods/clarifications:

1. Source code placed in c:\tornado_03\host\src\wtxtest (I hate that
$(WIND_BASE) reference). Example source did not show up in my install,
or anybody else's machine that I could check, as stated in the manual.

2. Source code modified as shown below (mostly signal handler code mods
and added includes, the other mods are 		specific to my application)

3. Added the following to Project settings, C/C++, Preprocessor:
	Additional Include directories:
C:\Tornado_03\share\src\wtx,C:\Tornado_03\host\include 
		(path was wrong)
	Preprocessor Definitions:  HOST (this wasn't listed anywhere, lots 'o'
pain finding, 
		absolutely required)

4. Added the following to Project settings, Link, General:
		C:\Tornado_03\host\x86-win32\lib\wtxapidll-d.lib (path was wrong)

5. Added the environment variable WIND_REGISTRY to Win NT, set to my
registry location.  Lots and lots of pain 	and digging to find this
one.  Manual refers to a wtxEnvironSet() call that doesn't exist and is 
listed out of order in various places . Windsurf says it doesn't exist,
and don't bother using it. The 	docs for the nonexistent function
provide a confused set of references to all other tools using this 
function.  WIND_REGISTRY was not set on my machine; all tornado tools
are happy without it; my tool 	could not find the registry.  Set it and
poof!, registry found, tool works.  Make sure you add the 	wtxProbe()
call to check for the registry (and the variable being set), it will
save a lot of grief.

6. Modified Source Code:

/*************************************************************
// WTX C API Tool Sample Code for Windows                   //
// C.A Leddy, 02/15/01                                      //
*************************************************************/
/*************************************************************
// Code is provided free for any use, just leave above      //
// header on any derived source                             // 
*************************************************************/
/*************************************************************
// Tornado 1.0.1, Win NT 4.0 SP6a, Visual Studio C++ 6.0    //
// target vxworks 5.3.1                                     //		
*************************************************************/

#include "stdafx.h"
// not shown in example
#include "host.h"
#include "wtx.h"
#include 
#include 

HWTX wtxh;

// WTX housecleaning when quit
void wtxAppTerminate(void)
	{
	wtxToolDetach(wtxh);
	wtxTerminate(wtxh);
	exit(0);
	}

// signal terminate call
void wtxAppSigHandler(int sig)
	{
	wtxAppTerminate();
	}

// install signal handlers
void wtxAppSigInit(void)
	{
	// modified windriver code for windows 32 api:
	signal(SIGABRT, &wtxAppSigHandler);
	signal(SIGFPE, &wtxAppSigHandler);
	signal(SIGILL, &wtxAppSigHandler);
	signal(SIGINT, &wtxAppSigHandler);
	signal(SIGSEGV, &wtxAppSigHandler);
	signal(SIGTERM, &wtxAppSigHandler);
	}

int main(int argc, char* argv[])
	{
	STATUS stats;
	int count=0;
	TGT_ADDR_T t_add;
    	// get server name from command line
	if (argc != 2)
		{
		printf("Error: Usage -> wtxcom targetservername\n");
		exit(0);
		}

	// set up signals like in the example
	// trap errors and terminate properly
	wtxAppSigInit();
	printf("Signals Initialized\n");

	// initialize WTX protocol
	if (wtxInitialize(&wtxh)!= OK)
		{
		printf("Error: WTX API did not Initialize\n");
		exit(0);
		}
	else
		printf("WTX Initialized\n");

	// check registry
	if (wtxProbe(wtxh)!= OK)
		{
		printf("Warning: Tornado Registry Not Found\n");
		printf("     add the environment variable 'WIND-REGISTRY' \npointing
to the registry to 				correct this!\n");
		}
	else
		printf("Tornado Registry Located\n");

	// attached to server on command line with my tool
	stats=wtxToolAttach(wtxh, argv[1], "ChrisTool");
	if (stats != OK)
		{
		printf("Error: WTX did not Attach to target #%s#\n",argv[1]);
		wtxTerminate(wtxh);
		exit(0);
		}
	else
		{
		printf("WTX Tool Attached\n");
		}

	// allocate memory for an integer on target
	t_add=wtxMemAlign(wtxh, 4, 4);
	if (t_add == NULL)
		{
		printf("Error: WTX could not allocate Memory for Integer\n");
		wtxAppTerminate();
		}
	else
		{
		printf("WTX Allocated Memory for Int\n");
		}

	 //add the symbol to target so I can check it on the target shell
	stats=wtxSymAdd(wtxh, "addedsymbol", t_add, V_INT32);
	if (stats != OK)
		{
		printf("Error: WTX did add symbol");
		wtxAppTerminate();
		}
	else
		{
		printf("added 'addedsymbol' to symbol table\n");
		}

	// we can terminate with a CTRL-C (signal handler will catch it)
	// or we dump with this call when done
	wtxAppTerminate();
	return 0;
	}

---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: Rebuilding gcc for ARM
Date: Thu, 15 Feb 2001 18:42:52 GMT
From: Bill Pringlemeir 
Organization: Factory of the mind
Message-ID: 
References: 
Sender: bpringlemeir@DeadDuck

>>>>> "Kevin" == Kevin Cline  writes:

 Kevin> I'm trying to build Dave Korn's gcc source for the ARM, but
 Kevin> I'm stumped on a basic question -- what is the target
 Kevin> architecture.  Is there a list?  Is there a way to find out
 Kevin> the target architecture for the WRS-supplied compilers and use
 Kevin> that?  -- Kevin Cline

objdump has the same target name (or so I believe).

This is from my mingw distribution,
objdump: supported targets: pe-i386 pei-i386 srec symbolsrec tekhex binary ihex

This is the Tornado version,
objdumparm: supported targets: coff-arm-little coff-arm-big srec 
                               symbolsrec tekhex binary ihex

I also believe that there are coff-arm-thumb ...  You should always build
coff-arm-little and/or coff-arm-thumb-little as this is what the vxWorks
library files are distributed as.  Other targets wouldn't be useful.

hth,
Bill

- -- 
Many people will ask you for advice about reading mail.  Listen to
your chicken tomorrow.  You will discover that you do not have a pet.
You will meet your elephant this week.  You'll never be sad.

---------------------------

Newsgroups: comp.os.vxworks
Subject: re: T2 man pages???
Date: Thu, 15 Feb 2001 07:39:22 -0800
From: "Sinn, Larry" 
Organization: Lawrence Berkeley National Laboratory
Message-ID: <96h9ta$kks$1@overload.lbl.gov>

I agree!!!
We upgraded from T1 to T2 and kept the T1 man page directories and created a
Tornado man command:
	alias tman "$WIND_BASE_T1/target/man:$WIND_BASE_T1/host/man"

Larry.

> 	Sounds like WRS can't tell the diff between a 
> feature and a major deficiency.  Menawhile, does anyone have
> the time to write a script to regen the man pages out of 
> all that masty html stuff???
> 
> Speaking only for myself,
> 
> Joe Durusau


> Bob Irwin wrote:
> > 
> > Is it just me or are the Unix style man pages missing in T2??
> > 
> > I got so used to finding libs and functions by using vapropos (man -k).
> > I find the html pages a pain, they take up so much room on the screen.
>>  
> > With the good old unix man pages, I could have a couple of man pages,
> > a couple of includes and a couple of .c files displayed on my monitor
> > with all visible, no overlap.
> > 
> > Is there a way to restore the man pages?  Does anyone miss them?
> > 
> >   tnx.
> > --
> > Bob Irwin | bobi@sis-dev.com | SIS Development | 281-484-6661 | Houston
> >    "Research causes cancer in rats."
> 

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


---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: WTX C API Interface to target server on a Windows Machine
Date: Thu, 15 Feb 2001 10:46:47 -0800
From: Christopher A Leddy 
Organization: Raytheon Company
Message-ID: <3A8C2417.120A26DD@west.raytheon.com>
References: <3A8C2315.78E0D3D9@west.raytheon.com>

Make that 5th line read:

"notebook for later use, but why not post it AND SPARE others some
grief?"

I guess I should read my posts over before sending....

			sorry,

			Chris


Christopher A Leddy wrote:
> 
> This isn't a request for help, just sharing some info.
> 
> I actually decided that I should write the info below for myself in my
> notebook for later use, but why not post it share others some grief?

---------------------------

Newsgroups: comp.os.vxworks
Subject: passFs problem
Date: Thu, 15 Feb 2001 20:05:48 GMT
From: "Eugene Zlobin" 
Organization: EarthLink Inc. -- http://www.EarthLink.net
Message-ID: 

Hi,

In case NT simulator and passFs for i/o operations should use prefix :
"/tgtsvr/" before real path. How i understand it's a bug :).
Does anyone know how to solve this problem (i mean avoid this prefix)?

Best regards,
Eugene.



---------------------------

Newsgroups: comp.os.vxworks
Subject: Can't create vxWorks_rom for mcp750
Date: Fri, 16 Feb 2001 01:39:55 +0530
From: "Balaji  Ramachandran" 
Organization: Lawrence Berkeley National Laboratory
Message-ID: <96hddr$mj9$1@overload.lbl.gov>

VxWorks 5.4
BSP:MCP750

Hi All,

I have tried to build "vxworks_rom" image from a bootable application
created from T-II project facility.

But it gives linker error "undefined reference to "SysGetBusSpd" and
"SysGetDramSpd" error in the "rominit.o" file.

Can any one throw light on this, how to solve it?.


Pl.. see below, the build output


Thanks in advance,

Balaji.

- --------------------------------------------------------------------------------------------------------------

