From vxwexplo-errs@csg.lbl.gov Wed Aug 1 04:04:06 2001 From: Vxworks Exploder Date: Wed Aug 1 04:04:09 PDT 2001 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Wed Aug 1 04:03:13 PDT 2001 Subject: Virtual Address Subject: Re: RealTek 8139A Driver Subject: Re: printer cartridge Subject: vxWork hang up Subject: Re: intLock() or intDisable() ? Subject: Re: error of the task Subject: Re: Error while downloading __eabi Subject: Puzzled by PPC assembler for varargs function Subject: vxhacks repository Subject: Re: vxhacks repository Subject: Re: prjConfig.c in BSP for Power PC Subject: Clearcase???? Subject: Re: Error while downloading __eabi Subject: Re: Error while downloading __eabi Subject: Re: Q: End driver, Network Interface and etherHooks Subject: Re: Exception Handler Question... [excVecSet() vs excHookAdd() on a Subject: Re: Does ACE/TAO work with GreenHills C++ and VxWorks Subject: Writing host files under serial/TSFS Subject: Re: tcl/tk Subject: Re: vxWork hang up Subject: WindView filters? Subject: Re: Where Is the FAQ and Archive of Code? Subject: Re: Q: End driver, Network Interface and etherHooks Subject: utilize Smalltalk classes on vxworks target Subject: tiny library for memory mapped graphic interface (MPC823) Subject: what's the usage of ... appears in source code Subject: Re: Does ACE/TAO work with GreenHills C++ and VxWorks Subject: Can not boot FADS860T from TFFS? Subject: Re: Q: End driver, Network Interface and etherHooks Subject: Pb with call to ioctl (FIOCANCEL) ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Virtual Address Date: Tue, 31 Jul 2001 12:11:42 +0000 (UTC) From: raghunath@controlnet.co.in (Raghunath Adhyapak) Organization: Mailgate.ORG Server - http://www.Mailgate.ORG Message-ID: Hi all, I want to convert a physical address to virtual address. Can anyone tell me how to do it? In cacheLib.h, there's are 2 macros named 1. CACHE_DRV_PHYS_TO_VIRT 2. CACHE_DMA_PHYS_TO_VIRT Can I use one of the above macros? If yes, what is the difference between the two? If no, then can anyone suggest a suitable alternative. Regards, Raghunath (O) 91-832-783615 ext 505 (R) 91-832-423024 - -- Posted from IDENT:netscape@dakiya.controlnet.co.in.116.54.202.in-addr.arpa [202.54.116.69] (may be forged) via Mailgate.ORG Server - http://www.Mailgate.ORG --------------------------- Newsgroups: comp.os.vxworks Subject: Re: RealTek 8139A Driver Date: Tue, 31 Jul 2001 12:51:18 GMT From: Chris Organization: EarthLink Inc. -- http://www.EarthLink.net Message-ID: References: Reply-To: zgreek_junk@bigfoot.com A driver can be gotten from Realtek Semiconductor Corp http://www.realtek.com.tw/htm/download/cgi/DLd1.cgi?model=RTL8139%28A%2FB%2FC%2F8130%29&type=2 Z On Thu, 26 Jul 2001 14:46:22 -0500, "Joel Hannah" wrote: >I am using a single board computer to host my VxWorks application on. I have >been having TCP/IP problems and I have narrowed the problem down to ReakTek >8139A ethernet controler driver. I have been using a driver that a Windriver >FAE supplied to me with the stipulation that Windriver does not support this >driver. He promised me that Windriver was going to support the 8139A but I >haven't seen it yet. > >I was wondering if someone had a later version of this driver (I am running >driver version: Release 1.0/EA3 > >I would really appreciate any help anyone could give me.... > >Joel --------------------------- Newsgroups: comp.os.vxworks Subject: Re: printer cartridge Date: Sat, 28 Jul 2001 8:50:23 GMT From: Ransley@hotmail.com (Ransley) Message-ID: <6Vx97.23529$Ne6.1076402@wagner.videotron.net> Check this out http://promos.batcave.net/ Ransley --------------------------- Newsgroups: comp.os.vxworks Subject: vxWork hang up Date: Tue, 31 Jul 2001 21:11:07 +0800 From: "Chew GimHeng" Organization: SCV Cablemodem News site Message-ID: <9k6a32$i3l$1@newsie.singa.pore.net> This is a multi-part message in MIME format. - ------=_NextPart_000_002C_01C11A05.50E18480 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable Hi all,=20 I had used the TCP/IP connection from a vxwork system to a Windows = sytem . The connection was used to send heart beat from the vxWork to = the Window System. But after approx. 10000 heart beat send, the whole = system hang.=20 Will that cause by the watchdog routine which I used to call the = heartbeat routine very second? or is there any other way beside using watchdog for peroidic sending of = heartbeat? Thank you. - ------=_NextPart_000_002C_01C11A05.50E18480 Content-Type: text/html; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable
Hi all,
    I had used the = TCP/IP connection=20 from a vxwork system to a Windows sytem . The connection was used to = send heart=20 beat from the vxWork to the Window System. But after approx. 10000 heart = beat=20 send, the whole system hang.
 
Will that cause by the watchdog = routine which=20 I used to call the heartbeat routine very second?
or is there any other way beside using = watchdog for=20 peroidic sending of heartbeat?
 
Thank you.
- ------=_NextPart_000_002C_01C11A05.50E18480-- --------------------------- Newsgroups: comp.os.vxworks Subject: Re: intLock() or intDisable() ? Date: Tue, 31 Jul 2001 11:46:37 +0100 From: vxworksintent@tao-group.com (David Given) Organization: I do not speak for anyone but myself, and barely that. Message-ID: References: Sender: dg@pearl.tao.co.uk In article , xgdong@sina.com (Docular Xu) writes: > I'm a beginner in VxWorks. > When entering an ISR, I know the interrupt should be locked first. In > VxWorks' docs, there are no directions told me how to do this. > There are two functions which seem approporate to this task : > intLock() and intDisable(). Which one should be used? According to my understanding, ISRs always run in the processor's interrupt mode with interrupts off. So you need take no explicit action to block interrupts. (Unless your system supports nested interrupts, but even then you would only need to explicitly call intLock() and intUnlock() if you wanted to block anything with a higher priority.) For reference: intLock() and intUnlock() block and unblock global interrupts. All systems have these. intDisable() and intEnable() block and unblock specific interrupt sources in a platform-dependent manner. Not all systems have these. They allow you to block, say, all interrupts with a priority less than 5 while allowing others to be executed. Which you use depends on the system you are running on and how portable you want your code to be. - -- David Given Tao Group Ltd. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: error of the task Date: Tue, 31 Jul 2001 10:11:24 -0400 From: "Kevin Livingston" Message-ID: References: <6ccda887.0107272250.95aea24@posting.google.com> try the printErrno command - remember the numbers are in Hex so the approiate usage would be printErrno 0x20100004 or printErrno 0x3006c I don't know if there is something you need to include in your kernel to get the printErrno command. when I ran those commands on my board I received the following output I don't know if it is safe to assume that they are the same on your board. printErrno 0x3006c S_taskLib_TASK_UNDELAYED value = 25 = 0x19 printErrno 0x20100004 Unknown errno = 0x20100004 value = 28 = 0x1c hope this helps Kevin "wang feng" wrote in message news:6ccda887.0107272250.95aea24@posting.google.com... > hi, guys, > > I am a newie to vxworks, when I use tonato2.0 shell to inspect my > task's status, the meaning of 'errno' column bother me a lot, just > like '20100004','3006c', how to check the error? > > thnks in advance, > > Wangfeng > > ps: here is what the screen looks like, > > > -> i > NAME ENTRY TID PRI STATUS PC SP > ERRNO DELAY > ---------- ------------ -------- --- ---------- -------- -------- > ------- ----- > tExcTask _excTask 7fb5134 0 PEND 18923a 7fb50a4 > 0 0 > tLogTask _logTask 7fb2830 0 PEND 18923a 7fb279c > 0 0 > tShell _shell 7efa11c 1 PEND 115424 7ef9df0 > 0 0 > tWdbTask 0x139c80 7efb2c0 3 READY 115424 7efb21c > 0 0 > TimerTask _TimerTask 7c36c40 20 DELAY 11479e 7c36bf0 > 0 1 > tNetTask _netTask 7f0f0f8 50 READY 11538b 7f0efd8 > 0 0 > tFtpdTask 0x13ff00 7efea18 55 PEND 115424 7efe94c > 0 0 > tMsgRecv0 _MsgRecvTask 7c224b8 75 PEND+T 115424 7c223a0 > 0 804 > tConnctChec_ConnectChec 7c1e324 85 DELAY 11479e 7c1e290 > 0 349 > SYSCTRL _TaskEntry 7c2ba34 200 PEND 18923a 7c2b9ac > 0 0 > BRDCTRL _TaskEntry 7c27840 200 PEND 18923a 7c277b8 > 0 0 > BRDMONI _TaskEntry 7c2364c 200 PEND 18923a 7c235c4 > 0 0 > DB_APP _TaskEntry 7bab0f8 200 PEND 18923a 7bab070 > 0 0 > AMAMS _TaskEntry 7ba6f04 200 PEND 18923a 7ba6e7c > 20100004 0 > AMACL _TaskEntry 7ba2d10 200 PEND 18923a 7ba2c88 > 20100004 0 > CLIS _TaskEntry 7b9eb1c 200 PEND 18923a 7b9ea94 > 0 0 > MDIA _TaskEntry 7b9a928 200 PEND 18923a 7b9a8a0 > 20100004 0 > AMCNF _TaskEntry 7b96734 200 PEND 18923a 7b966ac > 20100004 0 > tDcacheUpd _dcacheUpd 7f91148 250 DELAY 11479e 7f910f0 > 3006c 21 > value = 0 = 0x0 > -> --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Error while downloading __eabi Date: Tue, 31 Jul 2001 09:13:23 -0500 From: Dinker Charak Organization: Fermi National Accelerator Laboratory Message-ID: <3B66BD03.9E963A24@fnal.gov> References: <3B657868.71A55A33@fnal.gov> thanks for the help ... i understand the problem now ... dinker --------------------------- Newsgroups: comp.os.vxworks Subject: Puzzled by PPC assembler for varargs function Date: Tue, 31 Jul 2001 15:42:24 +0100 From: Hugh Duggan Organization: Lucent Technologies Message-ID: <3B66C3D0.2BCC6823@lucent.com> Hi, Here's hoping that there are some PPC assembler gurus out there who are willing to help me! I'm trying to understand how varargs functions work when called from an interrupt handler on a PPC-based board (because I suspect a compiler bug). So I disassembled a varargs function of my own [see below for a listing of the prologue]. I'm puzzled mainly about the branch in the function prologue (line 3455). Why is there a conditional statement during the setup of the stack frame and allocation of parameters? What does the condition depend upon? Why is it OK in some situations to simply skip over the conditional block? Is there any obvious reason why this code would cause problems if called from an ISR? We came across a similar problem where floating-point registers were being used to move 64 bits at a time, but these FP registers were not saved and restored across an ISR. Caused lots of fun... If it matters, this was compiled for ppc603 with -O2. I did experiment with changing the varargs function to an empty shell: this reduced the overall number of assembler instructions, but the branch & stfd statements were still present. Thanks in advance, Hugh 3437 .stabn 68,0,342,.LM269-traceVerboseReal 3438 .LM269: 3439 148c 9421FEF8 stwu 1,-264(1) 3440 1490 7C0802A6 mflr 0 3441 1494 936100F4 stw 27,244(1) 3442 1498 938100F8 stw 28,248(1) 3443 149c 93A100FC stw 29,252(1) 3444 14a0 93C10100 stw 30,256(1) 3445 14a4 93E10104 stw 31,260(1) 3446 14a8 9001010C stw 0,268(1) 3447 14ac 7C7D1B78 mr 29,3 3448 14b0 9081000C stw 4,12(1) 3449 14b4 90A10010 stw 5,16(1) 3450 14b8 90C10014 stw 6,20(1) 3451 14bc 90E10018 stw 7,24(1) 3452 14c0 9101001C stw 8,28(1) 3453 14c4 91210020 stw 9,32(1) 3454 14c8 91410024 stw 10,36(1) 3455 14cc 40860028 bc 4,6,.L1197 3456 14d0 D8210028 stfd 1,40(1) 3457 14d4 D8410030 stfd 2,48(1) 3458 14d8 D8610038 stfd 3,56(1) 3459 14dc D8810040 stfd 4,64(1) 3460 14e0 D8A10048 stfd 5,72(1 3461 14e4 D8C10050 stfd 6,80(1) 3462 14e8 D8E10058 stfd 7,88(1) 3463 14ec D9010060 stfd 8,96(1) 3464 14f0 D9210068 stfd 9,104(1) 3465 .L1197: 3466 .LBB84: 3467 14f4 7C9B2378 mr 27,4 3468 14f8 48000001 bl __get_eh_context 3469 14fc 7C7E1B78 mr 30,3 3470 1500 7FC0F378 mr 0,30 3471 .stabn 68,0,343,.LM270-traceVerboseReal 3472 .LM270: - ----------------------------------------------------------- Hugh Duggan | Home Office : +44 117 9626702 Lucent Technologies | Swindon : +44 1793 887278 Chippenham, UNITED KINGDOM | E-mail : hduggan@lucent.com --------------------------- Newsgroups: comp.os.vxworks Subject: vxhacks repository Date: Tue, 31 Jul 2001 17:08:46 +0200 From: "Shlomo Pongratz" Organization: NetVision Israel Message-ID: <9k6e60$5ej$1@news.netvision.net.il> Does anyone know what happend to vxhacks repository. Used to be on http://www.pso.com/vxhacks.html Thanks Shlomi --------------------------- Newsgroups: comp.os.vxworks Subject: Re: vxhacks repository Date: 31 Jul 2001 15:09:07 GMT From: "Johan Borkhuis" Organization: Agere Systems Message-ID: References: <9k6e60$5ej$1@news.netvision.net.il> "Shlomo Pongratz" wrote: > Does anyone know what happend to vxhacks repository. > > Used to be on http://www.pso.com/vxhacks.html Can be found on http://www.vxhacks.net/ or http://www.xs4all.nl/~borkhuis/vxworks/ftp/vxhacks/ 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: prjConfig.c in BSP for Power PC Date: Tue, 31 Jul 2001 16:20:19 +0100 From: Owain Phillips Organization: Siemens Inc. Message-ID: <3B66CCB3.3B4B4C6E@siemenscomms.co.uk> References: <3b659edb.851409982@news.demon.co.uk> This is a multi-part message in MIME format. - --------------65490DDF2EB0B8D5234873E1 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit I think we had this one.... We built a BSP for tornado 2.0, then got patched up to 2.0.2 and found the BSP wouldn't build. The problem was that our BSP used the architecture library librc32364sfgnu, which was not upgraded; the standard R3K, R4K libraries were upgraded by the patches....consequence unresloved externals. We have been told our problems will be solved when we get a new architecture library which comes as standard with tornado 2.1 for mips.....we are waiting fo this. Try looking on windsurf or contacting WRS. Regards, Owain Phillips Jan Larcombe wrote: > I'm having problems building my BSP, I end up with an unresolved > external muxMaxBinds. I've checked why this is and it seems to be > because the prjConfig.c which is generated is calling the routine > usrMuxLibInit. In earlier versons of my BSP prjConfig.c was calling > MuxLibInit instead. > Obviously there is no point in ediiting prjConfig.c but does anyone > know how to change configurations so that the prjConfig.c file which > is automatically generated will call the correct function. > > Thanks in advance - --------------65490DDF2EB0B8D5234873E1 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 - --------------65490DDF2EB0B8D5234873E1-- --------------------------- Newsgroups: comp.os.vxworks Subject: Clearcase???? Date: Tue, 31 Jul 2001 16:21:59 +0100 From: Owain Phillips Organization: Siemens Inc. Message-ID: <3B66CD17.9FBAC4D9@siemenscomms.co.uk> This is a multi-part message in MIME format. - --------------412652DF0E53A1EC6BE8720B Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Can anyone point me to resources for using clearcase under the tornado "IDE".... regards, Owain - --------------412652DF0E53A1EC6BE8720B 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 - --------------412652DF0E53A1EC6BE8720B-- --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Error while downloading __eabi Date: 31 Jul 2001 08:53:01 -0700 From: john_94501@yahoo.com (John) Organization: http://groups.google.com/ Message-ID: <488e459a.0107310753.305d57da@posting.google.com> References: <3B657868.71A55A33@fnal.gov> Hello, This one is simple; don't use main() as your function. When you compile functions called main the compiler does some extra things behind your back (hence the error you are seeing). You can call your entry point function anything you like with VxWorks since you are just going to call the function from the shell, or add it to the kernel's initialisation sequence (via USER_APPL_INIT). HTH, John... Dinker Charak wrote in message news:<3B657868.71A55A33@fnal.gov>... > Hello. > > I created a simple Hello.c with just > > main() > { > } > > and compiled (right click in workspace window, then compile Hello.c) it > and downloaded (using download object file icon in workspace) it, all > using tornado. I get this error ... > > Errors while downloading > /usr/products/VxWorks/TornadoII/target/proj/Project0/default/Hello.o: > __eabi > > any idea? > > regards, > dinker --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Error while downloading __eabi Date: Tue, 31 Jul 2001 09:55:16 -0500 From: "tazboy78" Organization: http://extra.newsguy.com Message-ID: <9k6gpb02pbc@enews2.newsguy.com> References: <3B657868.71A55A33@fnal.gov> You can't use main() as a function name. It is used in the vxWorks kernel and will conflict. "Dinker Charak" wrote in message news:3B657868.71A55A33@fnal.gov... > Hello. > > I created a simple Hello.c with just > > main() > { > } > > and compiled (right click in workspace window, then compile Hello.c) it > and downloaded (using download object file icon in workspace) it, all > using tornado. I get this error ... > > Errors while downloading > /usr/products/VxWorks/TornadoII/target/proj/Project0/default/Hello.o: > __eabi > > any idea? > > regards, > dinker --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Q: End driver, Network Interface and etherHooks Date: 31 Jul 2001 09:25:15 -0700 From: bmurr@my-deja.com (Brian Murphy) Organization: http://groups.google.com/ Message-ID: References: "Johan Borkhuis" wrote in message news:... > bmurr@my-deja.com (Brian Murphy) wrote: > > > Can someone explain to be the relationship between an End driver and > > network interface. > > > > For example there are 2 files associated with the AMD 7990 LANCE, > > these are:- > > > > target/src/drv/netif/if_ln.c > > target/src/drv/end/ln7990End.c > > > > If I want to develop a new ethernet driver, do I have to develop both > > the End driver and the network interface? > > > > Is it essential that I write both? > > No, just go for the END driver, and leave the netif file. The netif file is > an older type of network interface, and is replaced by the END style > driver. > > Groeten, > Johan Johan Thanks a lot for that. As you seem to be the guru can you tell me what I should be using instead of the etherxxxHooks functions? After writing the End driver my main is to be able to intercept messages as they are received from the driver before they are passed to the network layer. I want link layer to link layer traffic. I was hoping to do this using etherInputHook () and etherOutptuHook () however, these will not be supported in the future. How can I achieve the inteception of traffic without using these functions. How will the functionality of the etherxxxHook functions be replaced. Thanks, Brian --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Exception Handler Question... [excVecSet() vs excHookAdd() on a 8260] Date: Tue, 31 Jul 2001 18:11:43 +0100 From: James Marshall Organization: Agilent Technologies UK Ltd Message-ID: <3B66E6CF.7EC4B254@agilent.com> References: <3B51368C.7832D79F@home.com> <3B54F2F7.F0EADF7C@home.com> Todd, I used excHookAdd() on x86 and found that only tasks created AFTER the hook is installed could make use of it. James Marshall. Todd Sandor wrote: > It turned out that the excHookAdd() method was working all along. When I use this > method and then forced a "task" have an exception the specified routine was > called. My initial tests involved invoking commands/routines that performed an > exception in the context of the vxWorks "shell" which has special characteristics > [the shell is respawn'ed if an exception occurs]. Thanks for the responses... > > Cheers... > > Dan Gold wrote: > > > In article <3B51368C.7832D79F@home.com>, Todd Sandor says... > > > > > >Hi: > > >I'm looking for help - I'm having trouble getting excHookAdd() to work > > >on a 8260 (ppc) based system. > > > > > >I'm trying to perform some additional operations when a [task] exception > > >[machine check] occurs. I'ld prefer not to replace the existing default > > >exception handler (excIntHandler), unless I can determine how to get > > >information normally displayed by the default exception hander via the > > >exception stack frame pointer [Exception next instruction address, Task > > >name, etc]. > > > > It depends on the particular exception. External interrupt (0x500) and the > > decrementer (0x900) call excIntHandle(). The others (i.e. machine check, dsi, > > etc..) call excExcHandle(). > > > > >The method I'ld prefer [or I think I would prefer] to use is the > > >excHookAdd() method that is suppose to call the supplied routine after > > >normal exception handling is performed. I want to do this for machine > > >check exceptions and possibly data access exception - although I don't > > >really know the difference between a machine check and data access > > >exception on a 8260, anyone? The supplied routine is called with the > > >taskId, the exception vector number and the stack frame pointer as > > >parameters. > > > > Machine check can be caused by multiple things, i.e. TEA, etc... data access is > > usually called in conjunction with the MMU. Either there's no translation for > > the desired address, the write-protect bits were set on the page, > > user/supervisor access, etc... > > > > >When I try to use excHookAdd() method, via the call: > > >excHookAdd((FUNCPTR)handler); > > >the "hander" functions isn't called when an exception occurs - I'm not > > >sure why? > > >When I check the address pointed to by the "_func_excIntHook" variable, > > >it is 0 after the excHookAdd() call - I believe this is suppose to be > > >set after the excHookAdd() call is made. > > > > Because _func_excIntHook is checked in excIntHandle(). Try modifying > > _func_excBaseHook instead. > > > > >If anyone can give me any hints, it would be appreciated...Cheers... > > > > hth, > > > > Dan Gold gold@ensemble.com --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Does ACE/TAO work with GreenHills C++ and VxWorks Date: Tue, 31 Jul 2001 15:09:05 -0400 From: "Kevin Livingston" Message-ID: References: <3B6361B5.B63340D4@.no.spam.nc.rr.com> This might not be a direct answer to your question... but, We have successfully compilied and used ACE (not TAO) with VxWorks. we used the g++ compilier. we are targeting a pentium. and using Windows NT as a host. we could not use Tornado to build the library but had to do it manually. It should be noted that the version of make that came with tornado is not adequate, and for building on a Windows NT host we needed to get sh. So we needed to get the Cygwin tool set to get the utilities we needed, such as a make with the --unix option and then sh which some of the make commands seemed to require. There are some changes that need to be made to ACE to get it to function correctly with VxWorks, they are minimal though - at least on our end - many are documented on the internet, or I could try to find my list and send it to you. look at http://www.cs.wustl.edu/~schmidt/ACE_wrappers/ACE-INSTALL.html All in all it was a pretty straight forward process and easily duplicatable once you have all the pieces. Kevin "LB" wrote in message news:3B6361B5.B63340D4@.no.spam.nc.rr.com... > Hello, > > I am trying to find out if the GreenHills C++ compiler works with > VxWorks 5.3.1 (or VxWorks 5.4) can compile ACE/TAO. Has anyone ever got > it to work? Will ACE/TAO be crippled by any GreenHills limitations? > > Thanks, > Lance > --------------------------- Newsgroups: comp.os.vxworks Subject: Writing host files under serial/TSFS Date: 31 Jul 2001 12:17:16 -0700 From: majeb@yahoo.com (matt) Organization: http://groups.google.com/ Message-ID: <66845b46.0107311117.161b7aab@posting.google.com> Is it possible to create a file on the host network when connected over serial TSFS? Ive tried on my system and failed using fopen, open etc. Works fine when hooked up over Ethernet. Thanx a lot --------------------------- Newsgroups: comp.os.vxworks Subject: Re: tcl/tk Date: Tue, 31 Jul 2001 15:14:21 -0700 From: Anmol Organization: http://extra.newsguy.com Message-ID: <3B672DBD.25812A23@eternal-systems.com> References: <9jrfa6$fkb$1@rex.ip-plus.net> Mehmet Demirel wrote: > Hi, > > is there a version of tcl/tk for vxworks ? > > Where can I download it and has someone experiences in it ? > > Thanx > Mehmet Hi! vxWorks comes with wtxTcl. You can use that to run tcl scripts, thats how people write host based tools for vxWorks.. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: vxWork hang up Date: Tue, 31 Jul 2001 15:23:48 -0700 From: Anmol Organization: http://extra.newsguy.com Message-ID: <3B672FF4.26FE0DB0@eternal-systems.com> References: <9k6a32$i3l$1@newsie.singa.pore.net> Does it give any error message ?? How many heartbeats per second ? Also, your process on the windows side - does it timeout ?? You need to give out more info... Chew GimHeng wrote: > Hi all, I had used the TCP/IP connection from a vxwork system to a > Windows sytem . The connection was used to send heart beat from the > vxWork to the Window System. But after approx. 10000 heart beat send, > the whole system hang. Will that cause by the watchdog routine which I > used to call the heartbeat routine very second?or is there any other > way beside using watchdog for peroidic sending of heartbeat? Thank > you. --------------------------- Newsgroups: comp.os.vxworks Subject: WindView filters? Date: 31 Jul 2001 20:06:28 -0400 From: Bill Pringlemeir Organization: Bell Sympatico Message-ID: Sender: bpringlemeir@DeadDuck Is there some way to filter out tasks that are never given control from the WindView display? When I click on information for many tasks, there is only a `PENDING' status. I would like to eliminate all of these tasks. I thought that the `Attributes' dialog with the pattern could do this, but it looks like it is only a reg-expression on the task name. Any other WindView resources besides the FAQ and src directory? tia, Bill Pringlemeir. - -- ANNOYED BY ENORMOUS GNATS. CAPTAIN INTOXICATED. SINKING. U-47. vxWorks FAQ, "http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html" --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Where Is the FAQ and Archive of Code? Date: 31 Jul 2001 18:07:57 -0700 From: clusardi2k@aol.com (Christopher M. Lusardi) Organization: http://groups.google.com/ Message-ID: References: clusardi2k@aol.com (Christopher M. Lusardi) wrote in message news:... > Hello, > > Where are the archives for the source code in this newsgroup? > > Thanks, > Christopher Lusardi I found it, on my own, at ftp.atd.ucar.edu under pub/archive/vxworks/vx. Anybody have anything else? Christopher Lusardi --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Q: End driver, Network Interface and etherHooks Date: Tue, 31 Jul 2001 20:18:33 -0700 From: DrDiags Organization: laserlink.net Message-ID: <3B677509.86212EF8@covad.net> References: Reply-To: drdiags@covad.net Brian, Not the GURU here, but for the END or NPT devices you can use MUX_SNARF and MUX_OUTPUT protocols with the muxBind() or muxTkBind() routines. Brian Murphy wrote: > <....SNIP....> > > After writing the End driver my main is to be able to intercept > messages as they are received from the driver before they are passed > to the network layer. I want link layer to link layer traffic. I was > hoping to do this using etherInputHook () and etherOutptuHook () > however, these will not be supported in the future. How can I achieve > the inteception of traffic without using these functions. How will the > functionality of the etherxxxHook functions be replaced. You should be able to go onto WindSurf and download the NPT developer's ToolKit which has documentation and binaries, which you don't want to install in your development Tornado tree, but in a dummy directory (The NPT support extended the END driver and other significant changes - see Tornado 2.0.2 release notes if available). But the documentation discusses both the END and NPT drivers and the MUX. - -- http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html (vxWorks FAQ home page) --------------------------- Newsgroups: comp.os.vxworks Subject: utilize Smalltalk classes on vxworks target Date: Wed, 01 Aug 2001 07:47:56 +0200 From: Rainer Stelzer Organization: Stoye GmbH, =?iso-8859-1?Q?K=F6ln?= Message-ID: <3B67980C.6DC695DF@stoye.de> Hi, I'ld like to utilize some smalltalk classes on vxworks target, to avoid reprogramming the stuff. Has somebody out there an idea how to translate smalltalk to C/C++ ? thanx in advance. regards Rainer tornado, vxworks ===================================================== Gebrüder Stoye GmbH department: traffic control systems name: Rainer Stelzer adr: Longericher Strasse 177 city: Cologne, Germany zip: 50739 phone: ++49/221/1707019 fax: ++49/221/1707010 mailto: r.stelzer@stoye.de www: http://www.stoye.de ===================================================== --------------------------- Newsgroups: comp.os.vxworks Subject: tiny library for memory mapped graphic interface (MPC823) Date: Wed, 01 Aug 2001 08:14:57 +0200 From: Rainer Stelzer Organization: Stoye GmbH, =?iso-8859-1?Q?K=F6ln?= Message-ID: <3B679E61.23F5B7A5@stoye.de> Hi i'm looking for a tiny graphics library to output characters (and maybe simple graphics) to a pixel based / memory mapped grafic controller. The target hardware is a MPC823, but this problem should be quite common. (It was quite common in the computer stoneage ) Can somebody out there give me a hint were to start searching ? Thanx in advance. regards R.Stelzer vxworks, tornado ===================================================== Gebrüder Stoye GmbH department: traffic control systems name: Rainer Stelzer adr: Longericher Strasse 177 city: Cologne, Germany zip: 50739 phone: ++49/221/1707019 fax: ++49/221/1707010 mailto: r.stelzer@stoye.de www: http://www.stoye.de ===================================================== --------------------------- Newsgroups: comp.os.vxworks Subject: what's the usage of ... appears in source code Date: 31 Jul 2001 23:32:46 -0700 From: fjpan@cmmail.com (Baumann) Organization: http://groups.google.com/ Message-ID: <6603907b.0107312232.60dde0ef@posting.google.com> anyone knows this usage? retval = pciIntConnect((void(**)(...))INT_LVL_DOORBELL,(void(*)(...))Pci_DoorBell_ISR, 0); --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Does ACE/TAO work with GreenHills C++ and VxWorks Date: Wed, 1 Aug 2001 09:10:39 +0200 From: "Frank Wolf" Organization: T-Online Message-ID: References: <3B6361B5.B63340D4@.no.spam.nc.rr.com> Hi there! I've assembled some instructions on how to build ACE for VxWorks (on PowerPC and Hitachi SuperH). I think my document is quite easy to understand - if someone is interested I could mail it. Frank "Kevin Livingston" schrieb im Newsbeitrag news:BkD97.29$hN4.1069@client... > This might not be a direct answer to your question... but, > > We have successfully compilied and used ACE (not TAO) with VxWorks. > we used the g++ compilier. > we are targeting a pentium. > and using Windows NT as a host. > > we could not use Tornado to build the library but had to do it manually. It > should be noted that the version of make that came with tornado is not > adequate, and for building on a Windows NT host we needed to get sh. So we > needed to get the Cygwin tool set to get the utilities we needed, such as a > make with the --unix option and then sh which some of the make commands > seemed to require. > > There are some changes that need to be made to ACE to get it to function > correctly with VxWorks, they are minimal though - at least on our end - many > are documented on the internet, or I could try to find my list and send it > to you. > look at > http://www.cs.wustl.edu/~schmidt/ACE_wrappers/ACE-INSTALL.html > > All in all it was a pretty straight forward process and easily duplicatable > once you have all the pieces. > Kevin > > > "LB" wrote in message > news:3B6361B5.B63340D4@.no.spam.nc.rr.com... > > Hello, > > > > I am trying to find out if the GreenHills C++ compiler works with > > VxWorks 5.3.1 (or VxWorks 5.4) can compile ACE/TAO. Has anyone ever got > > it to work? Will ACE/TAO be crippled by any GreenHills limitations? > > > > Thanks, > > Lance > > > > --------------------------- Newsgroups: comp.os.vxworks Subject: Can not boot FADS860T from TFFS? Date: 1 Aug 2001 02:42:03 -0700 From: liuyw70@21cn.com (LiuYW) Organization: http://groups.google.com/ Message-ID: <838b088e.0108010142.3404de96@posting.google.com> Hi, We have encountered a problem when we boot loadable vxWorks image from TFFS on FADS860T. The case is: 1. Write the loadable vxWorks image in TFFS on FADS, name: "/tffs0/vxWorks" through TFTP; 2. Run bootrom, select boot parameter: $tffs=0,0(0,0)host:/tffs0/vxWorks e=192.168.88.200 o=motfec 3. Run command: l, it shows load successfully, entry=0x10000 4. Run command: g 0x10000, it shows start at 0x10000... 5. Then it stops, displays nothing, it cannot start! 6. If we use ICE to run the image loaded from TFFS, it starts correctly; if we boot the same loadable vxWorks image from FTP, it also starts correctly. We have added 'o=motfec', and also added the line with the '+' sign to the module target\config\comps\src\net\usrNetBoot.c (line 97): if ( (strncmp (sysBootParams.bootDev, "scsi", 4) == 0) || + (strncmp (sysBootParams.bootDev, "tffs", 4) == 0) || (strncmp (sysBootParams.bootDev, "ide" , 3) == 0) || (strncmp (sysBootParams.bootDev, "ata" , 3) == 0) || (strncmp (sysBootParams.bootDev, "fd" , 2) == 0)) We do not think the error is caused by network initialization. So what's wrong? Regards, - - LiuYW --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Q: End driver, Network Interface and etherHooks Date: 1 Aug 2001 10:45:15 GMT From: "Johan Borkhuis" Organization: Agere Systems Message-ID: References: bmurr@my-deja.com (Brian Murphy) wrote: > Johan > Thanks a lot for that. As you seem to be the guru can you tell me what > I should be using instead of the etherxxxHooks functions? I am not the guru here, I only collect at lot of information. You can take a look at this http://www.xs4all.nl/~borkhuis/vxworks/vxw_pt4.html#4.8-C, it is an example on how to use etherhooks, it works with END drivers 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: Pb with call to ioctl (FIOCANCEL) Date: 01 Aug 2001 09:56:47 GMT From: Fab Message-ID: <200181-115647-968773@foorum.com> Hi all ! (VxWorks 5.2) Has someone ever experienced trouble with a call to ioctl (and the FIOCANCEL option) on a tyCo device ? On Vx5.2, all subsequents call to read functions are cancelled. Is there a mean to "cancel" only ONE pending read or the next call to read function ? (This is the case on VxWorks 5.4 ! Where to search for this difference ?) Thanks. __________________________________________________________ Ce message a été posté via la plateforme Web club-Internet.fr This message has been posted by the Web platform club-Internet.fr http://forums.club-internet.fr/ --------------------------- End of New-News digest ********************** From vxwexplo-errs@csg.lbl.gov Wed Aug 1 04:19:19 2001 From: "Burkin, Valery" Date: Wed Aug 1 04:19:21 PDT 2001 Subject: RE: startup scripts >please throw more light on the startup scripts given >with bootChange. what are those names separated by >commas ? i see some ip address, some directory etc. i >have found vxworks crazy here. there are no sample >shell scripts, nothing. it is crazy documentation - no >logic in the information furnished. It's not my common behavior to stick to and share some opinions bundled with more or less flaming character, especially ANONYMOUS ones... But I must confess this case makes a bit sence. VxWorks documenation as it comes with Tornado 2 is in the VERY bad state!!! It doesn't worth to mention the importance of CLEARLY and UNAMBIGUOUSLY written manuals which could show straight logical way to get to the desirable results within a reasonable amount of time. I was a newbie in the vxworks field and what I've found is just a bunch of cross-(and self-referenced) documents with no clear logic in between! It doesn't surprise me if lots of people getting stuck in it in a few minutes. Most of highly important topics especially concerning kernel, network and boot stage configuration create a very silly view of the whole system. Explanation about how to manage with GUI stuff in the vxworks wizard instead of giving clear explanation about included modules, their COMPLETE description in the corresponding docs. and dependency looks a bit stupid. Comparing with other development environments I ever worked with I find Tornado pretty good, but the documentation seemed to be a part of a secretary's job. W.R.S must get a point of it. Tornado isn't in GPL sphere. valery. From vxwexplo-errs@csg.lbl.gov Thu Aug 2 04:03:31 2001 From: Vxworks Exploder Date: Thu Aug 2 04:03:33 PDT 2001 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Thu Aug 2 04:03:13 PDT 2001 Subject: problem with muxDevUnload Subject: nfs error Subject: Re: tiny library for memory mapped graphic interface (MPC823) Subject: Re: Can not boot FADS860T from TFFS? Subject: Re: Can not boot FADS860T from TFFS? Subject: Re: tiny library for memory mapped graphic interface (MPC823) Subject: Re: tiny library for memory mapped graphic interface (MPC823) Subject: Send a Ethernet frame with etherOutputHook Subject: Project tool and clearcase fouls up Subject: Re: Project tool and clearcase fouls up Subject: INT0 as displayed by windview Subject: Re: Puzzled by PPC assembler for varargs function Subject: Re: INT0 as displayed by windview Subject: Re: INT0 as displayed by windview Subject: VxWorks boot problem Subject: Re: what's the usage of ... appears in source code Subject: Re: INT0 as displayed by windview Subject: Re: INT0 as displayed by windview Subject: Exception Subject: Re: Send a Ethernet frame with etherOutputHook Subject: screen saving Subject: About RIPAuthHookAdd () cannot work? Subject: graphics library / somebody already ported "VOGLE" to vxworks ? Subject: Re: Puzzled by PPC assembler for varargs function Subject: How to send a Ethernet Frame Subject: why ripAuthHookAdd() cannot work ? Subject: STL Support Subject: Re: Exception Subject: Why WindView shows that VxWorks spent most time in ISR? Subject: Re: Project tool and clearcase fouls up ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: problem with muxDevUnload Date: Wed, 01 Aug 2001 13:10:56 +0200 From: "Fauris, Sebastien" Organization: Nortel Networks Message-ID: <3B67E3C0.CFBF6053@europem01.nt.com> Reply-To: sfauris@nortelnetworks.com Hi VxWorkers, I meet some problems using the muxDevLoad routine in VxWorks5.4 (Tornado 2.0.2, MPC8260 BSP). I get an error during a call to a free (memPartFree) and the task performing the muxDevUnload get suspended. - -> tt s1u0 1a6444 vxTaskEntry +60 : pseudoBootSbcPn ([]) 11aaab8 pseudoBootSbcPn+70 : muxDevUnload ([]) 191480 muxDevUnload +160: free ([]) 1496bc free +1c : memPartFree ([]) 1491d4 memPartFree +144: taskSuspend ([]) value = 0 = 0x0 Has someone already met a similar problem using muxDevUnload ? I am sure it is not my END pnDrvUnload routine attached to the muxDevLoad which generates the error but the code within muxDevLoad in the kernel. Thanks, Sebastien. --------------------------- Newsgroups: comp.os.vxworks Subject: nfs error Date: Wed, 01 Aug 2001 11:36:40 GMT From: Gururaj patkar Organization: Optimum Online Message-ID: <3B67E93E.21381184@optonline.net> Hi, I am posting this again for help. I need help to debug a embedded card emttting nfs error. below are the details. I have nfsServer on a slave card and nfsClient on a master card. I have need to do download some files from master to the slave under a particular CLI command. I am doing this using NFS. The slave card informs the master card of its "hostname","ipaddress","mount point dir" during its initial interaction with the master card (prior to this, the slave card has nfsExported the corresponding "mount point dir"). When the master card gets this info it tries to nfsMount the "mount point dir" to a local mount point. However, nfsMount fails with the error - RPC_PROGUNAVAIL. why am I gettin this error? The config.h for slave and master card has - INCLUDE_NFS INCLUDE_NFS_SERVER INCLUDE_RPC. Thanks for any help. gururaj --------------------------- Newsgroups: comp.os.vxworks Subject: Re: tiny library for memory mapped graphic interface (MPC823) Date: Wed, 01 Aug 2001 07:24:33 -0700 From: DrDiags Organization: laserlink.net Message-ID: <3B681121.A5267A2A@covad.net> References: <3B679E61.23F5B7A5@stoye.de> Reply-To: drdiags@covad.net Rainier, What about curses? There is some vxcurses source files which can be ported (You didn't say which version of vxWorks). If you search this newsgroup for the post detailing where to find these files, that should get you started. I did download them, but you had to run them through Unix "shar" to make the *.c and *.h files. If you are interested, I have the curses text files unconverted. Rainer Stelzer wrote: > > Hi > > i'm looking for a tiny graphics library to output characters (and maybe > simple graphics) > to a pixel based / memory mapped grafic controller. > The target hardware is a MPC823, but this problem should be quite > common. > (It was quite common in the computer stoneage ) > > Can somebody out there give me a hint were to start searching ? > > Thanx in advance. > > regards > > R.Stelzer > > vxworks, tornado > ===================================================== > Gebrüder Stoye GmbH > department: traffic control systems > name: Rainer Stelzer > adr: Longericher Strasse 177 > city: Cologne, Germany > zip: 50739 > phone: ++49/221/1707019 > fax: ++49/221/1707010 > mailto: r.stelzer@stoye.de > www: http://www.stoye.de > ===================================================== - -- http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html (vxWorks FAQ home page) --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Can not boot FADS860T from TFFS? Date: Wed, 01 Aug 2001 09:26:25 -0500 From: Dinker Charak Organization: Fermi National Accelerator Laboratory Message-ID: <3B681191.296AB17D@fnal.gov> References: <838b088e.0108010142.3404de96@posting.google.com> Try this: Comment out (void)muxDevStopAll() in function go() in bootConfig.h ... LiuYW wrote: > > Hi, > We have encountered a problem when we boot loadable vxWorks image > from TFFS on FADS860T. The case is: > 1. Write the loadable vxWorks image in TFFS on FADS, name: > "/tffs0/vxWorks" through TFTP; > 2. Run bootrom, select boot parameter: > $tffs=0,0(0,0)host:/tffs0/vxWorks e=192.168.88.200 o=motfec > 3. Run command: l, it shows load successfully, entry=0x10000 > 4. Run command: g 0x10000, it shows start at 0x10000... > 5. Then it stops, displays nothing, it cannot start! > 6. If we use ICE to run the image loaded from TFFS, it starts > correctly; if we boot the same loadable vxWorks image from FTP, it > also starts correctly. > > We have added 'o=motfec', and also added the line with the '+' sign > to the module target\config\comps\src\net\usrNetBoot.c (line 97): > > if ( (strncmp (sysBootParams.bootDev, "scsi", 4) == 0) || > + (strncmp (sysBootParams.bootDev, "tffs", 4) == 0) || > (strncmp (sysBootParams.bootDev, "ide" , 3) == 0) || > (strncmp (sysBootParams.bootDev, "ata" , 3) == 0) || > (strncmp (sysBootParams.bootDev, "fd" , 2) == 0)) > > We do not think the error is caused by network initialization. So > what's wrong? > > Regards, > - LiuYW - -- Computing Division/Online and Database Systems 630-840-8604 (office) Fermi National Accelerator Laboratory http://www.fnal.gov --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Can not boot FADS860T from TFFS? Date: Wed, 01 Aug 2001 09:27:45 -0500 From: Dinker Charak Organization: Fermi National Accelerator Laboratory Message-ID: <3B6811E1.3839D71C@fnal.gov> References: <838b088e.0108010142.3404de96@posting.google.com> <3B681191.296AB17D@fnal.gov> > comment out > > (void)muxDevStopAll() in function go() in bootConfig.h oops! typo .... in bootConfig.c hth, dinker --------------------------- Newsgroups: comp.os.vxworks Subject: Re: tiny library for memory mapped graphic interface (MPC823) Date: Wed, 01 Aug 2001 16:34:12 +0200 From: Rainer Stelzer Organization: Stoye GmbH, =?iso-8859-1?Q?K=F6ln?= Message-ID: <3B681364.1B9B443@stoye.de> References: <3B679E61.23F5B7A5@stoye.de> <3B681121.A5267A2A@covad.net> Hi DrDiags, > What about curses? Aren't curses limited to character terminal output devices ? What I'm looking for is a low level driver working on a memory array where one byte represents the colour of a pixel in a TFT Display. With other words: I do not have the capability to place a character, I need to draw hiw first ;-) regards rainer ===================================================== Gebrüder Stoye GmbH department: traffic control systems name: Rainer Stelzer adr: Longericher Strasse 177 city: Cologne, Germany zip: 50739 phone: ++49/221/1707019 fax: ++49/221/1707010 mailto: r.stelzer@stoye.de www: http://www.stoye.de ===================================================== --------------------------- Newsgroups: comp.os.vxworks Subject: Re: tiny library for memory mapped graphic interface (MPC823) Date: 01 Aug 2001 11:14:22 -0400 From: Bill Pringlemeir Organization: Bell Sympatico Message-ID: References: <3B679E61.23F5B7A5@stoye.de> Sender: bpringlemeir@DeadDuck >>>>> "Rainer" == Rainer Stelzer writes: Rainer> Hi i'm looking for a tiny graphics library to output Rainer> characters (and maybe simple graphics) to a pixel based / Rainer> memory mapped grafic controller. The target hardware is a Rainer> MPC823, but this problem should be quite common. (It was Rainer> quite common in the computer stoneage ) Rainer> Can somebody out there give me a hint were to start searching Rainer> ? Try these links, "http://www.wrs.com/csdocs/product/windML/drivers/index.shtml" "http://www.wrs.com/csdocs/product/driver/distrib/windml/motorola.tar" These drivers have code for `bit blitting'. As well, you can use this for fonts. "http://www.wrs.com/csdocs/product/windML/fonts/index.html" "http://www.wrs.com/csdocs/product/windML/utilities/index.html" Afaik, these are ok to use if you have a WindSurf login. There are other graphics libraries around. For what you want, display initialization, and bit-blitting are probably all you need. You can give the Motorola web sight a try to, the have several code samples. I have done enough work you you though... hth, Bill Pringlemeir. - -- Have you ever joined a church where everyone but you can sing with your endothermic therapsid? Or tried to make boxes of Borax generate ESP? You will. And the company that will bring it to you: AT&T. vxWorks FAQ, "http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html" --------------------------- Newsgroups: comp.os.vxworks Subject: Send a Ethernet frame with etherOutputHook Date: Wed, 1 Aug 2001 17:40:02 +0200 From: "News" Message-ID: <3b6821e7$1_2@news.vo.lu> Hi, I have to code with VxWorks, But I don't know it. I want to send a frame Ethernet using etherOutputHook mechanism. Output()->ether_output()->TxStartup()->etherOutputHook). But I don't understand how to implement this mechanism. I just want to have an simple example. Regards, Christophe. --------------------------- Newsgroups: comp.os.vxworks Subject: Project tool and clearcase fouls up Date: Wed, 01 Aug 2001 17:40:30 +0100 From: Owain Phillips Organization: Siemens Inc. Message-ID: <3B6830FE.5FD5869B@siemenscomms.co.uk> This is a multi-part message in MIME format. - --------------C4A3EBA6C7BBE80ADAD78F6D Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit We want to put the our new BSP under source control (clearcase). The project tool works fine if the BSP is on the local disk, but as soon as we move the BSP over to the clearcase remote drive we have problems to generate projects based on the BSP. I try to generate a new bootloadable project based on the BSP on the clearcase drive. The project tool apears to parse the BSP OK, but fails to include all the necessary BSP files in the project. Files such as sysLib.c are missing from the projects files, therfore the project cannot be built. Anyone got a workaround? Ideas??? Thanks in advance. Owain - --------------C4A3EBA6C7BBE80ADAD78F6D 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 - --------------C4A3EBA6C7BBE80ADAD78F6D-- --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Project tool and clearcase fouls up Date: 01 Aug 2001 14:06:05 -0400 From: Bill Pringlemeir Organization: Bell Sympatico Message-ID: References: <3B6830FE.5FD5869B@siemenscomms.co.uk> Sender: bpringlemeir@DeadDuck >>>>> "Owain" == Owain Phillips writes: [snip] Owain> Files such as sysLib.c are missing from the projects files, Owain> therfore the project cannot be built. Owain> Anyone got a workaround? Ideas??? Are you on a Windows system? You might check that the project files have been marked as binary [even though they are ASCII]. I think that the are expected to be Unix end of line (single LF) and the CR/LF with hose the tools up... just a guess. hth, Bill Pringlemeir. - -- Two wrongs don't make a right, but three lefts do. vxWorks FAQ, "http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html" --------------------------- Newsgroups: comp.os.vxworks Subject: INT0 as displayed by windview Date: Wed, 1 Aug 2001 17:58:28 GMT From: "James D. McDonald" Organization: Boeing Message-ID: <3B684344.8B8027D5@yahoo.com> Sender: nntp@news.boeing.com (Boeing NNTP News Access) Exactly what is the INT0 line displayed by windview? Can system calls cause a context switch to this "thread"? --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Puzzled by PPC assembler for varargs function Date: Wed, 1 Aug 2001 11:14:13 +0100 From: vxworksintent@tao-group.com (David Given) Organization: I do not speak for anyone but myself, and barely that. Message-ID: References: <3B66C3D0.2BCC6823@lucent.com> Sender: dg@pearl.tao.co.uk In article <3B66C3D0.2BCC6823@lucent.com>, Hugh Duggan writes: [...] > I'm puzzled mainly about the branch in the function prologue (line 3455). > Why is there a conditional statement during the setup of the stack frame > and allocation of parameters? What does the condition depend upon? Why > is it OK in some situations to simply skip over the conditional block? [...] > 3455 14cc 40860028 bc 4,6,.L1197 > 3456 14d0 D8210028 stfd 1,40(1) > 3457 14d4 D8410030 stfd 2,48(1) > 3458 14d8 D8610038 stfd 3,56(1) > 3459 14dc D8810040 stfd 4,64(1) > 3460 14e0 D8A10048 stfd 5,72(1 > 3461 14e4 D8C10050 stfd 6,80(1) > 3462 14e8 D8E10058 stfd 7,88(1) > 3463 14ec D9010060 stfd 8,96(1) > 3464 14f0 D9210068 stfd 9,104(1) > 3465 .L1197: Bit 6 is in CR1, which is used to store status about the floating-point unit. I'd hazard a guess that what it's doing is checking to see whether the current process is using the floating point registers, and only saving them if needed. This would save a lot of time in the common case of a process that doesn't use the FPU. - -- David Given Tao Group Ltd. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: INT0 as displayed by windview Date: Wed, 01 Aug 2001 21:00:09 +0100 From: molochai Message-ID: References: <3B684344.8B8027D5@yahoo.com> In message <3B684344.8B8027D5@yahoo.com> "James D. McDonald" wrote: > Exactly what is the INT0 line displayed by windview? Can system calls > cause a context switch to this "thread"? It is genuinely an interrupt context, and in general the system calls can't switch into it. In this case, think of interrupts as super-high priority threads. What exactly is attached to INT0 depends on your board and BSP. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: INT0 as displayed by windview Date: Wed, 1 Aug 2001 20:58:17 GMT From: "James D. McDonald" Organization: Boeing Message-ID: <3B686D69.EAE93C32@yahoo.com> References: <3B684344.8B8027D5@yahoo.com> Sender: nntp@news.boeing.com (Boeing NNTP News Access) molochai wrote: > > In message <3B684344.8B8027D5@yahoo.com> > "James D. McDonald" wrote: > > > Exactly what is the INT0 line displayed by windview? Can system calls > > cause a context switch to this "thread"? > > It is genuinely an interrupt context, and in general the system calls can't > switch into it. In this case, think of interrupts as super-high priority > threads. That was my impression as well. > What exactly is attached to INT0 depends on your board and BSP. All are doing is running some test programs (albeit complex test programs) using VxWorks on the PC. It appears that the system uses this for the system clock, and indeed this context is activated periodically for a short interval--presumably to handle the system clock code. However, periodically (although not regularly) instead of staying in this context for a very short time, it stays there for one full system clock cycle. Do you have any idea what could be causing this? Do I have the kernal configured incorrectly perhaps? Thanks! James McDonald --------------------------- Newsgroups: comp.os.vxworks Subject: VxWorks boot problem Date: 1 Aug 2001 14:49:02 -0700 From: gayathri_jay@yahoo.com (Gayathri) Organization: http://groups.google.com/ Message-ID: <16028583.0108011349.7f98463d@posting.google.com> Hi, I am trying to load Vxworks on a mtx604 board. I have a FTP server running on my NT machine. Boot is stuck while loading the image. I have attached my FTP log and the console info. console: [VxWorks Boot]: @ boot device : dc unit number : 0 processor number : 0 host name : pc217-41 file name : target\config\mtx604\vxWorks inet on ethernet (e) : 192.9.217.168 host inet (h) : 192.9.217.41 user (u) : gayathri ftp password (pw) : vxworks flags (f) : 0x0 Attached TCP/IP interface to dc0. Attaching network interface lo0... done. Loading... Ftp log: [L 0017] 08/01/01 14:38:43 Connection accepted from 192.9.217.168 [C 0017] 08/01/01 14:38:43 Command "USER gayathri" received [C 0017] 08/01/01 14:38:43 PASSword accepted [L 0017] 08/01/01 14:38:43 User gayathri logged in. [C 0017] 08/01/01 14:38:43 Command "TYPE I" received [C 0017] 08/01/01 14:38:43 TYPE set to I N [C 0017] 08/01/01 14:38:43 Command "PORT 192,9,217,168,4,1" received [C 0017] 08/01/01 14:38:43 PORT set to 192.9.217.168 - 1025 (4,1) [C 0017] 08/01/01 14:38:43 Command "RETR target\config\mtx604\vxWorks" received [C 0017] 08/01/01 14:38:43 RETRieve started on file target\config\mtx604\vxWorks [C 0017] 08/01/01 14:39:28 Command "PORT 192,9,217,168,4,2" received [C 0017] 08/01/01 14:39:28 PORT set to 192.9.217.168 - 1026 (4,2) [C 0017] 08/01/01 14:39:28 Command "RETR target\config\mtx604\vxWorks" received [C 0017] 08/01/01 14:39:28 RETRieve started on file target\config\mtx604\vxWorks Line "Command "RETR ...." repeats forever and the boot never completes. Could anybody please help me with this? Thanks in advance, Gayathri --------------------------- Newsgroups: comp.os.vxworks Subject: Re: what's the usage of ... appears in source code Date: 1 Aug 2001 21:55:25 GMT From: lied@nwsgpc.ih.lucent.com (Bob Lied) Organization: Lucent Technologies, Naperville, Illinois, USA Message-ID: <9k9tsd$i9d@nntpb.cb.lucent.com> References: <6603907b.0107312232.60dde0ef@posting.google.com> Reply-To: lied@lucent.com In article <6603907b.0107312232.60dde0ef@posting.google.com>, Baumann wrote: >anyone knows this usage? >retval = pciIntConnect((void(**)(...))INT_LVL_DOORBELL,(void(*)(...))Pci_DoorBell_ISR, >0); The three dots mean "takes a variable number of arguments". INT_LVL_DOORBELL is being cast to the address of a function pointer, pointing to a function that takes a variable number of arguments and returns void. Pci_DoorBell_ISR is being cast to a function pointer, pointing to a similar function. This is why typedef was invented. __ Bob Lied --------------------------- Newsgroups: comp.os.vxworks Subject: Re: INT0 as displayed by windview Date: Thu, 2 Aug 2001 00:01:34 +0200 From: "Werner Schiendl" Message-ID: <3b687ce4$0$28950$6e365a64@newsreader02.highway.telekom.at> References: <3B684344.8B8027D5@yahoo.com> <3B686D69.EAE93C32@yahoo.com> Hi, > > All are doing is running some test programs (albeit complex test > programs) using VxWorks on the PC. > INT 0 on a standard PC architecture is the system clock interrupt, as you stated correctly. > It appears that the system uses this for the system clock, and indeed > this context is activated periodically for a short interval--presumably > to handle the system clock code. However, periodically (although not > regularly) instead of staying in this context for a very short time, it > stays there for one full system clock cycle. > I would guess this is simple an error in WindView's display. Maybe the event got dropped because the buffer was full. Maybe it is an problem in the implementation of WindView to get the context switches correctly. I don't believe that the system locks up in the system clock interrupt for a whole cycle. Since this is a 1/60 Second on that architecture by default, that would certainly have ill effects and would have been found I guess. Did you contact WRS Support about the issue? > Do you have any idea what could be causing this? Do I have the kernal > configured incorrectly perhaps? > No, don't think you can mis-configure that. The only thing that would cause such behavior would be if you added code or changed the system clock interrupt otherwise. hth Werner --------------------------- Newsgroups: comp.os.vxworks Subject: Re: INT0 as displayed by windview Date: 01 Aug 2001 18:15:27 -0400 From: Bill Pringlemeir Organization: Bell Sympatico Message-ID: References: <3B684344.8B8027D5@yahoo.com> <3B686D69.EAE93C32@yahoo.com> Sender: bpringlemeir@DeadDuck >>>>> "James" == James D McDonald writes: [snip] James> It appears that the system uses this for the system clock, and James> indeed this context is activated periodically for a short James> interval--presumably to handle the system clock code. James> However, periodically (although not regularly) instead of James> staying in this context for a very short time, it stays there James> for one full system clock cycle. James> Do you have any idea what could be causing this? Do I have James> the kernal configured incorrectly perhaps? If you aren't running a `high speed timer' then your WindView display will just be a synthetic time value. It is not really a tick. It is just the time between events, all given equal spacing. Also, if you have timeout callbacks of some kind, they maybe running in the context of this interrupt. If you are running this in a simulator, then it might be multi-tasking or something else stealing time... a WindView display bug as Werner noted. hth, Bill Pringlemeir. - -- However, if you want the underlying mechanism, it isnt a semantic issue, because at some level, there is no more rug to shuffle things under. - Bilge vxWorks FAQ, "http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html" --------------------------- Newsgroups: comp.os.vxworks Subject: Exception Date: Thu, 2 Aug 2001 06:53:09 +0800 From: "Chet Wee" Organization: SCV Cablemodem News site Message-ID: <9ka1eq$aku$1@newsie.singa.pore.net> hi all, I am implementing the exception handling in C++ using try() & catch() mechanism Manage to compile and execute in MSC++ environment but when recompiled using Tornado, the try & catch mechanism did not catch the exception Please advise if I miss any setting for this Rgs --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Send a Ethernet frame with etherOutputHook Date: Wed, 01 Aug 2001 17:14:33 -0700 From: Anmol Organization: http://extra.newsguy.com Message-ID: <3B689B69.1F955CF2@eternal-systems.com> References: <3b6821e7$1_2@news.vo.lu> News wrote: > Hi, > > I have to code with VxWorks, But I don't know it. > I want to send a frame Ethernet using etherOutputHook mechanism. > Output()->ether_output()->TxStartup()->etherOutputHook). > But I don't understand how to implement this mechanism. > I just want to have an simple example. > > Regards, > Christophe. I'm assuming you know Ethernet and are only asking about vxWorks hooks so here goes. I used soWakeupHook once, so here is a general way to use hooks.. vxWorks has no processes, only threads so all memory is global. You would probably define a function frameEtherSend () and then do etherOutputHook = frameEtherSend Then each call to etherOutputHook would reference frameEtherSend, and your work is done --------------------------- Newsgroups: comp.os.vxworks Subject: screen saving Date: 1 Aug 2001 20:18:11 -0700 From: gcai@netease.com (cai) Organization: http://groups.google.com/ Message-ID: <31f1a7e2.0108011918.2468c29d@posting.google.com> I am programming on vxworks5.4 with windML2.0.Is there anyone can tell me how to save my screen(640X480,high color) image to memory, and how to restore it? I am appreciate for any help. thinks in advance. --------------------------- Newsgroups: comp.os.vxworks Subject: About RIPAuthHookAdd () cannot work? Date: 1 Aug 2001 21:07:05 -0700 From: jade_chen@21cn.com (shy) Organization: http://groups.google.com/ Message-ID: <937a20a8.0108012007.66c89fa2@posting.google.com> I use ripAuthHookAdd to see if the hook can take effect or not,but no effect. I use m2RipIfConfEntrySet to open the Authentication switch, which is the first condition the Hook function. what's the matter with RipAuthHookAdd()? --------------------------- Newsgroups: comp.os.vxworks Subject: graphics library / somebody already ported "VOGLE" to vxworks ? Date: Thu, 02 Aug 2001 08:06:45 +0200 From: Rainer Stelzer Organization: Stoye GmbH, =?iso-8859-1?Q?K=F6ln?= Message-ID: <3B68EDF5.6D456710@stoye.de> - --------------864BDBFDB71E6E09931C2D79 Content-Type: text/plain; charset=iso-8859-1 Content-Transfer-Encoding: 8bit Hi On my investigations for a tiny graphics library I found the VOGLE library in source format at ftp://munnari.oz.au/pub/graphics/vogle.tar.gz DESCRIPTION VOGLE is a library of C routines for doing line drawings and polygon fills in 2 and 3 Dimensions. It handles circles, curves, arcs, patches, polygons, and software text in a dev- ice independent fashion. Simple hidden line removal is also available via polygon backfacing. Access to hardware text and double buffering of drawings depends on the driver. Has somebody already ported "VOGLE" to vxworks ? regards Rainer Tornado vxworks ===================================================== Gebrüder Stoye GmbH department: traffic control systems name: Rainer Stelzer adr: Longericher Strasse 177 city: Cologne, Germany zip: 50739 phone: ++49/221/1707019 fax: ++49/221/1707010 mailto: r.stelzer@stoye.de www: http://www.stoye.de ===================================================== - --------------864BDBFDB71E6E09931C2D79 Content-Type: text/html; charset=us-ascii Content-Transfer-Encoding: 7bit Hi