vxrm *.o *.rpo ctdt.c symTbl.c vxWorks*
vxrm C:\Vx_test\Romable\prjComps.h C:\Vx_test\Romable\prjParams.h
C:\Vx_test\Romable\prjCo
nfig.c C:\Vx_test\Romable\linkSyms.c
vxrm C:\Vx_test\Romable\libs.nm C:\Vx_test\Romable\libs.size
wtxtcl C:\Tornado\host\src\hutils\configGen.tcl
C:\Vx_test\Romable\romable.wpj
ccppc -g -mstrict-align -ansi -nostdinc -fvolatile -fno-builtin
- -fno-for-scope -P -x assem
bler-with-cpp -IC:\Vx_test\Romable -IC:\Tornado\target\config\mcp750
- -IC:\Tornado\target\h
 -IC:\Tornado\target\config\comps\src -IC:\Tornado\target\src\drv
- -DCPU=PPC604 -DPRJ_BUILD
 -DMCP750 -DTARGET_DIR="\"mcp750\"" -c
C:\Tornado\target\config\mcp750\sysALib.s -o sysALi
b.o
C:\TEMP\cca00234.s: Assembler messages:
C:\TEMP\cca00234.s:4764: Warning: operand out of range (-32768 not
between 0 and 65535)
C:\TEMP\cca00234.s:4853: Warning: operand out of range (57343 not
between -32768 and 32767
)
ccppc -O2 -g -mstrict-align -ansi -nostdinc -DRW_MULTI_THREAD
- -D_REENTRANT -fvolatile -fno
- -builtin -fno-for-scope -IC:\Vx_test\Romable
- -IC:\Tornado\target\config\mcp750 -IC:\Tornad
o\target\h -IC:\Tornado\target\config\comps\src
- -IC:\Tornado\target\src\drv -DCPU=PPC604 -
DPRJ_BUILD -DMCP750 -DTARGET_DIR="\"mcp750\"" -c
C:\Tornado\target\config\mcp750\sysLib.c
ccppc -O2 -g -mstrict-align -ansi -nostdinc -DRW_MULTI_THREAD
- -D_REENTRANT -fvolatile -fno
- -builtin -fno-for-scope -IC:\Vx_test\Romable
- -IC:\Tornado\target\config\mcp750 -IC:\Tornad
o\target\h -IC:\Tornado\target\config\comps\src
- -IC:\Tornado\target\src\drv -DCPU=PPC604 -
DPRJ_BUILD -DMCP750 -DTARGET_DIR="\"mcp750\"" -c
C:\Tornado\target\config\mcp750\sysSpeed.
c
ccppc -O2 -g -mstrict-align -ansi -nostdinc -DRW_MULTI_THREAD
- -D_REENTRANT -fvolatile -fno
- -builtin -fno-for-scope -IC:\Vx_test\Romable
- -IC:\Tornado\target\config\mcp750 -IC:\Tornad
o\target\h -IC:\Tornado\target\config\comps\src
- -IC:\Tornado\target\src\drv -DCPU=PPC604 -
DPRJ_BUILD -DMCP750 -DTARGET_DIR="\"mcp750\"" -c
C:\Tornado\target\config\mcp750\usrLib.c
ccppc -O2 -g -mstrict-align -ansi -nostdinc -DRW_MULTI_THREAD
- -D_REENTRANT -fvolatile -fno
- -builtin -fno-for-scope -IC:\Vx_test\Romable
- -IC:\Tornado\target\config\mcp750 -IC:\Tornad
o\target\h -IC:\Tornado\target\config\comps\src
- -IC:\Tornado\target\src\drv -DCPU=PPC604 -
DPRJ_BUILD -DMCP750 -DTARGET_DIR="\"mcp750\"" -c
C:\Vx_test\Romable\usrAppInit.c
ccppc -O2 -g -mstrict-align -ansi -nostdinc -DRW_MULTI_THREAD
- -D_REENTRANT -fvolatile -fno
- -builtin -fno-for-scope -IC:\Vx_test\Romable
- -IC:\Tornado\target\config\mcp750 -IC:\Tornad
o\target\h -IC:\Tornado\target\config\comps\src
- -IC:\Tornado\target\src\drv -DCPU=PPC604 -
DPRJ_BUILD -DMCP750 -DTARGET_DIR="\"mcp750\"" -c
C:\Vx_test\Romable\prjConfig.c
ccppc -O2 -g -mstrict-align -ansi -nostdinc -DRW_MULTI_THREAD
- -D_REENTRANT -fvolatile -fno
- -builtin -fno-for-scope -IC:\Vx_test\Romable
- -IC:\Tornado\target\config\mcp750 -IC:\Tornad
o\target\h -IC:\Tornado\target\config\comps\src
- -IC:\Tornado\target\src\drv -DCPU=PPC604 -
DPRJ_BUILD -DMCP750 -DTARGET_DIR="\"mcp750\"" -c
C:\Vx_test\Romable\linkSyms.c
ccppc -O2 -g -mstrict-align -ansi -nostdinc -DRW_MULTI_THREAD
- -D_REENTRANT -fvolatile -fno
- -builtin -fno-for-scope -IC:\Vx_test\Romable
- -IC:\Tornado\target\config\mcp750 -IC:\Tornad
o\target\h -IC:\Tornado\target\config\comps\src
- -IC:\Tornado\target\src\drv -DCPU=PPC604 -
DPRJ_BUILD -DMCP750 -DTARGET_DIR="\"mcp750\"" -c
C:\Tornado\target\config\comps\src\versio
n.c
ccppc -O2 -g -mstrict-align -ansi -nostdinc -DRW_MULTI_THREAD
- -D_REENTRANT -fvolatile -fno
- -builtin -fno-for-scope -IC:\Vx_test\Romable
- -IC:\Tornado\target\config\mcp750 -IC:\Tornad
o\target\h -IC:\Tornado\target\config\comps\src
- -IC:\Tornado\target\src\drv -DCPU=PPC604 -
DPRJ_BUILD -DMCP750 -DTARGET_DIR="\"mcp750\"" -c
C:\Tornado\target\config\comps\src\usrEnt
ry.c
ccppc -BC:\Tornado\host\x86-win32\lib\gcc-lib\ -nostdlib -r -Wl,-X \
    usrEntry.o -Wl,@C:\Vx_test\Romable\prjObjs.lst  version.o
C:\Tornado\target\lib\libPP
C604gnuvx.a -o partialImage.o
nmppc partialImage.o @C:\Vx_test\Romable\prjObjs.lst | wtxtcl
C:\Tornado\host\src\hutils\m
unch.tcl -asm ppc > ctdt.c
ccppc -c -fdollars-in-identifiers -O2 -g -mstrict-align -ansi -nostdinc
- -DRW_MULTI_THREAD
- -D_REENTRANT -fvolatile -fno-builtin -fno-for-scope -IC:\Vx_test\Romable
- -IC:\Tornado\targ
et\config\mcp750 -IC:\Tornado\target\h
- -IC:\Tornado\target\config\comps\src -IC:\Tornado\t
arget\src\drv -DCPU=PPC604 -DPRJ_BUILD -DMCP750
- -DTARGET_DIR="\"mcp750\"" ctdt.c -o ctdt.o
ccppc -O2 -g -mstrict-align -ansi -nostdinc -DRW_MULTI_THREAD
- -D_REENTRANT -fvolatile -fno
- -builtin -fno-for-scope -IC:\Vx_test\Romable
- -IC:\Tornado\target\config\mcp750 -IC:\Tornad
o\target\h -IC:\Tornado\target\config\comps\src
- -IC:\Tornado\target\src\drv -DCPU=PPC604 -
DPRJ_BUILD -DMCP750 -DTARGET_DIR="\"mcp750\"" -c
C:\Tornado\target\config\comps\src\dataSe
gPad.c
ldppc -X -N -e _sysInit -Ttext 00100000  \
 dataSegPad.o partialImage.o ctdt.o -o vxWorks
C:\Tornado\host\x86-win32\bin\vxsize PPC -v 00300000  00100000  vxWorks
vxWorks: 799248(t) + 96244(d) +  35796(b) = 931288 (1165864 unused)
ccppc -c -g -mstrict-align -ansi -nostdinc -fvolatile -fno-builtin
- -fno-for-scope -P -x as
sembler-with-cpp -IC:\Vx_test\Romable -IC:\Tornado\target\config\mcp750
- -IC:\Tornado\targe
t\h -IC:\Tornado\target\config\comps\src -IC:\Tornado\target\src\drv
- -DCPU=PPC604 -DPRJ_BU
ILD -DMCP750 -DTARGET_DIR="\"mcp750\"" -DROM_COPY
- -DRAM_DATA_ADRS=0x00300000  -DRAM_DST_AD
RS=0x00100000  C:\Tornado\target\config\mcp750\romInit.s -o romInit.o
C:\TEMP\cca00304.s: Assembler messages:
C:\TEMP\cca00304.s:4745: Warning: operand out of range (-32768 not
between 0 and 65535)
C:\TEMP\cca00304.s:4823: Warning: operand out of range (57343 not
between -32768 and 32767
)
ccppc -c -O2 -g -mstrict-align -ansi -nostdinc -DRW_MULTI_THREAD
- -D_REENTRANT -fvolatile -
fno-builtin -fno-for-scope -IC:\Vx_test\Romable
- -IC:\Tornado\target\config\mcp750 -IC:\Tor
nado\target\h -IC:\Tornado\target\config\comps\src
- -IC:\Tornado\target\src\drv -DCPU=PPC60
4 -DPRJ_BUILD -DMCP750 -DTARGET_DIR="\"mcp750\"" -DROM_COPY
- -DRAM_DATA_ADRS=0x00300000  -D
RAM_DST_ADRS=0x00100000  C:\Tornado\target\config\comps\src\romStart.c
- -o romStart.o
copy vxWorks vxWorks.o
        1 file(s) copied.
C:\Tornado\host\x86-win32\bin\elfToBin < vxWorks.o > tmp.Z
C:\Tornado\host\x86-win32\bin\binToAsm tmp.Z > vxWorks.Z.s
ccppc -c -O2 -g -mstrict-align -ansi -nostdinc -DRW_MULTI_THREAD
- -D_REENTRANT -fvolatile -
fno-builtin -fno-for-scope -IC:\Vx_test\Romable
- -IC:\Tornado\target\config\mcp750 -IC:\Tor
nado\target\h -IC:\Tornado\target\config\comps\src
- -IC:\Tornado\target\src\drv -DCPU=PPC60
4 -DPRJ_BUILD -DMCP750 -DTARGET_DIR="\"mcp750\"" vxWorks.Z.s -o
vxWorks.Z.o
ccppc -O2 -g -mstrict-align -ansi -nostdinc -DRW_MULTI_THREAD
- -D_REENTRANT -fvolatile -fno
- -builtin -fno-for-scope -IC:\Vx_test\Romable
- -IC:\Tornado\target\config\mcp750 -IC:\Tornad
o\target\h -IC:\Tornado\target\config\comps\src
- -IC:\Tornado\target\src\drv -DCPU=PPC604 -
DPRJ_BUILD -DMCP750 -DTARGET_DIR="\"mcp750\"" -c
C:\Tornado\target\config\comps\src\versio
n.c
ldppc -X -N -e _romInit  -Ttext fff00100  -Tdata 00300000  \
    romInit.o romStart.o version.o
C:\Tornado\target\lib\libPPC604gnuvx.a vxWorks.Z.o -o
vxWorks_rom
romInit.o(.text+0x40c): undefined reference to `sysGetDramSpd'
romInit.o(.text+0x438): undefined reference to `sysGetBusSpd'
make: *** [vxWorks_rom] Error 0x1

Done.
- -------------------------------------------------------------------------------------------------------------------------

VxWorks / Tornado



---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: STDIN problem on target
Date: Thu, 15 Feb 2001 13:12:27 -0800
From: "Gang Wang" 
Organization: None
Message-ID: <0EXi6.7$s%1.559@newsfeed.avtel.net>
References: <96grrb$e8t$1@overload.lbl.gov>

Ken:

Thanks for your response.
How do you run it, exactly?
I tried the following two ways:
1) In winsh windows, type testStdIO:
I could not get character echo back when I am typing.
2) In winsh, run "sp(testStdIO)"
nothing happens.

What's wrong with my setting?

Thanks in advance.

Baron, Ken wrote in message <96grrb$e8t$1@overload.lbl.gov>...
>
> Your code below works fine for me. If you run testStdIO from the
>windsh the function receives input from the shell window.
> If you spawn testStdIO as a new task, then the task receives input
>from the console. (vxworks sim window or your primary serial port on your
>hardware.
>
>> Hi,
>> I am new to vxWorks and WinSh. I am using Tornado II. I am using the
WinSh
>> to exec the following code. ( i.e. in windsh , after load in the .o file,
>> type >testStdIO). I wish I could see the characters I typed in there
>> reflected in the window, just as xterm. However, it seems that the data
is
>> swallowed by the application.
>>
>> How can I do this.
>>
>> Gang Wang
>>
>> #include 
>>
>> int testStdIO()
>> {
>>    FILE *fp;
>>    char  data[256];
>>
>>    /*if ( ( fp = fopen(stdin, "r")) == NULL )
>>      {
>>        printf("ERROR in open stdin");
>>        return 0;
>>       }*/
>>    while (1)
>>    {
>>      fgets(data, 256, stdin);
>>      if ( data[0] == 'q' )
>>         return 1;
>>      else
>>        printf("I get something\n");
>>    }
>>    return 0 ;
>> }
>>



---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: How to spawn two target server for two different boards in one 
 machine?
Date: Thu, 15 Feb 2001 16:14:39 -0500
From: Peter Colby 
Organization: Raytheon Systems Company C3I
Message-ID: <3A8C46BF.922117EB@res.raytheon.com>
References: <90a6f9$1kku$1@news.cz.js.cn> <3A297659.6F795C24@flashcom.net>

Using the T2 IDE I was able to start 2 target servers - each pointing to
a separate processor in the
same target VME chassis. I was then able to start a separate target
shell for each server.

Windview was only able to connect to one processor at a time though - I
can't remember whether I
ever used the debugger.

All the sub-windows appeared in the large IDE window - course it was
sometimes hard to find the
one I wanted to view...

DrDiags wrote:
> 
> Tony,
> 
>    You should be able to create multiple target servers and then select the
> target you are interested in using the tools with by selecting it from the drop
> down list. If you want to have the ability to debug both at the same time, you
> should fire up another instance of the T2 IDE and connect to the second target.
> If you are trying to switch from one to the other in the same IDE with tools
> already attached, this may not work for you. Others may have a workaround, I am
> speaking only as a newbie, so take my input with a grain of salt.
> 
> TonyHuang wrote:
> 
> > Hi vxworkers,
> >
> > I am trying to make a communication test through ethernet  between two
> > PPC860T boards.
> >
> > I assigned the two boards with different IP address, different  MAC address,
> > different  user name, different  target name,
> > and booted them from the tornado host , the booting processes of both are
> > OK.
> >
> > then I tried to spawn the target server for each board with different IP
> > address that I assigned before.
> > the first spawned target sever is OK, but when the second is spawned, the
> > first is corrupted.
> >
> > I have paid attention to the two target board's ethernet LED indicators,
> > they always blinks synchronously.
> > I'm not sure the tornado tools takes them different target.
> >
> > Why can't I start two target servers at the same time? Tornado2 does not
> > support this?
> > or maybe my network configuration has problem?
> > Please give me some suggestion.
> > Thanks in advance.
> >
> > My testing environment:
> >
> > Tornado2 project facility;
> > vxWorks5.4;
> > MPC860T50B3;
> > motFecEnd driver;
> > Booting from ethernet using FTP protocol.
> >
> > Tony

---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: Can't create vxWorks_rom for mcp750
Date: Thu, 15 Feb 2001 13:55:17 -0800
From: Rajesh Tripathi 
Organization: Globix Corp.
Message-ID: <3A8C5045.FFAB587D@com21.com>
References: <96hddr$mj9$1@overload.lbl.gov>

Balaji Ramachandran wrote:

> VxWorks 5.4
> BSP:MCP750
>
> Hi All,
>
> I have tried to build "vxworks_rom" image from a bootable application
> created from T-II project facility.
>
> But it gives linker error "undefined reference to "SysGetBusSpd" and
> "SysGetDramSpd" error in the "rominit.o" file.
>
> Can any one throw light on this, how to solve it?.
>
> Pl.. see below, the build output
>
> Thanks in advance,
>
> Balaji.
>
> --------------------------------------------------------------------------------------------------------------
>
> vxrm *.o *.rpo ctdt.c symTbl.c vxWorks*
> vxrm C:\Vx_test\Romable\prjComps.h C:\Vx_test\Romable\prjParams.h
> C:\Vx_test\Romable\prjCo
> nfig.c C:\Vx_test\Romable\linkSyms.c
> vxrm C:\Vx_test\Romable\libs.nm C:\Vx_test\Romable\libs.size
> wtxtcl C:\Tornado\host\src\hutils\configGen.tcl
> C:\Vx_test\Romable\romable.wpj
> ccppc -g -mstrict-align -ansi -nostdinc -fvolatile -fno-builtin
> -fno-for-scope -P -x assem
> bler-with-cpp -IC:\Vx_test\Romable -IC:\Tornado\target\config\mcp750
> -IC:\Tornado\target\h
>  -IC:\Tornado\target\config\comps\src -IC:\Tornado\target\src\drv
> -DCPU=PPC604 -DPRJ_BUILD
>  -DMCP750 -DTARGET_DIR="\"mcp750\"" -c
> C:\Tornado\target\config\mcp750\sysALib.s -o sysALi
> b.o
> C:\TEMP\cca00234.s: Assembler messages:
> C:\TEMP\cca00234.s:4764: Warning: operand out of range (-32768 not
> between 0 and 65535)
> C:\TEMP\cca00234.s:4853: Warning: operand out of range (57343 not
> between -32768 and 32767
> )
> ccppc -O2 -g -mstrict-align -ansi -nostdinc -DRW_MULTI_THREAD
> -D_REENTRANT -fvolatile -fno
> -builtin -fno-for-scope -IC:\Vx_test\Romable
> -IC:\Tornado\target\config\mcp750 -IC:\Tornad
> o\target\h -IC:\Tornado\target\config\comps\src
> -IC:\Tornado\target\src\drv -DCPU=PPC604 -
> DPRJ_BUILD -DMCP750 -DTARGET_DIR="\"mcp750\"" -c
> C:\Tornado\target\config\mcp750\sysLib.c
> ccppc -O2 -g -mstrict-align -ansi -nostdinc -DRW_MULTI_THREAD
> -D_REENTRANT -fvolatile -fno
> -builtin -fno-for-scope -IC:\Vx_test\Romable
> -IC:\Tornado\target\config\mcp750 -IC:\Tornad
> o\target\h -IC:\Tornado\target\config\comps\src
> -IC:\Tornado\target\src\drv -DCPU=PPC604 -
> DPRJ_BUILD -DMCP750 -DTARGET_DIR="\"mcp750\"" -c
> C:\Tornado\target\config\mcp750\sysSpeed.
> c
> ccppc -O2 -g -mstrict-align -ansi -nostdinc -DRW_MULTI_THREAD
> -D_REENTRANT -fvolatile -fno
> -builtin -fno-for-scope -IC:\Vx_test\Romable
> -IC:\Tornado\target\config\mcp750 -IC:\Tornad
> o\target\h -IC:\Tornado\target\config\comps\src
> -IC:\Tornado\target\src\drv -DCPU=PPC604 -
> DPRJ_BUILD -DMCP750 -DTARGET_DIR="\"mcp750\"" -c
> C:\Tornado\target\config\mcp750\usrLib.c
> ccppc -O2 -g -mstrict-align -ansi -nostdinc -DRW_MULTI_THREAD
> -D_REENTRANT -fvolatile -fno
> -builtin -fno-for-scope -IC:\Vx_test\Romable
> -IC:\Tornado\target\config\mcp750 -IC:\Tornad
> o\target\h -IC:\Tornado\target\config\comps\src
> -IC:\Tornado\target\src\drv -DCPU=PPC604 -
> DPRJ_BUILD -DMCP750 -DTARGET_DIR="\"mcp750\"" -c
> C:\Vx_test\Romable\usrAppInit.c
> ccppc -O2 -g -mstrict-align -ansi -nostdinc -DRW_MULTI_THREAD
> -D_REENTRANT -fvolatile -fno
> -builtin -fno-for-scope -IC:\Vx_test\Romable
> -IC:\Tornado\target\config\mcp750 -IC:\Tornad
> o\target\h -IC:\Tornado\target\config\comps\src
> -IC:\Tornado\target\src\drv -DCPU=PPC604 -
> DPRJ_BUILD -DMCP750 -DTARGET_DIR="\"mcp750\"" -c
> C:\Vx_test\Romable\prjConfig.c
> ccppc -O2 -g -mstrict-align -ansi -nostdinc -DRW_MULTI_THREAD
> -D_REENTRANT -fvolatile -fno
> -builtin -fno-for-scope -IC:\Vx_test\Romable
> -IC:\Tornado\target\config\mcp750 -IC:\Tornad
> o\target\h -IC:\Tornado\target\config\comps\src
> -IC:\Tornado\target\src\drv -DCPU=PPC604 -
> DPRJ_BUILD -DMCP750 -DTARGET_DIR="\"mcp750\"" -c
> C:\Vx_test\Romable\linkSyms.c
> ccppc -O2 -g -mstrict-align -ansi -nostdinc -DRW_MULTI_THREAD
> -D_REENTRANT -fvolatile -fno
> -builtin -fno-for-scope -IC:\Vx_test\Romable
> -IC:\Tornado\target\config\mcp750 -IC:\Tornad
> o\target\h -IC:\Tornado\target\config\comps\src
> -IC:\Tornado\target\src\drv -DCPU=PPC604 -
> DPRJ_BUILD -DMCP750 -DTARGET_DIR="\"mcp750\"" -c
> C:\Tornado\target\config\comps\src\versio
> n.c
> ccppc -O2 -g -mstrict-align -ansi -nostdinc -DRW_MULTI_THREAD
> -D_REENTRANT -fvolatile -fno
> -builtin -fno-for-scope -IC:\Vx_test\Romable
> -IC:\Tornado\target\config\mcp750 -IC:\Tornad
> o\target\h -IC:\Tornado\target\config\comps\src
> -IC:\Tornado\target\src\drv -DCPU=PPC604 -
> DPRJ_BUILD -DMCP750 -DTARGET_DIR="\"mcp750\"" -c
> C:\Tornado\target\config\comps\src\usrEnt
> ry.c
> ccppc -BC:\Tornado\host\x86-win32\lib\gcc-lib\ -nostdlib -r -Wl,-X \
>     usrEntry.o -Wl,@C:\Vx_test\Romable\prjObjs.lst  version.o
> C:\Tornado\target\lib\libPP
> C604gnuvx.a -o partialImage.o
> nmppc partialImage.o @C:\Vx_test\Romable\prjObjs.lst | wtxtcl
> C:\Tornado\host\src\hutils\m
> unch.tcl -asm ppc > ctdt.c
> ccppc -c -fdollars-in-identifiers -O2 -g -mstrict-align -ansi -nostdinc
> -DRW_MULTI_THREAD
> -D_REENTRANT -fvolatile -fno-builtin -fno-for-scope -IC:\Vx_test\Romable
> -IC:\Tornado\targ
> et\config\mcp750 -IC:\Tornado\target\h
> -IC:\Tornado\target\config\comps\src -IC:\Tornado\t
> arget\src\drv -DCPU=PPC604 -DPRJ_BUILD -DMCP750
> -DTARGET_DIR="\"mcp750\"" ctdt.c -o ctdt.o
> ccppc -O2 -g -mstrict-align -ansi -nostdinc -DRW_MULTI_THREAD
> -D_REENTRANT -fvolatile -fno
> -builtin -fno-for-scope -IC:\Vx_test\Romable
> -IC:\Tornado\target\config\mcp750 -IC:\Tornad
> o\target\h -IC:\Tornado\target\config\comps\src
> -IC:\Tornado\target\src\drv -DCPU=PPC604 -
> DPRJ_BUILD -DMCP750 -DTARGET_DIR="\"mcp750\"" -c
> C:\Tornado\target\config\comps\src\dataSe
> gPad.c
> ldppc -X -N -e _sysInit -Ttext 00100000  \
>  dataSegPad.o partialImage.o ctdt.o -o vxWorks
> C:\Tornado\host\x86-win32\bin\vxsize PPC -v 00300000  00100000  vxWorks
> vxWorks: 799248(t) + 96244(d) +  35796(b) = 931288 (1165864 unused)
> ccppc -c -g -mstrict-align -ansi -nostdinc -fvolatile -fno-builtin
> -fno-for-scope -P -x as
> sembler-with-cpp -IC:\Vx_test\Romable -IC:\Tornado\target\config\mcp750
> -IC:\Tornado\targe
> t\h -IC:\Tornado\target\config\comps\src -IC:\Tornado\target\src\drv
> -DCPU=PPC604 -DPRJ_BU
> ILD -DMCP750 -DTARGET_DIR="\"mcp750\"" -DROM_COPY
> -DRAM_DATA_ADRS=0x00300000  -DRAM_DST_AD
> RS=0x00100000  C:\Tornado\target\config\mcp750\romInit.s -o romInit.o
> C:\TEMP\cca00304.s: Assembler messages:
> C:\TEMP\cca00304.s:4745: Warning: operand out of range (-32768 not
> between 0 and 65535)
> C:\TEMP\cca00304.s:4823: Warning: operand out of range (57343 not
> between -32768 and 32767
> )
> ccppc -c -O2 -g -mstrict-align -ansi -nostdinc -DRW_MULTI_THREAD
> -D_REENTRANT -fvolatile -
> fno-builtin -fno-for-scope -IC:\Vx_test\Romable
> -IC:\Tornado\target\config\mcp750 -IC:\Tor
> nado\target\h -IC:\Tornado\target\config\comps\src
> -IC:\Tornado\target\src\drv -DCPU=PPC60
> 4 -DPRJ_BUILD -DMCP750 -DTARGET_DIR="\"mcp750\"" -DROM_COPY
> -DRAM_DATA_ADRS=0x00300000  -D
> RAM_DST_ADRS=0x00100000  C:\Tornado\target\config\comps\src\romStart.c
> -o romStart.o
> copy vxWorks vxWorks.o
>         1 file(s) copied.
> C:\Tornado\host\x86-win32\bin\elfToBin < vxWorks.o > tmp.Z
> C:\Tornado\host\x86-win32\bin\binToAsm tmp.Z > vxWorks.Z.s
> ccppc -c -O2 -g -mstrict-align -ansi -nostdinc -DRW_MULTI_THREAD
> -D_REENTRANT -fvolatile -
> fno-builtin -fno-for-scope -IC:\Vx_test\Romable
> -IC:\Tornado\target\config\mcp750 -IC:\Tor
> nado\target\h -IC:\Tornado\target\config\comps\src
> -IC:\Tornado\target\src\drv -DCPU=PPC60
> 4 -DPRJ_BUILD -DMCP750 -DTARGET_DIR="\"mcp750\"" vxWorks.Z.s -o
> vxWorks.Z.o
> ccppc -O2 -g -mstrict-align -ansi -nostdinc -DRW_MULTI_THREAD
> -D_REENTRANT -fvolatile -fno
> -builtin -fno-for-scope -IC:\Vx_test\Romable
> -IC:\Tornado\target\config\mcp750 -IC:\Tornad
> o\target\h -IC:\Tornado\target\config\comps\src
> -IC:\Tornado\target\src\drv -DCPU=PPC604 -
> DPRJ_BUILD -DMCP750 -DTARGET_DIR="\"mcp750\"" -c
> C:\Tornado\target\config\comps\src\versio
> n.c
> ldppc -X -N -e _romInit  -Ttext fff00100  -Tdata 00300000  \
>     romInit.o romStart.o version.o
> C:\Tornado\target\lib\libPPC604gnuvx.a vxWorks.Z.o -o
> vxWorks_rom
> romInit.o(.text+0x40c): undefined reference to `sysGetDramSpd'
> romInit.o(.text+0x438): undefined reference to `sysGetBusSpd'
> make: *** [vxWorks_rom] Error 0x1
>
> Done.
> -------------------------------------------------------------------------------------------------------------------------
>
> VxWorks / Tornado

a wrs bsp file sysSpeed.c should be in the bsp directory.  just add sysSpeed.o to MACH_EXTRA in the make file.