On my investigations for a tiny graphics library I found the VOGLE
library in source format at

 ftp://munnari.oz.au/pub/graphics/vogle.tar.gz

DESCRIPTION
     VOGLE is a library of C routines for doing line drawings and
     polygon  fills  in  2  and 3 Dimensions. It handles circles,
     curves, arcs, patches, polygons, and software text in a dev-
     ice  independent fashion. Simple hidden line removal is also
     available via polygon backfacing. Access  to  hardware  text
     and  double  buffering  of  drawings  depends on the driver.
 

Has somebody already ported "VOGLE" to vxworks ?

regards

Rainer

Tornado vxworks
=====================================================
            Gebrüder Stoye GmbH
department: traffic control systems
      name: Rainer Stelzer
       adr: Longericher Strasse 177
      city: Cologne, Germany
       zip: 50739
     phone: ++49/221/1707019
       fax: ++49/221/1707010
    mailto: r.stelzer@stoye.de
       www: http://www.stoye.de
=====================================================
  - --------------864BDBFDB71E6E09931C2D79-- --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Puzzled by PPC assembler for varargs function Date: Thu, 02 Aug 2001 07:21:19 +0100 From: Hugh Duggan Organization: Lucent Technologies Message-ID: <3B68F15F.916B411C@lucent.com> References: <3B66C3D0.2BCC6823@lucent.com> Thanks for this explanation - but it just begs another question. If bit 6/CR1 set denotes that the current process is using FP instructions, how is this bit set? Does calling any FP instruction set this bit? Or is it up to the FP user to set it explicitly? The bug we found before (on a different PPC board) revolved about some code compiled without "-mstrict-align". Apparently under this regime the compiler used some FP registers to move 64-bit structures within a task. However these registers were not preserved by an ISR, resulting in a big headache for the task. The code below *was* compiled with "-mstrict-align", so either this flag doesn't have the control over the earlier bug that we thought it had, or we are looking at a brand new bug with similar symptoms. Once again, thanks for the assistance. Hugh David Given wrote: > > In article <3B66C3D0.2BCC6823@lucent.com>, > Hugh Duggan writes: > [...] > > I'm puzzled mainly about the branch in the function prologue (line 3455). > > Why is there a conditional statement during the setup of the stack frame > > and allocation of parameters? What does the condition depend upon? Why > > is it OK in some situations to simply skip over the conditional block? > [...] > > 3455 14cc 40860028 bc 4,6,.L1197 > > 3456 14d0 D8210028 stfd 1,40(1) > > 3457 14d4 D8410030 stfd 2,48(1) > > 3458 14d8 D8610038 stfd 3,56(1) > > 3459 14dc D8810040 stfd 4,64(1) > > ... > > 3465 .L1197: > > Bit 6 is in CR1, which is used to store status about the floating-point > unit. I'd hazard a guess that what it's doing is checking to see whether > the current process is using the floating point registers, and only saving > them if needed. This would save a lot of time in the common case of a > process that doesn't use the FPU. > > David Given > Tao Group Ltd. - ----------------------------------------------------------- Hugh Duggan | Home Office : +44 117 9626702 Lucent Technologies | Swindon : +44 1793 887278 Chippenham, UNITED KINGDOM | E-mail : hduggan@lucent.com --------------------------- Newsgroups: comp.os.vxworks Subject: How to send a Ethernet Frame Date: Thu, 2 Aug 2001 09:18:22 +0200 From: "News" Message-ID: <3b68fdd8$1_2@news.vo.lu> Hi, I have to code with VxWorks, But I don't know it. I want to send a frame Ethernet using etherOutputHook mechanism. Output()->ether_output()->TxStartup()->etherOutputHook). But I don't understand how to implement this mechanism. I just want to have an simple example. Regards, Christophe. --------------------------- Newsgroups: comp.os.vxworks Subject: why ripAuthHookAdd() cannot work ? Date: 2 Aug 2001 00:45:32 -0700 From: jade_chen@21cn.com (shy) Organization: http://groups.google.com/ Message-ID: <937a20a8.0108012345.75c32222@posting.google.com> I use ripAuthHookAdd to see if the hook can take effect or not,but no effect. I use m2RipIfConfEntrySet to open the Authentication switch, which is the first condition the Hook function. what's the matter with RipAuthHookAdd()? --------------------------- Newsgroups: comp.os.vxworks Subject: STL Support Date: 02 Aug 2001 08:22:41 GMT From: mhartman@grintek.com (Morne Hartman) Organization: Avitronics Maritime Message-ID: <3b690dd1$0$228@hades.is.co.za> Hi all, I am using Tornado 2 for our development. I have installed the second patch to be able to use the STL. However I have descovered that I can not use auto_ptrs. This is a great disadvantage to me. It seems that Tornado and VxWorks are a bit slow to support the latest c++ standards. Does anyone have a reliable solution of how to upgrade Tornado to use the latest stl facilities? Many Thanks Morne Hartman mhartman@grintek.com --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Exception Date: Thu, 2 Aug 2001 10:49:24 +0200 From: "Morne Hartman" Message-ID: <3b69140f$0$228@hades.is.co.za> References: <9ka1eq$aku$1@newsie.singa.pore.net> Hi, I have been able to implement exception handling in VxWorks, could you please provide some code to see what you are doing. Morne Hartman mhartman@grintek.com Chet Wee wrote in message news:9ka1eq$aku$1@newsie.singa.pore.net... > hi all, > > I am implementing the exception handling in C++ using try() & catch() > mechanism > > Manage to compile and execute in MSC++ environment but when recompiled using > Tornado, the try & catch mechanism did not catch the exception > > Please advise if I miss any setting for this > > Rgs > > --------------------------- Newsgroups: comp.os.vxworks Subject: Why WindView shows that VxWorks spent most time in ISR? Date: 2 Aug 2001 02:22:39 -0700 From: people@micro-web.co.kr (JinSoo Kim) Organization: http://groups.google.com/ Message-ID: Hi, I just tested WindView in Tornaod 2.0. I launched standard simulator(vxsim) and launched WindView. I downloaded no program to vxsim. (The only tasks running on vxsim are tExtTask, tLogTask, tWdbTask, tWvRBuffMgr, idle tasks.) I gathered data and uploaded it. The graph shows that vxworks spent most of time in INT1 ISR other than idle task. I think that vxworks should spent most of time in idle task if there are no user tasks running. Would you please let me know what's wrong with WindView? Or is there anything I missed? Thanks in advance... to your answer. Best Regards, JinSoo Kim --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Project tool and clearcase fouls up Date: Thu, 02 Aug 2001 10:26:09 +0100 From: Owain Phillips Organization: Siemens Inc. Message-ID: <3B691CB1.5FBB0EFF@siemenscomms.co.uk> References: <3B6830FE.5FD5869B@siemenscomms.co.uk> This is a multi-part message in MIME format. - --------------D6BD341E56BD5D143EE094D3 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Bill Pringlemeir wrote: > >>>>> "Owain" == Owain Phillips writes: > [snip] > Owain> Files such as sysLib.c are missing from the projects files, > Owain> therfore the project cannot be built. > > Owain> Anyone got a workaround? Ideas??? > > Are you on a Windows system? You might check that the project files > have been marked as binary [even though they are ASCII]. I think that > the are expected to be Unix end of line (single LF) and the CR/LF with > hose the tools up... just a guess. > Thanks for the suggestion Bill. I am working on a NT4 system. The BSP and project tool work when the BSP is located.... 1) local drive 2) remote samba mount and it fails if the BSP is located on a... 3) clearcase remote drive 4) nfs mount for each test I copied the base BSP to the filesystem under windows, then generated a new workspace and bootloadable project basing this on the BSP on the drive. In the failure cases the BSP wasn't parsed properly and not all core BSP files got included in the project. Ohhhh Dearrrrr Owain - --------------D6BD341E56BD5D143EE094D3 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 - --------------D6BD341E56BD5D143EE094D3-- --------------------------- End of New-News digest ********************** From vxwexplo-errs@csg.lbl.gov Thu Aug 2 06:16:56 2001 From: fscheurer@manz-acs.de Date: Thu Aug 2 06:16:59 PDT 2001 Subject: gcc 2.95.X Cygwin cross x86 - 'size_t' or 'operator new' problem! Hi-VxWorkers, (Please reply to my e-mail, since the rare answer get lost in the whole digest. I don't want to miss an important hint.) a few weeks ago I posted about my efforts to build a V. 2.95.3(4) gcc cygwin-based (WinNT 4) i386-wrs-vxworks. I use the contributions of Dave Korn and Jim Pisano, and want to thank them BTW. I got the hint of Dave attached below. I now had the time to try the whole thing once more. I didn't expect to : 1. I think I prepare the right options to the configure-process (I tried separate my Installation directory from the original T2-Directory which is $WIND_BASE_ORIG. $WIND_BASE/target is linked to $WIND_BASE_ORIG/target) export WIND_BASE=/Tornado2 export WIND_HOST_TYPE=winnt export VX_TARGET=i386-wrs-vxworks ./configure --prefix=$WIND_BASE/host/$WIND_HOST_TYPE --target=$VX_TARGET --with-gnu-as --with-gnu-ld --with-headers=$WIND_BASE/target/h --enable-languages=c,c++ -v 2>&1 | tee config.log 2. All patches are loaded and the fixinc should be prepared ... patching file `gcc/config/rs6000/rs6000.md' patching file `gcc/config/rs6000/sysv4.h' The next patch would create the file `gcc/config/rs6000/t-ppcgas-vx', which already exists! Assume -R? [n] y <-------------- Something double patching file `gcc/config/rs6000/t-ppcgas-vx' patching file `gcc/config/rs6000/vxppc.h' ... patching file `libstdc++/stl/stl_rope.h' patching file `libstdc++/stlinst.cc' AutoGen-ing inclhack.sh AutoGen-ing fixincl.x AutoGen-ing fixincl.sh .. 3. The headers are provided a) via the --with-headers option b) $WIND_BASE/host/$WIND_HOST_TYPE/$VX_TARGET/include points to $WIND_BASE/target/h c) $WIND_BASE/host/$WIND_HOST_TYPE/$VX_TARGET/sys-include points to $WIND_BASE_ORIG/host/x86-win32/i386-pc-mingw32/sys-include 4. I see that fixincl seems to be run although the binary fixincl fails due to a "hyphen"-sign used in a #define (The stdio-lock.h fix described in vx-ppc-gcc.diff is done!): .. In file included from ../../../gcc/fixinc/fixincl.c:140: ./../../gcc/fixinc/fixincl.x:3344: warning: missing white space after `#define apzVxworks_Stdio' ./../../gcc/fixinc/fixincl.x:3352: warning: missing white space after `#define VXWORKS_STDIO' ./../../gcc/fixinc/fixincl.x:3353: warning: missing white space after `#define VXWORKS_STDIO' ./../../gcc/fixinc/fixincl.x:3353: warning: `VXWORKS_STDIO' redefined ./../../gcc/fixinc/fixincl.x:3352: warning: this is the location of the previous definition In file included from ../../../gcc/fixinc/fixincl.c:140: ./../../gcc/fixinc/fixincl.x:3334: parse error before `-' ./../../gcc/fixinc/fixincl.x:3339: parse error before `-' ./../../gcc/fixinc/fixincl.x:3349: parse error before `-' ./../../gcc/fixinc/fixincl.x:3354: parse error before `-' ./../../gcc/fixinc/fixincl.x:3360: parse error before `-' ./../../gcc/fixinc/fixincl.x:4333: `zVxworks_Stdio' undeclared here (not in a function) ./../../gcc/fixinc/fixincl.x:4333: `Lock_Cxx_CompatName' undeclared here (not in a function) ./../../gcc/fixinc/fixincl.x:4333: initializer element is not constant ./../../gcc/fixinc/fixincl.x:4333: (near initialization for `fixDescList[91].fix_name') ./../../gcc/fixinc/fixincl.x:4333: `zVxworks_Stdio' undeclared here (not in a function) ./../../gcc/fixinc/fixincl.x:4333: `Lock_Cxx_CompatList' undeclared here (not in a function) ./../../gcc/fixinc/fixincl.x:4333: initializer element is not constant ./../../gcc/fixinc/fixincl.x:4333: (near initialization for `fixDescList[91].file_list') ./../../gcc/fixinc/fixincl.x:4334: parse error before `const' ./../../gcc/fixinc/fixincl.x:4334: warning: initialization from incompatible pointer type ./../../gcc/fixinc/fixincl.x:4335: `LOCK_CXX_COMPAT_RE_CT' undeclared here (not in a function) ./../../gcc/fixinc/fixincl.x:4335: initializer element is not constant ./../../gcc/fixinc/fixincl.x:4335: (near initialization for `fixDescList[91].unused') ./../../gcc/fixinc/fixincl.x:4335: parse error before `1' ./../../gcc/fixinc/fixincl.x:4336: `aVxworks_Stdio' undeclared here (not in a function) ./../../gcc/fixinc/fixincl.x:4336: `Lock_Cxx_CompatTests' undeclared here (not in a function) ./../../gcc/fixinc/fixincl.x:4336: initializer element is not constant ./../../gcc/fixinc/fixincl.x:4336: (near initialization for `fixDescList[91].fd_flags') ./../../gcc/fixinc/fixincl.x:4336: parse error before `const' ./../../gcc/fixinc/fixincl.c: In function `initialize': ./../../gcc/fixinc/fixincl.c:349: `SIGIOT' undeclared (first use in this function) ./../../gcc/fixinc/fixincl.c:349: (Each undeclared identifier is reported only once ./../../gcc/fixinc/fixincl.c:349: for each function it appears in.) make[2]: *** [fixincl.o] Error 1 make[2]: Leaving directory `/GNU_CrossGCC_Cygwin/DaveK_JimPisano/build/gcc-2.95.3/build/gcc/fixinc' Could not install binary fixincludes. Installing shell script instead. .. I removed the "Copied" label-file in $WIND_BASE/host/$WIND_HOST_TYPE/$VX_TARGET/sys-include to be sure that all headers are transfered. I wonder why the "arch/i86/*h" files aren't fixed in the fixincl-Procedure and why the rooted link (attached to directory gcc-2.95.3/build/gcc/include/root) of i86 doesn't cover all of the headers originally in arch/i86: .. Making symbolic directory links ./arch/i86.lnk -> root/Tornado_2.0/target/h/arch/i86 : Making rooted link <------SHOULD I AVOID THIS??? ./arch/i86.sim.lnk -> root/Tornado_2.0/target/h/arch/i86.sim : Making rooted link Fixing directory /Tornado_2.0/host/x86-win32/i386-pc-mingw32/sys-include into /GNU_CrossGCC_Cygwin/DaveK_JimPisano/build/gcc-2.95.3/ build/gcc/include Checking header files Fixed ./aio.h: machine_name Fixed ./aioSysDrv.h: machine_name Fixed ./algo.h: no_double_slash machine_name Fixed ./algobase.h: no_double_slash machine_name Fixed ./alloc.h: no_double_slash machine_name Fixed ./arch/simnt/dsmSimntLib.h: machine_name Fixed ./arch/simnt/excSimntLib.h: machine_name Fixed ./arch/simnt/fppSimntLib.h: machine_name Fixed ./arch/simnt/ntcontext.h: machine_name Fixed ./arch/simnt/stdargGnu.h: machine_name Fixed ./arch/simnt/va_simnt.h: machine_name Fixed ./arch/simnt/_G_configArch.h: machine_name Fixed ./arpLib.h: machine_name Fixed ./asm.h: machine_name Fixed ./assert.h: machine_name .. 5. Finally I run into another "size_t"-problem: .. make[2]: Entering directory `/GNU_CrossGCC_Cygwin/DaveK_JimPisano/build/gcc-2.95.3/build/gcc' CONFIG_HEADERS=auto-host.h:config.in LANGUAGES="c++" /bin/sh config.status creating Makefile creating intl/Makefile creating po/Makefile.in creating fixinc/Makefile creating cp/Makefile creating auto-host.h auto-host.h is unchanged linking ../../gcc/intl/libgettext.h to intl/libintl.h creating libintl.h /GNU_CrossGCC_Cygwin/DaveK_JimPisano/build/gcc-2.95.3/build/gcc/xgcc -B/GNU_CrossGCC_Cygwin/DaveK_JimPisano/build/gcc-2.95.3/build/g cc/ -B/Tornado2/host/winnt/i386-wrs-vxworks/bin/ -I/Tornado2/host/winnt/i386-wrs-vxworks/include -O2 -DCROSS_COMPILE -DIN_GCC - g -O2 -I./include -g1 -DHAVE_GTHR_DEFAULT -DIN_LIBGCC2 -D__GCC_FLOAT_NOT_NEEDED -g -O2 -I. -I../../gcc -I../../gcc/config -I../. ./gcc/../include \ -c ../../gcc/cp/tinfo.cc In file included from ../../gcc/cp/tinfo.cc:32: /Tornado2/host/winnt/i386-wrs-vxworks/include/new:32: `operator new' takes type `size_t' as first parameter /Tornado2/host/winnt/i386-wrs-vxworks/include/new:33: `operator new' takes type `size_t' as first parameter /Tornado2/host/winnt/i386-wrs-vxworks/include/new:36: `operator new' takes type `size_t' as first parameter /Tornado2/host/winnt/i386-wrs-vxworks/include/new:37: `operator new' takes type `size_t' as first parameter /Tornado2/host/winnt/i386-wrs-vxworks/include/new:42: `operator new' takes type `size_t' as first parameter /Tornado2/host/winnt/i386-wrs-vxworks/include/new:43: `operator new' takes type `size_t' as first parameter ./../gcc/cp/tinfo.cc:45: redefinition of `bool type_info::operator ==(const type_info &) const' /Tornado2/host/winnt/i386-wrs-vxworks/include/typeinfo:43: `bool type_info::operator ==(const type_info &) const' previously defined here make[2]: *** [tinfo.o] Error 1 make[2]: Leaving directory `/GNU_CrossGCC_Cygwin/DaveK_JimPisano/build/gcc-2.95.3/build/gcc' make[1]: *** [libgcc2.a] Error 1 make[1]: Leaving directory `/GNU_CrossGCC_Cygwin/DaveK_JimPisano/build/gcc-2.95.3/build/gcc' make: *** [all-gcc] Error 2 .. Still one of the headers not fixed? Any hints appreciated! Apologies for the long and detailed posting. How can we shed light on this without going into details. TIA MfG / Kind Regards Friedrich Scheurer, R&D -------------------------------- acs GmbH Steigaeckerstrasse 13 72768 Reutlingen Telefon (07121) 9019-23 Telefax (07121) 9019-98 Internet www.manz-acs.de e-mail: fscheurer@manz-acs.de -------------------------------- -------- Original Message -------- > Newsgroups: comp.os.vxworks > Subject: Re: Anyone any experiences building gcc 2.95.X Cygwin cross x86 > Date: Wed, 11 Jul 2001 16:51:00 +0100 > From: "Dave Korn" > Organization: Lumber Cartel (tinlc) Members #2234-2237 (owing to browser > refresh) > Message-ID: <8P%27.104$os1.458451@newsr2.u-net.net> > References: <9i145t$65d$1@overload.lbl.gov> > > fscheurer@manz-acs.de wrote in message <9i145t$65d$1@overload.lbl.gov>... > >Hi VxWorks-Gurus, > > > >I ran into the problem, that the integrated target-headers > >seem to be not confor with GNU expectations. > > Did you use the --with-headers= flag when configuring to supply the > target > headers, or did you copy them into $prefix/$target/sys-include, or into > $prefix/$target/include ? > > The first two of those methods will cause the Gcc build process to invoke > the 'fixincs' script, which runs through all the supplied headers, checking > for and fixing things that need to be made GNU-compatible. > > Although it is recommended in many pieces of documentation that you > install target headers into $prefix/$target/include when building a cross > compiler, doing so prevents the fixincs from being run: fixincs will only > ever process files from the sys-include directory. > > The error messages you posted look a lot like fixincs hasn't run. Can > you > confirm or deny this? > > >MY QUESTION: > >Where is the most central place in > >- the sources, > >- the configure-script > >or > >- the Makefile > > > >to "repair" it > >AND to get the right types to be compatible to objects build with > >older gcc-version? > > I suspect that the fixes already exist in the fixincs script, and merely > haven't been applied to your target headers in the course of building the > cross-compiler. However, the VxWorks targets for Gcc haven't been well > maintained of late, and it is possible that the WRS headers have changed in > such a way that the fixincs test-pattern no longer matches the WRS header > file, causing a fix to be missed; or that new code in the headers mean that > new fixes need to be added. Give it another try, being sure to use one of > the first two methods I mentioned above to supply the headers, and tell us > if it works any better. > > DaveK From vxwexplo-errs@csg.lbl.gov Fri Aug 3 04:03:41 2001 From: Vxworks Exploder Date: Fri Aug 3 04:03:44 PDT 2001 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Fri Aug 3 04:03:12 PDT 2001 Subject: INTEGRITY & VxWorks Subject: Re: prjConfig.c in BSP for Power PC Subject: WindView/TornadoII/Solaris Subject: Re: Exception Subject: Re: screen saving Subject: Re: Puzzled by PPC assembler for varargs function Subject: Re: Puzzled by PPC assembler for varargs function Subject: Re: vxWork hang up Subject: Cannot interrupt VxWorks target shell with Control-C Subject: RE: C++ and code size ... Subject: Re: Puzzled by PPC assembler for varargs function Subject: Re: Puzzled by PPC assembler for varargs function Subject: Re: STL Support Subject: Re: Why WindView shows that VxWorks spent most time in ISR? Subject: Re: Project tool and clearcase fouls up Subject: networking issues Subject: Re: INTEGRITY & VxWorks Subject: Re: problem with muxDevUnload Subject: Replacing CplusLib operator new() etc. Subject: Re: Why WindView shows that VxWorks spent most time in ISR? Subject: TCB in CrossWind Subject: Re: Replacing CplusLib operator new() etc. Subject: Re: Exception Subject: pSOS and vxworks Subject: Re: pSOS and vxworks Subject: Re: C++ and code size ... Subject: Re: vxWork hang up Subject: Re: Puzzled by PPC assembler for varargs function Subject: Re: networking issues Subject: Problem in starting Profilescope Subject: Re: Project tool and clearcase fouls up Subject: Re: Problem in scaning user input ! Subject: Re: Writing host files under serial/TSFS Subject: Re: How to send a Ethernet Frame Subject: Re: vxWork hang up ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: INTEGRITY & VxWorks Date: Thu, 2 Aug 2001 08:11:37 -0400 From: "Jizhong" Organization: Posted via Supernews, http://www.supernews.com Message-ID: Hello, has anybody compared green hills INTEGRITY with VxWorks/VxVMI. Runtime performance? How easy to move applications between integrity and vxworks. JL --------------------------- Newsgroups: comp.os.vxworks Subject: Re: prjConfig.c in BSP for Power PC Date: 2 Aug 2001 07:55:24 -0700 From: Dave.Sanderson@transmitton.co.uk (Dave Sanderson) Organization: http://groups.google.com/ Message-ID: References: <3b659edb.851409982@news.demon.co.uk> <3B66CCB3.3B4B4C6E@siemenscomms.co.uk> I am having a similar problem after installing the 2.0.2 patch. I found a workaround on Windsurf that involves 'make'ing the files related to the newly integrated NPT that is part of the patch. I dont have these files. We are using Intel StrongARM 1110, with a modified brutus BSP. Any ideas? Thanks in advance Dave Owain Phillips wrote in message news:<3B66CCB3.3B4B4C6E@siemenscomms.co.uk>... > I think we had this one.... > > We built a BSP for tornado 2.0, then got patched up to 2.0.2 > and found the BSP wouldn't build. > > The problem was that our BSP used the architecture library > librc32364sfgnu, which was not upgraded; the standard R3K, R4K > libraries were upgraded by the patches....consequence unresloved > externals. > > We have been told our problems will be solved when we get a > new architecture library which comes as standard with tornado > 2.1 for mips.....we are waiting fo this. > > Try looking on windsurf or contacting WRS. > > Regards, > Owain Phillips > > Jan Larcombe wrote: > > > I'm having problems building my BSP, I end up with an unresolved > > external muxMaxBinds. I've checked why this is and it seems to be > > because the prjConfig.c which is generated is calling the routine > > usrMuxLibInit. In earlier versons of my BSP prjConfig.c was calling > > MuxLibInit instead. > > Obviously there is no point in ediiting prjConfig.c but does anyone > > know how to change configurations so that the prjConfig.c file which > > is automatically generated will call the correct function. > > > > Thanks in advance > > -- --------------------------- Newsgroups: comp.os.vxworks Subject: WindView/TornadoII/Solaris Date: Thu, 02 Aug 2001 10:32:37 -0500 From: Dinker Charak Organization: Fermi National Accelerator Laboratory Message-ID: <3B697295.DE08B1F6@fnal.gov> Hello, I am unable to start WindView. In my Workspace Window, in VxWorks Tab, I did a right click on WindView and Included it. It showed a box in which these componets were selected rBuff library windview library windview class instrumentation file upload path initialization TSFS upload path initialization sequential timestamping I did ok ... a rebuild and reboot. On starting WindView, I click on Launch WindView button and anoterh window open in which it gives me a list of tasks in which Context switch is selected. i do an ok. but then i get the message that target does not support WindView. Any idea what else needs to be done? tia, dinker --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Exception Date: 02 Aug 2001 11:56:21 -0400 From: Bill Pringlemeir Organization: Bell Sympatico Message-ID: References: <9ka1eq$aku$1@newsie.singa.pore.net> <3b69140f$0$228@hades.is.co.za> Sender: bpringlemeir@DeadDuck >>> Chet Wee wrote: >> I am implementing the exception handling in C++ using try() & >> catch() mechanism >> Manage to compile and execute in MSC++ environment but when >> recompiled using Tornado, the try & catch mechanism did not catch >> the exception >>>>> "Morne" == Morne Hartman writes: Morne> Hi, I have been able to implement exception handling in Morne> VxWorks, could you please provide some code to see what you Morne> are doing. My experience is that exceptions and templates do not work as most C++ programmers might expect with vxWorks. The also tend to generate larger code bases. We have disabled support in our projects. We do use virtual functions, multiple inheritance and RTTI without problems. I believe that embedded C++ disables both templates and exceptions. Anyways, I won't analyze whether you should be using them or not, but they do seem to be problematic. I have used `setjmp' and `longjmp' to achieve similar effects without the stack penalties incurred by C++ exceptions. If this is your own code base, you might consider this as an alternative. hth, Bill Pringlemeir. - -- SDI Khaddafi munitions NORAD counter-intelligence plutonium kibo KGB explosion Ortega Nazi fissionable CIA domestic disruption FBI vxWorks FAQ, "http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html" --------------------------- Newsgroups: comp.os.vxworks Subject: Re: screen saving Date: 02 Aug 2001 12:07:32 -0400 From: Bill Pringlemeir Organization: Bell Sympatico Message-ID: References: <31f1a7e2.0108011918.2468c29d@posting.google.com> Sender: bpringlemeir@DeadDuck >>>>> "cai" == cai writes: cai> I am programming on vxworks5.4 with windML2.0.Is there anyone cai> can tell me how to save my screen(640X480,high color) image to cai> memory, and how to restore it? I am appreciate for any help. cai> thinks in advance. This works for me [change SCREEN_BASE define to match your video memory], #include #include #include #include #include #include #define SCREEN_BASE 0xc0000000 #define WIDTH 640 #define HEIGHT 480 void screenDump(unsigned char *name) { FILE *fp; if(name == NULL || name[0] == 0) name = "screen.out"; fp = fopen(name,"w+"); fwrite((char*)SCREEN_BASE, sizeof(char),WIDTH*HEIGHT*4,fp); fclose(fp); } void screenPump(unsigned char *name) { FILE *fp; if(name == NULL || name[0] == 0) name = "screen.out"; fp = fopen(name,"r"); fread((char*)SCREEN_BASE, sizeof(char),WIDTH*HEIGHT*4,fp); fclose(fp); } I call these from the shell. hth, Bill Pringlemeir. - -- Loaning money to your spouse will bring dividends. Do something nice for your parent tomorrow. Aliens from Mars will piss on the bed this month. You'll fall into a state of lossage, but eventually be gay. vxWorks FAQ, "http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html" --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Puzzled by PPC assembler for varargs function Date: Thu, 2 Aug 2001 16:00:25 +0100 From: vxworksintent@tao-group.com (David Given) Organization: Tao Group Ltd. Message-ID: <9upbk9.u77.ln@127.0.0.1> References: <3B66C3D0.2BCC6823@lucent.com> <3B68F15F.916B411C@lucent.com> Sender: dg@pearl.tao.co.uk In article <3B68F15F.916B411C@lucent.com>, Hugh Duggan writes: > Thanks for this explanation - but it just begs another question. If > bit 6/CR1 set denotes that the current process is using FP instructions, > how is this bit set? Does calling any FP instruction set this bit? Or > is it up to the FP user to set it explicitly? Disclaimer: I don't have a PowerPC. ...but there's a flag, VX_FP_TASK, that you pass to taskSpawn() that tells VxWorks to allow floating-point support for that task. Presumably, if this is set, VxWorks will make sure to save the FP register as needed. (This is certainly how other operating systems work.) > The bug we found before (on a different PPC board) revolved about some > code compiled without "-mstrict-align". Apparently under this regime the > compiler used some FP registers to move 64-bit structures within a task. > However these registers were not preserved by an ISR, resulting in a big > headache for the task. Quite likely, setting VX_FP_TASK would cause the ISR to save them. (Unless there was a bug in VxWorks itself --- and what do you think the chances of that are?) - -- David Given Tao Group Ltd. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Puzzled by PPC assembler for varargs function Date: Thu, 2 Aug 2001 16:53:57 +0100 From: "Dave Korn" Organization: Lumber Cartel (tinlc) Members #2234-2237 (owing to browser refresh) Message-ID: References: <3B66C3D0.2BCC6823@lucent.com> <3B68F15F.916B411C@lucent.com> Hugh Duggan wrote in message <3B68F15F.916B411C@lucent.com>... >Thanks for this explanation - but it just begs another question. If >bit 6/CR1 set denotes that the current process is using FP instructions, >how is this bit set? Does calling any FP instruction set this bit? Or >is it up to the FP user to set it explicitly? Strictly speaking, the meaning of the flag is not that 'the process is using FP instructions', but that "This particular function call to this particular varargs function has passed one or more arguments in the FP registers". The compiler adds the instruction to set or clear this bit when it sees a call to a function that it knows (from having seen the prototype) is a varargs function. It's not done on a per-process basis, but on a per-function-call basis when the code is compiled. This behaviour is part of the EABI standard (Sun SysV ABI PowerPC processor supplement, Sep. 95 edn, p3-21), and it *only* applies to varargs functions. This also means that if you call a varargs function that hasn't been prototyped, the compiler won't know it's a varargs function, and so won't set the bit appropriately. This in turn means that the varargs function won't save the FP regs on entry, which means that they may be corrupted or the FP values obtained when calling va_arg may be wrong. >The bug we found before (on a different PPC board) revolved about some >code compiled without "-mstrict-align". Apparently under this regime the >compiler used some FP registers to move 64-bit structures within a task. >However these registers were not preserved by an ISR, resulting in a big >headache for the task. If you want to leave out the VXWORKS_FP_TASK flag when you spawn the task and save the overhead of saving and restoring the FP regs when context-switching, it isn't sufficient to just not use floating point data types in your C code: you need to also use the -msoft-float argument to prevent the compiler from allocating the FP regs to store emporaries. -mstrict-align isn't enough on its own, since if the compiler knows that the data in question *is* correctly aligned (8 bytes) there's no reason (as far as it knows) why not to use the FP regs. Telling it that there is no FP hardware in the CPU is the only way to tell it not to use the FP regs. DaveK - -- Last time I checked, the UN Convention of Human Rights didn't say anything about the right to run a bot on Usenet. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: vxWork hang up Date: Thu, 2 Aug 2001 17:07:00 +0100 From: "Dave Korn" Organization: Lumber Cartel (tinlc) Members #2234-2237 (owing to browser refresh) Message-ID: References: <9k6a32$i3l$1@newsie.singa.pore.net> Chew GimHeng wrote in message <9k6a32$i3l$1@newsie.singa.pore.net>... >Hi all, > I had used the TCP/IP connection from a vxwork system to a Windows >sytem. The connection was used to send heart beat from the vxWork to the >Window System. But after approx. 10000 heart beat send, the whole system >hang. > >Will that cause by the watchdog routine which I used to call the heartbeat >routine very second? or is there any other way beside using watchdog for >peroidic sending of heartbeat? Aaaaaaaaaarrrrrggghh !! Do you mean you're calling a network send routine from a watchdog callback function ? From VxWorks Reference Manual : Libraries : wdLib " Note that the timeout routine is invoked at interrupt level, rather than in the context of the task. Thus, there are restrictions on what the routine may do. Watchdog routines are constrained to the same rules as interrupt service routines. For example, they may not take semaphores, issue other calls that may block, or use I/O system routines like printf( ). " From VxWorks Programmer's Guide 5.4, Edition 1, Chapter 2, para 2.5 Interrupt Service Code, section 2.5.3: Special Limitations of ISRs " Table 2-23: Routines that Can Be Called by Interrupt Service Routines - ---------------------------------------------------------------------------- - ---- Library Routines - ---------------------------------------------------------------------------- - ---- bLib All routines errnoLib errnoGet( ), errnoSet( ) fppArchLib fppSave( ), fppRestore( ) intLib intContext( ), intCount( ), intVecSet( ), intVecGet( ) intArchLib intLock( ), intUnlock( ) logLib logMsg( ) lstLib All routines except lstFree( ) mathALib All routines, if fppSave( )/fppRestore( ) are used msgQLib msgQSend( ) pipeDrv write( ) rngLib All routines except rngCreate( ) and rngDelete( ) selectLib selWakeup( ), selWakeupAll( ) semLib semGive( ) except mutual-exclusion semaphores, semFlush( ) sigLib kill( ) taskLib taskSuspend( ), taskResume( ), taskPrioritySet( ), taskPriorityGet( ), taskIdVerify( ), taskIdDefault( ), taskIsReady( ), taskIsSuspended( ), taskTcb( ) tickLib tickAnnounce( ), tickSet( ), tickGet( ) tyLib tyIRd( ), tyITx( ) vxLib vxTas( ), vxMemProbe( ) wdLib wdStart( ), wdCancel( ) " You need to restructure your heartbeat: the watchdog ISR should do nothing except give a mutex semaphore. You should also have a task running at a very high priority that waits on this semaphore, and each time it receives the semaphore from the watchdog ISR it will wake up; then it can send one network packet (from a task context!) and go back to waiting on the semaphore for next time. DaveK - -- Last time I checked, the UN Convention of Human Rights didn't say anything about the right to run a bot on Usenet. --------------------------- Newsgroups: comp.os.vxworks Subject: Cannot interrupt VxWorks target shell with Control-C Date: 2 Aug 2001 10:01:11 -0700 From: chuck.rosselli@latitude.com (Chuck Rosselli) Organization: http://groups.google.com/ Message-ID: In my application, I have redirected I/O to the VxWorks target shell to use a serial I/O system that I have developed. Basically, everything works fine. The commands I previously entered to the target shell via telnet access work identical with the serial I/O system I have developed. Everything EXCEPT Control-C (which I need for interrupting my application in continiuous output mode). When I connect to the target shell with telnet, a Control-C correctly restarts the target shell with the following message: cBIZ(m0 p0)> 60910 vxTaskEntry +28 : shell () abf7c shell +14c: abfbc () ac0d4 shell +2a4: ledRead () dadd8 ledRead +168: read () 89d38 read +c : iosRead () 9d294 iosRead +a4 : be068 () be074 ptyDevCreate +18c: semQPut () tShell restarted. cBIZ(m0 p0)> When I connect to the target shell with my own serial I/O package, a Control-C essentially does nothing and produces the following message at an input prompt: cBIZ(m0 p0)> ? syntax error cBIZ(m0 p0)> I have verified that I am doing the following: As per the VxWorks programmer's guide (section 9.2.4 - Aborting the Target Shell): a. dbgInit() is called when my application initializes. b. excTask() is initialized & is running in the task list. c. My driver does support Control-C by passing it through to the target shell. (I log every Control-C read event and I see this logged). d. tyAbortSet() is set to respond to the default Control-C event. To be sure, I even tried to redundantly set tyAbortSet() to Control-C again in case it was not defaulting to Control-C. (No effect). e. My device's abort option is enabled as per the following code snippet and I get a positive config completion message (no error occurs). /* Step 4: Enable our favorite serial port // options via an ioctl() call. */ VtermIoctlSetoptions = OPT_LINE /* Select line mode of operation. */ | OPT_ECHO /* Echo input chars to the output. */ | OPT_CRMOD /* Input CR=>LF translation, Output LF=>CR/LF translation. */ | OPT_TANDEM /* Allow X-on/X-off via CTRL-Q & S */ | OPT_7_BIT /* Don't let special chars through. */ | OPT_MON_TRAP /* Enable Ctrl-X to restart VxWorks.*/ | OPT_ABORT; /* Enable Ctrl-C to break shell command execution. */ ret_code = ioctl( VtermIoFileDescriptorNum, FIOSETOPTIONS, VtermIoctlSetoptions ); if( ret_code == ERROR ) { LOGERR( SEV_MAJOR, "Error: could not set ioctl file options for device %d.", VtermIoFileDescriptorNum, 0, 0, 0 ); return( ERROR ); } else { LOGERR( SEV_INFO, "Configured I/O set options of 0x%04x for dev num: %d.", VtermIoctlSetoptions, VtermIoFileDescriptorNum, 0, 0 ); } Any attempt by me to intercept Control-C in my character read routine and restart the target shell (via either shellRestart() or taskRestart ) causes the target shell with my I/O system to hang. (A few garbage chars are printed but otherwise the target shell seems to hang). My questions are: 1. What do I need to do to get the target shell to respond to Control-C with my serial I/O system? 