Rajesh


---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: Shared memory network problems
Date: Thu, 15 Feb 2001 16:49:18 -0500
From: Peter Colby 
Organization: Raytheon Systems Company C3I
Message-ID: <3A8C4EDE.9ED47981@res.raytheon.com>
References: <3a2c0d56.3444883@news.ic24.net>

We use the shared memory network stack on a Pentium over a VME
backplane.
The vxWorks version is 5.3.1 (T1.0.1). If TCP sockets are used the net
is reasonably
stable, however UDP just doesn't cut it, too many datagrams and the
shared memory
buffers just bleed away until the whole net hangs with ENOBUFS in
tNetTask.

The spin_lock mechanism vxWorks uses to control access to the buffer
pool can
timeout easily when there is a lot of interprocessor traffic and the
higher layer
protocols just can't properly recover from the timeout.

Peter Colby

Mark Lamb wrote:
> 
> Hi,
> 
> I'm currently having problems using TCP sockets over the shared memory
> network. I get panics from (usually) tNetTask and eventually mbufShow
> shows corruption of its mbuf statistics (the totals don't agree). This
> is often accompanied by a fail light coming on on my slot 0
> controller, a National Instruments VXIcpu-030 (68030-based). I'm
> currently running vxWorks 5.1.1 (can't change at the moment).
> 
> I have a couple of questions:
> 
> 1. Has anyone else had problems using the shared memory network?
> 
> 2. Has anyone developed or know of an alternative implementation of
> TCP sockets which doesn't use the shared memory?
> 
> Thanks,
> 
> Mark Lamb

---------------------------

Newsgroups: comp.os.vxworks
Subject: file I/O performance
Date: Thu, 15 Feb 2001 18:17:25 -0500
From: "Mike Kordik" 
Message-ID: <_pZi6.77$in6.941@client>

Is this normal?

I was trying to understand how the disk caching works as it sometimes
preempts a higher priority task so that it can write out it's cached data to
disk. I am using DOM and the write times are very slow. The read times are
very fast. But the is to be expected.

To further test I decided to write to a ram disk. To my surprise the write
is still much slower than the read. I am writing a a 2K string a 100 times.
The performance is better when I write larger blocks of data.

On my 233Mhz board the write is about 10Mb/s and the read is almost 500Mb/s.

Is this normal that the write is that much slower?

Thanks,
Mike



---------------------------

Newsgroups: comp.os.vxworks
Subject: File I/O through Serial Link on RAD6000
Date: Thu, 15 Feb 2001 16:19:14 -0800
From: "Anthony A. Salazar" 
Organization: University of Oregon, Eugene
Message-ID: <96ho3h$928$1@pith.uoregon.edu>

VxWorks Tornado


Hi folks,


My current configuration is a RAD6000 booting VxWorks 5.3.1 (RAD6000 BSP)
over a serial line (ttya) from a Sun. I use Tornado V1.0.1 to communicate
with VxWorks on a Boeing RAD6000. I belive Tornado uses Wind Rivers WDB
protocol to communicate with the host via Virtual I/O Channels.


Is there a way to access files on the host(sun) from the target (RAD6000)
over the serial line?  Are the WTX host tools a possible avenue
or are they just for use on the host side (sun)?


Any suggestions would be greatly appreciated.


Thanks,
Anthony Salazar




More Details:


At login we redirect Stdin, Stdout and Stderr
via a windsh.tcl script:


    shParse {ioGlobalStdSet (0,vf0)} ;
    shParse {ioGlobalStdSet (1,vf0)} ;
    shParse {ioGlobalStdSet (2,vf0)} ;
    shParse {logFdAdd (vf0)} ;



When I issue devs and iosDrvShow I get:
- -> devs
drv name
  0 /null
  3 /vio
value = 0 = 0x0


- -> iosDrvShow
drv    create     delete       open      close       read      write
ioctl
  1     7f160          0      7f160          0      7fc40      7fb30 7f168
  2         0          0      6f014          0      6f080      6f0fc 6f240
  3     8a0ac          0      8a0ac      8a228      7fc40      7fb30 8a2bc
value = 0 = 0x0
- -


I can do things like:


#include "vxWorks.h"
#include 


int CFErun(void)
{
        while(1)
        {
          fgets(data,256,stdin);
          if(data[0] == 'q')
            return 1;
          else
            printf("I get something\n");
         }
}


but if I try to do file I/O instead I never get a valid file descriptor:


        if ((Fd=fopen("test2","r")) != NULL)
                printf("Fd = %d\n",Fd);
        else
                printf(" Couldn't open file !\n");


        numread=fread(&dummy,sizeof(unsigned long),1,Fd);
        printf("Numread = %d     dummy = %#010x\n",numread,dummy);


        mystatus=fclose(Fd);
        printf("Close status = %d\n",mystatus);





****************************************************************************
**
Anthony A. Salazar                                   NIS-3 \ DAPS \ CIBOLA
Los Alamos National Laboratory                       Telephone: (505)
667-2508
Nonproliferation International Security (NIS)        Fax: (505) 665-4197
P.O. Box 1663, MS:D440
E-Mail:aasalazar@lanl.gov
Los Alamos, NM     87545
****************************************************************************
**




---------------------------

Newsgroups: comp.os.vxworks
Subject: Question on SNMP agent getBulk operation
Date: Thu, 15 Feb 2001 23:35:54 -0800
From: "Joe Zhou" 
Organization: Lawrence Berkeley National Laboratory
Message-ID: <96injq$cic$1@overload.lbl.gov>

This is a multi-part message in MIME format.

- ------=_NextPart_000_0007_01C097A8.0A27BC10
Content-Type: text/plain;
	charset="iso-8859-1"
Content-Transfer-Encoding: quoted-printable

Hi, all

I am new to this, so bear with me.
I am trying to implement GETBULK operation on SNMP agent, using Tornado2 =
WindNet SNMPV1/v2c.
I've finished GET,GETNEXT and SET operation, but somehow I am a little =
bit confused with GETBULK.=20
Using a SNMP manager, I send a GETBUK to the SNMP agent. It seems that =
agent can interpret the packet correctly (showing the pdu type=3D5, =
correct max_reps and non_reps) and seemingly, the result returned to the =
manager is corret. So my question is -- I never customerily implemented =
my getbulk functions, has it been taken care of by Tornado internally?

From the backtrace window, I can see that following procedures have been =
invoked in sequence:
snmpdMain-->snmpIoMain-->snmpIoBody-->snmpdPktProcess-->Process_Rcvd_SNMP=
_Packet_Asyn-->process_packet_one-->process_packet_twon-->snmpdContinue--=
> my_own_getnext_procedure
So does it mean that Tornado treat GETBULK as getnext internally? Do I =
still need to do anything?

I am sure there are people out there are much more experienced than I =
am, please help! Thanks in advance.

Regards

Joe Zhou


- ------=_NextPart_000_0007_01C097A8.0A27BC10
Content-Type: text/html;
	charset="iso-8859-1"
Content-Transfer-Encoding: quoted-printable









Hi, all


 


I am new to this, so bear with =
me.


I am trying to implement GETBULK =
operation on=20
SNMP agent, using Tornado2 WindNet SNMPV1/v2c.


I've finished GET,GETNEXT and SET =
operation,=20
but somehow I am a little bit confused with GETBULK. 


Using a SNMP manager, I send a GETBUK =
to the SNMP=20
agent. It seems that agent can interpret the packet correctly (showing =
the pdu=20
type=3D5, correct max_reps and non_reps) and seemingly, the result returned to the manager is corret. So my =
question is=20
- -- I never customerily implemented my getbulk functions, has it been =
taken care=20
of by Tornado internally?


 


From the backtrace window, I can see =
that following=20
procedures have been invoked in sequence:


snmpdMain-->snmpIoMain-->snmpIoBody-->snmpdPktProcess--=
>Process_Rcvd_SNMP_Packet_Asyn-->process_packet_one-->process_pa=
cket_twon-->snmpdContinue-->=20
my_own_getnext_procedure


So does it mean that Tornado treat =
GETBULK as=20
getnext internally? Do I still need to do anything?


 


I am sure there are people out there =
are much more=20
experienced than I am, please help! Thanks in advance.


 


Regards


 


Joe Zhou


- ------=_NextPart_000_0007_01C097A8.0A27BC10--


---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: (beginer) Problem "catching" a cPCI interrupt
Date: Fri, 16 Feb 2001 11:12:58 +0100
From: Wim Mosterman 
Organization: PMSN
Message-ID: <3A8CFD2A.900A38E1@philips.com>
References: <3a8a4aab@news.bezeqint.net>

David,

If I recall correctly, you should add INT_NUM_IRQ0 (defined in config.h) to your
interrupt number like this:
INUM_TO_IVEC(lvl + INT_NUM_IRQ0)

Hope this helps...

Wim Mosterman


David Ganor wrote:

> Hello,
>
> Some backgroug info, first:
> 1. VxWorks 5.4
> 2. Target: Motorola with CPU PPC750
> 3. BSP version: 1.2/0
> 4. Tornado II
>
> I have a system board with a cPCI interface, on which a PLX9054 chip is the
> PCI bridge (the board has a local bus...).
> I have configured the CPU_PCI_MEM_SIZE to be 32MBytes and the board uses
> about 10MBytes.
> I have wrote an ISR to be activated when the board, througth the PLX chip,
> drives an interrupt. There are no problems configuring the PLX and the
> interrupt
> line is "active".
> When running the code the ISR isn't activated, thought the intConnect()
> returns
> with OK. I used intConnecte() like that:
>     lvl = PCI_cfg->DEV_INT_LINE;
>     ...
>     if (intConnect (INUM_TO_IVEC (lvl), (VOIDFUNCPTR) ISR, (int) this) !=
> OK)
>     {
>         printf("[PLX] ERROR - Failed to connect interrupt to ISR!\n");
>     }
> Could anyone sujest an insight? What am I doing wrong?
> Maybe the PLX9054 isn't configured right? -> What are the main configuration
> regarding
> the described problem?
>
> Thanks in advance.
> David Ganor


---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: Syncing symbol tables between target-resident and host shells
Date: 16 Feb 2001 11:11:11 GMT
From: borkhuis@agere.com (Johan Borkhuis)
Organization: Lucent Technologies
Message-ID: 
References: <3A8C1A69.58396313@ferndown.tt.slb.com>

Tom Ko  wrote:
>In the vxWorks, you can use the startup script to load modules. To do
>this, you need to use the target-resident shell. Once loaded, the symbol
>information are stored in the symbol table of the target-resident shell.
>Now if you launch the host shell from NT, these modules/symbols you
>loaded during the startup cannot be seen from the host shell. For
>example, 'moduleShow' on the host shell will not list modules loaded
>during the startup. However, 'moduleShow' in the target-resident shell
>i.e. the hyper-terminal window, will list the loaded modules.
>
>Is there a way to sync these tables?

Yes, enable Synchronise symbol tabel in your project and in the 
configuration of the target server.

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: T2 man pages???
Date: Fri, 16 Feb 2001 06:42:58 -0800
From: Joe Durusau 
Organization: Lockheed Martin Corporation
Message-ID: <3A8D3C72.FBF1BECF@lmco.com>
References: <96h9ta$kks$1@overload.lbl.gov>

	True, I've kept the old pages as well, but I assume that there might
be some differences in T2 that are worthwhile.  In that case, I have to 
write something to gen up new man pages or somehow get around the
problem.  
I wondered why so many strange questions about functions were seen in
this ng
until I realized that the man pages had been stupidly deleted.