2. Why does the target shell respond to Control-C with the telnet I/O system but not my I/O system? I thought the target shell was the same with either I/O system and the issue was simply getting Control-C delivered to the target shell. (In fact, I can switch the I/O back and forth to the target shell in the same application session and Control-C works while I'm in telnet but does not work while I'm using my own I/O). Any help would be very much appreciated, Sincerely, Chuck Rosselli --------------------------- Newsgroups: comp.os.vxworks Subject: RE: C++ and code size ... Date: Thu, 2 Aug 2001 12:41:47 -0400 From: Todd Johnson Message-ID: <94BAF40039B0D411BC150002B315275A0CDB7B@mailserv.hatteras.com> Sender: Todd Johnson Thanks. I'll take a look. - -tdj > -----Original Message----- > From: Srikumar Natarajan > Sent: Thursday, August 02, 2001 11:38 AM > To: Nancy Adams; Todd Johnson > Subject: C++ and code size ... > > Hi, > > FYI. This is a discussion of code size when the C++ STL > (Standard Template Library) is used, in a statically linked > executable. Not exactly the same as ours, but very informative. > > sri > > http://groups.google.com/groups?hl=en&safe=off&th=84ff22c969b7 > 6b8c,12&start=0 > > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Puzzled by PPC assembler for varargs function Date: Thu, 02 Aug 2001 20:52:55 +0100 From: molochai Message-ID: <11e2a34a%molochai@vapour-trail.demon.co.uk> References: <3B66C3D0.2BCC6823@lucent.com> <3B68F15F.916B411C@lucent.com> <9upbk9.u77.ln@127.0.0.1> In message <9upbk9.u77.ln@127.0.0.1> vxworksintent@tao-group.com (David Given) wrote: > In article <3B68F15F.916B411C@lucent.com>, > Hugh Duggan writes: > > Thanks for this explanation - but it just begs another question. If > > bit 6/CR1 set denotes that the current process is using FP instructions, > > how is this bit set? Does calling any FP instruction set this bit? Or > > is it up to the FP user to set it explicitly? > > Disclaimer: I don't have a PowerPC. > > ...but there's a flag, VX_FP_TASK, that you pass to taskSpawn() that tells > VxWorks to allow floating-point support for that task. Presumably, if this > is set, VxWorks will make sure to save the FP register as needed. (This is > certainly how other operating systems work.) > > > The bug we found before (on a different PPC board) revolved about some > > code compiled without "-mstrict-align". Apparently under this regime the > > compiler used some FP registers to move 64-bit structures within a task. > > However these registers were not preserved by an ISR, resulting in a big > > headache for the task. > > Quite likely, setting VX_FP_TASK would cause the ISR to save them. (Unless > there was a bug in VxWorks itself --- and what do you think the chances of > that are?) No. ISTR that FP registers are *not* saved in an ISR. If you need FP in one, you save and restore registers yourself, and take suitable precautions against FP exceptions. If it did, the performance hit of saving and restoring 32 registers for *every* interrupt would be fairly unpleasent. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Puzzled by PPC assembler for varargs function Date: 2 Aug 2001 12:57:15 -0700 From: john_94501@yahoo.com (John) Organization: http://groups.google.com/ Message-ID: <488e459a.0108021157.6e136b3c@posting.google.com> References: <3B66C3D0.2BCC6823@lucent.com> <3B68F15F.916B411C@lucent.com> Hello Hugh, > Thanks for this explanation - but it just begs another question. If > bit 6/CR1 set denotes that the current process is using FP instructions, > how is this bit set? Does calling any FP instruction set this bit? Or > is it up to the FP user to set it explicitly? It should be set for all tasks running on hardware with an FP unit (i.e. anything using the PPC603 or PPC604 VxWorks CPU types, which includes all the newer 750 and 7400 series I believe). The VxWorks task initialisation sequence for PPC automatically sets this bit for CPUs with an FPU. Also, if you have floating point support on your hardware I would recommend using VX_FP_TASK for all your tasks too, just in case future changes to the code introduce anything that makes use of the FPU or its registers. That will make sure that the FP regs are correctly saved and restored on context switches. > The bug we found before (on a different PPC board) revolved about some > code compiled without "-mstrict-align". Apparently under this regime the > compiler used some FP registers to move 64-bit structures within a task. > However these registers were not preserved by an ISR, resulting in a big > headache for the task. Remember that there is a big difference between the prologues and the ISR code: that difference is that the prologue is generated by the compiler, the ISR code is not. So, if the compiler is planning on using the FP regs for something other than FP work, I would assume that it is aware of that, and will save and restore them as needed. Of course, if this kind of thing is going on, the VX_FP_TASK flag will be essential to make sure that the context switches save the FP regs too. > The code below *was* compiled with "-mstrict-align", so either this flag > doesn't have the control over the earlier bug that we thought it had, or > we are looking at a brand new bug with similar symptoms. All that has done is stop the use of the FP regs for non-FP operations. It doesn't stop their use for true floating point operations. My guess is that the compiler is generating a prologue that safe guards against possible FP usage ;-) Is there any use of the FP regs later in the function? HTH, John... > Once again, thanks for the assistance. > > Hugh > > > > David Given wrote: > > > > In article <3B66C3D0.2BCC6823@lucent.com>, > > Hugh Duggan writes: > > [...] > > > I'm puzzled mainly about the branch in the function prologue (line 3455). > > > Why is there a conditional statement during the setup of the stack frame > > > and allocation of parameters? What does the condition depend upon? Why > > > is it OK in some situations to simply skip over the conditional block? > [...] > > > 3455 14cc 40860028 bc 4,6,.L1197 > > > 3456 14d0 D8210028 stfd 1,40(1) > > > 3457 14d4 D8410030 stfd 2,48(1) > > > 3458 14d8 D8610038 stfd 3,56(1) > > > 3459 14dc D8810040 stfd 4,64(1) > > > ... > > > 3465 .L1197: > > > > Bit 6 is in CR1, which is used to store status about the floating-point > > unit. I'd hazard a guess that what it's doing is checking to see whether > > the current process is using the floating point registers, and only saving > > them if needed. This would save a lot of time in the common case of a > > process that doesn't use the FPU. > > > > David Given > > Tao Group Ltd. > > ----------------------------------------------------------- > Hugh Duggan | Home Office : +44 117 9626702 > Lucent Technologies | Swindon : +44 1793 887278 > Chippenham, UNITED KINGDOM | E-mail : hduggan@lucent.com --------------------------- Newsgroups: comp.os.vxworks Subject: Re: STL Support Date: Thu, 02 Aug 2001 21:10:35 +0100 From: molochai Message-ID: <189fe3a34a%molochai@vapour-trail.demon.co.uk> References: <3b690dd1$0$228@hades.is.co.za> In message <3b690dd1$0$228@hades.is.co.za> mhartman@grintek.com (Morne Hartman) wrote: > Hi all, > > I am using Tornado 2 for our development. I have installed the second patch > to be able to use the STL. However I have descovered that I can not use > auto_ptrs. This is a great disadvantage to me. It seems that Tornado and > VxWorks are a bit slow to support the latest c++ standards. Does anyone > have a reliable solution of how to upgrade Tornado to use the latest stl > facilities? A bit slow? Apparently there is a yet another cummulative patch on its way from WindRiver with a newer GCC. However, I've been hearing that it's very close for about 6 months, so you might be waiting for a while longer. What platform are you using? There are several 3rd party compilers that can integrate with VxWorks and Tornado II. The Diab compiler (PowerPC and perhaps others?) is fully integrated, also being from WindRiver. If you're using a PowerPC platform, there's a binary release of GCC 2.95.4 courtesy of Dave Korn at: http://www.lysergic.u-net.com/ With it, there is a more uptodate STL and IOSTREAMS implememtation. You might find you need to work out how to integrate them. If you're using a non-PowerPC platform, you can probably build a GCC from source for your target, but this quite involved. Otherwise, start looking for 3rd party compilers. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Why WindView shows that VxWorks spent most time in ISR? Date: Thu, 02 Aug 2001 21:18:57 +0100 From: molochai Message-ID: <6c63e4a34a%molochai@vapour-trail.demon.co.uk> References: In message people@micro-web.co.kr (JinSoo Kim) wrote: > Hi, > > I just tested WindView in Tornaod 2.0. > I launched standard simulator(vxsim) and launched WindView. > I downloaded no program to vxsim. (The only tasks running on vxsim > are tExtTask, tLogTask, tWdbTask, tWvRBuffMgr, idle tasks.) > I gathered data and uploaded it. The graph shows that vxworks spent most > of time in INT1 ISR other than idle task. > I think that vxworks should spent most of time in idle task > if there are no user tasks running. > Would you please let me know what's wrong with WindView? > Or is there anything I missed? Yep.. Please RTFM. VxSim (at least on WinNT) uses the incrementor timestamp driver rather than a real clock. This timer increments upon *each event*. As the only thing your system is doing is servicing timer interrupts and idling, you'll find that the ratio of idle to INT1 is about 1:1. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Project tool and clearcase fouls up Date: Thu, 02 Aug 2001 21:26:30 +0100 From: molochai Message-ID: <1814e5a34a%molochai@vapour-trail.demon.co.uk> References: <3B6830FE.5FD5869B@siemenscomms.co.uk> <3B691CB1.5FBB0EFF@siemenscomms.co.uk> In message <3B691CB1.5FBB0EFF@siemenscomms.co.uk> Owain Phillips wrote: > > > Bill Pringlemeir wrote: > > > > > > > > "Owain" == Owain Phillips > > > > > > > writes: > > [snip] > > Owain> Files such as sysLib.c are missing from the projects files, > > Owain> therfore the project cannot be built. > > > > Owain> Anyone got a workaround? Ideas??? > > > > Are you on a Windows system? You might check that the project files have > > been marked as binary [even though they are ASCII]. I think that the are > > expected to be Unix end of line (single LF) and the CR/LF with hose the > > tools up... just a guess. > > > > Thanks for the suggestion Bill. > > I am working on a NT4 system. > > The BSP and project tool work when the BSP is located.... > 1) local drive > 2) remote samba mount > > and it fails if the BSP is located on a... > 3) clearcase remote drive > 4) nfs mount As an asside you are aware that some NT4 NFS clients don't work with Clearcase aren't you? > > for each test I copied the base BSP to the filesystem under windows, then > generated a new workspace and bootloadable project basing this on the BSP > on the drive. In the failure cases the BSP wasn't parsed properly and not > all core BSP files got > Arguably when using a fully blown CM tool, you should probably place the entire Tornado II installation under its control. You can probably either run it from there, or use a network installation. I suspect Tornado II will then find the BSP in the place it expects. I'll be moving over to Clearcase in a few weeks, so will be trying this myself, so would be interested in any experiences you have with it. --------------------------- Newsgroups: comp.os.vxworks Subject: networking issues Date: Thu, 02 Aug 2001 16:31:19 -0400 From: Jeremiah Frick Organization: Improv Systems, Inc. Message-ID: <3B69B897.BBA85C91@improvsys.com> This is a multi-part message in MIME format. - --------------BA2961942C15AF9BBFD00ECD Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Hello all! My ongoing saga with usrNetInit continues.... The issue at hand is that on my MIPS Malta card, usrNetInit will sparadically cause the system to hang when it is called from my code. Basically, the system waits for the user to choose an IP address, then using usrNetInit, initializes the system with the specified IP. Unfortunately, this sometimes causes the system to seemingly lock up immediately after the following lines: Attached TCP/IP interface to lnPci unit 0 Attaching network interface lo0... done. So, in testing it I've tried a number of things. First off, I noticed that lockups typically occur when I attempt to change the IP to something different than what it was when I was doing my initial development on the board. Secondly, I've noticed that if I disconnect the network entirely (remove the CAT5 from the card connector), the system will allow me to change the IP to whatever I want each time, no hangups at all, but if I leave the network connection in until the system prompts me to enter the IP and then remove it, the system will still hang. Please note that the IP is only one time configurable per boot, in other words, when the system initializes, it asks for the IP, the user inputs it, and that's it. If they want to change the IP the system has to be reset (don't want to give the idea that I'm successfully calling usrNetInit more than once per run, as that isn't what I'm trying). Thanks for any help! Jeremy - --------------BA2961942C15AF9BBFD00ECD Content-Type: text/x-vcard; charset=us-ascii; name="jeremyf.vcf" Content-Transfer-Encoding: 7bit Content-Description: Card for Jeremiah Frick Content-Disposition: attachment; filename="jeremyf.vcf" begin:vcard n:Frick;Jeremiah tel;work:716-756-5411 x45 x-mozilla-html:FALSE url:www.improvsys.com org:Improv Systems Inc.;PDG version:2.1 email;internet:jeremyf@improvsys.com title:Hardware Engineer adr;quoted-printable:;;3136 Winton Road South=0D=0ASuite 100;Rochester;New York;14623;United States of America fn:Jeremiah Frick end:vcard - --------------BA2961942C15AF9BBFD00ECD-- --------------------------- Newsgroups: comp.os.vxworks Subject: Re: INTEGRITY & VxWorks Date: 2 Aug 2001 13:54:35 -0700 From: eeyore@dial.pipex.com (Eeyore) Organization: http://groups.google.com/ Message-ID: <2c7641d6.0108021254.7639db23@posting.google.com> References: Hello, If you are looking at VxVMI you might be better off investigating VxWorks AE since that seems to have better MMU support. JAT, Eeyore. "Jizhong" wrote in message news:... > Hello, has anybody compared green hills INTEGRITY with VxWorks/VxVMI. > Runtime performance? How easy to move applications between integrity and > vxworks. > > JL --------------------------- Newsgroups: comp.os.vxworks Subject: Re: problem with muxDevUnload Date: 2 Aug 2001 14:36:17 -0700 From: du_bin@yahoo.com (billy du) Organization: http://groups.google.com/ Message-ID: <340587ac.0108021336.5a00bb0b@posting.google.com> References: <3B67E3C0.CFBF6053@europem01.nt.com> I had the same problem before. You have to finish stop the device by calling MuxDevStop(). If you called muxDevStop(void* pCookie), there is a trick there. "pCookie is a pointer to the END_OBJ returned from the muxDevLoad()" Don't use the pointer from muxShow or others. "Fauris, Sebastien" wrote in message news:<3B67E3C0.CFBF6053@europem01.nt.com>... > Hi VxWorkers, > > I meet some problems using the muxDevLoad routine in VxWorks5.4 (Tornado > 2.0.2, MPC8260 BSP). > I get an error during a call to a free (memPartFree) and the task > performing the muxDevUnload get suspended. > > -> tt s1u0 > 1a6444 vxTaskEntry +60 : pseudoBootSbcPn ([]) > 11aaab8 pseudoBootSbcPn+70 : muxDevUnload ([]) > 191480 muxDevUnload +160: free ([]) > 1496bc free +1c : memPartFree ([]) > 1491d4 memPartFree +144: taskSuspend ([]) > value = 0 = 0x0 > > Has someone already met a similar problem using muxDevUnload ? > I am sure it is not my END pnDrvUnload routine attached to the > muxDevLoad which generates the error > but the code within muxDevLoad in the kernel. > > Thanks, > > Sebastien. --------------------------- Newsgroups: comp.os.vxworks Subject: Replacing CplusLib operator new() etc. Date: 02 Aug 2001 17:58:32 -0400 From: Bill Pringlemeir Organization: Bell Sympatico Message-ID: Sender: bpringlemeir@DeadDuck Has anyone replaced the C++ allocators from the CplusLib? I see the function `set_new_handler()'. However, the description, set_new_handler() - ----------------- NAME: set_new_handler() - set new_handler to user-defined function (C++) SYNOPSIS: extern void (*set_new_handler (void(* pNewNewHandler)())) () DESCRIPTION: This function is used to define the function that will be called when operator new cannot allocate memory. The new_handler acts for all threads in the system; you cannot set a different handler for different tasks. RETURNS: A pointer to the previous value of new_handler. leaves a bit to be desired. This does not say what if any parameters are taken by the function returning a `void*'. I think there should be a size to allocate or is this retrieved through some other mechanism. Perhaps the prototype should be, extern void (*set_new_handler (void(* pNewNewHandler)(int))) (int); or something like that. The prototype is in the file "../target/h/new". I would like to try out different allocators (Kingly BSD, Doug Leah, GlibC, Hans Boehm GC, etc) The source for the `C++' new/delete aren't located anywhere are they? I might think that the source for the gcc tool would have them. thanks for any leads, Bill Pringlemeir. - -- Hmmm... a PINHEAD, during an EARTHQUAKE, encounters an ALL-MIDGET FIDDLE ORCHESTRA... ha.. ha.. vxWorks FAQ, "http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html" --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Why WindView shows that VxWorks spent most time in ISR? Date: 2 Aug 2001 15:15:17 -0700 From: TheDoctor68467@yahoo.com (DoctorDoctor) Organization: http://groups.google.com/ Message-ID: References: JinSoo, When using WindView with the simulator, the horizontal axis is an event count, not time. (This is called sequential timestamping in WindView). So you aren't seeing percentage of time that was spent in the ISR vs. idle, you are just seeing which events occur before and after each other, such as interrupt exit and entry. If you were running on a real target with high-resolution timestamping enabled, you would see the little time is spent in the ISRs. people@micro-web.co.kr (JinSoo Kim) wrote in message news:... > Hi, > > I just tested WindView in Tornaod 2.0. > I launched standard simulator(vxsim) and launched WindView. > I downloaded no program to vxsim. (The only tasks running on vxsim > are tExtTask, tLogTask, tWdbTask, tWvRBuffMgr, idle tasks.) > I gathered data and uploaded it. The graph shows that vxworks spent most > of time in INT1 ISR other than idle task. > I think that vxworks should spent most of time in idle task > if there are no user tasks running. > Would you please let me know what's wrong with WindView? > Or is there anything I missed? > Thanks in advance... to your answer. > > Best Regards, > JinSoo Kim --------------------------- Newsgroups: comp.os.vxworks Subject: TCB in CrossWind Date: Thu, 02 Aug 2001 17:34:59 -0500 From: Dinker Charak Organization: Fermi National Accelerator Laboratory Message-ID: <3B69D593.26842311@fnal.gov> Hello, I want to see contents of a task's TCB using crosswind. Any way of doing so? i have used trace32/lauterbach before for this but not crosswind. Also, I started two tasks t1 and t2 from function veryIdle() which is just a while (i==1) loop with a taskDelay() in it. i is set to 1. So, it is a forever loop. Now in CrossWind, I say list veryIdle and set a break point by a left click. Then as I step I am in the while (i==1) loop. Is this debugging different "run" then the two tasks, or am I stepping through t1 or t2? How can I debug a task, say t2? I want to step through t2 and in the middle change value of i to 0 so it comes out of forever loop. can it be done using crosswind? regards, dinker --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Replacing CplusLib operator new() etc. Date: Thu, 02 Aug 2001 23:30:21 +0100 From: molochai Message-ID: References: In message Bill Pringlemeir wrote: > > Has anyone replaced the C++ allocators from the CplusLib? I > see the function `set_new_handler()'. However, the description, > > set_new_handler() > ----------------- > > NAME: set_new_handler() - set new_handler to user-defined function (C++) > SYNOPSIS: extern void (*set_new_handler (void(* pNewNewHandler)())) () > > DESCRIPTION: > This function is used to define the function that will be called when > operator new cannot allocate memory. The new_handler acts for all > threads in the system; you cannot set a different handler for > different tasks. > > RETURNS: A pointer to the previous value of new_handler. > > leaves a bit to be desired. This does not say what if any parameters > are taken by the function returning a `void*'. I think there should be > a size to allocate or is this retrieved through some other mechanism. > I think you fundamentally misunderstand what set_new_handler() does - it does *not* replace the memory allocation function used by operator new(), but rather sets a function to be executed when memory exhaustion occurs. set_new_handler() is a C++ runtime function, and most good C++ texts will provide a more thorough explanation. > Perhaps the prototype should be, > > extern void (*set_new_handler (void(* pNewNewHandler)(int))) (int); > > or something like that. The prototype is in the file > "../target/h/new". I would like to try out different allocators > (Kingly BSD, Doug Leah, GlibC, Hans Boehm GC, etc) The source for the > `C++' new/delete aren't located anywhere are they? I might think that > the source for the gcc tool would have them. operator new() and operator delete() in most systems are implemented using malloc() and free(). As C++ allow operator overloading, you can implement your own strategy merely by overloading operator new() and operator delete(), either globally or on class-by-class basis. Pool-allocation can be very easily implemented this way and provides for highly deterministic allocation and deallocation performance - something that generalised allocation algorithms rarely gurantee. There are some very neat ways of implementing this (nearly) invisible using templates and inheritance in C++. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Exception Date: Fri, 03 Aug 2001 00:17:11 +0100 From: molochai Message-ID: <8bb4f4a34a%molochai@vapour-trail.demon.co.uk> References: <9ka1eq$aku$1@newsie.singa.pore.net> <3b69140f$0$228@hades.is.co.za> In message Bill Pringlemeir wrote: > >>> Chet Wee wrote: > > >> I am implementing the exception handling in C++ using try() & > >> catch() mechanism > > >> Manage to compile and execute in MSC++ environment but when > >> recompiled using Tornado, the try & catch mechanism did not catch > >> the exception > > >>>>> "Morne" == Morne Hartman writes: > > Morne> Hi, I have been able to implement exception handling in > Morne> VxWorks, could you please provide some code to see what you > Morne> are doing. > > My experience is that exceptions and templates do not work as most C++ > programmers might expect with vxWorks. The also tend to generate > larger code bases. We have disabled support in our projects. We do > use virtual functions, multiple inheritance and RTTI without problems. IME Templates work as well on VxWorks as they do elsewhere - which in all cases is a question of compiler support. Template handling is difficult for all compilers, and really requires linker support. Luckily, newer GCC/EGCS releases appear to now have this and should be able to reduce bloat. As for exceptions, yes they produce a lot of code, but is absolutely necessary if you wish to properly handle exceptional program flow in constructors. If your system is not calling constructors other than at initialisation time, sure, don't use exceptions, otherwise you'll find yourself writing *huge* amounts of error handling code. Incidentally, why not check how much code the compiler synthesis when RTTI is enabled. You might be surprised. Exception handling uses RTTI. In what way do exceptions not work as expected? Yes, there is a suspision that with GCC2.7.2/VxWorks C++ runtime that exception handlers are not as thread safe as they should be, but I've never seen this proved or otherwise. > > I believe that embedded C++ disables both templates and exceptions. > Anyways, I won't analyze whether you should beem or not, but > they do seem to be problematic. > Why not learn how to use C++ properly? Seriously, the benefits greatly outweight the extra code size - especially with the price of memory plumetting again > I have used `setjmp' and `longjmp' to achieve similar effects without > the stack penalties incurred by C++ exceptions. If this is your own > code base, you might consider this as an alternative. This is poor-man's excepetions - it provides exceptional program flow, but leave you to the stack-unwinding you would otherwise get for free with proper exceptions. Using exceptions automates the generation of error handling code in many cases, which would otherwise need writing manually (or more likely will not get written.....) --------------------------- Newsgroups: comp.os.vxworks Subject: pSOS and vxworks Date: Thu, 02 Aug 2001 23:17:29 GMT From: Gururaj patkar Organization: Optimum Online Message-ID: <3B69DF05.89701CDC@optonline.net> Hi all, Was curious to know if there is an article which compares pSOS and vxworks. Which is better? thanks for any reply. gururaj --------------------------- Newsgroups: comp.os.vxworks Subject: Re: pSOS and vxworks Date: Fri, 03 Aug 2001 00:43:11 +0100 From: molochai Message-ID: <2816f7a34a%molochai@vapour-trail.demon.co.uk> References: <3B69DF05.89701CDC@optonline.net> In message <3B69DF05.89701CDC@optonline.net> Gururaj patkar wrote: > Hi all, > > Was curious to know if there is an article which compares pSOS and > vxworks. Which is better? > It's kind of irrelevent in commerical terms as pSOS is pretty much dead following WindRiver's acquisition of it. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: C++ and code size ... Date: Fri, 03 Aug 2001 12:40:12 +1200 From: Ian Collins Organization: Masuma Message-ID: <3B69F2EC.714B3216@masuma.com> References: <94BAF40039B0D411BC150002B315275A0CDB7B@mailserv.hatteras.com> Todd Johnson wrote: > > Thanks. I'll take a look. > > -tdj > > > -----Original Message----- > > From: Srikumar Natarajan > > Sent: Thursday, August 02, 2001 11:38 AM > > To: Nancy Adams; Todd Johnson > > Subject: C++ and code size ... > > > > Hi, > > > > FYI. This is a discussion of code size when the C++ STL > > (Standard Template Library) is used, in a statically linked > > executable. Not exactly the same as ours, but very informative. > > > > sri > > > > http://groups.google.com/groups?hl=en&safe=off&th=84ff22c969b7 > > 6b8c,12&start=0 > > > > My comment in my reply on this thread still hold. If gcc is used, you can reduce template bloat with the interface and implementation pragmas. It all depends on their use in the STL files. - -- Ian Collins Masuma Ltd, Christchurch New Zealand. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: vxWork hang up Date: 2 Aug 2001 17:56:25 -0700 From: john_94501@yahoo.com (John) Organization: http://groups.google.com/ Message-ID: <488e459a.0108021656.6a34a803@posting.google.com> References: <9k6a32$i3l$1@newsie.singa.pore.net> Hello all, > You need to restructure your heartbeat: the watchdog ISR should do > nothing except give a mutex semaphore. You should also have a task running > at a very high priority that waits on this semaphore, and each time it > receives the semaphore from the watchdog ISR it will wake up; then it can > send one network packet (from a task context!) and go back to waiting on the > semaphore for next time. Wow! It would work, but why make life so difficult... here's a much simpler solution that should have the same effect if spawned at a high priority (i.e the priority that would have been needed for the task in the above idea): void heartbeatTask (void) { FOREVER { sendHeartBeat(); taskDelay (sysClkRateGet()); } } Simple is often better ;-) John... --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Puzzled by PPC assembler for varargs function Date: 2 Aug 2001 22:25:33 -0700 From: john_94501@yahoo.com (John) Organization: http://groups.google.com/ Message-ID: <488e459a.0108022125.673b74dc@posting.google.com> References: <3B66C3D0.2BCC6823@lucent.com> <3B68F15F.916B411C@lucent.com> <488e459a.0108021157.6e136b3c@posting.google.com> Hello again, > > Thanks for this explanation - but it just begs another question. If > > bit 6/CR1 set denotes that the current process is using FP instructions, > > how is this bit set? Does calling any FP instruction set this bit? Or > > is it up to the FP user to set it explicitly? > > It should be set for all tasks running on hardware with an FP unit > (i.e. anything using the PPC603 or PPC604 VxWorks CPU types, which > includes all the newer 750 and 7400 series I believe). The VxWorks > task initialisation sequence for PPC automatically sets this bit for > CPUs with an FPU. OK, ignore me... I was thinking of a different bit (the FP bit in the MSR) which is set when all tasks are started to leave the FPU enabled. Sorry for any confusion there, John... --------------------------- Newsgroups: comp.os.vxworks Subject: Re: networking issues Date: 2 Aug 2001 22:40:41 -0700 From: john_94501@yahoo.com (John) Organization: http://groups.google.com/ Message-ID: <488e459a.0108022140.149341fc@posting.google.com> References: <3B69B897.BBA85C91@improvsys.com> Hello Jeremy, > The issue at hand is that on my MIPS Malta card, usrNetInit will > sparadically cause the system to hang when it is called from my code. > Basically, the system waits for the user to choose an IP address, then > using usrNetInit, initializes the system with the specified IP. > Unfortunately, this sometimes causes the system to seemingly lock up > immediately after the following lines: > > Attached TCP/IP interface to lnPci unit 0 > Attaching network interface lo0... done. > > So, in testing it I've tried a number of things. First off, I noticed > that lockups typically occur when I attempt to change the IP to > something different than what it was when I was doing my initial > development on the board. This sounds like ARP related problems... often if you use network booting, but use a different IP address in the kernel startup (or the MAC/ethernet address for that matter) than the bootrom caused these delays can come in. What should the target be doing next? Is it going to be accessing the network for anything? Check the ARP table on your host (for unix arp -a shows you the info, for windoze I don't have a clue, but there must be a way ;-). HTH, John... --------------------------- Newsgroups: comp.os.vxworks Subject: Problem in starting Profilescope Date: 2 Aug 2001 22:50:33 -0700 From: pritamganguly@hotmail.com (Pritam) Organization: http://groups.google.com/ Message-ID: <166770ed.0108022150.4670f6b@posting.google.com> Hi All, I was trying to start the RTI tools on Tornado. Its getting stuck at the start of RTI Tools invocation. Here is the exact sequence of what is happening. 1. I start the target server. 2. Then I select that server in the tornado toolbar (before this the ScopeTools buttons are greyed out AFTER this they are enabled). The "memShow" at this point shows the following: - -> memShow SUMMARY: status bytes blocks avg block max block ------ --------- -------- ---------- ---------- current free 26699920 4 6674980 26303280 alloc 2948560 3350 880 - cumulative alloc 3361232 3708 906 - value = 0 = 0x0 - -> 3. Now I try to start ProfileScope 4. I get a dialogbox with "ProfileScope Setup Options" Caption - which asks for size of the queue,call-stack depth,Verbosity etc. I don't change any values here and press "OK" 5.Then the Shell shows these lines: - -> Exception number 10: Task: 0x81d55b20 (t1) Reserved Insn program counter: 0x803966c0 processor status: 0x1000ab01 cause register: 0x00000028 80168764 wdbFuncCallLibInit+154: ProfileScopeInit (eeeeeeee, eeeeeeee, eeeeeeee, eeeeeeee) And the ProfileScope window with the Error DialogBox which says "Target not intialized. Please use the ProfileScope Launcher button" 6. "memShow" at this point shows: - -> memShow SUMMARY: status bytes blocks avg block max block ------ --------- -------- ---------- ---------- current free 26678784 4 6669696 26303280 alloc 2969696 3352 885 - cumulative alloc 3382368 3710 911 - value = 0 = 0x0 - -> Hope this clarifies the stituation. Thanks and Regards, Pritam --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Project tool and clearcase fouls up Date: Fri, 03 Aug 2001 09:01:06 +0100 From: Owain Phillips Organization: Siemens Inc. Message-ID: <3B6A5A42.7E3A18DB@siemenscomms.co.uk> References: <3B6830FE.5FD5869B@siemenscomms.co.uk> <3B691CB1.5FBB0EFF@siemenscomms.co.uk> <1814e5a34a%molochai@vapour-trail.demon.co.uk> This is a multi-part message in MIME format. - --------------6530E7988EE8CD3D8B722EE3 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit molochai wrote: > > > As an asside you are aware that some NT4 NFS clients don't work with > Clearcase aren't you? > I have seen the descrption on windsurf. I only tried the nfs mount to to narrow down the problems. The problem symptoms are eactly the same. > > > > > for each test I copied the base BSP to the filesystem under windows, then > > generated a new workspace and bootloadable project basing this on the BSP > > on the drive. In the failure cases the BSP wasn't parsed properly and not > > all core BSP files got > > > > Arguably when using a fully blown CM tool, you should probably place the > entire Tornado II installation under its control. You can probably either run > it from there, or use a network installation. I suspect Tornado II will then > find the BSP in the place it expects. > Other projects in this building place the complete Tornado tree and sources a single clearcase VOB and this appears to work. We were trying to organise things a little more differently using multiple VOBs to store different sub-systems. Owain - --------------6530E7988EE8CD3D8B722EE3 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 - --------------6530E7988EE8CD3D8B722EE3-- --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Problem in scaning user input ! Date: 3 Aug 2001 01:37:45 -0700 From: mridulgupta1@yahoo.com (Mridul Gupta) Organization: http://groups.google.com/ Message-ID: References: <9k5q3t$8g$1@news2.kornet.net> "Ogre" wrote in message news:<9k5q3t$8g$1@news2.kornet.net>... > If you don't need "Shell", subtract SHELL in your configuration. > And don't use scanf. > in usrConfig.c, vxWorks use fioRdString. > > "Mridul Gupta" wrote in message > news:fef31d0c.0107302333.1b62f612@posting.google.com... > > hi all, > > i am a new user to VxWorks and i wrote a interactive program > > which takes inputs from user. the problem is that the scanf doesnt > > works ... it seems like the shell is eating the input in between. I > > tried suspending the shell task but it didnt work. i also tried > > changing priorities but nothing worked ... please tell me how do i > > take user input ?? > > > > Thanx > > > > Mridul hi, thanx for your reply ... I also tried fioRdString() function but it didnt worked ... so please tell something else ... did you get my problem ? Mridul --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Writing host files under serial/TSFS Date: Fri, 3 Aug 2001 09:44:52 +0100 From: "timallen" Organization: ntlworld News Service Message-ID: References: <66845b46.0107311117.161b7aab@posting.google.com> Reply-To: "timallen" Matt, matt wrote in message news:66845b46.0107311117.161b7aab@posting.google.com... > Is it possible to create a file on the host network > when connected over serial TSFS? Have you remembered to set write access in the target server configuration for TSFS? Regards, Tim. - -------------------- Tim Allen Ltd Real Time Embedded Software Consultancy --------------------------- Newsgroups: comp.os.vxworks Subject: Re: How to send a Ethernet Frame Date: 3 Aug 2001 09:04:30 GMT From: SIMONE@linac.ikp.physik.tu-darmstadt.de (RICHTER SIMONE) Organization: Technical University Darmstadt, FRG (Nuclear Physics Institute) Message-ID: <9kdpeu$594$1@sun27.hrz.tu-darmstadt.de> References: <3b68fdd8$1_2@news.vo.lu> Hi Christophe. I hope the following example will help you understand how the ethernet hooks can work for you. When you set up your system add the following lines to your code #include status = etherOutputHookAdd(FUNCPTR MyEthernetHoohkRoutine); The MyEthernetHoohkRoutine must return a boolean value, this is the routine that is called immediately before any transmission, initiated by etherOutput takes place. BOOL MyEthernetHoohkRoutine(struct ifnet *pIf, char *buffer, int length) It should return TRUE when the routine has handled the packet an no further processing should be done. It should return FALSE if the packet should be transmitted the normal way. Then in your own MyEthernetHookRoutine you can do anything you want with your ethernet packet, for example implement another proctocol. For a first times user it may sound a bit complicated, but if you used it once it's quite easy. Regards Simone - ------------------------------------------------------------------------------- Dr. Simone Richter S.Richter@gsi.de (old address: simone@ikp.tu-darmstadt.de) GSI Department Acclerator Physics Tel +49 6159 712170 Planckstrasse 1 Wixhausen --------------------------- Newsgroups: comp.os.vxworks Subject: Re: vxWork hang up Date: Fri, 3 Aug 2001 12:54:46 +0200 From: "Werner Schiendl" Message-ID: <3b6a839c$0$25324$6e365a64@newsreader02.highway.telekom.at> References: <9k6a32$i3l$1@newsie.singa.pore.net> <488e459a.0108021656.6a34a803@posting.google.com> Hi, > Wow! It would work, but why make life so difficult... here's a much > simpler solution that should have the same effect if spawned at a high > priority (i.e the priority that would have been needed for the task in > the above idea): > > void heartbeatTask (void) > { > FOREVER > { > sendHeartBeat(); > > taskDelay (sysClkRateGet()); > } > } > > Simple is often better ;-) This one is not equivalent to using what Dave suggested. If sendHeartBeat takes more time on one invocation all further heartbeats will be delayed, too. regards Werner --------------------------- End of New-News digest ********************** From vxwexplo-errs@csg.lbl.gov Fri Aug 3 09:26:45 2001 From: karunakaran nair Date: Fri Aug 3 09:26:47 PDT 2001 Subject: comp.os.vxworks newsdigest how to lock target from tornado ? what is "reserve" and "unreserve" ? how to prevent somebody from telnetting into your target ? thanks, KN __________________________________________________ Do You Yahoo!? Make international calls for as low as $.04/minute with Yahoo! Messenger http://phonecard.yahoo.com/ From vxwexplo-errs@csg.lbl.gov Fri Aug 3 18:06:01 2001 From: "Chang, Mark" Date: Fri Aug 3 18:06:03 PDT 2001 Subject: Help - strange task delay problem Hi, Running Vxworks 5.3.1 I have one spawned task that pends indefinitely. The reason being is that the delay timer starts out negative. This is the output from