Speaking only for myself,

Joe Durusau


"Sinn, Larry" wrote:
> 
> I agree!!!
> We upgraded from T1 to T2 and kept the T1 man page directories and created a
> Tornado man command:
>         alias tman "$WIND_BASE_T1/target/man:$WIND_BASE_T1/host/man"
> 
> Larry.
> 
> >       Sounds like WRS can't tell the diff between a
> > feature and a major deficiency.  Menawhile, does anyone have
> > the time to write a script to regen the man pages out of
> > all that masty html stuff???
> >
> > Speaking only for myself,
> >
> > Joe Durusau
> 
> > Bob Irwin wrote:
> > >
> > > Is it just me or are the Unix style man pages missing in T2??
> > >
> > > I got so used to finding libs and functions by using vapropos (man -k).
> > > I find the html pages a pain, they take up so much room on the screen.
> >>
> > > With the good old unix man pages, I could have a couple of man pages,
> > > a couple of includes and a couple of .c files displayed on my monitor
> > > with all visible, no overlap.
> > >
> > > Is there a way to restore the man pages?  Does anyone miss them?
> > >
> > >   tnx.
> > > --
> > > Bob Irwin | bobi@sis-dev.com | SIS Development | 281-484-6661 | Houston
> > >    "Research causes cancer in rats."
> >
> 
> 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

---------------------------

End of New-News digest
**********************


From vxwexplo-errs@csg.lbl.gov  Sun Feb  4 04:03:07 2001
From: Vxworks Exploder 
Date: Mon Feb 19 02:04:23 PST 2001
Subject: comp.os.vxworks newsdigest

Comp.Os.Vxworks Daily Digest    Sun Feb  4 04:03:04 PST 2001

        Subject: MPC823 Rev A --> Rev B problem
        Subject: Simple Tornado WTX programming
        Subject: Re: GDBARM and JEENI embeddedICE, problems, help!
        Subject: Re: target "forwards" all packets sent to it ???
        Subject: Re: pmc750 works?

-------------------------------------------------------

Newsgroups: comp.os.vxworks,comp.sys.powerpc.tech
Subject: MPC823 Rev A --> Rev B problem
Date: Sat, 3 Feb 2001 11:03:14 -0700
From: "Amy Michaelis" 
Organization: Dimensional Communications
Message-ID: 

Help!

Motorola has recently discontinued the Rev A silicon for the 823, and now we
can only get the Rev B silicon.  This should all be seamless of course, but
after some testing with the Rev B chip, we've found a problem.   The Rev B
chip causes random machine checks in our system.

We're using vxWorks as an operating system, and a check with Wind River
didn't turn up anything.  I've been trying to find out what's going on here,
but so far no luck.   We're using quite a bit of CPM, and I tend to believe
that's were the problem lies.  We have the SCC in Ethernet mode, and the SPI
and I2C are also being used.  These two guys have been relocated using the
downloaded microcode patch.  My sense is that since Motorola cleaned up a
lot of the problems in the I2C driver for the Rev B silicon, something has
gone wrong in the downloaded microcode.

The problem could also be the new interrupt for SCC3, but we're not using
the new SCC on the rev B chip, so I don't know how that would be a problem.

Any ideas would be greatly appreciated.

Thanks,

Matt Michaelis
amich@dimensional.com



---------------------------

Newsgroups: comp.os.vxworks
Subject: Simple Tornado WTX programming
Date: Sat, 3 Feb 2001 21:14:52 +0100
From: "Jørgen Jakobsen" 
Organization: Posted Courtesy of Tele Danmark or one of its European Subsidiaries
Message-ID: <95hp08$kl1$1@news.inet.tele.dk>

Dear VxWork'ers

I'm trying to make a simple tool which monitors the value of selected
variables and structures on my target by accessing it through my
targetserver utilising a simple WTX based C program. It is basicly my own
implementation of the watch window in the Tornado debugger ;)

I have written a small piece of code which attaches to the targetserver,
looks up symbols in the symboltable and gets the address.using
wtxSymListGet(), and I can also read memory from the target using
wtxMemRead().

My question is about how I know the type of the variable to probe, the size
of memory to read from target, and how I interpret the memory read ? I guess
I will then have to cast the block of memory read to the actual type to
display the value properly ?

Best regards
Jorgen





---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: GDBARM and JEENI embeddedICE, problems, help!
Date: 4 Feb 2001 00:50:32 GMT
From: Diamand@btinternet.com (Luke Diamand)
Organization: UKLINUX NET
Message-ID: 
References: 

We get pretty good results from canonical gdb from the FSF. Download,
configure and make. This is for arm.

HTH!
Luke Diamand

On Sun, 14 Jan 2001 17:28:46 GMT, george najarian  wrote:
>i am running into the situation where it appears that gdbarm
>is generating a null pointer exception because of input
>received by the jeeni.dll target. Tech support at Embedded Performance
>told me that the dll is generating the correct output, and that
>the trap is being generated in gdbarm.
>
>I am running Win2000.
>
>Does anybody have any suggestions on how to pursue this issue? I
>know that gdbarm is public domain, but doesn't Windriver put in
>custom hooks? Where can I get the source and load it into WinIce to debug?
>Does Windriver support gdb issues?
>
>I am very hesitant to contact Windriver again until i have tried everything
>i can because of the very arrogant and condescending treatment i received the 
>last time i talked to them. Any and all comments and suggestions will be
>gratefully accepted.
>
>George Najarian
>Principal Engineer
>Conet Communications, Inc.


- -- 

---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: target "forwards" all packets sent to it ???
Date: 4 Feb 2001 00:50:33 GMT
From: Diamand@btinternet.com (Luke Diamand)
Organization: UKLINUX NET
Message-ID: 
References: 

If you've written the ethernet driver yourself, have you
checked that the ip portion of the packets (start of
ethernet + 14) is word aligned?

If it isn't, the IP stack will fail to recognise the 
packets, and (IIRC) forward them.

HTH!
Luke Diamand

On Mon, 22 Jan 2001 16:09:19 GMT, Stuart J Adams  wrote:
>We have written a BSP for an SA1110 strongARM board
>and are seeing a weird problem with the ethernet
>and tcp/ip (Tornado 2.0.1 for ARM)
>
>For some reason the tcp/ip stack
>forwards packets that are sent to the target back
>out the same ethernet connection. For example if I try
>to ping the target (say 192.168.1.50), the target
>receives the packet but for some reasons does not
>process it but "forwards" the packet back out the
>ethernet connect. The target does respond to arp
>requests okay. I think this must be some weird 
>configuration problem where the target does not
>realize that it is 192.168.1.50.  
>
>(using etherInputHookAdd we are able to print all
> incoming packets and they look fine. Also, the
> network analyzer shows that packets are transmitted
> fine by the target too)
>
>Any ideas ???
>
>Thanks,
>  Stuart
>


- -- 

---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: pmc750 works?
Date: Sun, 04 Feb 2001 11:25:46 GMT
From: Bill Dennen 
Organization: Motorola Computer Group
Message-ID: <3A7D3826.1A53D65B@home.com>
References: 
Reply-To: wdennen@home.com

Kelly Hornsby wrote:
> 
> Hello all
> 
> I have a PrPMC750 from Motorola that I want to make work with a MVM2603
> system (short term).  Has anyone performed any integration with a setup like
> this?
> 
> Any information, or similiar experience, will be appreciated.

You'll need to obtain the PPMC750 Slave Mode patch for the PrPMC750 BSP
from
your MCG support representative.  This patch only works with the PPMC750
1.2/0
BSP released for Tornado 2.

Regards
- -- 
>@<
Bill Dennen		william.dennen@motorola.com
Motorola Computer Group, TOPAZ

---------------------------

End of New-News digest
**********************
From vxwexplo-errs@csg.lbl.gov  Mon Feb 19 02:04:20 2001
Received: from postal1.lbl.gov (postal1.lbl.gov [128.3.7.82])
	by csg.lbl.gov (8.9.3/8.9.3) with ESMTP id CAA04178
	for ; Mon, 19 Feb 2001 02:04:20 -0800 (PST)
Received: from SpamWall.lbl.gov (localhost [127.0.0.1])
	by postal1.lbl.gov (8.9.3/8.9.3) with ESMTP id CAA26614
	for ; Mon, 19 Feb 2001 02:04:19 -0800 (PST)
Received: from web4005.mail.yahoo.com (web4005.mail.yahoo.com [216.115.104.39])
	by SpamWall.lbl.gov (8.9.3/8.9.3) with SMTP id CAA26611
	for ; Mon, 19 Feb 2001 02:04:19 -0800 (PST)
Message-ID: <20010219100417.22248.qmail@web4005.mail.yahoo.com>
Received: from [203.127.32.40] by web4005.mail.yahoo.com; Mon, 19 Feb 2001 02:04:17 PST
Date: Mon, 19 Feb 2001 02:04:17 -0800 (PST)
From: prabhala rajendra 
Subject: VxMP on CPCI back plane
To: vxwexplo@lbl.gov
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii

vxworks tornado

Hi vxworkers,
I am planing to use VxMP for remote function
invocation from one telcom board to another one over a
cpci backplane using shared message queues.

I have never used VxMP before. I have checked the
manual for some details about this package
but the following things are not clear to me. 

Could somebody please clarify the following regarding
VxMP ? 

1.Do we have to allocate and reserve this
shared-memory partition (dedicated to shared memory
system pool) during the init phase by calling the
routine smMemAddToPool()? 
	
2.Is there any restriction on the type of back plane
(VME / PCI/CPCI) used when we use VxMP ? For non-VME
backplanes (eg.CPCI) are there any extra
settings/issues to be considered ? Manual seems to be
very specific about VME.

3.Has anybody tried using VxMP in typical call
processing applications ? How is the performance ?

Thanks and best regards,
rajendra.


__________________________________________________
Do You Yahoo!?
Get personalized email addresses from Yahoo! Mail - only $35 
a year!  http://personal.mail.yahoo.com/


From vxwexplo-errs@csg.lbl.gov  Mon Feb 19 10:13:31 2001
From: Atkinson Andrew 
Date: Mon Feb 19 10:13:34 PST 2001
Subject: Handling Interrupt - 

Hi All

VxWorks 5.3.1 / T2.

	We are currently writing a device driver for a VMEBus device.
This device is capable of generating interrupts to report status
information held in registers on-board the device. When an interrupt
occurs we want to read these registers to ascertain the status of the
device. The problem is that the status registers must be read some time
after the interrupt has occured. Now as a lengthy ISR is undesireable,
what is the best way to handle the interrupt and get the status
information. The most straightforward approach appears to be to spawn a
task to wait on a semaphore, then give the semaphore from the ISR. The
task can then implement the necessary delays and report the status
information.

	Does anyone else have any ideas on how to acheive this, or is
the above a good idea?

Any guidance appreciated.

Regards

Andy




-- 
The Information contained in this E-Mail and any subsequent correspondence
is private and is intended solely for the intended recipient(s).
For those other than the recipient any disclosure, copying, distribution, 
or any action taken or omitted to be taken in reliance on such information is
prohibited and may be unlawful.


From vxwexplo-errs@csg.lbl.gov  Mon Feb 19 10:52:28 2001
From: Andreas Wolf 
Date: Mon Feb 19 10:52:31 PST 2001
Subject: VxSim on Solaris5.8

Hi,
    we are having trouble starting vxWorks as VxSIM on
    Solaris 5.8 (64bit).

    We use PPP as the connection to the host network. The kernel
    can not setup any good link to the host, and
    it seems that the pipe "/tmp/.asppp.fifo" is missing.
    All our 5.7 machines have that file, but none of our 5.8's
    although PPP was setup the same way. Does any VxSIM/solaris
    user have any insight?

    Thanks a bunch!!

        Andreas =:-)





From vxwexplo-errs@csg.lbl.gov  Mon Feb 19 12:04:05 2001
From: John_W_Cosgrove@raytheon.com
Date: Mon Feb 19 12:04:07 PST 2001
Subject: Re: VxSim on Solaris5.8

Andreas,

The 5.3.1 we are using does not support 64 bit PPP.  We found we had to use the
32 bit implementation to get vxSim/PPP to work together.

John Cosgrove




To:   vxworks_users@csg.lbl.gov
cc:    (bcc: John W Cosgrove/RES/Raytheon/US)
Subject:  VxSim on Solaris5.8

Submitted-by vxwexplo-errs@csg.lbl.gov  Mon Feb 19 10:52:28 2001
Submitted-by: Andreas Wolf 

Hi,
    we are having trouble starting vxWorks as VxSIM on
    Solaris 5.8 (64bit).

    We use PPP as the connection to the host network. The kernel
    can not setup any good link to the host, and
    it seems that the pipe "/tmp/.asppp.fifo" is missing.
    All our 5.7 machines have that file, but none of our 5.8's
    although PPP was setup the same way. Does any VxSIM/solaris
    user have any insight?