tornado NAME ENTRY TID PRI STATUS PC SP ERRNO DELAY ShdslAgtTasshdslagt_mai 83d31a20 100 PEND+T 80131c28 83d316d0 0 -22975 So the DELAY just keeps getting more negative and this task never gets out of PEND. Another curious thing is that the PC for this task is always in the routine of noSemQPutEvt. Has anyone seen this? Or can shed some light. Where is the DELAY value kept and incremented? It's really baffling me. Thanks, Mark ---------------------------------------------------------- Mark Chang Senior Engineer Network Management Broadband Access and Transport Group ADC Wireline Systems Division Mark_Chang@ADC.com (714)730-2827 Learn about ADC - The Broadband Company at www.adc.com From vxwexplo-errs@csg.lbl.gov Fri Aug 3 23:54:03 2001 From: Lourens Geldenhuys Date: Fri Aug 3 23:54:06 PDT 2001 Subject: Minimal memory footprint? Has anybody ran VxWorks on an M-CORE (or any other arch for that matter) from less than 128KB of FLASH with less than 80KB RAM? This would be a very stripped down VxWorks, no network, no I/O system, i.e. really only the scheduler. Would WindRiver publish minimal memory footprint requirements, and if so where? Regards, Lourens From vxwexplo-errs@csg.lbl.gov Sat Aug 4 04:03:25 2001 From: Vxworks Exploder Date: Sat Aug 4 04:03:28 PDT 2001 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Sat Aug 4 04:03:09 PDT 2001 Subject: Re: vxWork hang up Subject: vxWorks_romCompressed.bin? Subject: About ping large packet on vxWorks? Subject: Far-End Fault-Indecation for Ethernet 100MB-FX Subject: Re: Replacing CplusLib operator new() etc. Subject: VxSim on Solaris 7/8 revisited Subject: Re: Puzzled by PPC assembler for varargs function Subject: Re: Exception Subject: Re: Exception Subject: Re: TCB in CrossWind Subject: Java application performance on VxWorks Subject: Importing WinView Graphs into Word Document Subject: help - one of my task's delay is in negative numbers Subject: vmipmc-5790 on mvme5100 anyone Subject: Re: prjConfig.c in BSP for Power PC Subject: Re: Exception Subject: binding socket to device Subject: How to get idle TCP/UDP port number which not used by the system Subject: Re: prjConfig.c in BSP for Power PC Subject: Re: help - one of my task's delay is in negative numbers - Part II Subject: Re: help - one of my task's delay is in negative numbers - btw - we're using Vxworks 5.3.1 Subject: Re: Importing WinView Graphs into Word Document Subject: Re: Exception ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: vxWork hang up Date: 3 Aug 2001 11:15:10 GMT From: "Johan Borkhuis" Organization: Agere Systems Message-ID: References: <9k6a32$i3l$1@newsie.singa.pore.net> <488e459a.0108021656.6a34a803@posting.google.com> <3b6a839c$0$25324$6e365a64@newsreader02.highway.telekom.at> >> void heartbeatTask (void) >> { >> FOREVER >> { >> sendHeartBeat(); >> >> taskDelay (sysClkRateGet()); } } > > This one is not equivalent to using what Dave suggested. > If sendHeartBeat takes more time on one invocation all further heartbeats > will be delayed, too. How about this then: void heartbeatTask (void) { int clkValue; FOREVER { clkValue = tickGet(); sendHeartBeat(); taskDelay (sysClkRateGet()- (tickGet() - clkValue); } } Groeten, Johan - -- o o o o o o o . . . _____________________________ o _____ || Johan Borkhuis | .][__n_n_|DD[ ====_____ | borkhuis@agere.com | >(________|__|_[_________]_|__________________________| _/oo OOOOO oo` ooo ooo 'o!o!o o!o!o` === VxWorks FAQ: http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html === --------------------------- Newsgroups: comp.os.vxworks Subject: vxWorks_romCompressed.bin? Date: Fri, 03 Aug 2001 13:23:21 GMT From: Jaz Organization: Bell Sympatico Message-ID: Hi Any ideas how to create a binary from the elf format (for vxWorks_romCompress)? I tried "objcopy -O binary vxWorks_romCompress vxWorks_romCompress.bin" but I got this error: objcopyppc: vxWorks.bin: Invalid argument objcopyppc: vxWorks.bin: Invalid argument I tried elftobin but it seems to hang. "objcopy -O binary vxWorks vxWorks.bin" works fine (ie uncompressed vxWorks image) - -- Regards Jaz PS: Please remove 'X' in my email address to obtain the correct address. --------------------------- Newsgroups: comp.os.vxworks Subject: About ping large packet on vxWorks? Date: Fri, 3 Aug 2001 21:58:25 -0700 From: "Xiao Boyu" Organization: nanwang Message-ID: <9keakm$njs$1@mail.cn99.com> ICBXZSBpc3N1ZSBmb2xsb3dpbmcgY29tbWFuZCJwaW5nIDE3Mi4xNi42NC4xOTIgLXQgLWwgMzUw MDAiIGZyb20gb3VyIGhvc3QgdG8gdGFyZ2V0IGJvYXJkDQooRkFEUzg2MFQgZnJvbSBtb3QpIHJ1 bm5pbmcgdnh3b3JrcywgdnhXb3JrcyB3YXMgY29uZmlndXJlIHRvIHVzZSBGRUMgb2YgODYwVCB3 aXRoIDEwME0gY29ubmVjdGlvbiBhbmQgdnhXb3JrcyBjYW4ndCByZXNwb25kIHRvIHRoaXMgY29y cmVjdGx5LldlIHRlc3QgdnhXb3JrcyB1c2luZyBTQ0MxIG9mIENQTSBydW5uaW5nDQoxME0gY29u bmVjdGlvbiBhbmQgaXQgd29ya3Mgd2VsbC5GdXJ0aGVyIHRlc3Qgc2hvdyB0aGUgcHJvYmxlbSBp cyByZWxhdGVkIHdpdGggdnhXb3JrcyBub3QNCjg2MFQuIENhbiBhbnlvbmUgaGVscCB1cyBvbiB0 aGlzIHF1ZXN0aW9uPw0KICBUaGFuayB5b3UgaW4gYWR2YW5jZS4NCg0KU2luY2VyZWx5IHlvdXJz LA0KWGlhbw0K --------------------------- Newsgroups: comp.os.vxworks Subject: Far-End Fault-Indecation for Ethernet 100MB-FX Date: Fri, 3 Aug 2001 16:16:03 +0200 From: "Michael Koestner" Organization: Siemens AG Karlsruhe Message-ID: <9kebn5$pn5@parzival.khe.siemens.de> Hi, does anyone have a implementation for a Far-End Fault-Indecation ethernet fiber optic connection. the problem work with the phy device, which have latched bit implemented to signal a dynamic state. Mfg Michael Koestner --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Replacing CplusLib operator new() etc. Date: 03 Aug 2001 10:38:56 -0400 From: Bill Pringlemeir Organization: Bell Sympatico Message-ID: References: Sender: bpringlemeir@DeadDuck >>>>> "molochai" == molochai writes: >> or something like that. The prototype is in the file >> "../target/h/new". I would like to try out different allocators >> (Kingly BSD, Doug Leah, GlibC, Hans Boehm GC, etc) The source for >> the `C++' new/delete aren't located anywhere are they? I might >> think that the source for the gcc tool would have them. molochai> operator new() and operator delete() in most systems are molochai> implemented using malloc() and free(). molochai> As C++ allow operator overloading, you can implement your molochai> own strategy merely by overloading operator new() and molochai> operator delete(), either globally or on class-by-class molochai> basis. Thanks, but no thanks. I will use ar$(ARCH) to manipulate the libraries and write my own code. I wish to replace all C++ allocators using new() and delete(). It would be very troublesome (and error prone) to replace each base class with allocation functions. You would be replicating code. Perhaps if you had a `purer' object oriented environment like Java or Small talk where all classes have a common base class. I have rarely seen C++ code like this and it certainly isn't like this in the code base that I have to support. Also, many of the allocators that I have source for are written in `C'. It would be more difficult to re-implement these in C++. I think that "slab" like memory management would do well in C++, but I have never seen this used by any C++ programmers. regards, Bill Pringlemeir. - -- vxWorks FAQ, "http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html" --------------------------- Newsgroups: comp.os.vxworks Subject: VxSim on Solaris 7/8 revisited Date: 3 Aug 2001 09:39:44 -0700 From: kplloyd@lucent.com (Kurt Lloyd) Organization: http://groups.google.com/ Message-ID: This is in response to a question Iulian Ciulinet (ciuli_net@yahoo.com) had back in June: "Has anybody actually got VxSim (with the networking features) to work on Solaris 7 or 8, using either the ULIP or PPP network backend ?" I managed to get it working on Solaris 8 with PPP, but had to do two things to get it to work that I didn't have to do on Solaris 2.5.1 (both as root): 1. chmod 666 /tmp/.asppp.fifo (Seems security had been tightened-up.) 2. ndd -set /dev/ip ip_forwarding 1 (IP forwarding was not on by default.) - -Kurt Lloyd, kplloyd@lucent.com --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Puzzled by PPC assembler for varargs function Date: Fri, 3 Aug 2001 15:03:07 +0100 From: vxworksintent@tao-group.com (David Given) Organization: I do not speak for anyone but myself, and barely that. Message-ID: References: <3B66C3D0.2BCC6823@lucent.com> <3B68F15F.916B411C@lucent.com> Sender: dg@pearl.tao.co.uk In article , "Dave Korn" writes: > Strictly speaking, the meaning of the flag is not that 'the process is > using FP instructions', but that "This particular function call to this > particular varargs function has passed one or more arguments in the FP > registers". > > The compiler adds the instruction to set or clear this bit when it sees a > call to a function that it knows (from having seen the prototype) is a > varargs function. It's not done on a per-process basis, but on a > per-function-call basis when the code is compiled. This behaviour is part > of the EABI standard (Sun SysV ABI PowerPC processor supplement, Sep. 95 > edn, p3-21), and it *only* applies to varargs functions. Ah. Yes, that makes sense. (The PowerPC manual I have doesn't refer to that bit in any meaningful sense; while I do have a copy of the ABI --- somewhere --- it didn't occur to me to look.) Varargs functions are a pain, aren't they? Do other platforms, such as the MIPS (which I'm currently working on, although without an FPU) do anything similar? - -- David Given vxworksintent@tao-group.com --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Exception Date: 3 Aug 2001 10:50:34 -0700 From: ryan_kinney@agilent.com (Ryan Kinney) Organization: http://groups.google.com/ Message-ID: References: <9ka1eq$aku$1@newsie.singa.pore.net> "Chet Wee" wrote in message news:<9ka1eq$aku$1@newsie.singa.pore.net>... > hi all, > > I am implementing the exception handling in C++ using try() & catch() > mechanism > > Manage to compile and execute in MSC++ environment but when recompiled using > Tornado, the try & catch mechanism did not catch the exception > > Please advise if I miss any setting for this > > Rgs I've had problems with try/catch. Specifically, code like this: class Exception {}; class DerivedException : public Exception {}; try { throw DerivedException(); } catch (DerivedException &) { // Exception handler never called } catch (Exception &) { // Exception handler never called } catch (...) { // Exception caught here } The default Tornado compiler is rather archaic and does have bad/non-existant support for templates/STL, exceptions, namespaces, and dozens of other minor changes to the C++ language that have since been standardized. I am impatiently waiting for WindRiver's next compiler update, which has been "almost ready" for quite some time now. Ryan Kinney Agilent Technolgies --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Exception Date: 03 Aug 2001 14:53:04 -0400 From: Bill Pringlemeir Organization: Bell Sympatico Message-ID: References: <9ka1eq$aku$1@newsie.singa.pore.net> <3b69140f$0$228@hades.is.co.za> <8bb4f4a34a%molochai@vapour-trail.demon.co.uk> Sender: bpringlemeir@DeadDuck >> I have used `setjmp' and `longjmp' to achieve similar effects >> without the stack penalties incurred by C++ exceptions. If this >> is your own code base, you might consider this as an alternative. molochai> This is poor-man's excepetions - it provides exceptional molochai> program flow, but leave you to the stack-unwinding you molochai> would otherwise get for free with proper exceptions. Using molochai> exceptions automates the generation of error handling code molochai> in many cases, which would otherwise need writing manually molochai> (or more likely will not get written.....) Really? I don't find the code that much more complex. Here is a comparison. [C++ exceptions] class ImportantStuff {...}; ... try { // Processing... do_stuff(); } catch(...) { // Exception processing. } void ImportantStuff::do_stuff(void) { throw... } [setjmp exceptions] class ImportantStuff {...; jmp_buf error; }; ... if(0 == setjmp(error)) { // Processing... do_stuff(); } else { // Exception processing. } void ImportantStuff::do_stuff(void) { longjmp(error,-1);} That doesn't seem to be too different to me and you don't have stack winding etc as you pass through the different levels of function calls, if the `do_stuff()' function is deep in a call hierarchy. This is what exceptions are for. I have seen this technique used in several `C' and `C++' libraries. Obviously there are some extra features to exceptions, but generally I think that people only use this core functionality. There is a direct co-relation between the jmp_buf and exceptions in my example. [none] => jmp_buf error; try => if(!setjmp(error)) catch => else throw => longjmp(error,-1); I think this shows that it is not that different at all to achieve the same effects. I don't see how the statement "which would otherwise need writing manually (or more likely will not get written.....)" is applicable. The effort seems similar to me. If you really like multiple catch statements, then you can code the longjmp to indicate an exception number or add a data member to note the specific error. I personally don't find this that useful as an error is always bad and I usually wish to do the same thing whatever the case [several testing methodologies say this is better]. I would be interested to know what features you use that make the `C++' exceptions much better than this as I am sure to learn something. regards, Bill Pringlemeir. - -- TORPEDOED BY SPACE-TIME VORTEX 57W 24N. SINKING. U-990. vxWorks FAQ, "http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html" --------------------------- Newsgroups: comp.os.vxworks Subject: Re: TCB in CrossWind Date: Fri, 03 Aug 2001 19:13:15 GMT From: "Peter Marqui" Organization: Verio Message-ID: References: <3B69D593.26842311@fnal.gov> "Dinker Charak" wrote in message news:3B69D593.26842311@fnal.gov... > Hello, > > I want to see contents of a task's TCB using crosswind. Any way of doing > so? i have used trace32/lauterbach before for this but not crosswind. > > Also, I started two tasks t1 and t2 from function veryIdle() which is > just a while (i==1) loop with a taskDelay() in it. i is set to 1. So, it > is a forever loop. > > Now in CrossWind, I say list veryIdle and set a break point by a left > click. Then as I step I am in the while (i==1) loop. Is this debugging > different "run" then the two tasks, or am I stepping through t1 or t2? > > How can I debug a task, say t2? I want to step through t2 and in the > middle change value of i to 0 so it comes out of forever loop. can it be > done using crosswind? use Attach from the Debug to pick which task use wish to work with > > regards, > dinker --------------------------- Newsgroups: comp.os.vxworks Subject: Java application performance on VxWorks Date: Fri, 03 Aug 2001 19:54:29 GMT From: Hong Hsu Message-ID: <3B6B020B.4040700@bellatlantic.net> I am wondering someone is using Java application on the top of VxWorks. I also heard that FastJ compiler can convert Java source into native executable for VxWorks. Does any one have such an experience? How is the Java application's performance with comparison of using Java VM? Thanks a lot, - -Hong honghsu@bellatlantic.net --------------------------- Newsgroups: comp.os.vxworks Subject: Importing WinView Graphs into Word Document Date: Fri, 3 Aug 2001 17:21:54 -0400 From: "Tom Kocur" Organization: Nortel Networks Message-ID: <9kf4dm$48h$1@zrtph05m.us.nortel.com> I'm sure this has been done before but I can't figure out a way. I'd like to take a WinView Graph and include it in a Microsoft Word document. The problem I'm having is that I can only seem to store these as .wvr files. Can anybody help me out? Tom Kocur --------------------------- Newsgroups: comp.os.vxworks Subject: help - one of my task's delay is in negative numbers Date: 3 Aug 2001 15:25:27 -0700 From: mchang@cisdesign.com (Mark Chang) Organization: http://groups.google.com/ Message-ID: <65464ea6.0108031425.54237c27@posting.google.com> I have a strange problem. One of the task that I spawned is in PEND+T but the delay count is negative. It looks like this in Tornado NAME ENTRY TID PRI STATUS PC SP ERRNO DELAY - ---------- ------------ -------- --- ---------- -------- -------- ------- ----- ShdslAgtTasshdslagt_mai 83d31a20 100 PEND+T 80131be8 83d316d8 0 -13793 and since it is negative, the delay count never reaches 0, just goes more negative. Actually, after about 15 mintues or so, it finally gets out of PEND runs and goes right back into a negative delay. The PC counter of this task is always in the routine noSemQPutEvt I have no clue as to why this is happening. Only that it has something to do with select. The timeout for the select seems fine and I don't believe you can have a negative timeout. Can you? Has anyone seen this before? Any and I mean any idea, insights, ESP - please share with me. Thanks, Mark --------------------------- Newsgroups: comp.os.vxworks Subject: vmipmc-5790 on mvme5100 anyone Date: Fri, 03 Aug 2001 23:14:10 GMT From: pat.dirienzo@verizon.net Message-ID: <3B6B303F.40200231@verizon.net> Reply-To: pat.dirienzo@verizon.net Has anyone used the 5790 pmc scsi card from VMI on the mvme5100 Motorola card. I have the vxworks bsp from VMI, it looks generic enough, but it does not compile with vxworks. There are missing constants (pmc/pci addresses & interrupt levels) which when supplied does allow the VMI bsp to compile. We do not have the hardware yet to test it, and I am not sure of the correct addresses and interrupt levels to use for the missing constants. Any experience here would be appreciated as I would like to determine if this combination will fly. pat.dirienzo@verizon.net pdirienzo@sierradatasys.com --------------------------- Newsgroups: comp.os.vxworks Subject: Re: prjConfig.c in BSP for Power PC Date: Fri, 03 Aug 2001 23:30:42 GMT From: pat.dirienzo@verizon.net Message-ID: <3B6B341F.A17A5427@verizon.net> References: <3b659edb.851409982@news.demon.co.uk> Reply-To: pat.dirienzo@verizon.net I would search thru the Tornado directories for this function, it may expose the reason for it being included. Also the function may be part of your bsp/vxworks library (libxxxxgnu.a). You can use the Tornado archive tool to see if it is. Jan Larcombe wrote: > I'm having problems building my BSP, I end up with an unresolved > external muxMaxBinds. I've checked why this is and it seems to be > because the prjConfig.c which is generated is calling the routine > usrMuxLibInit. In earlier versons of my BSP prjConfig.c was calling > MuxLibInit instead. > Obviously there is no point in ediiting prjConfig.c but does anyone > know how to change configurations so that the prjConfig.c file which > is automatically generated will call the correct function. > > Thanks in advance --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Exception Date: 03 Aug 2001 20:20:04 -0400 From: Bill Pringlemeir Organization: Bell Sympatico Message-ID: References: <9ka1eq$aku$1@newsie.singa.pore.net> <3b69140f$0$228@hades.is.co.za> <8bb4f4a34a%molochai@vapour-trail.demon.co.uk> Sender: bpringlemeir@DeadDuck >>>>> "Bill" == Bill Pringlemeir writes: [snip] Bill> I would be interested to know what features you use that make Bill> the `C++' exceptions much better than this as I am sure to Bill> learn something. Sorry, I forgot that you may have destroy objects created on the stack. I was assuming that the code wasn't doing this. I guess that is why you were talking about the stack unwinding. regards, Bill Pringlemeir. - -- ANNOYED BY TINY GNATS. CREW UNMOTIVATED. SINKING. U-35. vxWorks FAQ, "http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html" --------------------------- Newsgroups: comp.os.vxworks Subject: binding socket to device Date: Sat, 4 Aug 2001 00:22:06 +0000 (UTC) From: nvpopov@mail.com (Nikolay Popov) Organization: Mailgate.ORG Server - http://www.Mailgate.ORG Message-ID: <20010804002158.5440.qmail@mail.com> I need to launch several copies (several task) of server program (not VxWorks's DHCP server). And every task should serve only one ethernet link. On Linux system I can use SO_BINDTODEVICE socket option. Of course the first answer - use etherInputHook(). But in this case I have to modify program and will loss some advantages like packet buffering on socket level. Most of UNIX systems allow bind socket to interface. Are there any ways to for VxWorks ? Is it possible to do that by ioctl() ? - -- _______________________________________________ FREE Personalized E-mail at Mail.com http://www.mail.com/?sr=signup Talk More, Pay Less with Net2Phone Direct(R), up to 1500 minutes free! http://www.net2phone.com/cgi-bin/link.cgi?143 - -- Posted from 205-158-62-57.outblaze.com [205.158.62.57] via Mailgate.ORG Server - http://www.Mailgate.ORG --------------------------- Newsgroups: comp.os.vxworks Subject: How to get idle TCP/UDP port number which not used by the system Date: 3 Aug 2001 19:31:36 -0700 From: zds914@263.net (Swimmer) Organization: http://groups.google.com/ Message-ID: How to get idle TCP/UDP port number which not used by the system? thanks --------------------------- Newsgroups: comp.os.vxworks Subject: Re: prjConfig.c in BSP for Power PC Date: Fri, 03 Aug 2001 21:05:35 -0700 From: DrDiags Organization: laserlink.net Message-ID: <3B6B748F.25749892@covad.net> References: <3b659edb.851409982@news.demon.co.uk> Reply-To: drdiags@covad.net Jan, There should be a 00networks.cdf in the $(WIND_BASE)/target/config/comps/src/vxWorks where you will find the following: Component INCLUDE_MUX { NAME network mux initialization SYNOPSIS network driver to protocol multiplexer MODULES muxLib.o INIT_RTN muxLibInit (); REQUIRES INCLUDE_NET_SETUP } Place your INIT_RTN in place of the other routine. The caveat here is, why didn't you get the newer *.cdf files when you upgraded your Tornado installation? There may be other symptoms you will have to consider. Anyway, in the Tornado BSP developer kit guide, there is a section that describes the Component definition file syntax and how to add your own Tornado IDE extensions to support your installation. HTH. Jan Larcombe wrote: > > I'm having problems building my BSP, I end up with an unresolved > external muxMaxBinds. I've checked why this is and it seems to be > because the prjConfig.c which is generated is calling the routine > usrMuxLibInit. In earlier versons of my BSP prjConfig.c was calling > MuxLibInit instead. > Obviously there is no point in ediiting prjConfig.c but does anyone > know how to change configurations so that the prjConfig.c file which > is automatically generated will call the correct function. > > Thanks in advance - -- http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html (vxWorks FAQ home page) --------------------------- Newsgroups: comp.os.vxworks Subject: Re: help - one of my task's delay is in negative numbers - Part II Date: Sat, 04 Aug 2001 04:28:30 GMT From: "News Chang" Organization: Excite@Home - The Leader in Broadband http://home.com/faster Message-ID: References: <65464ea6.0108031425.54237c27@posting.google.com> I didn't want to believe that the 'delay' value was negative. So I looked at the TCB of the task and it was zero!! But this task is still pending for some reason. So now I have 2 problems. 1) The task is pending for unknown reasons 2) The information from the 'i' command is incorrect. I traced the function and it basically fills the 'task_desc' structure of td_delay with the value from node_ticks from the structure WIND_TCB. But as I said, a dump of WIND_TCB structure for that task, the node_ticks is 0. So where is td_delay getting that negative number. My only guess is that there is a memory corruption for the structure 'task_desc' so the value there is negative. But the task scheduler does not use 'task_desc' for tasking. I'm sure it uses WIND_TCB structure. So again, why is the task PEND+T (T = timeout) This is getting really weird. Again, any insights greatly appreciated. Mark --------------------------- Newsgroups: comp.os.vxworks Subject: Re: help - one of my task's delay is in negative numbers - btw - we're using Vxworks 5.3.1 Date: Sat, 04 Aug 2001 04:34:39 GMT From: "News Chang" Organization: Excite@Home - The Leader in Broadband http://home.com/faster Message-ID: References: <65464ea6.0108031425.54237c27@posting.google.com> --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Importing WinView Graphs into Word Document Date: 4 Aug 2001 00:37:39 -0700 From: john_94501@yahoo.com (John) Organization: http://groups.google.com/ Message-ID: <488e459a.0108032337.21613b45@posting.google.com> References: <9kf4dm$48h$1@zrtph05m.us.nortel.com> Hello, "Tom Kocur" wrote in message news:<9kf4dm$48h$1@zrtph05m.us.nortel.com>... > I'm sure this has been done before but I can't figure out a way. I'd like > to take a WinView Graph and include it in a Microsoft Word document. The > problem I'm having is that I can only seem to store these as .wvr files. > Can anybody help me out? You could always "screen shot" the window. For Windoze, ALT + the print screen button should copy the current window to the clipboard as a bitmap. Paste that into your document (or into another graphics app if you want to play with it first). For Unix users, xwd gets you an image of a window (click to select); then use a graphics conversion tool, such as xv, to convert to a more usable format. HTH, John... --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Exception Date: Sat, 04 Aug 2001 10:52:19 +0100 From: molochai Message-ID: <6cb0b2a44a%molochai@vapour-trail.demon.co.uk> References: <9ka1eq$aku$1@newsie.singa.pore.net> <3b69140f$0$228@hades.is.co.za> <8bb4f4a34a%molochai@vapour-trail.demon.co.uk> In message Bill Pringlemeir wrote: > >>>>> "Bill" == Bill Pringlemeir writes: > [snip] > Bill> I would be interested to know what features you use that make > Bill> the `C++' exceptions much better than this as I am sure to > Bill> learn something. > > Sorry, I forgot that you may have destroy objects created on the > stack. I was assuming that the code wasn't doing this. I guess that > is why you were talking about the stack unwinding. Stack undwinding refers not only to temporary objects, but also those those that are aggregates or superclasses, all of which need to be constructed (and destructed) atomically with their containing objects, and for which the construction code is compiler generated. This all happens before the programmer-defined costructor for the containing class get exwcuted. If your code is compiled using exceptions, the compiler generates the cleanup code to cope with any one of those aggragate objects or base-class objects failing construction. Without this, there is potential for having partially constructed objects left in a state of limbo. It is just about possible to emulate this behaviour without exception support using some kind of status flags, and writing destructor code which is aware of the possibility of partially costructed objects, but this is a lot of work. Can you really tell me that you would get full test coverage for all the possible failure modes? I'd rather pay the price of a small amount of code bloat, let the compiler take the strain, and have the peace of mind that it works. --------------------------- End of New-News digest ********************** From vxwexplo-errs@csg.lbl.gov Sun Aug 5 04:03:34 2001 From: Vxworks Exploder Date: Sun Aug 5 04:03:37 PDT 2001 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Sun Aug 5 04:03:04 PDT 2001 Subject: Re: VxWorks boot problem Subject: Re: Puzzled by PPC assembler for varargs function Subject: about optimization ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: VxWorks boot problem Date: Sat, 4 Aug 2001 13:40:02 -0700 From: "News" Organization: SBC Internet Services Message-ID: References: <16028583.0108011349.7f98463d@posting.google.com> I've seen similar problem before, in our case, it was the duplicated MAC address that caused the FTP to hang up, a common problem in early stage of development where the target boards don't have their unique MAC address assigned yet.... "Gayathri" wrote in message news:16028583.0108011349.7f98463d@posting.google.com... > Hi, > > I am trying to load Vxworks on a mtx604 board. I have a FTP server > running on my NT machine. Boot is stuck while loading the image. I > have attached my FTP log and the console info. > > console: > [VxWorks Boot]: @ > > boot device : dc > unit number : 0 > processor number : 0 > host name : pc217-41 > file name : target\config\mtx604\vxWorks > inet on ethernet (e) : 192.9.217.168 > host inet (h) : 192.9.217.41 > user (u) : gayathri > ftp password (pw) : vxworks > flags (f) : 0x0 > > Attached TCP/IP interface to dc0. > Attaching network interface lo0... done. > Loading... > > Ftp log: > [L 0017] 08/01/01 14:38:43 Connection accepted from 192.9.217.168 > [C 0017] 08/01/01 14:38:43 Command "USER gayathri" received > [C 0017] 08/01/01 14:38:43 PASSword accepted > [L 0017] 08/01/01 14:38:43 User gayathri logged in. > [C 0017] 08/01/01 14:38:43 Command "TYPE I" received > [C 0017] 08/01/01 14:38:43 TYPE set to I N > [C 0017] 08/01/01 14:38:43 Command "PORT 192,9,217,168,4,1" received > [C 0017] 08/01/01 14:38:43 PORT set to 192.9.217.168 - 1025 (4,1) > [C 0017] 08/01/01 14:38:43 Command "RETR target\config\mtx604\vxWorks" > received > [C 0017] 08/01/01 14:38:43 RETRieve started on file > target\config\mtx604\vxWorks > [C 0017] 08/01/01 14:39:28 Command "PORT 192,9,217,168,4,2" received > [C 0017] 08/01/01 14:39:28 PORT set to 192.9.217.168 - 1026 (4,2) > [C 0017] 08/01/01 14:39:28 Command "RETR target\config\mtx604\vxWorks" > received > [C 0017] 08/01/01 14:39:28 RETRieve started on file > target\config\mtx604\vxWorks > > Line "Command "RETR ...." repeats forever and the boot never > completes. Could anybody please help me with this? > > Thanks in advance, > Gayathri > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Puzzled by PPC assembler for varargs function Date: 4 Aug 2001 15:15:20 -0700 From: Chris Torek Organization: None of The Above Message-ID: <9khs5o$4cv$1@elf.eng.bsdi.com> References: <3B66C3D0.2BCC6823@lucent.com> <3B68F15F.916B411C@lucent.com> Sender: torek@bsdi.com In article David Given wrote: >Varargs functions are a pain, aren't they? Do other platforms, such as the >MIPS (which I'm currently working on, although without an FPU) do anything >similar? Yes, and yes. I am not sure about ABIs for MIPS, but it is normal to pass FP arguments in FPU registers there (for non-varargs/stdarg functions). The "set a bit to mark FP-passing stdarg calls" seems like a good idea, especially with C99's new requirement that all functions be declared before use. Otherwise every stdarg-like function will appear to "use" the FP registers, which will hurt performance on heavier-weight OSes like BSD/OS... :-) - -- In-Real-Life: Chris Torek, Wind River Systems (BSD engineering) El Cerrito, CA, USA Domain: torek@bsdi.com +1 510 234 3167 http://claw.eng.bsdi.com/torek/ (not always up) I report spam to abuse@. --------------------------- Newsgroups: comp.os.vxworks Subject: about optimization Date: 4 Aug 2001 22:17:02 -0700 From: cimsroom@263.net (cimsroom) Organization: http://groups.google.com/ Message-ID: <1c4059c5.0108042117.3a89cf82@posting.google.com> Hi All: There are some limits on optimization of code? I use option -O2 to compile code and found strange error. Vxworks 5.4, Tornado 2.0. target 8260. BR Cims . Room --------------------------- End of New-News digest ********************** From vxwexplo-errs@csg.lbl.gov Mon Aug 6 04:04:04 2001 From: Vxworks Exploder Date: Mon Aug 6 04:04:06 PDT 2001 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Mon Aug 6 04:03:04 PDT 2001 Subject: Re: about optimization Subject: Re: about optimization Subject: Re: script file error in Tornado 1 Subject: Implementation Dependent Software Emulation? Subject: eev_44_drv ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: about optimization Date: Sun, 05 Aug 2001 13:46:18 -0400 From: joe durusau Organization: BELLSOUTH.net & WebUseNet Corp - "ReInventing the UseNet" Message-ID: <3B6D866A.B0FA74CB@bellsosuth.net> References: <1c4059c5.0108042117.3a89cf82@posting.google.com> The short answer is yes. In many cases, for example using a system based on a vme box, most or all optimizations need to be turned off to avoid failures at run time. The general reason is that the compiler cannot have any insight into whether some external device is going to change some data location it has written. This is partially the reason that many vxworks progs are compiled with -fvolatile. Speaking only for myself, Joe Durusau cimsroom wrote: > Hi All: > There are some limits on optimization of code? > I use option -O2 to compile code and found strange error. > Vxworks 5.4, Tornado 2.0. target 8260. > BR > Cims . Room --------------------------- Newsgroups: comp.os.vxworks Subject: Re: about optimization Date: Sun, 05 Aug 2001 21:21:43 +0100 From: molochai Message-ID: <242670a54a%molochai@vapour-trail.demon.co.uk> References: <1c4059c5.0108042117.3a89cf82@posting.google.com> <3B6D866A.B0FA74CB@bellsosuth.net> In message <3B6D866A.B0FA74CB@bellsosuth.net> joe durusau wrote: > The short answer is yes. In many cases, for example using a system > based on a vme box, most or all optimizations need to be turned off > to avoid failures at run time. The general reason is that the compiler > cannot No. This is not a compiler problem - but rather one of system architecture, and possibly poor programming of device drivers. > have any insight into whether some external device is going to change > some data location it has written. Respectfully, this is precisely why C has a volatile modifier, and all MMUs allow memory to be marked as uncachable. > This is partially the reason that many vxworks progs are compiled with > -fvolatile. > Perhaps some device drivers are badly written, but the main reason for - -fvolatile is as a workaround for a fairly well know optimizer bug in GCC2.7.2. > > There are some limits on optimization of code? > > I use option -O2 to compile code and found strange error. > > Vxworks 5.4, Tornado 2.0. target 8260. We used -O3 without -fvolatile with no apparent problems, except for one source code file that caused GCC to crash during compilation with GCC 2.7.2 targetted at PPC603. FWIW we're now using GCC 2.95.4 --------------------------- Newsgroups: comp.os.vxworks Subject: Re: script file error in Tornado 1 Date: Mon, 06 Aug 2001 06:49:45 GMT From: Jerry Hudgins Organization: Western Digital Design, Marin County, California Message-ID: <3B6E3E09.D3F8B2A9@e-farm.com> References: <9jlo4t$1vo$1@hiline.shinbiro.com> Reply-To: Jerry Hudgins This is a problem with the Windows development environment. When the line in the startup script containing the 'ld' command is parsed, the carriage return portion of the line terminator is being erroneously included in the file name. Try replacing the command: ld < hello.o with the equivalent: ld 1,0,"hello.o" The use of quotation marks to delimit the file name will prevent the line termination problem. - -jch --------------------------- Newsgroups: comp.os.vxworks Subject: Implementation Dependent Software Emulation? Date: 6 Aug 2001 01:29:49 -0700 From: chwin@china.com (Adom) Organization: http://groups.google.com/ Message-ID: <92dbccc1.0108060029.21f8c3ab@posting.google.com> Hi,all : we are implemeting ipsec and ike on vxworks . the hardware is ppc860. now it is to be finished ,we are testing it. yesterday,i encounter a problem when the system have run a long time. the exception information is: Implementation Dependent Software Emulation Exception current instruction address: 0x01ae4d78 Machine Status Register: 0x00009030 Condition Register: 0x24020040 Task: 0x1f86970 "tNetTask" i don't know what this means and the reasons. Can you help me? Any advice is appreciated! Thanks! Adom 8/6/2001 --------------------------- Newsgroups: comp.os.vxworks Subject: eev_44_drv Date: 6 Aug 2001 02:31:50 -0700 From: lalitsc@hotmail.com (lalit) Organization: http://groups.google.com/ Message-ID: I am trying to build a project which has "$(BSP_DIR)/eeV/EEV_44_DRV.o" as a value for "Extra Module" macro. Does anybody know what it is and where can I get EEV_44_DRV.o file Thanks. --------------------------- End of New-News digest ********************** From vxwexplo-errs@csg.lbl.gov Mon Aug 6 07:13:45 2001 Date: Mon Aug 6 07:13:47 PDT 2001 Subject: AUTOREPLY comp.os.vxworks newsdigest This is the Mailbox of Friedrich Scheurer. R&D - ACS GmbH - D-72768 Reutlingen Sorry, I am out of office until August 20th 2001! In urgent cases please contact one of the following persons at ACS GmbH: Otto Angerhofer (Managing director) e-mail: oangerhofer@manz-acs.de Eduard Ams (Technical manager) e-mail: eams@manz-acs.de Yvonne Anhorn (Marketing) e-mail: yanhorn@manz-acs.de Margit Arnold (Sales) e-mail: marnold@manz-acs.de Christina Sixel (Sales) e-mail: csixel@manz-acs.de MfG / Kind Regards F. Scheurer ---------------------------------------------------------------- Friedrich Scheurer, Dipl.-Ing. phone +49 7121 9019-23(0) ACS GmbH - R&D fax +49 7121 9019-98 Steigaeckerstrasse 13 web www.manz-acs.de D-72768 Reutlingen (Germany) e-mail: fscheurer@manz-acs.de ---------------------------------------------------------------- From vxwexplo-errs@csg.lbl.gov Mon Aug 6 19:29:29 2001 From: "chris rigano" Date: Mon Aug 6 19:29:32 PDT 2001 Subject: Writting BSP for Power PC and DSP code examples Employment/Consulting This is a multi-part message in MIME format. ------=_NextPart_000_0041_01C11EC8.86614A60 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable Hi all, I have been doing mostly VoIP stack work on VxWorks. I would like to = know where I could find: (1) Public Domain Information on how to write BSPs. I'm especially = interested in Power PC and ARM-7. I would also like to know where to = find more information on the processors themselves. (2) Public Domain information on writing DSP code. (3) Writing SCSI and PCIA interfaces. (4) Gigabit Ethernet code examples. (5) Consulting/Full-time companies that I could look to for VxWorks = relate employment.=20 Many thanks for your time, Christopher ------=_NextPart_000_0041_01C11EC8.86614A60 Content-Type: text/html; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable

Hi all,
 
I have been doing mostly VoIP stack = work on=20 VxWorks. I would like to know where I could find:
(1) Public Domain Information on how = to write=20 BSPs. I'm especially interested in Power PC and ARM-7. I would also like = to know=20 where to find more information on the processors = themselves.
(2) Public Domain information on = writing DSP=20 code.
(3) Writing SCSI and PCIA=20 interfaces.
(4) Gigabit Ethernet code = examples.
(5) Consulting/Full-time companies = that I could=20 look to for VxWorks relate employment.
 
Many thanks for your = time,
 
Christopher
------=_NextPart_000_0041_01C11EC8.86614A60-- From vxwexplo-errs@csg.lbl.gov Mon Aug 6 21:18:59 2001 From: "chris rigano" Date: Mon Aug 6 21:19:01 PDT 2001 Subject: FWritting BSP for Power PC and DSP code examples Employment/Consulting This is a multi-part message in MIME format. ------=_NextPart_000_001E_01C11ED7.D24195C0 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable Hi all, =20 I have been doing mostly VoIP stack work on VxWorks. I would like to = know where I could find: (1) Public Domain Information on how to write BSPs. I'm especially = interested in Power PC and ARM-7. I would also like to know where to = find more information on the processors themselves. (2) Public Domain information on writing DSP code. (3) Writing SCSI and PCIA interfaces. (4) Gigabit Ethernet code examples. (5) Consulting/Full-time companies that I could look to for VxWorks = relate employment.=20 =20 Many thanks for your time, =20 Christopher ------=_NextPart_000_001E_01C11ED7.D24195C0 Content-Type: text/html; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable

 
Hi all,
 
I have been doing mostly VoIP stack = work on=20 VxWorks. I would like to know where I could find:
(1) Public Domain Information on how = to write=20 BSPs. I'm especially interested in Power PC and ARM-7. I would also like = to know=20 where to find more information on the processors = themselves.
(2) Public Domain information on = writing DSP=20 code.
(3) Writing SCSI and PCIA=20 interfaces.
(4) Gigabit Ethernet code = examples.
(5) Consulting/Full-time companies = that I could=20 look to for VxWorks relate employment.
 