    Thanks a bunch!!

        Andreas =:-)






From vxwexplo-errs@csg.lbl.gov  Mon Feb 19 23:22:14 2001
From: "R R, Rajendra Prabhala (Rajendra)** CTR **" 
Date: Mon Feb 19 23:22:18 PST 2001
Subject: Back plane communication and VxMP

vxworks

Hi vxworkers,

Could somebody please clarify the following regarding the back plane
communication driver supported by vxworks and this optional VxMP component ?

	1.	I have gone through the manual and I understand that the
back plane communication driver comes free 
		with the vxworks image and there is no need to have the
optional package VxMP running to enable this 	back plane communication
along multiple processors.  I suppose that this VxMP is required only to
enable the usage of shared memory objects. Am I correct ?

	2.	Are there any back plane type restrictions (VME /  PCI/CPCI
) for this optional package VxMP ?   	Programmers manual is very specific
about VME.

	Thanks and best regards,
	rajendra.




From vxwexplo-errs@csg.lbl.gov  Tue Feb 20 01:49:50 2001
From: Nagasesha Reddy 
Date: Tue Feb 20 01:49:52 PST 2001
Subject: memDrv and memDevCreate in VxWorks

Hi VxWork'ers,
     I have the following problem. When I try to use
"memDrv" or "memDevCreate" routine at the shell
prompt,
I get the error "undefined symbol:"
      Could some body please tell me, what should be
defined in configAll.h/config.h/configdb.h, so that
these routines are available in the vxworks kernel?

thanks in advance,
seshu

__________________________________________________
Do You Yahoo!?
Get personalized email addresses from Yahoo! Mail - only $35 
a year!  http://personal.mail.yahoo.com/


From vxwexplo-errs@csg.lbl.gov  Tue Feb 20 07:10:51 2001
From: "bao" 
Date: Tue Feb 20 07:10:53 PST 2001
Subject: VCPU/VxSim debugging problem


Hi All,

I am currently working on Vxworks device driver using VCPU/VxSim for
simulation.
I have a problem during debugging my project by using CrossWind. I have 2
projects built under 1 workspace.
The first one is device driver plus API code like test function called by
users, the second project is my software functional model written in C as
well since I have no way to import my RTL into the test environment. This
project 2 is targeted to a .out file that will be used by VCPU agent in a
specific configuration.
When I try to step into any C code in project 2 from a function in project
1, it take in assembly code then stop debugging process.

Any one has a idea how I debug this situation ?

Bao Pham
MOTOROLA -Global Software Group
Austin,TX
Tel:(512) 996 7770
Fax: (512) 996 7320



From vxwexplo-errs@csg.lbl.gov  Tue Feb 20 08:52:04 2001
From: "Baron, Ken" 
Date: Tue Feb 20 08:52:06 PST 2001
Subject: Re: STDIN problem on target


	[Baron, Ken]   I ran it just like you did. I used the simulator and
built your code as a downloadable project for SIMNTgnu.
	Note that because the ttyDrv is set for terminal mode, fgets will
not return until a  is typed in the windsh for case 1 and in the VxWorks
Simulator window for case 2. The ttyDrv can be set for different modes. See
the documentation.

> Ken:
> 
> Thanks for your response.
> How do you run it, exactly?
> I tried the following two ways:
> 1) In winsh windows, type testStdIO:
> I could not get character echo back when I am typing.
> 2) In winsh, run "sp(testStdIO)"
> nothing happens.
> 
> What's wrong with my setting?
> 
> Thanks in advance.
> 
> Baron, Ken wrote in message <96grrb$e8t$1@overload.lbl.gov>...
> >
> > Your code below works fine for me. If you run testStdIO from the
> >windsh the function receives input from the shell window.
> > If you spawn testStdIO as a new task, then the task receives input
> >from the console. (vxworks sim window or your primary serial port on your
> >hardware.
> >
> >> Hi,
> >> I am new to vxWorks and WinSh. I am using Tornado II. I am using the
> WinSh
> >> to exec the following code. ( i.e. in windsh , after load in the .o
> file,
> >> type >testStdIO). I wish I could see the characters I typed in there
> >> reflected in the window, just as xterm. However, it seems that the data
> is
> >> swallowed by the application.
> >>
> >> How can I do this.
> >>
> >> Gang Wang
> >>
> >> #include 
> >>
> >> int testStdIO()
> >> {
> >>    FILE *fp;
> >>    char  data[256];
> >>
> >>    /*if ( ( fp = fopen(stdin, "r")) == NULL )
> >>      {
> >>        printf("ERROR in open stdin");
> >>        return 0;
> >>       }*/
> >>    while (1)
> >>    {
> >>      fgets(data, 256, stdin);
> >>      if ( data[0] == 'q' )
> >>         return 1;
> >>      else
> >>        printf("I get something\n");
> >>    }
> >>    return 0 ;
> >> }
> >>
> 


From vxwexplo-errs@csg.lbl.gov  Wed Feb 21 08:08:34 2001
From: "T. Hariharan" 
Date: Wed Feb 21 08:08:38 PST 2001
Subject: Flash memory questions

Hello,

A few questions on embedded and removable flash memory.
The box is x86 based.

Embedded:
o Looks like TrueFFS is being widely used out there. But a simpler
solution would be to use an embedded chip/card that acts as an IDE.
(Silicon Storage Tech's ATA-Disk is a chip product. Sandisk makes
FlashDrive card which you can plug like an IDE hard drive).
Then you don't need TrueFFS at all. So the question is, what are
the advantages of TrueFFS? There must be something I'm missing.

The VxWorks TrueFFS doc says that if you format a flash under
TrueFFS, you need to leave some "fallow" area where you copy the
boot image. I guess you'll copy just the boot-rom (which has the
TrueFFS/dosFs code) to this fallow area and keep the VxWorks image
on the TrueFFS formatted filesystem. Is this correct?

Since BIOS can boot from IDE drives, it seems simpler to
not use TrueFFS.

I don't have anything against TrueFFS - just want to know
its advantages.

Removable:
o The choice here appears to be straight-forward. Use a PC Card ATA.
The vxworks PCMCIA doc says that socket and chip services are not
supported. Does this mean that vxworks does not support hot-swapping
the removable flash card?

Does anyone have good/bad experiences with specific card readers?

Thanks,
Hari



From vxwexplo-errs@csg.lbl.gov  Wed Feb 21 11:12:49 2001
From: Charles Johnson 
Date: Wed Feb 21 11:12:51 PST 2001
Subject: RE: Flash memory questions

Our experience with TFFS vs ATA flash is that ATA flash has some distinct
disadvantages:

1.	No wear leveling.
2.	Unable to survive powerfail.  Often we end up with a read error on a
sector which can only be fixed if you re-write the sector.

We're using the Silicon Tech ATA drives and it is very easy to damage the
drive.

--Charles Johnson
Oresis Communications, Inc.


		-----Original Message-----
		From:	vxwexplo@lbl.gov [mailto:vxwexplo@lbl.gov]
		Sent:	Wednesday, February 21, 2001 8:12 AM
		To:	vxworks_users@csg.lbl.gov
		Subject:	Flash memory questions

		Submitted-by vxwexplo-errs@csg.lbl.gov  Wed Feb 21 08:08:34
2001
		Submitted-by: "T. Hariharan" 

		Hello,

		A few questions on embedded and removable flash memory.
		The box is x86 based.

		Embedded:
		o Looks like TrueFFS is being widely used out there. But a
simpler
		solution would be to use an embedded chip/card that acts as
an IDE.
		(Silicon Storage Tech's ATA-Disk is a chip product. Sandisk
makes
		FlashDrive card which you can plug like an IDE hard drive).
		Then you don't need TrueFFS at all. So the question is, what
are
		the advantages of TrueFFS? There must be something I'm
missing.

		The VxWorks TrueFFS doc says that if you format a flash
under
		TrueFFS, you need to leave some "fallow" area where you copy
the
		boot image. I guess you'll copy just the boot-rom (which has
the
		TrueFFS/dosFs code) to this fallow area and keep the VxWorks
image
		on the TrueFFS formatted filesystem. Is this correct?

		Since BIOS can boot from IDE drives, it seems simpler to
		not use TrueFFS.

		I don't have anything against TrueFFS - just want to know
		its advantages.

		Removable:
		o The choice here appears to be straight-forward. Use a PC
Card ATA.
		The vxworks PCMCIA doc says that socket and chip services
are not
		supported. Does this mean that vxworks does not support
hot-swapping
		the removable flash card?

		Does anyone have good/bad experiences with specific card
readers?

		Thanks,
		Hari


		**********

		    This is a user group mailing list for vxWorks related
topics
		    see http://csg.lbl.gov/vxworks for posting and other
info


From vxwexplo-errs@csg.lbl.gov  Wed Feb 21 11:38:02 2001
From: Diego Warszawski 
Date: Wed Feb 21 11:38:04 PST 2001
Subject: taskDelay problem

Hi,	
	I am using Tornado 2.0 - VxWorks 5.4
	I am trying to debug a problem we have in our SW. 
	We are running several tasks. We have interrupts, tasks changing
their priorities, and timers. One of the tasks is very simple:

void MyTask(int,char**)
{
   while(1)
   {
     DoMyStaff();
     taskDelay( DELAY_1_SEC); // 100 ticks
   }
}

     This stack is getting stuck in the taskDelay call. The target shell
goes on running. The task trace shows:

485230  _taskDelay   +122: idleLoop([b860a8,b86270,b84bf0,468cf4,64,b860a8])

	The task state is READY all the time instead of changing to DELAY.
This blocks all the tasks with lower priority. I checked the assembly code
and at taskDelay+122 there is a call to windExit().
	What could have happened to cause this problem? Any pointers?
	Thanks for your help!

		Diego W.
	


From vxwexplo-errs@csg.lbl.gov  Thu Feb 22 10:30:50 2001
From: "Tan, Tat Kin" 
Date: Thu Feb 22 10:30:52 PST 2001
Subject: Unix to Win platform

Greetings all,

Previously we developed some Apps and compiled the image under a project,
say Proj1. It's all done in Unix environment (makefile). Now what i want to
do is, to use T2 Win base instead of Unix. Basically to port the entire
collections of source files into T2 Win, maintaining the original directory
structure. I understand that in Windows based Tornado2.0, there is a
facility that allow you to add projects to workspace (under File) but that
will require ".pwj" file to describe the structure. In Unix, there is no
such thing.

My problem is, instead of rebuilding the entire project, how can i port
Proj1 to Win T2? And so that when building the image, it works exactly the
same as Unix make. Any taker? Million thanks in advance.

Rgds,
tatkin



From vxwexplo-errs@csg.lbl.gov  Thu Feb 22 20:38:55 2001
From: VxWorks User Group 
Date: Thu Feb 22 20:38:57 PST 2001
Subject: VxWorks: Servo Control System Job Opportunity


Triad Project Management Services supports major engineering projects for 
the government and commercial accounts.  Our clients seek creative 
solutions to unique engineering problems involving astronomy, physics and 
sophisticated hi-tech applications.  

If this kind of work interests you, Triad Project Management is currently 
recruiting Sr. Electrical Engineers with real-time servo control  
experience using VxWorks. 

As an engineer on this project you will design the control system to 
operate a six-axis optical positioning system involving 20+ interrelated 
feedback loops. 

Candidates must have an understanding of the interface between electrical 
hardware and the programming of the real time control software written in 
C in a VxWorks environment. 

Candidates need to be able to design, develop, implement and test low 
noise/high dynamic range feedback control electronics, including design, 
implement and troubleshooting of real-time control systems, especially 
those with VME based microprocessors, embedded DSP technology and VxWorks 
operating systems. 

Experience with C or C++ programming is desired. Experience in one or more 
of the following is desirable: design of front-end interface electronics 
including low noise analog electronic circuits, VME based systems, UNIX, 
embedded microprocessor hardware and software systems, remote control and 
monitoring systems. Some travel of about 20% will be needed. 

If you would like to know more about this job opportunity call our offices 
or contact us via email immediately.   Of course, if you would prefer not 
to receive email regarding this or other VxWorks job opportunities, we do 
understand.  Just send me a note containing   "remove"  in the subject and  
"VxWorks Jobs"  in the body and you won't be contacted in the future when 
other VxWorks opportunities arise.  Thanks.


Michael Green  
mgreen@triadpm.com

Triad Project Management Services, Inc.
3452 E Foothill Blvd  Suite 530
Pasadena, CA 91107

Office: 626-793-7314  
Fax: 626-793-7432 




From vxwexplo-errs@csg.lbl.gov  Thu Feb 22 21:43:53 2001
From: "Krishna Kishore Reddy Sureddy" 
Date: Thu Feb 22 21:43:55 PST 2001
Subject: VxWorks(64 Bit Addition).

Hi VxWorks Gurus,
How to perform addition of 64 bit integers on a 32 bit processor.(MIPS,
R4000).
Can someone pass me code in "C"....
Thanks and Regards,
S.K.K.Reddy.




From vxwexplo-errs@csg.lbl.gov  Fri Feb 23 13:55:31 2001
From: Alan Biocca 
Date: Fri Feb 23 13:55:33 PST 2001
Subject: Bad Email Address Purging

VxWorks Users Group Exploder Bad Address Purging

Periodically we purge bouncing addresses from the Exploder list. We had not 
done so for a year or so, and the number of bad addresses reached more than 
200.

Our current policy on the VxWorks list is to periodically purge addresses 
that we find in the bounce files, after they have earned at least 15 
(mostly consecutive) days of bounces. Since we do this once a week or so, 
it often means they have earned more than 20 days of bounce points before 
being removed. Each day they bounce, they gain one point (regardless of how 
many bounces that day). Each day they don't, they lose one point. When the 
score exceeds 15 they are filtered to a list of bad addresses. 
Approximately once a week the current daily bad address list is used to 
reduce the list (if they are still above the 15 point score).

There has apparently recently been an issue of an address that we removed 
from the list that was working, but was sending something back to our 
server that was scored as a bounce. This is the first time I know of that 
this has been an issue. I copied part of the users message below, with 
identity removed, for reference.

Auto-Reply Engines

Some folks have automatic reply programs set up for vacations, etc. The 
usual policy for reply programs is ONLY to reply if your own actual address 
is in the To: field. This avoids replies to lists. Many lists will remove 
recipients that bounce even once.

In addition, auto-replies are normally sent back only once to each distinct 
address - the address is kept in a list and not replied to again. The list 
is cleared when the vacation feature is re-enabled for the next trip, when 
presumably the message has been changed.

There are many programs that do auto-reply, and they do not all work the 
same. Many are configurable (and can be misconfigured), and many are 
written by folks who don't understand all the issues of mailing lists, etc.

I hope that members of the list feel that our policy is reasonable. We get 
megabytes of bounces, so we cannot look at them, and this simple program is 
our attempt to keep it under control without expending much effort on the 
task. If a machine sends us back email 15 days running, there is something 
not right. We can easily adjust the threshold to a different point value if 
necessary.

Feel free to send comments directly to me (AKBiocca@lbl.gov), or to the 
list if you wish an open discussion.

-- Alan Biocca
VxWorks User Group Mailing List Management

> >Thanks for your help. I have a question for you. When I go out of office
> >(infrequently but it does happen) I set my [Microsoft Outlook] Out-Of-Office
> >Assistant to send back a message to whomever emailed me. It sounds like
> >there's a conflict between my need to notify business associates and the
> >management of the exploder.
> >
> >I'm thinking this might be a problem that many people share. Perhaps this is
> >an opportunity for the list managers to come up with a policy that meets
> >everyone's needs, including your need to minimize list management tasks.
> >
> >Again, thanks for your help. This list has saved my hours/days of
> >frustration!
> >...



From vxwexplo-errs@csg.lbl.gov  Sat Feb 24 10:31:08 2001
From: "Deja User" 
Date: Sat Feb 24 10:31:10 PST 2001
Subject: VxWorks hang up in mcp750

Tornado vxworks
Hi guys / gals,

Host:WinNT
Target:MCP750, BSP V1.2
Tornado - II

We have inserted Lucent access board on Compact PCI chassis, where MCP750 board is used as CPU.

When we execute lucent driver commands from target shell prompt, the system getting hang up.

This problem is occuring randomly, not specifically during any single driver command. we have tried to increase target shell stack space from default 20K to 50K, defined all board register variables as volatile, given few microsecond delay before accessing each register variable, etc..

Still the problem persists.

Can any one faced up with similar problem and can suggest the solution found..

Thanks in advance.

Bye...

EmbLov




------------------------------------------------------------
--== Sent via Deja.com ==--
http://www.deja.com/




From vxwexplo-errs@csg.lbl.gov  Sun Feb 25 19:53:37 2001
From: "LakshmiNarayan" 
Date: Sun Feb 25 19:53:39 PST 2001
Subject: RE: comp.os.vxworks newsdigest

This is a multi-part message in MIME format.

------=_NextPart_000_016B_01C09D7A.2885C440
Content-Type: text/plain;
	charset="iso-8859-1"
Content-Transfer-Encoding: 7bit

Hi 
I am on leave , I will be back in office on 26th Feb .

For NIT192 project details contact shashank.kulkarni@wipro.com

rgds
Lakshminarayan S
Wipro Technologies ..
#	30 ,I Mn Road ,S.R Nagar ,Mission Road
	Bangalore,India.
@: 91-80-2275590/2241730 Extn 1131



------=_NextPart_000_016B_01C09D7A.2885C440
Content-Type: application/ms-tnef;
	name="winmail.dat"
Content-Transfer-Encoding: base64
Content-Disposition: attachment;
	filename="winmail.dat"

eJ8+IiMDAQaQCAAEAAAAAAABAAEAAQeQBgAIAAAA5AQAAAAAAADoAAEIgAcAGAAAAElQTS5NaWNy
b3NvZnQgTWFpbC5Ob3RlADEIAQ2ABAACAAAAAgACAAEGgAMADgAAANEHAgAXAAkAEgAAAAUAEQEB
A5AGADAGAAAoAAAACwACAAEAAAALACMAAAAAAAMAJgAAAAAACwApAAAAAAADADYAAAAAAB4AcAAB
AAAAHwAAAFJFOiBjb21wLm9zLnZ4d29ya3MgbmV3c2RpZ2VzdAAAAgFxAAEAAAAWAAAAAcCdS3O3
t/yXwQlsEdW1hwCASNhh/AAAAgEdDAEAAAAiAAAAU01UUDpMQUtTSE1JTkFSQVlBTi5TSElAV0lQ
Uk8uQ09NAAAACwABDgEAAABAAAYOANhsakudwAECAQoOAQAAABgAAAAAAAAACqAlemnB1BG1hQBg
ZwIeTMKAAAALAB8OAQAAAAMABhCGn1aNAwAHENEAAAAeAAgQAQAAAGUAAABISUlBTU9OTEVBVkUs
SVdJTExCRUJBQ0tJTk9GRklDRU9OMjZUSEZFQkZPUk5JVDE5MlBST0pFQ1RERVRBSUxTQ09OVEFD
VFNIQVNIQU5LS1VMS0FSTklAV0lQUk9DT01SR0RTAAAAAAIBCRABAAAAdwEAAHMBAADBAQAATFpG
dVuFGFEDAAoAcmNwZzEyNRYyAPgLYG4OEDAzM50B9yACpAPjAgBjaArAYHNldDAgBxMCgH0ZCoF1
YwBQCwN1bG7FAiBlC6YgSGkK4wqAIEkgYW0gAiAgbIBlYXZlICwgFEBVA/BsAyBiFQBiANBrfiAL
gBSAASAN4BUAFJEyADZ0aCBGZWIgFi4T5BPkRgWxTklUoDE5MiBwA2BqBZD/BUABAAGQAxAEIAWg
AjAA0A0FQHMRABqRbmsua20SwGsKwAMAQAPwGREuxQWgbReqcmdkEMAT82hMYWsakG0LgArAYSp5
A5FTE+RXG8IgVG0FkGgS4AkAZwiQBCAupReVIwyCIDMRYCwUQCZNA6AIAGFkFRBTLpJSB7BhZwrB
LE0EAd5pFJEhohelINJCDxEHQGUFsGUhQG5kBzAXlUAAOiA5MS04MC0AMjI3NTU5MC8RJhA0MTch
EUV4dPEDoDExMxJTF+YT5BHxAgApAAADABAQAAAAAAMAERAAAAAACwABgAggBgAAAAAAwAAAAAAA
AEYAAAAAA4UAAAAAAAADABCACCAGAAAAAADAAAAAAAAARgAAAABShQAA8BMAAB4AEYAIIAYAAAAA
AMAAAAAAAABGAAAAAFSFAAABAAAABAAAADguNQALABWACCAGAAAAAADAAAAAAAAARgAAAAAGhQAA
AAAAAAMAFoAIIAYAAAAAAMAAAAAAAABGAAAAAAGFAAAAAAAAQAAYgAggBgAAAAAAwAAAAAAAAEYA
AAAAYIUAAADQduXR//8fCwAfgAggBgAAAAAAwAAAAAAAAEYAAAAADoUAAAAAAAADACCACCAGAAAA
AADAAAAAAAAARgAAAAAQhQAAAAAAAAMAIYAIIAYAAAAAAMAAAAAAAABGAAAAABGFAAAAAAAAAwAj
gAggBgAAAAAAwAAAAAAAAEYAAAAAGIUAAAAAAAAeADOACCAGAAAAAADAAAAAAAAARgAAAAA2hQAA
AQAAAAEAAAAAAAAAHgA0gAggBgAAAAAAwAAAAAAAAEYAAAAAN4UAAAEAAAABAAAAAAAAAB4ANYAI
IAYAAAAAAMAAAAAAAABGAAAAADiFAAABAAAAAQAAAAAAAAALAD2ACCAGAAAAAADAAAAAAAAARgAA
AACChQAAAQAAAAsAP4ALIAYAAAAAAMAAAAAAAABGAAAAAACIAAAAAAAACwBBgAsgBgAAAAAAwAAA
AAAAAEYAAAAABYgAAAAAAAACAfgPAQAAABAAAAAKoCV6acHUEbWFAGBnAh5MAgH6DwEAAAAQAAAA
CqAlemnB1BG1hQBgZwIeTAIB+w8BAAAAcwAAAAAAAAA4obsQBeUQGqG7CAArKlbCAABQU1RQUlgu
RExMAAAAAAAAAABOSVRB+b+4AQCqADfZbgAAAEM6XFdJTkRPV1NcQXBwbGljYXRpb24gRGF0YVxN
aWNyb3NvZnRcT3V0bG9va1xvdXRsb29rLnBzdAAAAwD+DwUAAAADAA00/TcAAAIBfwABAAAAMQAA
ADAwMDAwMDAwMEFBMDI1N0E2OUMxRDQxMUI1ODUwMDYwNjcwMjFFNENBNEFDMjEwMAAAAACIKA==

------=_NextPart_000_016B_01C09D7A.2885C440--



From vxwexplo-errs@csg.lbl.gov  Mon Feb 26 08:26:07 2001
From: "Anil Dadlani" 
Date: Mon Feb 26 08:26:09 PST 2001
Subject: How do I get the MAC address using vxWorks API?

> Hi,
>
> I need to read the MAC address of a Ethernet chip.  How do I get this
using
> vxWorks APIs?
> We do have a lower level BSP API which stores this 6 byte address into the
> driver control structure (etherEndDevice)
> I found:
> ifunit ("ln0") function which returns the ifnet structure.  Which field
has
> the MAC address?
>
>
> Thanks
>
> Anil
>
>



From vxwexplo-errs@csg.lbl.gov  Mon Feb 26 11:06:15 2001
From: Jim Way 
Date: Mon Feb 26 11:06:17 PST 2001
Subject: RE: How do I get the MAC address using vxWorks API?

Here's how I do it:

	UINT8 macBuffer[8];
	sysCpmEnetAddrGet(0, macBuffer);
	MacAddr = nlprintf("0x%02X%02X%02X%02X%02X%02X", 
						macBuffer[0],
						macBuffer[1],
						macBuffer[2],
						macBuffer[3],
						macBuffer[4],
						macBuffer[5]);

Jim
-----------------------------------------
Jim Way, Software Engineer
Datum Austin (Austron Inc.)
voice: 512.721.4170
fax  : 512.990.9712
email: jwayATdatumDOTcom (no spam please)
-----------------------------------------
vxWorks Tornado II


From vxwexplo-errs@csg.lbl.gov  Tue Feb 27 04:11:28 2001
From: "Gokul Gopalasamy" 
Date: Tue Feb 27 04:11:31 PST 2001
Subject: VxWorks-zbuf

Hi all,

I'm a beginner in VxWorks. Anyone pls. suggest some documents/links or
send the struct zbuf contents? Thanks in advance.