Many thanks for your = time,
 
Christopher
------=_NextPart_000_001E_01C11ED7.D24195C0-- From vxwexplo-errs@csg.lbl.gov Tue Aug 7 01:48:27 2001 From: "David Anderson" Date: Tue Aug 7 01:48:29 PDT 2001 Subject: Network performance VxWorks 5.3.1 & Tornado 1.0.1 Using SENS 1.1 Dear Xperts, Can anyone point me to any area(s) of vxWorks which cause apparently slow network activity, I'm not entirely sure what sort of performance to expect over a 10 M Ethernet link, but we seem to see about a third of the transfer rate that uCLinux is producing. We are using the CS8900 Ethernet controller. Any ideas - general or specific , would be appreciated. David Anderson. NEW E-MAIL ADDRESS: mailto:david.anderson@dtrack.com ------------------------------------- FF: 2B + ~2B , that is the question ? ------------------------------------- Disclaimer ---------- This email and any files transmitted with it are confidential and intended solely for the use of the individual or entity to whom they are addressed. If you have received this email in error please destroy and notify Data Track Technology Plc +44 1425 271900. ------------------------------------------------------- From vxwexplo-errs@csg.lbl.gov Tue Aug 7 04:03:51 2001 From: Vxworks Exploder Date: Tue Aug 7 04:03:54 PDT 2001 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Tue Aug 7 04:03:17 PDT 2001 Subject: License prices Subject: Re: Can not boot FADS860T from TFFS? Subject: etherOutput pb ? Subject: Re: License prices Subject: ifunit function Subject: Re: License prices Subject: Re: Implementation Dependent Software Emulation? Subject: Re: Far-End Fault-Indecation for Ethernet 100MB-FX Subject: Re: eev_44_drv Subject: Re: eev_44_drv Subject: Re: eev_44_drv Subject: OPC for VxWorks Subject: How does gcc -g0 option work? Subject: Re: How to measure CPU utilization of a task at run time Subject: Re: OPC for VxWorks Subject: 'LOADER_RELOCATION_TOO_LARGE' ERROR Subject: Main difference between ATA / IDE drive? Subject: How to use the Intel 82559ER Lan Controller in the vxworks system? Subject: MPC857T : UT bit set in PDPAR register Subject: Write to disk files ASAP Subject: Re: About ping large packet on vxWorks? Subject: Re: Main difference between ATA / IDE drive? Subject: can i use socket that is ipc with non-blocking mode? Subject: how can i use select() as non-blocking mode? Subject: how can i use select() as non-blocking mode? Subject: Re: eev_44_drv Subject: how can i use select() as non-blocking mode? Subject: Re: Write to disk files ASAP Subject: Virtual Memory Subject: how to write net driver Subject: question about taskspawn Subject: sp in shell Subject: std_in input Subject: Network performance Subject: Looking for driver example Subject: memShow (memory leaks) Subject: Re: eev_44_drv Subject: graphic library ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: License prices Date: 6 Aug 2001 05:26:12 -0700 From: brahim_mouhdi@hotmail.com (Brahim) Organization: http://groups.google.com/ Message-ID: <899dceb5.0108060426.38a9084b@posting.google.com> Hello, I've heard that VxWorks licenses are very expensive, but we want to learn developing with VxWorks and cannot buy a license because we don't have a client requiring it. Isn't it possible to obtain a license for studying purposes?? Greetings. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Can not boot FADS860T from TFFS? Date: Mon, 06 Aug 2001 08:34:26 -0500 From: Dinker Charak Organization: Fermi National Accelerator Laboratory Message-ID: <3B6E9CE2.661E8A13@fnal.gov> References: <838b088e.0108010142.3404de96@posting.google.com> <3B681191.296AB17D@fnal.gov> <3B6811E1.3839D71C@fnal.gov> >Hi, > >Thanks for your help. We have solved the problem according to your reply. But we >wonder if it would cause some bugs when we commented the >muxDevStopAll function. i had looked deeper into the problem and had found that in motFec.c there was some declaration where it had set an wait for operation to forever and we changed it to soe fixed value, 5, i guess and it used to time out and the muxDevStopAll never got hung up ... i do not exactly remember what the change was ... if possible send me this file and i will point it out for you dinker --------------------------- Newsgroups: comp.os.vxworks Subject: etherOutput pb ? Date: Mon, 6 Aug 2001 15:41:03 +0200 From: "News" Message-ID: <3b6e9d78$1_2@news.vo.lu> Hi, I want to use etherOutput function, but a I have a little pb. I use bcopy or memcpy to add ethernet adresse into struct ether_header. No pb with the compilator. But when I execute this prog, an error data access appears. The target is a Ramix 694 card. data access Exception current instruction address: 0x00088c50 Machine Status Register: 0x0000b030 Data Access Register: 0x10000015 Condition Register: 0x82400082 Data storage interrupt Register: 0x0000b030 e5f1c vxTaskEntry +60 : shell () d1a00 shell +18c: d1a2c () d1c50 shell +3dc: execute () d1dd4 execute +d8 : yyparse () 1045e4 yyparse +7a8: 10252c () 1026a4 yystart +8f8: memcpy () 9a720 memcpy +28 : bcopy () Regards, Christophe --------------------------- Newsgroups: comp.os.vxworks Subject: Re: License prices Date: 6 Aug 2001 13:46:38 GMT From: "Johan Borkhuis" Organization: Agere Systems Message-ID: References: <899dceb5.0108060426.38a9084b@posting.google.com> brahim_mouhdi@hotmail.com (Brahim) wrote: > Hello, > > I've heard that VxWorks licenses are very expensive, but we want to > learn developing with VxWorks and cannot buy a license because we > don't have a client requiring it. Isn't it possible to obtain a > license for studying purposes?? Yes, try Tornado prototyper, that is a simulator running on a PC. 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: ifunit function Date: Mon, 6 Aug 2001 16:38:52 +0200 From: "News" Message-ID: <3b6eab06$1_2@news.vo.lu> Hi, I want to recuperate pIf pointeur, because I still want to send an Ethernet trame. So I use this following function ifunit( ). But I don't know which variable I have to put in ( rx or lo or ln ). I don't the differents between these variables. I also use rx for etherInputHookAdd function. Regards. christophe --------------------------- Newsgroups: comp.os.vxworks Subject: Re: License prices Date: Mon, 06 Aug 2001 14:24:49 GMT From: Tim Sohacki Organization: Road Runner - NC Message-ID: <3B6EA9CA.7EDC12CB@nc.rr.com> References: <899dceb5.0108060426.38a9084b@posting.google.com> Johan Borkhuis wrote: > brahim_mouhdi@hotmail.com (Brahim) wrote: > > I've heard that VxWorks licenses are very expensive, but we > > want to learn developing with VxWorks and cannot buy a license > > because we don't have a client requiring it. Isn't it possible > > to obtain a license for studying purposes?? > > Yes, try Tornado prototyper, that is a simulator running on a PC. > For more info on the prototyper see: http://www.windriver.com/prototyper/introsheet.html --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Implementation Dependent Software Emulation? Date: 6 Aug 2001 08:45:37 -0700 From: john_94501@yahoo.com (John) Organization: http://groups.google.com/ Message-ID: <488e459a.0108060745.43c561d1@posting.google.com> References: <92dbccc1.0108060029.21f8c3ab@posting.google.com> Hello, The exception means that the network task has tried to execute an instruction that doesn't exist in the PPC instruction set (i.e. an illegal opcode). The reasons for this could be many, including the following: 1) An inappropriate pointer passed to tNetTask via netJobAdd() 2) Corruption of the task's stack 3) Corruption of the kernel code and/or data spaces 4) A bug in the network stack code First thing to check is what routine, if any, is at 0x01ae4d78 (the address that caused the exception). Then try a stack trace on tNetTask (using tt from the shell for example) to see if that provides any insight as to the cause. HTH, John... chwin@china.com (Adom) wrote in message news:<92dbccc1.0108060029.21f8c3ab@posting.google.com>... > Hi,all : > we are implemeting ipsec and ike on vxworks . > the hardware is ppc860. > now it is to be finished ,we are testing it. > yesterday,i encounter a problem when the system > have run a long time. > > the exception information is: > > Implementation Dependent Software Emulation > Exception current instruction address: 0x01ae4d78 > Machine Status Register: 0x00009030 > Condition Register: 0x24020040 > Task: 0x1f86970 "tNetTask" > > i don't know what this means and the reasons. > Can you help me? > Any advice is appreciated! > > Thanks! > Adom > 8/6/2001 --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Far-End Fault-Indecation for Ethernet 100MB-FX Date: Mon, 06 Aug 2001 16:57:30 +0100 From: James Marshall Organization: Agilent Technologies UK Ltd Message-ID: <3B6EBE6A.3AAD7D37@agilent.com> References: <9kebn5$pn5@parzival.khe.siemens.de> Have you tried reading the PHY status register twice? I do not have experience with fibre optic ethernet, but 100BaseT PHYs have similar latched-until-read status bits. James Marshall. Michael Koestner wrote: > Hi, > does anyone have a implementation for a Far-End Fault-Indecation ethernet > fiber optic connection. the problem work with the phy device, which have > latched bit implemented to signal a dynamic state. > Mfg > Michael Koestner --------------------------- Newsgroups: comp.os.vxworks Subject: Re: eev_44_drv Date: Mon, 06 Aug 2001 17:01:27 +0100 From: James Marshall Organization: Agilent Technologies UK Ltd Message-ID: <3B6EBF57.BFE4F4FD@agilent.com> References: It's Intel's driver for the 82557/8/9 LAN controller. You can pick up a copy from developer.intel.com My experience with this driver is only temporary as it did not behave and I was able to return to WindRiver's fei END driver which seems better. The other problem with EEV is that Intel direct you to WindRiver for support, but WRS don't support it. James Marshall. lalit wrote: > I am trying to build a project which has "$(BSP_DIR)/eeV/EEV_44_DRV.o" > as a value for "Extra Module" macro. Does anybody know what it is and > where can I get EEV_44_DRV.o file > > Thanks. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: eev_44_drv Date: Mon, 6 Aug 2001 18:01:30 +0200 From: "Werner Schiendl" Message-ID: <3b6ec000$0$22316$6e365a64@newsreader02.highway.telekom.at> References: Hi, We have this file too, it is the driver for the Intel Ether Express PRO100B PCI. You can get it from the Intel Website. I'm afraid I don't have an URL at hand... hth Werner --------------------------- Newsgroups: comp.os.vxworks Subject: Re: eev_44_drv Date: Mon, 06 Aug 2001 20:38:08 +0100 From: molochai Message-ID: References: <3B6EBF57.BFE4F4FD@agilent.com> In message <3B6EBF57.BFE4F4FD@agilent.com> James Marshall wrote: > It's Intel's driver for the 82557/8/9 LAN controller. You can pick up a > copy from developer.intel.com > > My experience with this driver is only temporary as it did not behave and I > was able to return to WindRiver's fei END driver which seems better. The > other problem with EEV is that Intel direct you to WindRiver for support, > but WRS don't support it. Further more, it is a binary only release... So absolutely no use if you don't use x86 hardware... The WindRiver driver seems to be fine. > > > I am trying to build a project which has "$(BSP_DIR)/eeV/EEV_44_DRV.o" > > as a value for "Extra Module" macro. Does anybody know what it is and > > where can I get EEV_44_DRV.o file --------------------------- Newsgroups: comp.os.vxworks Subject: OPC for VxWorks Date: Mon, 6 Aug 2001 16:48:05 -0400 From: "Kevin Livingston" Message-ID: I was courious if anyone else was useing OPC in conjunction with VxWorks, specifically using the VxWorks platform as an OPC server? Are their any good toolkit for VxWorks. I know WindRiver has a product VxOPC, however, it's price makes it a little bit prohibitive to use. Are their any other options? what are your experiences? thanks Kevin --------------------------- Newsgroups: gnu.gcc.help,comp.os.vxworks Subject: How does gcc -g0 option work? Date: 6 Aug 2001 14:59:34 -0700 From: jhowellatwork@yahoo.com (John Howell) Organization: http://groups.google.com/ Message-ID: I noticed several messages in gnu.gcc.help which include -g0 in the command line. In the source for gcc, it appears that -g0 is the same as if -g was never specified. Is this true? Is gcc -g -g0 the same as gcc The gcc documentation only mentions -g, -g1, -g2, and -g3. Is -g0 documented anywhere? John --------------------------- Newsgroups: comp.os.vxworks Subject: Re: How to measure CPU utilization of a task at run time Date: Mon, 30 Jul 2001 07:52:24 +0200 From: "Roderik Wildenburg" Organization: MAN Roland Druckmaschinen AG Message-ID: <9k2smo$ord$1@aras.mra.man.de> References: <3B5F22AE.70C159C1@ntl.inf.com> Simply use Spylib (WRS library) . You have to activate it in your BSP. Sunil Kulkarni schrieb in im Newsbeitrag: 3B5F22AE.70C159C1@ntl.inf.com... > How to measure the CPU utilization of a task in runtime > > so that some action can be taken based on the % CPU > > utilization of the task. > > Thanks, > > Sunil > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: OPC for VxWorks Date: Tue, 07 Aug 2001 00:57:05 GMT From: Jerry Hudgins Organization: Western Digital Design, Marin County, California Message-ID: <3B6F3CE1.2FC27307@e-farm.com> References: Reply-To: Jerry Hudgins Kevin Livingston wrote: > > I was courious if anyone else was useing OPC in conjunction with > VxWorks, specifically using the VxWorks platform as an OPC server? It's possible to roll your own server using VxDCOM - I've done it - but the server toolkit bundled with VxOPC would likely make server implementation much faster and easier. If you do use VxDCOM alone, make sure that you're up-to-date patch-wise. Early builds had some significant OPC-related problems that have since been addressed. - -jch --------------------------- Newsgroups: comp.os.vxworks Subject: 'LOADER_RELOCATION_TOO_LARGE' ERROR Date: 6 Aug 2001 18:17:03 -0700 From: ellin_lin@263.net (ellin) Organization: http://groups.google.com/ Message-ID: <44a01eba.0108061717.6f3a6db2@posting.google.com> I built a project with T2 of kernel PPCEC602 for 8260 target, but when I tried to download it to target, I got error of "Not fit in 24bit" and "LOADER_RELOCATION_TOO_LARGE". What can cause this and how to avoid it? Thanks in advance. ellin --------------------------- Newsgroups: comp.os.vxworks Subject: Main difference between ATA / IDE drive? Date: 6 Aug 2001 18:29:18 -0700 From: romid@brightlink.com (ramin omid) Organization: http://groups.google.com/ Message-ID: <1fa60c7c.0108061729.1bb0aca@posting.google.com> What is the main difference between ATA and IDE drives? thanks, Ramin --------------------------- Newsgroups: comp.os.vxworks Subject: How to use the Intel 82559ER Lan Controller in the vxworks system? Date: 6 Aug 2001 18:36:02 -0700 From: srs_zh@263.net (srs_zh) Organization: http://groups.google.com/ Message-ID: I build the Intel 82559ER Lan Controller for X86,but the system can't work rightly!Who can help me? I will greatly appriciate your help! --------------------------- Newsgroups: comp.os.vxworks Subject: MPC857T : UT bit set in PDPAR register Date: Tue, 7 Aug 2001 10:42:28 +0900 From: "Ogre" Organization: A poorly-installed InterNetNews site Message-ID: <9kngm3$fvk$1@news2.kornet.net> UT bit in PDPAR register determines whether the parameter RAM¡¯s page 4 (SCC4) operates in serial or UTOPIA mode. I want to use MPC857T in simultaneous MII and UTOPIA operation (UTOPIA muxed mode). My source is following as .. - -------------------------------------- sysFecEnetEnable() { 01 : /* mask IRQ7 off, as it is shared with MII_TX_CLK */ 02 : *SIMASK &= ~SIMASK_IRM7 ; 03 : /* also clear any pending interrupt */ 04 : *SIPEND |= SIPEND_IRQ7 ; 05 : *SDCR |= SDCR_FAID_BR6 ; 06 : *PDPAR = 0xC000 ; or 06 : *PDPAR = 0x4000 ; 07 : *PDPAR = 0x0080 ; 08 : *PDDIR = 0x0080 ; 09 : *UTMODE &= 0xFFFFFF7F ; } motFecStart() { ..... 55 : sysFecEnetEnable() ; 56 : MOT_FEC_PINMUX_ENABLE_REVD ; ... 70 : MOT_FEC_ETH_ENABLE_REVD ; } - ----------------------------------------- 01 ~ 05 is default. 06 set UT bit and ATM bit. or 06 set UT bit. 07 set MDC for ethernet management clock. 09 set muxed mode. 56 set FEC_PINMUX bit 70 set FEC_PINMUX and ETHER_EN bit. but, at 06, system crashed. To use simultaneous MII and UTOPIA, MPC857T has UT = 1, FEC_PINMUX = 1, SPLIT = 0(MUX) configuration. but, crashed.. -_-; I hope this problem will be solved. thanks any response. Ogre --------------------------- Newsgroups: comp.os.vxworks Subject: Write to disk files ASAP Date: Mon, 06 Aug 2001 21:50:24 -0400 From: Klein family Message-ID: <3B6F495B.9800BF9E@erols.com> I am trying to write to disk files as fast as possible without losing any of the input data. I can expect to be generating many GB of data in a single start/stop sequence, which I will be putting into a series of 2GB files. I do not want to lose any data at the start or during file switch. We are using DOSFS2. An additional problem is that software does not recieve a "prepare to exit" signal, so it always has to be prepared to begin recording. I have found that allocating disk space takes longer as the disk fills, gradually reducing our throughput rate until we lose data. So far I have tried: 1. sequentially opening, writing and closing files. The throughput rate gradually decreases. 2. allocating contiguous disk space [ioctl(fd, FIOCONTIG,0x7fffffff)] when the start command is recieved. This can take seconds for each file. 3. opening files and allocating contiguous disk space when the program starts; followed by repeated after the stop command is recieved. I found that the disk becomes unusable after a number of cycles because disk space does not get deallocated and chkdsk does not do a very good job of fixing the disk allocation table. (dosFsShow says that I have 140,000GB free on a 73GB disk.) My current scheme is to create 2GB files on the disk in advance, either before running the program or at program start. I am doing this as follows: until the disk is full: create a file allocate 2GB contiguous disk space lseek to 2GB - 512 write a 512 bytes block close the file The problem with this is that the write call takes a long time, during which the "in use" light on the disk flashes. It is as if the write is writing the entire 2GB of the file. Even though it can be done on an off-line basis, I don't want to spend hours preparing every disk. Also, I don't know of any other probelms with this scheme as I just finished coding it today and have only given it a minimal exercize so far. Can anyone suggest any other schemes or an improvement to what I've already tried. Thanks. Thanks Scott Klein Northrop Grumman Baltimore, MD 410-765-0910 scott_b_klein@md.northgrum.com --------------------------- Newsgroups: comp.os.vxworks Subject: Re: About ping large packet on vxWorks? Date: Tue, 7 Aug 2001 10:52:42 +0900 From: "Ogre" Organization: A poorly-installed InterNetNews site Message-ID: <9knh97$hs4$1@news2.kornet.net> References: <9keakm$njs$1@mail.cn99.com> I think that your 100M ethernet source does not have buffer enough for ping response. Ogre "Xiao Boyu" wrote in message news:9keakm$njs$1@mail.cn99.com... > We issue following command"ping 172.16.64.192 -t -l 35000" from our host to target board > (FADS860T from mot) running vxworks, vxWorks was configure to use FEC of 860T with 100M connection and vxWorks can't respond to this correctly.We test vxWorks using SCC1 of CPM running > 10M connection and it works well.Further test show the problem is related with vxWorks not > 860T. Can anyone help us on this question? > Thank you in advance. > > Sincerely yours, > Xiao > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Main difference between ATA / IDE drive? Date: Tue, 07 Aug 2001 02:51:52 GMT From: Harvey Taylor Organization: MTS Internet Message-ID: <3B6F731F.6A70@despam.autobahn.mb.ca> References: <1fa60c7c.0108061729.1bb0aca@posting.google.com> In article <1fa60c7c.0108061729.1bb0aca@posting.google.com>, ramin omid wrote: > > What is the main difference between ATA and IDE drives? > Check the FAQ: http://thef-nym.sci.kun.nl/~pieterh/storage.html - -het - -- Everyone has the right to life, liberty and security of person. - -UDHR:Article 3: http://www.autobahn.mb.ca/~het/udhr.html H.E. Taylor http://www.autobahn.mb.ca/~het/ --------------------------- Newsgroups: comp.os.vxworks Subject: can i use socket that is ipc with non-blocking mode? Date: Tue, 07 Aug 2001 04:52:57 GMT From: "±èµ¿Çö" Organization: Dacom Message-ID: hi everyone. i use socket as ipc. i made some protocol stack. and use sockets as communication equipment between Lays. but each process get into ready state. How can i solve this problem? - -- - ------------------------------------ LG ÀüÀÚ µðÁöÅÐÀ̵¿Åë½Å ¿¬±¸¼Ò À̵¿Åë½Å ±â±â½Ç ¼±ÀÓ ¿¬±¸¿ø ±èµ¿Çö(bigstar@lge.com) Kim Dong-Hyun(bigstar@lge.com) Research Engineer Mobile Communication Equipment DEPT. Digital Mobile COMM. LAB. LG R&D Complex 533,Hogye-dong,Dongan-gu,Anyang-shi, Kyongki-do, 431-080 , Korea Phone: 82-31-450-4013 Fax : 82-31-450-4059 - ------------------------------------ --------------------------- Newsgroups: comp.os.vxworks Subject: how can i use select() as non-blocking mode? Date: Tue, 07 Aug 2001 05:01:11 GMT From: "±èµ¿Çö" Organization: Dacom Message-ID: hi i use socket at ipc(interprocess control) between Layer 2 and Layer 3. I/O mode is asynchronous mode, and use timval. There are no problem socket as ipc at UNIX. But at Vxworks( cpu is mpc8260), each process get into ready state and system is holted. how can i solve this problem? is Vxwork support non-blocking mode. --------------------------- Newsgroups: comp.os.vxworks Subject: how can i use select() as non-blocking mode? Date: Tue, 07 Aug 2001 05:01:12 GMT From: "±èµ¿Çö" Organization: Dacom Message-ID: hi i use socket at ipc(interprocess control) between Layer 2 and Layer 3. I/O mode is asynchronous mode, and use timval. There are no problem socket as ipc at UNIX. But at Vxworks( cpu is mpc8260), each process get into ready state and system is holted. how can i solve this problem? is Vxwork support non-blocking mode. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: eev_44_drv Date: 6 Aug 2001 22:12:02 -0700 From: lalitsc@hotmail.com (lalit) Organization: http://groups.google.com/ Message-ID: References: <3b6ec000$0$22316$6e365a64@newsreader02.highway.telekom.at> Hi, Thanks for the reply, as you said you have this file could you please email me the file. I tried searching it on intel's site but could not find it. My email id is lalitsc@hotmail.com. Thanks, Lalit. "Werner Schiendl" wrote in message news:<3b6ec000$0$22316$6e365a64@newsreader02.highway.telekom.at>... > Hi, > > We have this file too, it is the driver for the Intel Ether Express PRO100B > PCI. > You can get it from the Intel Website. > > I'm afraid I don't have an URL at hand... > > hth > Werner --------------------------- Newsgroups: comp.os.vxworks Subject: how can i use select() as non-blocking mode? Date: Tue, 07 Aug 2001 05:32:19 GMT From: "±èµ¿Çö" Organization: Dacom Message-ID: hi i use socket at ipc(interprocess control) between Layer 2 and Layer 3. I/O mode is asynchronous mode, and use timval. There are no problem socket as ipc at UNIX. But at Vxworks( cpu is mpc8260), each process get into ready state and system is holted. how can i solve this problem? is Vxwork support non-blocking mode. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Write to disk files ASAP Date: 6 Aug 2001 23:11:05 -0700 From: john_94501@yahoo.com (John) Organization: http://groups.google.com/ Message-ID: <488e459a.0108062211.7d7c2587@posting.google.com> References: <3B6F495B.9800BF9E@erols.com> Hello, Have you looked at the raw file system? It sounds like your solution so far is to try to replicate the behaviour of the raw file system under DOSFS - you may as well just remove the DOS layer and work directly with the disk via rawFs. I seem to remember that others have also used this approach in the past for streaming data apps... HTH, John... Klein family wrote in message news:<3B6F495B.9800BF9E@erols.com>... > I am trying to write to disk files as fast as possible without losing > any of the input data. I can expect to be generating many GB of data in > a single start/stop sequence, which I will be putting into a series of > 2GB files. I do not want to lose any data at the start or during file > switch. We are using DOSFS2. An additional problem is that software does > not recieve a "prepare to exit" signal, so it always has to be prepared > to begin recording. > > I have found that allocating disk space takes longer as the disk fills, > gradually reducing our throughput rate until we lose data. So far I have > tried: > 1. sequentially opening, writing and closing files. The throughput rate > gradually decreases. > 2. allocating contiguous disk space [ioctl(fd, FIOCONTIG,0x7fffffff)] > when the start command is recieved. This can take seconds for each file. > > 3. opening files and allocating contiguous disk space when the program > starts; followed by repeated after the stop command is recieved. I found > that the disk becomes unusable after a number of cycles because disk > space does not get deallocated and chkdsk does not do a very good job of > fixing the disk allocation table. (dosFsShow says that I have 140,000GB > free on a 73GB disk.) > > My current scheme is to create 2GB files on the disk in advance, either > before running the program or at program start. I am doing this as > follows: > until the disk is full: > create a file > allocate 2GB contiguous disk space > lseek to 2GB - 512 > write a 512 bytes block > close the file > The problem with this is that the write call takes a long time, during > which the "in use" light on the disk flashes. It is as if the write is > writing the entire 2GB of the file. Even though it can be done on an > off-line basis, I don't want to spend hours preparing every disk. Also, > I don't know of any other probelms with this scheme as I just finished > coding it today and have only given it a minimal exercize so far. > > Can anyone suggest any other schemes or an improvement to what I've > already tried. Thanks. > > > Thanks > > Scott Klein > Northrop Grumman > Baltimore, MD > 410-765-0910 > scott_b_klein@md.northgrum.com --------------------------- Newsgroups: comp.os.vxworks Subject: Virtual Memory Date: Tue, 7 Aug 2001 16:20:41 +0800 From: "john tian" Organization: Lucent Technologies, Columbus, Ohio Message-ID: <9ko831$787@nntpa.cb.lucent.com> How to distinguish system memory pool, memory partition, no-cachable memory. thamks; John Tian tian_jun@yahoo.com --------------------------- Newsgroups: comp.os.vxworks Subject: how to write net driver Date: Tue, 7 Aug 2001 16:39:11 +0800 From: "john tian" Organization: Lucent Technologies, Columbus, Ohio Message-ID: <9ko95q$7b3@nntpa.cb.lucent.com> hello, There is a confusing problem about net driver: TX interrupt--- Send all data in buffers, after interface card transfer data totally, then produce the next TX interrupt, but no data is transminted for some time. who triger a new TX interrupt, if it is timer, how long? --------------------------- Newsgroups: comp.os.vxworks Subject: question about taskspawn Date: 7 Aug 2001 01:47:05 -0700 From: fengwk@bhnec.nec.co.jp (bhnecvxworks) Organization: http://groups.google.com/ Message-ID: How can I set a break point,in a task, which is initized by taskspawn? because, In debug mode, the break point can never take effect. --------------------------- Newsgroups: comp.os.vxworks Subject: sp in shell Date: 7 Aug 2001 01:52:56 -0700 From: fengwk@bhnec.nec.co.jp (bhnecvxworks) Organization: http://groups.google.com/ Message-ID: I created a bootable task, for example, sp myinittask, arg1, arg2 however, I can not spawn myinittask from target shell, I had scaled symbol table in my project. --------------------------- Newsgroups: comp.os.vxworks Subject: std_in input Date: 7 Aug 2001 01:59:59 -0700 From: fengwk@bhnec.nec.co.jp (bhnecvxworks) Organization: http://groups.google.com/ Message-ID: In downloadable project WDBrpc debug mode, I want to get a char input from keyboard that is char inchar = getchar(), how can I input STD_IN from keyboard. --------------------------- Newsgroups: comp.os.vxworks Subject: Network performance Date: 7 Aug 2001 02:46:54 -0700 From: david.anderson@dtrack.com (David Anderson) Organization: http://groups.google.com/ Message-ID: VxWorks 5.3.1 & Tornado 1.0.1 Using SENS 1.1 Dear Xperts, Can anyone point me to any area(s) of vxWorks which cause apparently slow network activity, I'm not entirely sure what sort of performance to expect over a 10 M Ethernet link, but we seem to see about a third of the transfer rate that uCLinux is producing. We are using the CS8900 Ethernet controller. Any ideas - general or specific , would be appreciated. David Anderson. david.anderson@dtrack.com --------------------------- Newsgroups: comp.os.vxworks Subject: Looking for driver example Date: Tue, 7 Aug 2001 11:45:44 +0200 From: "Rui" Organization: KGB Corp Message-ID: <9koder$9nt$1@news1.xs4all.nl> Hi : I'm looking for an example on how to implement a driver in VxWorks. One that it's better than the one that comes with Tornado. Something with also the read and write operations. Poupose? Just for training. The pourpoise is to make a virtual device and read with that virtual driver. So, no interrupt routine necessary. If anyone knows where I can find fin an example or has one please inform me. Regards Rui --------------------------- Newsgroups: comp.os.vxworks Subject: memShow (memory leaks) Date: 7 Aug 2001 02:52:07 -0700 From: ike@1freeemail.com (Ike Robinson) Organization: http://groups.google.com/ Message-ID: <1a86250f.0108070152.440d2e7@posting.google.com> Hello, I have some kind of newbie question. First, I'm using VxWorks (for Motorola MCP750 - MPC 750) version 5.4. Kernel: WIND version 2.5. I have a bit complicated program which allocates some memory for internal structures and then release it. My problem is that after first run of this program memShow shows that I have 80 bytes less free memory. Next calls of this program doesn't show free memory decrease. My question is: is it memory leaks in program or vxWorks initially reserves these 80 bytes for some reason? I started to suspect that vxWorks can eats these 80 bytes when I ran memShow before 'cd' command and after. I realized that 'cd' eats 8 bytes (probably for some internal heap/stack). Or my guess is wrong and it definitely memory leaks in program (actually third-party, so I can't view it's source)? Thanks in advance, Ike --------------------------- Newsgroups: comp.os.vxworks Subject: Re: eev_44_drv Date: Tue, 7 Aug 2001 11:24:22 +0200 From: "Werner Schiendl" Message-ID: <3b6fbab4$0$4126$6e365a64@newsreader02.highway.telekom.at> References: <3b6ec000$0$22316$6e365a64@newsreader02.highway.telekom.at> Hi, look at http://developer.intel.com/platforms/applied/software/d559vx.htm http://developer.intel.com/platforms/applied/software/b559mux.htm the first is an 'old-style' BSD ethernet driver, the second uses VxWorks' END driver model. Unfortunately, they BOTH contain a file with that very name :-( hth Werner --------------------------- Newsgroups: comp.os.vxworks Subject: graphic library Date: Tue, 7 Aug 2001 12:02:08 +0200 From: "Stig Nordmoen" Organization: Kongsberg Defence & Aerospace, Norway Message-ID: <9koe80$8kf$1@spiten.kongsberg.com> I am loocking for a graphic library for the C&T 65550 video controller. I want to write text/lines and rectangels on the screen (flat panel), defined with pixel coordinates. Do I have other choices then WindML. Stig Nordmoen --------------------------- End of New-News digest ********************** From vxwexplo-errs@csg.lbl.gov Tue Aug 7 06:28:48 2001 From: "Suresh T" Date: Tue Aug 7 06:28:50 PDT 2001 Subject: reg Makefile Hi All, We want to run the make file generated from the Tornado Workspace (downloadable project) from the Command prompt. We have made sure the Environment Variables are set properly. While running make command we get the following error: make: *** No rule to make target `linkedObjs.o', needed by `exe'. Stop Please provide inputs on this. Advanced Thanks, Regards, Suresh From vxwexplo-errs@csg.lbl.gov Tue Aug 7 06:52:38 2001 Date: Tue Aug 7 06:52:40 PDT 2001 Subject: AUTOREPLY comp.os.vxworks newsdigest This is the Mailbox of Friedrich Scheurer. R&D - ACS GmbH - D-72768 Reutlingen Sorry, I am out of office until August 20th 2001! In urgent cases please contact one of the following persons at ACS GmbH: Otto Angerhofer (Managing director) e-mail: oangerhofer@manz-acs.de Eduard Ams (Technical manager) e-mail: eams@manz-acs.de Yvonne Anhorn (Marketing) e-mail: yanhorn@manz-acs.de Margit Arnold (Sales) e-mail: marnold@manz-acs.de Christina Sixel (Sales) e-mail: csixel@manz-acs.de MfG / Kind Regards F. Scheurer ---------------------------------------------------------------- Friedrich Scheurer, Dipl.-Ing. phone +49 7121 9019-23(0) ACS GmbH - R&D fax +49 7121 9019-98 Steigaeckerstrasse 13 web www.manz-acs.de D-72768 Reutlingen (Germany) e-mail: fscheurer@manz-acs.de ---------------------------------------------------------------- From vxwexplo-errs@csg.lbl.gov Tue Aug 7 07:38:55 2001 From: Kumaraswamy Mudide Date: Tue Aug 7 07:38:57 PDT 2001 Subject: Re: Network performance I do not think the problem lies in the vxWorks. It could be because of subnet overloaded. This can happen when more hosts are in one subnet and leads slow response. Bye swamy --- the vxWorks Users Group Exploder wrote: > Submitted-by vxwexplo-errs@csg.lbl.gov Tue Aug 7 > 01:48:27 2001 > Submitted-by: "David Anderson" > > > VxWorks 5.3.1 & Tornado 1.0.1 > Using SENS 1.1 > > Dear Xperts, > > Can anyone point me to any area(s) of vxWorks which > cause apparently slow > network activity, I'm not > entirely sure what sort of performance to expect > over a 10 M Ethernet link, > but we seem to see about > a third of the transfer rate that uCLinux is > producing. We are using the > CS8900 Ethernet controller. > > Any ideas - general or specific , would be > appreciated. > > > David Anderson. > NEW E-MAIL ADDRESS: > mailto:david.anderson@dtrack.com > > ------------------------------------- > FF: 2B + ~2B , that is the question ? > ------------------------------------- > > Disclaimer > ---------- > This email and any files transmitted with it are > confidential and intended > solely for the use of the individual or entity to > whom they are addressed. > > If you have received this email in error please > destroy and notify Data > Track Technology Plc +44 1425 271900. > ------------------------------------------------------- > > > ********** > > This is a user group mailing list for vxWorks > related topics > see http://www-vxw.lbl.gov/vxworks for posting > and subscription info __________________________________________________ Do You Yahoo!? Make international calls for as low as $.04/minute with Yahoo! Messenger http://phonecard.yahoo.com/ From vxwexplo-errs@csg.lbl.gov Wed Aug 8 04:03:38 2001 From: Vxworks Exploder Date: Wed Aug 8 04:03:41 PDT 2001 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Wed Aug 8 04:03:17 PDT 2001 Subject: Re: How to use the Intel 82559ER Lan Controller in the vxworks system? Subject: I Like PowerPC and Vxworks Subject: sending float value from window to vxwork over TCP/IPsocket Subject: Re: Write to disk files ASAP Subject: Multiple VME slave windows Subject: Re: eev_44_drv Subject: Re: I Like PowerPC and Vxworks Subject: Re: I Like PowerPC and Vxworks Subject: Re: Write to disk files ASAP Subject: Re: Exception Subject: Re: graphic library Subject: Re: OPC for VxWorks Subject: Re: Network performance Subject: Varible name conflicts Subject: Re: Exception Subject: A question for VxWorks Web server Subject: info on writing device drivers Subject: Re: question about taskspawn Subject: Re: how to write net driver Subject: Connect() Subroutine Subject: Re: Not able to ping to the board Subject: dosFs2Lib: Anyone using it? Subject: Re: 'LOADER_RELOCATION_TOO_LARGE' ERROR Subject: Re: OPC for VxWorks Subject: vme bus errors Subject: Re: problem with booting vxworks as standalone Subject: Use Flash file System Or Not Subject: Re: dosFs2Lib: Anyone using it? Subject: Re: vme bus errors Subject: Re: info on writing device drivers Subject: Re: Varible name conflicts Subject: Re: sending float value from window to vxwork over TCP/IPsocket Subject: Re: memShow (memory leaks) Subject: Re: dosFs2Lib: Anyone using it? Subject: Re: how to write net driver Subject: Now here's some interesting fodder dodder Subject: WTX Error 0x1000f (SVR_DOESNT_RESPOND) Subject: Re: How to use the Intel 82559ER Lan Controller in the vxworks system? Subject: Re: Far-End Fault-Indecation for Ethernet 100MB-FX Subject: Re: Far-End Fault-Indecation for Ethernet 100MB-FX Subject: Using Sym8xx SCSI adapter on PPC Subject: any HDLC driver for VxWorks on ARM base? ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: How to use the Intel 82559ER Lan Controller in the vxworks system? Date: Tue, 7 Aug 2001 13:07:35 +0200 From: "Werner Schiendl" Message-ID: <3b6fcc9f$0$32924$6e365a64@newsreader02.highway.telekom.at> References: Hi, Intel has a driver for the 82559 series Ethernet controller. I don't know what is special about the ..ER type. maybe you give that driver a try look at http://developer.intel.com/platforms/applied/software/d559vx.htm http://developer.intel.com/platforms/applied/software/b559mux.htm the first is an 'old-style' BSD ethernet driver, the second uses VxWorks' END driver model. hth Werner "srs_zh" wrote in message news:b5c0d5b7.0108061736.43df277@posting.google.com... > I build the Intel 82559ER Lan Controller for X86,but the system can't > work rightly!Who can help me? I will greatly appriciate your help! --------------------------- Newsgroups: comp.os.vxworks,comp.sys.powerpc.tech,comp.arch.embedded Subject: I Like PowerPC and Vxworks Date: 7 Aug 2001 04:48:30 -0700 From: zhaochun@wri.com.cn (zhaochun) Organization: http://groups.google.com/ Message-ID: Say Something about PPC --------------------------- Newsgroups: comp.os.vxworks Subject: sending float value from window to vxwork over TCP/IPsocket Date: Tue, 7 Aug 2001 19:59:37 +0800 From: "Chew GimHeng" Organization: SCV Cablemodem News site Message-ID: <9kokgj$hch$1@newsie.singa.pore.net> This is a multi-part message in MIME format. - ------=_NextPart_000_001E_01C11F7B.7C4DDA80 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable Dear all,=20 I encounter a problem when sending a floating point value from window to = vxwork. no matter how I swap the byte I still could not get back the = value on the vxWork side.=20 Beside doing ntohl() on the windows is there any other way to swap the = byte order before sending.=20 Or what is your implementation.=20 Thank you gimheng - ------=_NextPart_000_001E_01C11F7B.7C4DDA80 Content-Type: text/html; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable
Dear all,
I encounter a problem when sending a = floating point=20 value from window to vxwork. no matter how I swap the byte I still could = not get=20 back the value on the vxWork side.
 
Beside doing ntohl() on the windows is = there any=20 other way to swap the byte order before sending.
Or what is your implementation. =
 
 
Thank you
gimheng
 
- ------=_NextPart_000_001E_01C11F7B.7C4DDA80-- --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Write to disk files ASAP Date: 7 Aug 2001 05:31:10 -0700 From: scott_b_klein@md.northgrum.com (Scott Klein) Organization: http://groups.google.com/ Message-ID: <69517bb8.0108070431.7d4da755@posting.google.com> References: <3B6F495B.9800BF9E@erols.com> <488e459a.0108062211.7d7c2587@posting.google.com> I want to use FAT32 so that we can move the disk to a PC & read the files there. Thanks. Scott B. Klein Senior Engineer Electronics Sensors and Systems Sector Northrop Grumman Corporation Post Office Box 746 - Mail Stop 285A Baltimore, Maryland 21203-0746 Telephone (410) 765-0910 Fax: (410) 981-2045 scott_b_klein@md.northgrum.com john_94501@yahoo.com (John) wrote in message news:<488e459a.0108062211.7d7c2587@posting.google.com>... > Hello, > > Have you looked at the raw file system? It sounds like your solution > so far is to try to replicate the behaviour of the raw file system > under DOSFS - you may as well just remove the DOS layer and work > directly with the disk via rawFs. > > I seem to remember that others have also used this approach in the > past for streaming data apps... > > HTH, > > John... > > Klein family wrote in message news:<3B6F495B.9800BF9E@erols.com>... > > I am trying to write to disk files as fast as possible without losing > > any of the input data. I can expect to be generating many GB of data in > > a single start/stop sequence, which I will be putting into a series of > > 2GB files. I do not want to lose any data at the start or during file > > switch. We are using DOSFS2. An additional problem is that software does > > not recieve a "prepare to exit" signal, so it always has to be prepared > > to begin recording. > > > > I have found that allocating disk space takes longer as the disk fills, > > gradually reducing our throughput rate until we lose data. So far I have > > tried: > > 1. sequentially opening, writing and closing files. The throughput rate > > gradually decreases. > > 2. allocating contiguous disk space [ioctl(fd, FIOCONTIG,0x7fffffff)] > > when the start command is recieved. This can take seconds for each file. > > > > 3. opening files and allocating contiguous disk space when the program > > starts; followed by repeated after the stop command is recieved. I found > > that the disk becomes unusable after a number of cycles because disk > > space does not get deallocated and chkdsk does not do a very good job of > > fixing the disk allocation table. (dosFsShow says that I have 140,000GB > > free on a 73GB disk.) > > > > My current scheme is to create 2GB files on the disk in advance, either > > before running the program or at program start. I am doing this as > > follows: > > until the disk is full: > > create a file > > allocate 2GB contiguous disk space > > lseek to 2GB - 512 > > write a 512 bytes block > > close the file > > The problem with this is that the write call takes a long time, during > > which the "in use" light on the disk flashes. It is as if the write is > > writing the entire 2GB of the file. Even though it can be done on an > > off-line basis, I don't want to spend hours preparing every disk. Also, > > I don't know of any other probelms with this scheme as I just finished > > coding it today and have only given it a minimal exercize so far. > > > > Can anyone suggest any other schemes or an improvement to what I've > > already tried. Thanks. > > > > > > Thanks > > > > Scott Klein > > Northrop Grumman > > Baltimore, MD > > 410-765-0910 > > scott_b_klein@md.northgrum.com --------------------------- Newsgroups: comp.os.vxworks Subject: Multiple VME slave windows Date: Tue, 07 Aug 2001 12:35:30 GMT From: "Mark Fanara" Organization: Excite@Home - The Leader in Broadband http://home.com/faster Message-ID: I am looking for a detailed example or explaination on how to setup multiple slave windows on a VME bus. There were posts on this topic back in May, but the information offered was somewhat generic. We are currently using the vxMP add-on option and shared memory message queues to transfer data between processors. Our application consists of a lot of slave to slave communication. Therefore, we would really like to share memory on each board. BTW - It does appear that some BSPs (Radstone boards) provide this capability. Ours, (Motorola 2305) does not. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: eev_44_drv Date: 7 Aug 2001 05:49:26 -0700 From: lalitsc@hotmail.com (lalit) Organization: http://groups.google.com/ Message-ID: References: <3b6ec000$0$22316$6e365a64@newsreader02.highway.telekom.at> <3b6fbab4$0$4126$6e365a64@newsreader02.highway.telekom.at> Thanks for the links. Its really a great help. Now I can proceed my work further. "Werner Schiendl" wrote in message news:<3b6fbab4$0$4126$6e365a64@newsreader02.highway.telekom.at>... > Hi, > > look at > > http://developer.intel.com/platforms/applied/software/d559vx.htm > http://developer.intel.com/platforms/applied/software/b559mux.htm > > the first is an 'old-style' BSD ethernet driver, the second uses VxWorks' > END driver model. > Unfortunately, they BOTH contain a file with that very name :-( > > hth > Werner --------------------------- Newsgroups: comp.os.vxworks,comp.sys.powerpc.tech,comp.arch.embedded Subject: Re: I Like PowerPC and Vxworks Date: 7 Aug 2001 12:58:26 GMT From: "Mark A. Odell" Organization: Embedded Firmware Design, Inc. Message-ID: References: zhaochun@wri.com.cn (zhaochun) wrote in news:a8ffc95e.0108070348.42263cae@posting.google.com: > Say Something about PPC [Subject] That's nice. Come on over to news:comp.sys.powerpc.tech when you're ready to discuss the PowerPC. - -- - - Mark --------------------------- Newsgroups: comp.os.vxworks,comp.sys.powerpc.tech,comp.arch.embedded Subject: Re: I Like PowerPC and Vxworks Date: Tue, 07 Aug 2001 12:54:38 GMT From: "Keith Brafford" Organization: Road Runner - NC Message-ID: References: Reply-To: "Keith Brafford" Uh, how's this: PPC has a cool EIEIO instruction I like it too! - --Keith Brafford "zhaochun" wrote in message news:a8ffc95e.0108070348.42263cae@posting.google.com... > Say Something about PPC --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Write to disk files ASAP Date: Tue, 7 Aug 2001 15:28:02 +0200 From: "Werner Schiendl" Message-ID: <3b6fed89$0$25324$6e365a64@newsreader02.highway.telekom.at> References: <3B6F495B.9800BF9E@erols.com> <488e459a.0108062211.7d7c2587@posting.google.com> <69517bb8.0108070431.7d4da755@posting.google.com> Hi, may be you could consider writing an utility to read the raw disk for Windows. Another possibility would be the use of a disk editor. You can of course also consider using *nix to do the work, simply with the dd command. All these methods provide you with a convinient access while getting maximum performance to write the data. And obviously writing raw data is MUCH more predictable than using a file system. Especially such a poor-designed one than DosFS. The Dos FS is - in my point of view - in no way real-time suitable. (Expect that it sometimes really needs a lot of time ;-) ) regards Werner --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Exception Date: 07 Aug 2001 09:57:06 -0400 From: Bill Pringlemeir Organization: Bell Sympatico Message-ID: References: <9ka1eq$aku$1@newsie.singa.pore.net> <3b69140f$0$228@hades.is.co.za> <8bb4f4a34a%molochai@vapour-trail.demon.co.uk> <6cb0b2a44a%molochai@vapour-trail.demon.co.uk> Sender: bpringlemeir@DeadDuck >>>>> "molochai" == molochai writes: molochai> Stack undwinding refers not only to temporary objects, but molochai> also those those that are aggregates or superclasses, all molochai> of which need to be constructed (and destructed) atomically molochai> with their containing objects, and for which the molochai> construction code is compiler generated. This all happens molochai> before the programmer-defined costructor for the containing molochai> class get exwcuted. Yes, I was aware of this. I generally check for the existence of an object before destroying it in my `catch/setjmp' code. The objects that these other functions create are always contained (has a) in the object whose member function contains the exception. molochai> It is just about possible to emulate this behaviour without molochai> exception support using some kind of status flags, and molochai> writing destructor code which is aware of the possibility molochai> of partially costructed objects, but this is a lot of molochai> work. Can you really tell me that you would get full test molochai> coverage for all the possible failure modes? I'd rather pay molochai> the price of a small amount of code bloat, let the compiler molochai> take the strain, and have the peace of mind that it works. The setjmp/longjmp method forces the exception to have one common catch statement. The same could be a `coding standard' issue and the C++ exceptions would also work well. I think I will take a look at the type of code generated by exceptions. There are still many pitfalls with exceptions. Such as, ... int level = intLock(); ... bigObject *large = new bigObject; ... intUnlock(level); If an "Out Of Memory" exception is thrown here, it is easy to forget that an interrupt lock has been obtained. The same is true of semaphore, mutexes and any other shared resources. These problems also exist without exceptions; I just have found it to be more likely to handles these cases when exceptions aren't being used. The solution to this problem is to guard the allocation with a try and then catch and unlock and throw the exception again. However, this creates a new code path that should be tested. The conditional test in the common exception will catch any `intLock'/`intUnlock' whereas this try/catch construct might be duplicated in several places. regards, Bill Pringlemeir. - -- TORPEDOED BY SPACE-TIME VORTEX 122W 41N. EXPLODING. U-241. vxWorks FAQ, "http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html" --------------------------- Newsgroups: comp.os.vxworks Subject: Re: graphic library Date: 07 Aug 2001 10:03:29 -0400 From: Bill Pringlemeir Organization: Bell Sympatico Message-ID: References: <9koe80$8kf$1@spiten.kongsberg.com> Sender: bpringlemeir@DeadDuck >>>>> "Stig" == Stig Nordmoen writes: Stig> I am loocking for a graphic library for the C&T 65550 video Stig> controller. I want to write text/lines and rectangels on the Stig> screen (flat panel), defined with pixel coordinates. Do I have Stig> other choices then WindML. "http://sal.kachinatech.com/F/4/" - -- Finish your beer -- there are sober people in China! vxWorks FAQ, "http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html" --------------------------- Newsgroups: comp.os.vxworks Subject: Re: OPC for VxWorks Date: Tue, 7 Aug 2001 10:14:12 -0400 From: "Kevin Livingston" Message-ID: <6GSb7.7$hX5.424@client> References: <3B6F3CE1.2FC27307@e-farm.com> thanks for the pointer, could you estimate how long it took you to complete this task? weeks, months? I am trying to get a feel for the amount of work involved in doing such a task. thanks Kevin "Jerry Hudgins" wrote in message news:3B6F3CE1.2FC27307@e-farm.com... > Kevin Livingston wrote: > > > > I was courious if anyone else was useing OPC in conjunction with > > VxWorks, specifically using the VxWorks platform as an OPC server? > > It's possible to roll your own server using VxDCOM - I've done it - > but the server toolkit bundled with VxOPC would likely make server > implementation much faster and easier. If you do use VxDCOM alone, > make sure that you're up-to-date patch-wise. Early builds had some > significant OPC-related problems that have since been addressed. > > -jch --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Network performance Date: 7 Aug 2001 07:44:30 -0700 From: john_94501@yahoo.com (John) Organization: http://groups.google.com/ Message-ID: <488e459a.0108070644.564566d7@posting.google.com> References: Hello David, > Can anyone point me to any area(s) of vxWorks which cause apparently > slow network activity, I'm not > entirely sure what sort of performance to expect over a 10 M Ethernet > link, but we seem to see about > a third of the transfer rate that uCLinux is producing. We are using > the CS8900 Ethernet controller. Have you looked at the sizes of the buffers allocated for the network? These are not really set up for network intensive apps by default. You might find that increasing their size helps. There are some tech notes on WindSurf about this (WTN51 and WTN53 - both under T2, but also applies to T1 with SENS), but in essence you need to increase the following: 1) If your driver supports it, the receive and transmit buffer sizes. These are normally specified in the END load string (in your BSP, look for configNet.h). 2) The network stack's internal buffering. These are controlled via a collection of macros of the form NUM_n and NUM_SYS_n where 'n' is replaced by the size of the blocks (it is a slab allocation scheme). HTH, John... --------------------------- Newsgroups: comp.os.vxworks Subject: Varible name conflicts Date: Tue, 7 Aug 2001 11:19:55 -0400 From: "Joe Ott" Organization: Verio Message-ID: We are porting legacy C code to VxWorks and have ran into a problem creating a single downloadable module. In our code different variables in different modules with the same name have been declared. When linked and downloaded seperately the different modules seem to each have their own copy of the variable, which is the desired way. However if we link the modules into one downloadable image there seems to be only one variable and all modules reference it, even if the variable means different thing to different modules. As can be imagined this causes many bad things to happen. Is there a way to force the linker to create seperate copies of the variables or to at least throw warnings when combining declarations? We looked at the -warn-common linker switch but could not seem to get it to work under the Tornado workspace. We added it to the linker tab in the build properties but there was no indication it was used in the build output. We are using windows NT. Being legacy code we are not sure we could catch all the comon names by hand and hope there is a way to use the tools to help us. Thanks in advance for your help Joe --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Exception Date: Tue, 7 Aug 2001 17:20:08 +0200 From: "Geurt Vos" Organization: Rohill Technologies BV Message-ID: References: <9ka1eq$aku$1@newsie.singa.pore.net> <3b69140f$0$228@hades.is.co.za> <8bb4f4a34a%molochai@vapour-trail.demon.co.uk> <6cb0b2a44a%molochai@vapour-trail.demon.co.uk> Sender: rohill2@rohill.nl > There are still many pitfalls with exceptions. Such as, > > ... > int level = intLock(); > ... > bigObject *large = new bigObject; > ... > intUnlock(level); > > If an "Out Of Memory" exception is thrown here, it is easy to forget > that an interrupt lock has been obtained. The same is true of > semaphore, mutexes and any other shared resources. Yes, and that's the reason you should never lock resources this way when exception are enabled. Note that the same is true for successive memory allocations: void Func() { SmallObject *small = new SmallObject; LargObject *large = new LargeObject; } if 'new' throws an exceptions, 'small' is never deallocated. > These problems > also exist without exceptions; I just have found it to be more likely > to handles these cases when exceptions aren't being used. > > The solution to this problem is to guard the allocation with a try and > then catch and unlock and throw the exception again. However, this > creates a new code path that should be tested. The conditional test > in the common exception will catch any `intLock'/`intUnlock' whereas > this try/catch construct might be duplicated in several places. > Yes, you need to guard resource locking (and memory allocations), but you should definitely not guard them with try & catch blocks. Instead, to solve these kind of problems, you have to write guard classes, which lock the resource in the constructor and unlock (if necessary) the resource in the destructor. When an excpetions is thrown, it is guaranteed the destructor of your guard gets called, which will cleanly unlock the resource. So for intLock it could be something like: class IntLockGuard { public: IntLockGuard (): level(intLock()) { } ~IntLockGuard () { intUnlock(level); } private: int level; }; ... IntLockGuard guard; HugeObject *huge = new HugeObject; ... Regardless of whether or not 'new' will throw, ~IntLockGuard() will always get called. To solve the memory allocation problem, you can use std::auto_ptr... which is nicely #if 0'd on VxWorks, though it's quite trivial to, um, steal it from some other compiler and port it to GCC 2.7.2 :) Geurt --------------------------- Newsgroups: comp.os.vxworks Subject: A question for VxWorks Web server Date: Tue, 7 Aug 2001 09:31:45 -0400 From: "Jing Ma" Message-ID: <3b6fedb0.0@alognews.analogic.com> Hi, Does anybody know how to use a java URL class to set and get data from VxWorks web server? Thanks! --------------------------- Newsgroups: comp.os.vxworks Subject: info on writing device drivers Date: 7 Aug 2001 08:58:03 -0700 From: chrisanc1967@yahoo.com (Christopher Sanchez) Organization: http://groups.google.com/ Message-ID: Can anyone tell me if there exsists some info. on writing device drivers for the vxworks operating system. Thanks Christopher --------------------------- Newsgroups: comp.os.vxworks Subject: Re: question about taskspawn Date: Tue, 7 Aug 2001 17:55:55 +0200 From: "Werner Schiendl" Message-ID: <3b701032$0$21460$6e365a64@newsreader02.highway.telekom.at> References: Hi, I assume you use Tornado II environment. You can set a global breakpoint. That one works, regardless of which task runs into it. You need to attach to the task manually by default with the debugger, after it hit the breakpoint. You can change VxWorks Options to attach automatically. hth Werner --------------------------- Newsgroups: comp.os.vxworks Subject: Re: how to write net driver Date: Tue, 7 Aug 2001 09:48:03 -0700 From: "Dan Gold" Message-ID: <3b701c21$0$280$e2e8da3@nntp.cts.com> References: <9ko95q$7b3@nntpa.cb.lucent.com> Your (rough) description looks like operation for a tty driver, where each Tx interrupt sends a new character, thus causing a new Tx interrupt, etc... until all characters have been sent. For network drivers, the model is different. Usually network drivers send their data at task level. If you're an END driver, xxxEndSend(), which is called by the Mux layer. The Tx interrupt generated then is usually just used to clean up any buffers, check the status of the transmission, possibly re-enable transmission if it's been blocked (i.e. out of buffer descriptors or something) hth, Dan Gold gold@ensemble.com "john tian" wrote in message news:9ko95q$7b3@nntpa.cb.lucent.com... > hello, > There is a confusing problem about net driver: > TX interrupt--- Send all data in buffers, after interface card > transfer data totally, then produce the next TX interrupt, but no data is > transminted for some time. who triger a new TX interrupt, if it is timer, > how long? > > > > --------------------------- Newsgroups: comp.os.vxworks Subject: Connect() Subroutine Date: Tue, 07 Aug 2001 17:14:03 GMT From: yan_can_cook@hotmail.com Organization: @Work Internet powered by @Home Network Message-ID: I have been trying to debug some VxWorks code that hangs at the Connect() subroutine. The main problem is that the VxWorks 5.3 Reference manuals are fairly vague and do not include all possible options. Does anyone have a more in-depth description of this subroutine or have a good on-line resource? --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Not able to ping to the board Date: 7 Aug 2001 11:43:18 -0700 From: chrisanc1967@yahoo.com (Christopher Sanchez) Organization: http://groups.google.com/ Message-ID: References: <95ccca9f.0107260402.66a211c5@posting.google.com> try putting 'dc' into your 'other' bootup field. This should initialize your ethernet driver. nav@sasken.com (naveen) wrote in message news:<95ccca9f.0107260402.66a211c5@posting.google.com>... > Hi, > > I have created a standalone vxworks image. This image is > burnt into the flash. A valid ipaddress has been provided > to the board. The problem is > > a)i am not able to ping to this board from a pc in the > same network. > b) loopback with 127.0.0.1 or with the specified ipaddress of > the board happen fine. > > > I am using tornado II. Is there any settings problem. > > Any solutions to the problem. > > Thanks > > B.Naveen --------------------------- Newsgroups: comp.os.vxworks Subject: dosFs2Lib: Anyone using it? Date: Tue, 07 Aug 2001 15:02:12 -0400 From: Chris Williamson Organization: ncsc Message-ID: <3B703B34.1208E51F@nsa.gov> I'm currently using dosFsLib, but I've heard of something called dosFs2Lib. Does anyone know anything about it? Anyone using it? Are there any URLs related to it? Thanks. Chris Williamson --------------------------- Newsgroups: comp.os.vxworks Subject: Re: 'LOADER_RELOCATION_TOO_LARGE' ERROR Date: 7 Aug 2001 19:16:22 GMT From: cymyers@ecst.csuchico.edu (Cy H. Myers) Organization: California State University, Chico Message-ID: <9kpeq6$9at$1@hubble.csuchico.edu> References: <44a01eba.0108061717.6f3a6db2@posting.google.com> Hi Ellin, Most PPC branch instructions are only 24 bit relocations from the current program counter. This error means that the code you are trying to download is too far away (more than +/- 23 bits) from the code it is trying to call (probably kernel code.) Is the module very large? Are you doing something between booting and downloading the module that might allocate a few meg of memory? There are two solutions to this: get the module closer to the code it's calling by reordering things a bit, or splitting the parts of the module that make kernel calls. Or, rebuild the module to use full 32-bit relocations. There's a compiler option to do this, which you can add in the CFLAGS of your project. Bye, Cy In article <44a01eba.0108061717.6f3a6db2@posting.google.com>, ellin wrote: >I built a project with T2 of kernel PPCEC602 for 8260 target, but when >I tried to download it to target, I got error of "Not fit in 24bit" >and "LOADER_RELOCATION_TOO_LARGE". What can cause this and how to >avoid it? >Thanks in advance. > >ellin --------------------------- Newsgroups: comp.os.vxworks Subject: Re: OPC for VxWorks Date: Tue, 07 Aug 2001 19:59:23 GMT From: Jerry Hudgins Organization: Western Digital Design, Marin County, California Message-ID: <3B70489B.E8D1C714@e-farm.com> References: <3B6F3CE1.2FC27307@e-farm.com> <6GSb7.7$hX5.424@client> Reply-To: Jerry Hudgins Kevin Livingston wrote: > > could you estimate how long it took you to complete this task? > weeks, months? > I am trying to get a feel for the amount of work involved in doing > such a task. Well, my experience is probably a poor example. I didn't know a thing about DCOM or OPC when I started, and as I was an early adopter, I spent quite a bit of time waiting for VxDCOM patches for the problems that turned up. It took a long time for me to get everything working completely. If you understand DCOM and have the latest VxDCOM release, however, you might put together a useful DA server in a month or less. - -jch --------------------------- Newsgroups: comp.os.vxworks Subject: vme bus errors Date: Tue, 07 Aug 2001 15:51:05 -0500 From: James Toland Organization: Raytheon Company Message-ID: <3B7054B9.5B493C93@raytheon.com> How do I detect a VME bus error on a Motorola MVME2700 single board computer running vxWorks 5.4 RTOS? I don't plan on enabling VME interrupts.