Gokul




From vxwexplo-errs@csg.lbl.gov  Tue Feb 27 19:56:51 2001
From: "Ajay Agrawal" 
Date: Tue Feb 27 19:56:53 PST 2001
Subject: The routing and interface table information

Hi,

I am looking at developing a routing protocol module on Vxworks. The Vxworks
stack is used with BSD complaint (non-MUX) network interfaces.
The routing module would require all the existing routing and interface
table information during it's initialization. Also any dynamic updates on
the routing and interface table information should be made available to this
routing module.

What kind of interfaces/hooks does Vxworks provide to acheive this?

Thanks and regards
ajay




From vxwexplo-errs@csg.lbl.gov  Tue Feb 27 20:47:26 2001
From: Rick Davis 
Date: Tue Feb 27 20:47:28 PST 2001
Subject: Re: VxWorks-zbuf




> Submitted-by: "Gokul Gopalasamy" 
> 
> Hi all,
> 
> I'm a beginner in VxWorks. Anyone pls. suggest some documents/links or
> send the struct zbuf contents? Thanks in advance.

The struct zbuf is not really documented; it's private to the zbuf
API.  You use the zbuf ID ( which is a pointer to the zbuf structure )
as an argument to the zbufXXX routines, and they manipulate it as
required. 

Zbufs are basically collections of 'segments', each of which contains
some data.  You create them with zbufCreate,  add to them with 
zbufInsert, and have a couple of options for getting the data out
of them, using zbufExtractCopy to get a copy of the data and
zbufGetData ( ? ) to get a pointer to the data contained in the
zbuf.  The zbuf section of the programmer's guide goes into all 
this in enough detail to get started. 

The advantages to using zbufs are that they allow you to do
zero-copy buffer passing if used judiciously.  The disadvantages
are that you incur some insert/extract processing and storage
overhead.  It's a big win for large buffers, not so great for
small ones.  

The other issue with zbufs is that they use a scarce system resource, 
the mblock.  If you tie up too many of these, zbufCreate will fail
until you free up some.  

The number of mblocks is configurable, but you need to consider 
your worst-case load before using zbufs.

Regards;

-- Rick Davis


From vxwexplo-errs@csg.lbl.gov  Wed Feb 28 00:24:21 2001
From: "Gokul Gopalasamy" 
Date: Wed Feb 28 00:24:23 PST 2001
Subject: Re: VxWorks-zbuf

Hi Davis

Thanks a lot. Can i contact you in case i need some more details?

Warm Regards
Gokul



From vxwexplo-errs@csg.lbl.gov  Wed Feb 28 07:14:02 2001
From: Roger Cruz 
Date: Wed Feb 28 07:14:04 PST 2001
Subject: RE: Vx Exploder Digest


Do you have the source code?  Can you use Tornado to debug the problem?  Or
is the hang so bad that you can't even ping the board?  Your description
does not give enough information for me to be able to give you any valuable
advice.   The one thing I can say, having written drivers for the MCP750, is
that if the driver developer did not take into account that instructions on
the 750 can execute out of order, then "weird" problems will occur.  EIEIO
instructions need to be place at the right places.  Also, I/O space should
be marked as "guarded".  We saw a problem where some bytes were missing from
an FTP transfer.  We started by suspecting the HD driver, then Enet driver
and finally after we had looked and confirmed both drivers were OK, we found
out it was a simple GUARDED mask missing in the BSP.

Roger

**********

From vxwexplo-errs@csg.lbl.gov  Sat Feb 24 10:31:08 2001
From: "Deja User" 
Date: Sat Feb 24 10:31:10 PST 2001
Subject: VxWorks hang up in mcp750

Tornado vxworks
Hi guys / gals,

Host:WinNT
Target:MCP750, BSP V1.2
Tornado - II

We have inserted Lucent access board on Compact PCI chassis, where MCP750
board is used as CPU.

When we execute lucent driver commands from target shell prompt, the system
getting hang up.

This problem is occuring randomly, not specifically during any single driver
command. we have tried to increase target shell stack space from default 20K
to 50K, defined all board register variables as volatile, given few
microsecond delay before accessing each register variable, etc..

Still the problem persists.

Can any one faced up with similar problem and can suggest the solution
found..

Thanks in advance.

Bye...

EmbLov


From vxwexplo-errs@csg.lbl.gov  Wed Feb 28 07:47:19 2001
From: "Brown, Matthew" 
Date: Wed Feb 28 07:47:22 PST 2001
Subject: RE: How do I get the MAC address using vxWorks API?

VxWorks
I couldn't find any kind of direct call in the vxWorks API to get it, so I
used the arpcom struct.  Use a memcpy and copy the MAC from the ac_enaddr
into a 6 byte buffer.

Something like this :

char macAddr [6];
pIf = ifunit ("ln0");
struct arpcom *pArpcom;
pArpcom = (struct arpcom*) pIf;
memcpy (macaddr, (pArpcom->ac_ecaddr),6);


==========================
Matt Brown
Lockheed Martin - Software
Baltimore, MD

> ----------
> From: 	vxwexplo@lbl.gov[SMTP:vxwexplo@lbl.gov]
> Sent: 	Monday, February 26, 2001 11:32 AM
> To: 	vxworks_users@csg.lbl.gov
> Subject: 	How do I get the MAC address using vxWorks API?
> 
> Submitted-by vxwexplo-errs@csg.lbl.gov  Mon Feb 26 08:26:07 2001
> Submitted-by: "Anil Dadlani" 
> 
> > Hi,
> >
> > I need to read the MAC address of a Ethernet chip.  How do I get this
> using
> > vxWorks APIs?
> > We do have a lower level BSP API which stores this 6 byte address into
> the
> > driver control structure (etherEndDevice)
> > I found:
> > ifunit ("ln0") function which returns the ifnet structure.  Which field
> has
> > the MAC address?
> >
> >
> > Thanks
> >
> > Anil
> >
> >
> 
> 
> **********
> 
> 


From vxwexplo-errs@csg.lbl.gov  Wed Feb 28 08:05:29 2001
From: gtunali@hns.com
Date: Wed Feb 28 08:05:31 PST 2001
Subject: multicast



Hi

I am working with Udp sockets in vxworks and I simply want to add a socket to a
multicast group.. I included multicast in VxWorks image and I see MULTICAST in
flags section

After using "socket" and "bind" ,  I call:
----------------------
ipMreq.imr_multiaddr.s_addr = inet_addr("224.10.10.15");      //multicast
address that I want to add myself
ipMreq.imr_interface.s_addr = inet_addr("139.85.xx.xxx");       //my machine's
IP address

::setsockopt(myFileDescriptor, IPPROTO_IP, IP_ADD_MEMBERSHIP,
(char*)&ipMreq,sizeof(ipMreq))
----------------
It gives error 0x31( ADDRNOAVAIL??) , just at setsockopt.

What should I do before I use IP_ADD_MEMBERSHIP, is there some other steps?
Should I do something to create 224.10.10.15 multicast address interface before
I join it?

Regards

Gamze Tunali




From vxwexplo-errs@csg.lbl.gov  Wed Feb 28 08:21:31 2001
From: "zz Fa.NewlinkSystems01" 
Date: Wed Feb 28 08:21:33 PST 2001
Subject: PMC 663J Ethernet Card

Hi,



The company that I am working for is looking to use the PMC 663J Ethernet
Card with VxWorks, I have searched on WindSurf and found no mention of this
card and have an enquiry already ongoing with WRS, but just wondered :

1) whether there are any drivers available for VxWorks??
2) Anyone had any experience with it??



Thanks for any help you can provide,



David Yates.


-----------------------------------------------------------
This Mail has been checked for Viruses
Attention: Encrypted Mails can NOT be checked !

***

Diese Mail wurde auf Viren ueberprueft
Hinweis: Verschluesselte Mails koennen NICHT geprueft werden!
------------------------------------------------------------


From vxwexplo-errs@csg.lbl.gov  Wed Feb 28 08:53:49 2001
From: "Crozier, Bruce" 
Date: Wed Feb 28 08:53:51 PST 2001
Subject: RE: PMC 663J Ethernet Card


Hello David,

We have the Ramix PMC661 and PMC663 cards working on PMC sites on 
several of our CPCI and VME boards.  vxWorks drivers are available 
directly from Ramix.  (http://www.ramix.com)

Regards,
Bruce Crozier
Force Computers
http://www.forcecomputers.com

-----Original Message-----
Sent: Wednesday, February 28, 2001 10:32 AM
Subject: PMC 663J Ethernet Card

Submitted-by: "zz Fa.NewlinkSystems01" 

Hi,

The company that I am working for is looking to use the PMC 663J Ethernet
Card with VxWorks, I have searched on WindSurf and found no mention of this
card and have an enquiry already ongoing with WRS, but just wondered :

1) whether there are any drivers available for VxWorks??
2) Anyone had any experience with it??


From vxwexplo-errs@csg.lbl.gov  Wed Feb 28 09:43:38 2001
From: "Microsoft Exchange Connector for Lotus cc:Mail" 
Date: Wed Feb 28 09:43:40 PST 2001
Subject: Undeliverable: RE: How do I get the MAC address using vxWorks API

Your message was not delivered to all recipients.

     Subject:  RE: How do I get the MAC address using vxWorks API?
     Sent:     2/28/01 10:52:00 AM

The cause of the failure(s) is listed below:

     ***  Cannot store the message  ***


The message was not delivered to the following receipient(s):

     Rajagopalan, Chandrasekar at GBP-HCL-INDIA01


From vxwexplo-errs@csg.lbl.gov  Wed Feb 28 11:07:59 2001
From: "Guna Venkat" 
Date: Wed Feb 28 11:08:02 PST 2001
Subject: DHCP Client and Symbol table

This is a multi-part message in MIME format.

------=_NextPart_000_006D_01C0A176.34BDA2F0
Content-Type: text/plain;
	charset="iso-8859-1"
Content-Transfer-Encoding: quoted-printable

Hello All,
We are trying to use DHCP Client of Vxworks for our testing , we use =
Tornado for building the Vxworks image , in this case if we enable =
"built-in symbol table" component then DHCP Client is not functioning =
properly. Whereas if we go in for "downloadable symbol table" the same =
scenario works fine.=20

We intend to know what is the bondage between Symbol table and DHCP =
Client. I request the Vxworks users to throw some light on  this.
Best Regards

Gunaseelan Venkatachary
Ph:408 588 6672

------=_NextPart_000_006D_01C0A176.34BDA2F0
Content-Type: text/html;
	charset="iso-8859-1"
Content-Transfer-Encoding: quoted-printable









Hello All,


We are trying to use DHCP Client of =
Vxworks for our=20
testing , we use Tornado for building the Vxworks image , in this case =
if we=20
enable "built-in symbol table" component then DHCP Client is not =
functioning=20
properly. Whereas if we go in for "downloadable symbol table" the same =
scenario=20
works fine. 


 


We intend to know what is the bondage =
between=20
Symbol table and DHCP Client. I request the Vxworks users to throw some =
light=20
on  this.


Best Regards


 


Gunaseelan Venkatachary
Ph:408 588=20
6672


------=_NextPart_000_006D_01C0A176.34BDA2F0--



From vxwexplo-errs@csg.lbl.gov  Wed Feb 28 17:37:38 2001
From: Alan Biocca 
Date: Wed Feb 28 17:37:41 PST 2001
Subject: New Exploder Web Interface and other administrivia

The Web Interface for the VxWorks Mailing List Exploder Subscription 
Management has been put into production. This allows users to subscribe, 
unsubscribe, and select Immediate and/or Digest list options using a web 
browser.

Visit the website at the new URL: http://www-vxw.lbl.gov/vxworks to access 
these features, as well as our article archives and posting information.

Please use the new web interface to reduce our administration effort, and 
get near-realtime response.

The interface uses the email name and password to protect against 
tampering. To set your password the first time an emailed authentication 
code is used to protect against hijacking.

There was one bounce that leaked through the list today, it came within one 
character of an existing filter deleting it. I have fixed that. Just don't 
use the word 'undeliverable' in your subject lines anymore.

There is a problem with the digests. Our local news server is delivering a 
bad message to our newsreader program and crashing it, thus no 
comp.os.vxworks digests. This clears itself when the bad article times out, 
and the remaining articles come through then in a large batch. Sorry for 
the inconvenience.

One last request - please everyone set your email for PLAIN TEXT, do not 
use HTML and/or MIME as this wastes bandwidth and doesn't work well.

Thanks,

Alan K Biocca
AKBiocca@lbl.gov