Thanks --------------------------- Newsgroups: comp.os.vxworks Subject: Re: problem with booting vxworks as standalone Date: 7 Aug 2001 14:25:53 -0700 From: intmain@yahoo.com (Bruce) Organization: http://groups.google.com/ Message-ID: <2b26d6bb.0108071325.52f61532@posting.google.com> References: <95ccca9f.0107250100.27328092@posting.google.com> Bill Pringlemeir wrote in message news:... > In my image I have, > > ULONG standTblSize = 9796; > > Your problem could be that the symbol table array size and > standTblSize do not match. Check out your symTbl.c file. This is a > generated file. You might do a clean build and retry. There is also a utility called stripARCH.exe that can be used to selectively remove unneeded symbols. Bruce --------------------------- Newsgroups: comp.os.vxworks Subject: Use Flash file System Or Not Date: 7 Aug 2001 14:30:28 -0700 From: ryeung@earthlink.net (Raymond Yeung) Organization: http://groups.google.com/ Message-ID: Can anyone share his thinking of when it would make good technical and/or business sense to use Flash file systems, instead of writing home- grown ad-hoc naked flash drivers to store: + Executable + Configuration data How much are these file systems package we're talking about? Thanks, Raymond ryeung@earthlink.net --------------------------- Newsgroups: comp.os.vxworks Subject: Re: dosFs2Lib: Anyone using it? Date: Tue, 07 Aug 2001 21:49:54 GMT From: Andray Kaganovsky Organization: @Home Message-ID: References: <3B703B34.1208E51F@nsa.gov> Chris Williamson wrote in news:3B703B34.1208E51F@nsa.gov: > I'm currently using dosFsLib, but I've heard of something called > dosFs2Lib. Does anyone know anything about it? Anyone using it? Are > there any URLs related to it? > > Thanks. > Chris Williamson > > [AndrayK]. A lot of people are using dosFs-2, and it's the standard filesystem included with Tornado-2/VxWorks AE. The dosFs-2 is major improvement over old dosFs, in many aspects. As far as I know, it is available from WRS as free upgrade for all registered Tornado users. It also backward compatible with old dosFs, so existing applications will work. Based on my personal experience, I would say that it's very good idea to upgrade to dosFs-2. Andray --------------------------- Newsgroups: comp.os.vxworks Subject: Re: vme bus errors Date: Tue, 07 Aug 2001 17:17:53 -0500 From: Andrew Johnson Organization: APS, Argonne National Laboratory Message-ID: <3B706911.9767630D@aps.anl.gov> References: <3B7054B9.5B493C93@raytheon.com> James Toland wrote: > > How do I detect a VME bus error on a Motorola MVME2700 single board > computer running vxWorks 5.4 RTOS? I don't plan on enabling VME > interrupts. Add this code to your sysHwInit2() routine for starters: /* * Make VME Bus Errors generate a CPU exception by enabling * Received Target Abort Machine check exceptions in Raven */ mpcErrEnbl = sysIn16 ((UINT16 *)(RAVEN_BASE_ADRS + RAVEN_MPC_MEREN)); sysOut16 ((UINT16 *)(RAVEN_BASE_ADRS + RAVEN_MPC_MEREN), mpcErrEnbl | RAVEN_MPC_MEREN_RTAM); /* Enable Machine Check pin of CPU */ vxHid0Set(vxHid0Get() | _PPC_HID0_EMCP); Note that this won't cause the Bus Error location to be reported, but it will cause a Machine Check Exception to be generated when a Bus Error occurs. You can query the Universe chip for the location, although it doesn't record all cycles. This isn't a complete solution either as it only works once, you need to intercept the Machine Check Exception and clear the exception status (I have that code for an MVME2100, but not for the 2700 yet). Note that Machine Check exceptions are imprecise, and non-recoverable, so you won't be able to resume after one. - - Andrew - -- The world is such a cheerful place when viewed from upside-down It makes a rise of every fall, a smile of every frown --------------------------- Newsgroups: comp.os.vxworks Subject: Re: info on writing device drivers Date: 7 Aug 2001 16:25:46 -0700 From: john_94501@yahoo.com (John) Organization: http://groups.google.com/ Message-ID: <488e459a.0108071525.55427063@posting.google.com> References: Hello, chrisanc1967@yahoo.com (Christopher Sanchez) wrote in message news:... > Can anyone tell me if there exsists some info. on writing device > drivers for the vxworks operating system. Thanks There is information in the VxWorks Programmer's Guide on doing this (see section "3.9 Internal Structure" for a discussion and some example code). For VxWorks AE, the process is the same, but the chapter reference is 5.9 instead (there are a couple of new chapters in the VxWorks AE Programmer's Guide). Also, if you have an installed tree, check under target/src/drv for some examples. There should be some serial drivers there, and perhaps some others depending on your architecture. HTH, John... PS You can download the 5.4 programmer's guide from the web site - here's the URL for it: http://www.wrs.com/pdf/vxworks_guide.pdf --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Varible name conflicts Date: 7 Aug 2001 16:37:11 -0700 From: john_94501@yahoo.com (John) Organization: http://groups.google.com/ Message-ID: <488e459a.0108071537.7839177e@posting.google.com> References: Hello Joe, Couple of ideas... The compiler should be able to convert the commons into normal bss variables (use -fno-common on your compile line), but I suspect that this will just cause you link errors when you link all the objects together since there will now be multiple definitions of each 'common' symbol. The -warn-common linker switch should work; if you are having trouble from the IDE, you can always invoke the linker from the command line to generate the .o file and the associated list of common variables that have actually been combined. For example, on PPC: % ldppc -r -o image.o -warn-common obj1.o obj2.o obj3.o Once you have the list though, you still have some problems. If the variables are globals, then I assume that they are being shared by two or more .o files legitimately, even if not by all the file. That means that you need to identify which ones are meant to be collapsed into commons, and which are not. If the globals are not being shared then they can of course be converted to statics, instantly removing the problem. HTH, John... "Joe Ott" wrote in message news:... > We are porting legacy C code to VxWorks and have ran into a problem creating > a single downloadable module. > > In our code different variables in different modules with the same name have > been declared. When linked and downloaded seperately the different modules > seem to each have their own copy of the variable, which is the desired way. > However if we link the modules into one downloadable image there seems to be > only one variable and all modules reference it, even if the variable means > different thing to different modules. As can be imagined this causes many > bad things to happen. > > Is there a way to force the linker to create seperate copies of the > variables or to at least throw warnings when combining declarations? We > looked at the -warn-common linker switch but could not seem to get it to > work under the Tornado workspace. We added it to the linker tab in the build > properties but there was no indication it was used in the build output. We > are using windows NT. > > Being legacy code we are not sure we could catch all the comon names by hand > and hope there is a way to use the tools to help us. > > Thanks in advance for your help > Joe --------------------------- Newsgroups: comp.os.vxworks Subject: Re: sending float value from window to vxwork over TCP/IPsocket Date: 7 Aug 2001 16:51:34 -0700 From: john_94501@yahoo.com (John) Organization: http://groups.google.com/ Message-ID: <488e459a.0108071551.7bd05583@posting.google.com> References: <9kokgj$hch$1@newsie.singa.pore.net> Hello, You are using the routine for converting a long INTEGER to network order to deal with floating point values. This is not likely to work! You have some options though: 1) XDR (this should be part of the VxWorks installation since NFS uses it). There are XDR routines for floating point values. See http://www.unet.univie.ac.at/aix/aixprggd/progcomc/xdr_ref.htm#A6gDFg190blai for some more information on XDR. 2) Pass the values as strings, and convert them back using, for example, sscanf(), atof() or strtod(). HTH, John... "Chew GimHeng" wrote in message news:<9kokgj$hch$1@newsie.singa.pore.net>... > Dear all, > I encounter a problem when sending a floating point value from window to > vxwork. no matter how I swap the byte I still could not get back the > value on the vxWork side. > > Beside doing ntohl() on the windows is there any other way to swap the > byte order before sending. > Or what is your implementation. > > > Thank you > gimheng > > > -- --------------------------- Newsgroups: comp.os.vxworks Subject: Re: memShow (memory leaks) Date: 7 Aug 2001 16:54:15 -0700 From: john_94501@yahoo.com (John) Organization: http://groups.google.com/ Message-ID: <488e459a.0108071554.38931d80@posting.google.com> References: <1a86250f.0108070152.440d2e7@posting.google.com> Hello, If you are passing a string literal to the cd command, the shell will be allocating some memory for that string. This is a common source of "memory leaks" when using the target shell. To check if this is the cause, proceed as follows: - -> newDirectory = "my/new/location/on/the/disk" - -> memShow - -> cd newDirectory - -> memShow HTH, John... ike@1freeemail.com (Ike Robinson) wrote in message news:<1a86250f.0108070152.440d2e7@posting.google.com>... > Hello, > > I have some kind of newbie question. > First, I'm using > VxWorks (for Motorola MCP750 - MPC 750) version 5.4. > Kernel: WIND version 2.5. > > I have a bit complicated program which allocates some memory for > internal structures and then release it. > My problem is that after first run of this program memShow shows that > I have 80 bytes less free memory. Next calls of this program doesn't > show free memory decrease. > > My question is: > is it memory leaks in program or > vxWorks initially reserves these 80 bytes for some reason? > > I started to suspect that vxWorks can eats these 80 bytes when I ran > memShow before 'cd' command and after. I realized that 'cd' eats 8 > bytes (probably for some internal heap/stack). > > Or my guess is wrong and it definitely memory leaks in program > (actually third-party, so I can't view it's source)? > > > Thanks in advance, > Ike --------------------------- Newsgroups: comp.os.vxworks Subject: Re: dosFs2Lib: Anyone using it? Date: Wed, 08 Aug 2001 05:31:38 GMT From: "ggang1" Organization: Dacom Message-ID: <_84c7.15298$2k2.673686@news.bora.net> References: <3B703B34.1208E51F@nsa.gov> Hi.. I used to handle dosFs2 with PCMCIA interface to use both Windows98 and VxWorks simultaneously. dosFs isn't that functionality...so After I upgraded to dosFs2, I could do it. But I think that dosFs2 is difficult than dosfs to use.however it's more useful. "Chris Williamson" wrote in message news:3B703B34.1208E51F@nsa.gov... > I'm currently using dosFsLib, but I've heard of something called > dosFs2Lib. Does anyone know anything about it? Anyone using it? Are > there any URLs related to it? > > Thanks. > Chris Williamson > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: how to write net driver Date: Wed, 8 Aug 2001 14:21:24 +0800 From: "john tian" Organization: Lucent Technologies, Columbus, Ohio Message-ID: <9kqlf9$f08@nntpb.cb.lucent.com> References: <9ko95q$7b3@nntpa.cb.lucent.com> <3b701c21$0$280$e2e8da3@nntp.cts.com> Hi, firstly, thank Gold for your reply. following is my knowledage about net driver: Maybe net driver can be divided into three category: I/O port style, memory map directly and DMA style. and the last two is prevailing. in Memory Map Directly, driver writes ony ONE data packet to this memory segment once, once this packet is written to map memory, the system considers this packet is transfered to peer successfully. no feedback is provided to driver. in DMA style, driver notify the interface card, there is ONE packet to transfer and this packet's physical address and len. the interface card request a DMA to tranfer this packet. when a packet is transfered successfully, it trigger a TX interrupt and notify driver transition success. driver release ring buffer. if not be transfered successfully, after some time, driver will timeout, retransmite the last packet. but for net task, firstly, it check ring buffer,if there is free ring buffer, it copy data to ring buffer, if this packet is the first packet in ring buffer, then start driver to send packet. if ring buffer is full, it will be blocked herein, when there is a free buffer, the driver will wake up it. above is my analysis, many errors exist, can you help me to correct it, thank you. tianjun "Dan Gold" wrote in message news:3b701c21$0$280$e2e8da3@nntp.cts.com... > Your (rough) description looks like operation for a tty driver, where each > Tx interrupt sends a new character, thus causing a new Tx interrupt, etc... > until all characters have been sent. > > For network drivers, the model is different. Usually network drivers send > their data at task level. If you're an END driver, xxxEndSend(), which is > called by the Mux layer. The Tx interrupt generated then is usually just > used to clean up any buffers, check the status of the transmission, > possibly re-enable transmission if it's been blocked (i.e. out of buffer > descriptors or something) > > hth, > Dan Gold gold@ensemble.com > > "john tian" wrote in message > news:9ko95q$7b3@nntpa.cb.lucent.com... > > hello, > > There is a confusing problem about net driver: > > TX interrupt--- Send all data in buffers, after interface card > > transfer data totally, then produce the next TX interrupt, but no data is > > transminted for some time. who triger a new TX interrupt, if it is timer, > > how long? > > > > > > > > > > --------------------------- Newsgroups: comp.os.vxworks Subject: Now here's some interesting fodder dodder Date: Tue, 7 Aug 2001 23:29:19 -0700 From: "Chris Conlon" Organization: Posted via Supernews, http://www.supernews.com Message-ID: http://www.xs4all.nl/~borkhuis/vxworks/vxfaq.html#3.2 --------------------------- Newsgroups: comp.os.vxworks Subject: WTX Error 0x1000f (SVR_DOESNT_RESPOND) Date: Wed, 8 Aug 2001 07:56:50 +0000 (UTC) From: "Chih-Ming.Liao" Organization: Dept. of Computer & Information Science, NCTU, Taiwan Message-ID: <9kqrc2$21i8$1@news.cis.nctu.edu.tw> Sender: Chih-Ming.Liao Does anybody can help me to solve the following problems? My development System configuration is the following: Tornado 2.01 run in Win98 SOC1B00 BSP (StrongArm SA1110 Evaluation board) The HOSTPC is peer connection to the Board Machine HOSTPC is running the Registry and the Tornado tools, when I start the Target Server and enable the Shell I got the following message: - --------------------------------------------------------- TGTSVR (bsp@HOSTPC): Wed Aug 08 15:31:57 2001 Wind River Systems Target Server: NT/Win95 version Connecting to target agent... succeeded. Attaching C++ interface... succeeded. Attaching coff OMF reader for ARM CPU family... succeeded. Error: Couldn't open C:\TORNADO\vxWorks: status = 0x2 Error during target server core initialization Target Server will exit "Target Server "bsp@HOSTPC" is not responding. This may be due to slow network response." Target Information "Tool attach failed. WTX Error 0x1000f (SVR_DOESNT_RESPOND)" - ----------------------------------------------------------- I have tried to chage the Local Registry to Remote Registry, But I still got those above messages, does anybody know what else can be causeing it? Thanks! ChihMing --------------------------- Newsgroups: comp.os.vxworks Subject: Re: How to use the Intel 82559ER Lan Controller in the vxworks system? Date: 8 Aug 2001 00:58:07 -0700 From: srs_zh@263.net (srs_zh) Organization: http://groups.google.com/ Message-ID: References: <3b6fcc9f$0$32924$6e365a64@newsreader02.highway.telekom.at> Werner: Thank for your replay! But the b559mux is not suit to the Lan Controller! Can you have any new method! Thank you again ! --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Far-End Fault-Indecation for Ethernet 100MB-FX Date: Wed, 8 Aug 2001 09:32:48 +0200 From: "Michael Koestner" Organization: Siemens AG Karlsruhe Message-ID: <9kqpvd$pc2@parzival.khe.siemens.de> References: <9kebn5$pn5@parzival.khe.siemens.de> <3B6EBE6A.3AAD7D37@agilent.com> the problem is, that a dynamic state is signaled by a latched bit. So, if a FEFI/Remote-Fault is detected, you read the bit (which will than reseted) and you see the fault. But you don't know when this state ends, because the bit is already reseted by your reading action befor :-( Regards Michael Koestner "James Marshall" schrieb im Newsbeitrag news:3B6EBE6A.3AAD7D37@agilent.com... > Have you tried reading the PHY status register twice? I do not have experience > with fibre optic ethernet, but 100BaseT PHYs have similar latched-until-read > status bits. > > James Marshall. > > > Michael Koestner wrote: > > > Hi, > > does anyone have a implementation for a Far-End Fault-Indecation ethernet > > fiber optic connection. the problem work with the phy device, which have > > latched bit implemented to signal a dynamic state. > > Mfg > > Michael Koestner > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Far-End Fault-Indecation for Ethernet 100MB-FX Date: Wed, 8 Aug 2001 09:34:32 +0200 From: "Michael Koestner" Organization: Siemens AG Karlsruhe Message-ID: <9kqq1t$pe2@parzival.khe.siemens.de> the problem is, that a dynamic state is signaled by a latched bit. So, if a FEFI/Remote-Fault is detected, you read the bit (which will than reseted) and you see the fault. But you don't know when this state ends, because the bit is already reseted by your reading action befor :-( Regards Michael Koestner "James Marshall" schrieb im Newsbeitrag news:3B6EBE6A.3AAD7D37@agilent.com... > Have you tried reading the PHY status register twice? I do not have experience > with fibre optic ethernet, but 100BaseT PHYs have similar latched-until-read > status bits. > > James Marshall. > > > Michael Koestner wrote: > > > Hi, > > does anyone have a implementation for a Far-End Fault-Indecation ethernet > > fiber optic connection. the problem work with the phy device, which have > > latched bit implemented to signal a dynamic state. > > Mfg > > Michael Koestner > --------------------------- Newsgroups: comp.os.vxworks Subject: Using Sym8xx SCSI adapter on PPC Date: Wed, 08 Aug 2001 08:30:46 GMT From: "Koen Van Hulle" Organization: Xeikon Message-ID: Hi, Has any one been able to use a Symbis Logic SCSI controller with VxWorks on the PPC? When trying it on a commercial SYM875x adapter in a PPC405 evaluation board, the call to ncr810CtrlCreate() returns a NULL pointer and no error code. On the console, the message "0x1f3fdf8 (tRootTask): Invalid NCR8XX device type. Not supported" is shown. Thanks, Koen Van Hiulle --------------------------- Newsgroups: comp.os.vxworks Subject: any HDLC driver for VxWorks on ARM base? Date: 8 Aug 2001 01:44:45 -0700 From: xgdong@sina.com (Docular Xu) Organization: http://groups.google.com/ Message-ID: Is there anyone who knows where can I get an HDLC driver for VxWorks on an ARM base? For my work, I need to implement HDLC communications between two boards, which are connected by E1 line. We do have HDLC chips mounted on our boards ( both comes from infoneon). But, those chips don't implement HDLC protocols as I know. So, we need .... HDLC DRIVER Does Windriver company have this? How to contact them? Thanks. --------------------------- End of New-News digest ********************** From vxwexplo-errs@csg.lbl.gov Wed Aug 8 06:57:48 2001 Date: Wed Aug 8 06:57:51 PDT 2001 Subject: AUTOREPLY comp.os.vxworks newsdigest This is the Mailbox of Friedrich Scheurer. R&D - ACS GmbH - D-72768 Reutlingen Sorry, I am out of office until August 20th 2001! In urgent cases please contact one of the following persons at ACS GmbH: Otto Angerhofer (Managing director) e-mail: oangerhofer@manz-acs.de Eduard Ams (Technical manager) e-mail: eams@manz-acs.de Yvonne Anhorn (Marketing) e-mail: yanhorn@manz-acs.de Margit Arnold (Sales) e-mail: marnold@manz-acs.de Christina Sixel (Sales) e-mail: csixel@manz-acs.de MfG / Kind Regards F. Scheurer ---------------------------------------------------------------- Friedrich Scheurer, Dipl.-Ing. phone +49 7121 9019-23(0) ACS GmbH - R&D fax +49 7121 9019-98 Steigaeckerstrasse 13 web www.manz-acs.de D-72768 Reutlingen (Germany) e-mail: fscheurer@manz-acs.de ---------------------------------------------------------------- From vxwexplo-errs@csg.lbl.gov Thu Aug 9 04:03:33 2001 From: Vxworks Exploder Date: Thu Aug 9 04:03:35 PDT 2001 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Thu Aug 9 04:03:17 PDT 2001 Subject: Re: any HDLC driver for VxWorks on ARM base? Subject: Getting Volume info from dosFs Subject: Re: dosFs2Lib: Anyone using it? Subject: Re: 'LOADER_RELOCATION_TOO_LARGE' ERROR Subject: Re: Getting Volume info from dosFs Subject: Disk Corruption Subject: Galileo GT-64240 Experience Subject: Re: WTX Error 0x1000f (SVR_DOESNT_RESPOND) Subject: Re: Getting Volume info from dosFs Subject: install prototype of vxworks? Subject: Re: install prototype of vxworks? Subject: Re: Write to disk files ASAP Subject: reply to "Write to disk files ASAP" Subject: Hard disk problems.....malformed boot sector Subject: Re: Getting Volume info from dosFs Subject: Re: 'LOADER_RELOCATION_TOO_LARGE' ERROR Subject: MicroMonitor (Tiny File System) Subject: a problem when build VxWorks_rom Subject: Re: How to use the Intel 82559ER Lan Controller in the vxworks system? Subject: Re: How to use the Intel 82559ER Lan Controller in the vxworks system? Subject: Re: I Like PowerPC and Vxworks Subject: Unresolved simbols in c++ Templates Subject: Re: How to use the Intel 82559ER Lan Controller in the vxworks system? ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: any HDLC driver for VxWorks on ARM base? Date: Wed, 08 Aug 2001 12:56:22 +0100 From: Paul Caswell Organization: Land & Sea Systems Message-ID: <3B7128E6.5E566A6C@baesystems.com> References: You could try contacting Spider Software. They supply a number of protocol stacks for use in vxWorks. Their website is www.spider.com. Regards. --------------------------- Newsgroups: comp.os.vxworks Subject: Getting Volume info from dosFs Date: Wed, 8 Aug 2001 12:16:34 -0400 From: "Mostafa Kassem" Organization: AT&T Canada IES Message-ID: <%Bdc7.51781$TW.261060@tor-nn1.netcom.ca> Greetings: I am looking for a function similar to dosFsConfigShow() except that instead of directing the output to the std output, this function should populate a structure of the type DOS_VOL_CONFIG. I am aware of the function dosFsConfigGet, however I can't use it because I have no access to the volume DOS_VOL_DESC structure. Any idea how to accomplish that? Thanks in advance, Mostafa --------------------------- Newsgroups: comp.os.vxworks Subject: Re: dosFs2Lib: Anyone using it? Date: Wed, 08 Aug 2001 16:28:04 GMT From: Andray Kaganovsky Organization: @Home Message-ID: References: <3B703B34.1208E51F@nsa.gov> Andray Kaganovsky wrote in news:Xns90F6B5A6BA091andreykhomecom@24.2.9.58: > Chris Williamson wrote in news:3B703B34.1208E51F@nsa.gov: > >> I'm currently using dosFsLib, but I've heard of something called >> dosFs2Lib. Does anyone know anything about it? Anyone using it? Are >> there any URLs related to it? >> >> Thanks. >> Chris Williamson >> >> > > [AndrayK]. A lot of people are using dosFs-2, and it's the standard > filesystem included with Tornado-2/VxWorks AE. The dosFs-2 is major > improvement over old dosFs, in many aspects. As far as I know, > it is available from WRS as free upgrade for all registered Tornado > users. It also backward compatible with old dosFs, so existing > applications will work. Based on my personal experience, I would say > that it's very good idea to upgrade to dosFs-2. > > Andray > [AndrayK 2]. The above should read "Tornado-3/VxWorks AE". I apologize for the typo. Andray --------------------------- Newsgroups: comp.os.vxworks Subject: Re: 'LOADER_RELOCATION_TOO_LARGE' ERROR Date: 8 Aug 2001 10:22:37 -0700 From: tlksk@hotmail.com (Kumar) Organization: http://groups.google.com/ Message-ID: References: <44a01eba.0108061717.6f3a6db2@posting.google.com> <9kpeq6$9at$1@hubble.csuchico.edu> Hi Use -mlongcall in your CFLAGS option. Kumar cymyers@ecst.csuchico.edu (Cy H. Myers) wrote in message news:<9kpeq6$9at$1@hubble.csuchico.edu>... > Hi Ellin, > Most PPC branch instructions are only 24 bit relocations from the > current program counter. This error means that the code you are trying to > download is too far away (more than +/- 23 bits) from the code it is trying > to call (probably kernel code.) Is the module very large? Are you doing > something between booting and downloading the module that might allocate > a few meg of memory? > There are two solutions to this: get the module closer to the code > it's calling by reordering things a bit, or splitting the parts of the module > that make kernel calls. Or, rebuild the module to use full 32-bit > relocations. There's a compiler option to do this, which you can add in > the CFLAGS of your project. > > Bye, > Cy > > In article <44a01eba.0108061717.6f3a6db2@posting.google.com>, > ellin wrote: > >I built a project with T2 of kernel PPCEC602 for 8260 target, but when > >I tried to download it to target, I got error of "Not fit in 24bit" > >and "LOADER_RELOCATION_TOO_LARGE". What can cause this and how to > >avoid it? > >Thanks in advance. > > > >ellin --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Getting Volume info from dosFs Date: Wed, 08 Aug 2001 17:52:57 GMT From: "David Bryan" Organization: EarthLink Inc. -- http://www.EarthLink.net Message-ID: References: <%Bdc7.51781$TW.261060@tor-nn1.netcom.ca> Mostafa - Use iosDevFind() to return a pointer to the DOS_VOL_DESC structure. Then you can use dosFsConfigGet() HTH Dave - -- ===================================================== T h e P T R G r o u p, I n c. ===================================================== Embedded, Real-Time Solutions David Bryan Office: (703)726-0498 Manager, Embedded Services Cell: (703)627-1735 www.ThePTRGroup.com dave@ThePTRGroup.com ===================================================== "Mostafa Kassem" wrote in message news:%Bdc7.51781$TW.261060@tor-nn1.netcom.ca... > Greetings: > > I am looking for a function similar to dosFsConfigShow() except that instead > of directing the output to the std output, this function should populate a > structure of the type DOS_VOL_CONFIG. > > I am aware of the function dosFsConfigGet, however I can't use it because I > have no access to the volume DOS_VOL_DESC structure. > > Any idea how to accomplish that? > > Thanks in advance, > > Mostafa > > > --------------------------- Newsgroups: comp.os.vxworks Subject: Disk Corruption Date: 8 Aug 2001 10:54:09 -0700 From: mgf_julio@lycos.com (Shanan) Organization: http://groups.google.com/ Message-ID: <3818bea1.0108080954.674bb978@posting.google.com> I have a corrupt hard disk. Accessing on particular file in write mode wipes out all other entries in the dir below it and causes lost chains. I would like to be able to view the FAT and the Dir file to see how it is changed and where the starting clusters are. Can any body help me. I am trying to figure out how and why the corruption occurs. Shanan --------------------------- Newsgroups: comp.os.vxworks Subject: Galileo GT-64240 Experience Date: Wed, 08 Aug 2001 13:58:54 -0400 From: Doug Dotson Organization: Verio Message-ID: <3B717DDE.ECD2E90F@digidata.com> Does anyone have experience using the Galileo Gt-64240 chip. Specifically, using the MPCS as a UART? Thanks, Doug --------------------------- Newsgroups: comp.os.vxworks Subject: Re: WTX Error 0x1000f (SVR_DOESNT_RESPOND) Date: Wed, 08 Aug 2001 18:49:42 GMT From: "David Bryan" Organization: EarthLink Inc. -- http://www.EarthLink.net Message-ID: References: <9kqrc2$21i8$1@news.cis.nctu.edu.tw> From the looks of it, the core file your telling the target server to use does not exist, or is in a different directory on your host. In the Tools->Target Server-> Configure dialog, choose "Core Files and Symbols" in the Target Server Properties list box. You may have to explicity tell the target server where the symbol file is on your system. Choose "File" and enter the path and filename for your image for example" c:\Tornad\target\proj\SOC1B00\default\vxWorks HTH Dave - -- ===================================================== T h e P T R G r o u p, I n c. ===================================================== Embedded, Real-Time Solutions David Bryan Office: (703)726-0498 Manager, Embedded Services Cell: (703)627-1735 www.ThePTRGroup.com dave@ThePTRGroup.com ===================================================== "Chih-Ming.Liao" wrote in message news:9kqrc2$21i8$1@news.cis.nctu.edu.tw... > Does anybody can help me to solve the following problems? > > My development System configuration is the following: > Tornado 2.01 run in Win98 > SOC1B00 BSP (StrongArm SA1110 Evaluation board) > The HOSTPC is peer connection to the Board > > Machine HOSTPC is running the Registry and the Tornado tools, > when I start the Target Server and enable the Shell I got the > following message: > > --------------------------------------------------------- > TGTSVR (bsp@HOSTPC): Wed Aug 08 15:31:57 2001 > Wind River Systems Target Server: NT/Win95 version > Connecting to target agent... succeeded. > Attaching C++ interface... succeeded. > Attaching coff OMF reader for ARM CPU family... succeeded. > Error: Couldn't open C:\TORNADO\vxWorks: status = 0x2 > Error during target server core initialization > Target Server will exit > > > "Target Server "bsp@HOSTPC" is not responding. > This may be due to slow network response." > > > Target Information > "Tool attach failed. > WTX Error 0x1000f > (SVR_DOESNT_RESPOND)" > ----------------------------------------------------------- > > I have tried to chage the Local Registry to Remote Registry, > But I still got those above messages, does anybody know > what else can be causeing it? > > Thanks! > > ChihMing > > > > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Getting Volume info from dosFs Date: Wed, 8 Aug 2001 15:08:05 -0400 From: "Mostafa Kassem" Organization: AT&T Canada IES Message-ID: References: <%Bdc7.51781$TW.261060@tor-nn1.netcom.ca> Yo Dave: iosDevFind() does not return a pointer to DOS_VOL_DESC structure, rather it returns a pointer to DEV_HDR which is not really useful for what I want to do. Any other suggestion? Regards, Mostafa "David Bryan" wrote in message news:Z%ec7.800$%o4.75380@newsread1.prod.itd.earthlink.net... > Mostafa - > > Use iosDevFind() to return a pointer to the DOS_VOL_DESC structure. Then > you can use dosFsConfigGet() > > HTH > > Dave > > -- > ===================================================== > T h e P T R G r o u p, I n c. > ===================================================== > Embedded, Real-Time Solutions > David Bryan Office: (703)726-0498 > Manager, Embedded Services Cell: (703)627-1735 > www.ThePTRGroup.com dave@ThePTRGroup.com > ===================================================== > "Mostafa Kassem" wrote in message > news:%Bdc7.51781$TW.261060@tor-nn1.netcom.ca... > > Greetings: > > > > I am looking for a function similar to dosFsConfigShow() except that > instead > > of directing the output to the std output, this function should populate a > > structure of the type DOS_VOL_CONFIG. > > > > I am aware of the function dosFsConfigGet, however I can't use it because > I > > have no access to the volume DOS_VOL_DESC structure. > > > > Any idea how to accomplish that? > > > > Thanks in advance, > > > > Mostafa > > > > > > > > --------------------------- Newsgroups: comp.os.vxworks Subject: install prototype of vxworks? Date: Wed, 08 Aug 2001 14:19:08 -0500 From: Junhong Liu Organization: Ericsson North America Inc. Message-ID: <3B7190AC.3C98CC17@exu.ericsson.se> I deleted all the directories of vxworks several days ago and tried to install the prototype again today, but after I entered my key, the installation stoped there. Can any body help me with this? Thans! Junhong --------------------------- Newsgroups: comp.os.vxworks Subject: Re: install prototype of vxworks? Date: Wed, 8 Aug 2001 21:42:40 +0200 From: "Werner Schiendl" Message-ID: <3b7196d7$0$9872$6e365a64@newsreader02.highway.telekom.at> References: <3B7190AC.3C98CC17@exu.ericsson.se> Hi, you should at least specify exactly WHAT you want to setup/install on WHICH host. Further, I think WRS will be glad to get you installed and bying runtimes ;-) regards Werner --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Write to disk files ASAP Date: Wed, 8 Aug 2001 21:36:40 +0200 From: "Werner Schiendl" Message-ID: <3b7196d6$0$9872$6e365a64@newsreader02.highway.telekom.at> References: <3B6F495B.9800BF9E@erols.com> <488e459a.0108062211.7d7c2587@posting.google.com> <69517bb8.0108070431.7d4da755@posting.google.com> <3b6fed89$0$25324$6e365a64@newsreader02.highway.telekom.at> Hi Scott, Sorry my mail server is just not willing to cooperate :-( hope you get it here I meant using Linux or some other Unix flavor. It comes with the dd command line utility (shell command, if you prefer). under linux, with a syntax somewhat like dd if=/dev/hdb of=~/first-two-gig bs=512 count=4194304 you can easily copy the first 4194304 sectors, i. e. the first 2 GB from the driver to a file /dev/hdb means the raw, slave driver on the first controller I'm not quite sure about the exact syntax, have no Linux box here in the office. Not sure about the size limits either, but that should be solveable by glueing the files together. hth Werner - ----- Original Message ----- From: "Klein, Scott B." To: Sent: Wednesday, August 08, 2001 2:06 PM Subject: reply to "Write to disk files ASAP" > Werner, > Thanks for responding to my cry for help. > > What are *nix & the dd command? > > Thanks. > > > Scott B. Klein > Senior Engineer > Electronics Sensors and Systems Sector > Northrop Grumman Corporation > Post Office Box 746 - Mail Stop 285A > Baltimore, Maryland 21203-0746 > Telephone (410) 765-0910 > Fax: (410) 981-2045 > scott_b_klein@md.northgrum.com > > > --------------------------- Newsgroups: comp.os.vxworks Subject: Hard disk problems.....malformed boot sector Date: Wed, 08 Aug 2001 22:00:01 GMT From: Martijn Kamerbeek Organization: Sierra Data Systems Message-ID: <3B719610.6A88E12E@sierradatasys.com> Hi all!! I am trying to use a Quantum Atlas V Ultra 160 (36.7 GB) SCSI hard disk with Vxworks on a MVME1604, but when I try to open the whole device or format the device I get the error message "Malformed boot sector: Offset 0, value 246." I've been trying so many things such as: - - trying the disk on a Windows platform -> works fine! - - trying to use the disk with pSOS+ -> works fine! - - changing/adjusting all parameters during SCSI initialization - - specifying an offset for the scsiBlkDevCreate - - etc.... Trying dosFsShow says at the bottom: "Can't mount volume." So it seems that VxWorks can talk to the firmware but not to the actual storage space. Has anybody any suggestions, or can anybody recommend a hard disk (greater than 30 Gb) that works great for them? Thank you very much!! Martijn Kamerbeek mkamerbeek@sierradatasys.com --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Getting Volume info from dosFs Date: Wed, 08 Aug 2001 22:28:05 GMT From: "David Bryan" Organization: EarthLink Inc. -- http://www.EarthLink.net Message-ID: References: <%Bdc7.51781$TW.261060@tor-nn1.netcom.ca> In fact, DEV_HDR is the first entry in the DOS_VOL_DESC structure. Just as it's the first entry in the TY_DEV, TAPE_VOL_DESC, RAW_VOL_DESC, and LPT_DEV structures. Notice that in these structures, DEV_HDR is not a pointer, it is the actual location of the DEV_HDR for the device. Therefore, a pointer to the DEV_HDR for any of these devices is also a pointer to the larger structure. Try it Dave "Mostafa Kassem" wrote in message news:k7gc7.51814$TW.261099@tor-nn1.netcom.ca... > Yo Dave: > > iosDevFind() does not return a pointer to DOS_VOL_DESC structure, rather it > returns a pointer to DEV_HDR which is not really useful for what I want to > do. Any other suggestion? > > Regards, > > Mostafa > > > "David Bryan" wrote in message > news:Z%ec7.800$%o4.75380@newsread1.prod.itd.earthlink.net... > > Mostafa - > > > > Use iosDevFind() to return a pointer to the DOS_VOL_DESC structure. Then > > you can use dosFsConfigGet() > > > > HTH > > > > Dave > > > > -- > > ===================================================== > > T h e P T R G r o u p, I n c. > > ===================================================== > > Embedded, Real-Time Solutions > > David Bryan Office: (703)726-0498 > > Manager, Embedded Services Cell: (703)627-1735 > > www.ThePTRGroup.com dave@ThePTRGroup.com > > ===================================================== > > "Mostafa Kassem" wrote in message > > news:%Bdc7.51781$TW.261060@tor-nn1.netcom.ca... > > > Greetings: > > > > > > I am looking for a function similar to dosFsConfigShow() except that > > instead > > > of directing the output to the std output, this function should populate > a > > > structure of the type DOS_VOL_CONFIG. > > > > > > I am aware of the function dosFsConfigGet, however I can't use it > because > > I > > > have no access to the volume DOS_VOL_DESC structure. > > > > > > Any idea how to accomplish that? > > > > > > Thanks in advance, > > > > > > Mostafa > > > > > > > > > > > > > > > > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: 'LOADER_RELOCATION_TOO_LARGE' ERROR Date: 8 Aug 2001 16:19:21 -0700 From: john_94501@yahoo.com (John) Organization: http://groups.google.com/ Message-ID: <488e459a.0108081519.44fa1fde@posting.google.com> References: <44a01eba.0108061717.6f3a6db2@posting.google.com> Hello, In addition to using the -mlong-call flag (which will decrease the performance of your code), you might also like to consider the following trick: 1) Set up your BSP with LOCAL_MEM_SIZE set to 32Mb. 2) Boot your system, and load any code modules you need to load 3) Use memAddToPool() to add the remainder of your memory to the system memory pool ('memSysPartId' is the partition ID for the system memory pool) Assuming that you have less than 32Mb of code (including the kernel's image), that will allow you to load the module happily and then add the remaining memory to the heap for use dynamically by the program. Just an idea, John... ellin_lin@263.net (ellin) wrote in message news:<44a01eba.0108061717.6f3a6db2@posting.google.com>... > I built a project with T2 of kernel PPCEC602 for 8260 target, but when > I tried to download it to target, I got error of "Not fit in 24bit" > and "LOADER_RELOCATION_TOO_LARGE". What can cause this and how to > avoid it? > Thanks in advance. > > ellin --------------------------- Newsgroups: comp.os.vxworks Subject: MicroMonitor (Tiny File System) Date: 8 Aug 2001 17:53:18 -0700 From: ryeung@earthlink.net (Raymond Yeung) Organization: http://groups.google.com/ Message-ID: Has anybody used/ported the freeware MicroMonitor? What's the experience with the porting effort required, and its quality? My interest with it is mainly with its Flash File System. Any comments would be much appreciated. Raymond --------------------------- Newsgroups: comp.os.vxworks Subject: a problem when build VxWorks_rom Date: 8 Aug 2001 18:47:22 -0700 From: jade_chen@21cn.com (shy) Organization: http://groups.google.com/ Message-ID: <937a20a8.0108081747.6bfb791b@posting.google.com> I build vxworks_rom in the tornado enviroment ,but get the following message .(part of). ldppc -X -N -e _romInit -Ttext 02800100 -Tdata 00200000 \ romInit.o romStart.o version.o C:\TORNADO\target\lib\libPPC860gnuvx.a vxWorks.Z.o -o vxWorks_rom romStart.o: In function `romStart': C:\Tornado\target\proj\netSoftware\default\C:\TORNADO\target\config\comps\src\romStart.c:3 59: undefined reference to `binArrayStart' C:\Tornado\target\proj\netSoftware\default\C:\TORNADO\target\config\comps\src\romStart.c:3 59: undefined reference to `binArrayStart' C:\Tornado\target\proj\netSoftware\default\C:\TORNADO\target\config\comps\src\romStart.c:3 59: undefined reference to `binArrayEnd' C:\Tornado\target\proj\netSoftware\default\C:\TORNADO\target\config\comps\src\romStart.c:3 59: undefined reference to `binArrayEnd' C:\Tornado\target\proj\netSoftware\default\C:\TORNADO\target\config\comps\src\romStart.c:3 59: undefined reference to `binArrayStart' C:\Tornado\target\proj\netSoftware\default\C:\TORNADO\target\config\comps\src\romStart.c:3 59: undefined reference to `binArrayStart' make: *** [vxWorks_rom] Error 0x1 I make vxworks well,but where can it find binArrayStart & binArrayEnd?? Thanks Jade --------------------------- Newsgroups: comp.os.vxworks Subject: Re: How to use the Intel 82559ER Lan Controller in the vxworks system? Date: 9 Aug 2001 00:46:24 -0700 From: srs_zh@263.net (srs_zh) Organization: http://groups.google.com/ Message-ID: References: <3b6fcc9f$0$32924$6e365a64@newsreader02.highway.telekom.at> Who can help me? --------------------------- Newsgroups: comp.os.vxworks Subject: Re: How to use the Intel 82559ER Lan Controller in the vxworks system? Date: Thu, 09 Aug 2001 09:43:39 +0100 From: James Marshall Organization: Agilent Technologies UK Ltd Message-ID: <3B724D3B.5717B4C6@agilent.com> References: In your BSP's sysNetif.c file, change the line in sys557PciInit(): if (pciFindDevice (PRO100B_PCI_VENDOR_ID, PRO100B_PCI_DEVICE_ID, unit, &pciBus, &pciDevice, &pciFunc) != OK) break; to: if (pciFindDevice (PRO100B_PCI_VENDOR_ID, PRO100B_PCI_DEVICE_ID, unit, &pciBus, &pciDevice, &pciFunc) != OK) /* Look for 82559ER as well */ if (pciFindDevice (PRO100B_PCI_VENDOR_ID, 0x1209, unit, &pciBus, &pciDevice, &pciFunc) != OK) break; The 559ER has a different PCI Device ID (0x1209) than the 82557/8/9 (0x1229). James Marshall. srs_zh wrote: > I build the Intel 82559ER Lan Controller for X86,but the system can't > work rightly!Who can help me? I will greatly appriciate your help! --------------------------- Newsgroups: comp.os.vxworks,comp.sys.powerpc.tech,comp.arch.embedded Subject: Re: I Like PowerPC and Vxworks Date: Thu, 09 Aug 2001 09:50:26 +0100 From: James Marshall Organization: Agilent Technologies UK Ltd Message-ID: <3B724ED2.5216940D@agilent.com> References: > PPC has a cool EIEIO instruction > Yeah - I liked that one too! Maybe they got it from Old MacDonald's farm ;-) James Marshall. --------------------------- Newsgroups: comp.os.vxworks Subject: Unresolved simbols in c++ Templates Date: 9 Aug 2001 02:50:33 -0700 From: Santiago.Vazquez@casa.eads.net (santiago vazquez) Organization: http://groups.google.com/ Message-ID: Hello. I'm a user of Tornado 2.0.2, developing SW cross-compiling on a NT host and a PPC target. The problem I have is that when I use c++, I can compile code to generate a .out without any problem, and download it to the target. But, when I use Templates (trying to use ACE), during download (while linking) unresolved symbols appear. I think all paths are OK, cause the problem arise only with templates, even if defined on the .h files, so the problem isn't a path one. Tornado 2.0.2 uses a cygnus gcc version 2.7.2, and we think this is the problem, cause it's and old compiler version and may be it doesn't manage correctly templates, but we still don't have a newer gcc version, to verify this point. I wish to know if compiler is really our problem and in that case, where could I download a Tornado patch or a newer cross compiler version (NT-PPC) to fix the problem. I would appreciate a place to find the final executable for cross-compiling in order not to have to fight with gcc cross-compiler building. Thanks in advance. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: How to use the Intel 82559ER Lan Controller in the vxworks system? Date: Thu, 9 Aug 2001 12:26:58 +0200 From: "Werner Schiendl" Message-ID: <3b72661a$0$33474$6e365a64@newsreader02.highway.telekom.at> References: <3b6fcc9f$0$32924$6e365a64@newsreader02.highway.telekom.at> Hi, do you know, what is special about the 82559ER type controller? What is the exact reason, why the driver will not work? Maybe there is an updated driver on the Intel site that also support the controller. Otherwise, maybe you can get the source code for either the Intel eeE driver or the WindRiver fei (IIRC) one. regards Werner --------------------------- End of New-News digest ********************** From vxwexplo-errs@csg.lbl.gov Thu Aug 9 07:10:31 2001 Date: Thu Aug 9 07:10:34 PDT 2001 Subject: AUTOREPLY comp.os.vxworks newsdigest This is the Mailbox of Friedrich Scheurer. R&D - ACS GmbH - D-72768 Reutlingen Sorry, I am out of office until August 20th 2001! In urgent cases please contact one of the following persons at ACS GmbH: Otto Angerhofer (Managing director) e-mail: oangerhofer@manz-acs.de Eduard Ams (Technical manager) e-mail: eams@manz-acs.de Yvonne Anhorn (Marketing) e-mail: yanhorn@manz-acs.de Margit Arnold (Sales) e-mail: marnold@manz-acs.de Christina Sixel (Sales) e-mail: csixel@manz-acs.de MfG / Kind Regards F. Scheurer ---------------------------------------------------------------- Friedrich Scheurer, Dipl.-Ing. phone +49 7121 9019-23(0) ACS GmbH - R&D fax +49 7121 9019-98 Steigaeckerstrasse 13 web www.manz-acs.de D-72768 Reutlingen (Germany) e-mail: fscheurer@manz-acs.de ---------------------------------------------------------------- From vxwexplo-errs@csg.lbl.gov Thu Aug 9 14:38:49 2001 From: "Michael Morrison" Date: Thu Aug 9 14:38:52 PDT 2001 Subject: Compile windoze T2 makefiles in cygwin bash shell? Howdy, I have a windoze install of tornado II, but want to use the cygwin bash shell to run command-line make. Unfotunatly, when I do this I get the following error: $ make TARGET_BASE=iscsi ccppc -B/cygdrive/c/Tornado/host/x86-win32/lib/gcc-lib/ -g -nostdinc -f no-builtin -nostdlib -fno-defer-pop -W -Wall -I/cygdrive/c/Tornado/target/h -I/c ygdrive/c/Tornado/target/config/hsiSBC8260 -I. -I../include -c iscsi.cc cpp: -lang-c++: linker input file unused since linking not done as: unrecognized option `-u' ccppc.exe: Internal compiler error: program as got fatal signal 1 make: *** [iscsi.o] Error 1 The same makefile runs ok in the windoze DOS shell.... Anyone have any ideas on how to do this? Do you think the cygwin 'cpp' and 'as' be getting run instead of the vxworks version? TIA Mikem From vxwexplo-errs@csg.lbl.gov Fri Aug 10 04:03:40 2001 From: Vxworks Exploder Date: Fri Aug 10 04:03:43 PDT 2001 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Fri Aug 10 04:03:13 PDT 2001 Subject: Re: a problem when build VxWorks_rom Subject: Re: Write to disk files ASAP Subject: ata flash disk versus ide flash disk Subject: Re: Write to disk files ASAP Subject: ethernet link reset problem Subject: Re: MicroMonitor (Tiny File System) Subject: Re: a problem when build VxWorks_rom Subject: Re: Unresolved simbols in c++ Templates Subject: Send Trame Subject: support for explicit instantiation Subject: STL and Implicit Instantiation Subject: STL and Implicit Instantiation Subject: STL and Implicit Instantiation Subject: STL and Implicit Instantiation Subject: Re: STL and Implicit Instantiation Subject: Memory Protection (x86) and vxWorks? Subject: IP over Atm in Vxworks Subject: Questions on VxWorks boot loading Subject: high level question on menus Subject: Re: Questions on VxWorks boot loading Subject: Re: Questions on VxWorks boot loading Subject: MCP750 Subject: System level debugging Subject: Re: support for explicit instantiation Subject: what's wrong with this? please help Subject: Re: Unresolved simbols in c++ Templates Subject: Re: ata flash disk versus ide flash disk Subject: Re: Unresolved simbols in c++ Templates ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: a problem when build VxWorks_rom Date: Thu, 9 Aug 2001 13:36:47 +0200 From: "Rui" Organization: KGB Corp Message-ID: <9ktsmp$l2m$1@news1.xs4all.nl> References: <937a20a8.0108081747.6bfb791b@posting.google.com> Greatings: binArrayStart and binArrayEnd are defined in the Bootinit.c. They being defined depends on the the fact of the CPU_FAMILY being defined as either MIPS or PPC . Since you're using ldppc I assume you are using a PPC target environment. In my opinion you should make sure that you have CPU_FAMILY defined with the correct value at compile time by for example adding the flag '-DCPUFAMILY = PPC' to the compiler flags Just my 2 cents Rui "shy" wrote in message news:937a20a8.0108081747.6bfb791b@posting.google.com... > I build vxworks_rom in the tornado enviroment ,but get the following > message .(part of). > ldppc -X -N -e _romInit -Ttext 02800100 -Tdata 00200000 \ > romInit.o romStart.o version.o > C:\TORNADO\target\lib\libPPC860gnuvx.a vxWorks.Z.o -o > vxWorks_rom > romStart.o: In function `romStart': > C:\Tornado\target\proj\netSoftware\default\C:\TORNADO\target\config\comps\sr c\romStart.c:3 > 59: undefined reference to `binArrayStart' > C:\Tornado\target\proj\netSoftware\default\C:\TORNADO\target\config\comps\sr c\romStart.c:3 > 59: undefined reference to `binArrayStart' > C:\Tornado\target\proj\netSoftware\default\C:\TORNADO\target\config\comps\sr c\romStart.c:3 > 59: undefined reference to `binArrayEnd' > C:\Tornado\target\proj\netSoftware\default\C:\TORNADO\target\config\comps\sr c\romStart.c:3 > 59: undefined reference to `binArrayEnd' > C:\Tornado\target\proj\netSoftware\default\C:\TORNADO\target\config\comps\sr c\romStart.c:3 > 59: undefined reference to `binArrayStart' > C:\Tornado\target\proj\netSoftware\default\C:\TORNADO\target\config\comps\sr c\romStart.c:3 > 59: undefined reference to `binArrayStart' > make: *** [vxWorks_rom] Error 0x1 > > I make vxworks well,but where can it find binArrayStart & > binArrayEnd?? > > Thanks > Jade --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Write to disk files ASAP Date: 9 Aug 2001 05:09:37 -0700 From: scott_b_klein@md.northgrum.com (Scott Klein) Organization: http://groups.google.com/ Message-ID: <69517bb8.0108090409.417bd8dd@posting.google.com> References: <3B6F495B.9800BF9E@erols.com> <488e459a.0108062211.7d7c2587@posting.google.com> <69517bb8.0108070431.7d4da755@posting.google.com> <3b6fed89$0$25324$6e365a64@newsreader02.highway.telekom.at> <3b7196d6$0$9872$6e365a64@newsreader02.highway.telekom.at> Werner, Thanks, but management has already been sold on the idea that both recording & playback can be done without the middle-ware to re-process the data. Scott --------------------------- Newsgroups: comp.os.vxworks Subject: ata flash disk versus ide flash disk Date: Thu, 09 Aug 2001 12:21:46 GMT From: Gururaj patkar Organization: Optimum Online Message-ID: <3B727FEB.CC10FAF2@optonline.net> Hi all, I need pointers to the performance of using ata flash disk versus ide flash disk with vxworks. Looking at the ide and ata driver code supplied by wrs, the writes are done in polling mode (by checking the status register values on the flash controller). The usage of intConnect() in the ideDrv.c and ataDrv.c code is further intriguing as to why you poll the status register values and when STATUS_READY is TRUE then only provide interrupt driven functionality. One uses interrupt driven process to enhance the read/write performance to flash disk, then why use polling mode concurrent with it? Some benchmark figures in terms of read/write access time for files of varying sizes will be appreciated. gururaj --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Write to disk files ASAP Date: Thu, 9 Aug 2001 14:47:27 +0200 From: "Werner Schiendl" Message-ID: <3b728707$0$44634$6e365a64@newsreader02.highway.telekom.at> References: <3B6F495B.9800BF9E@erols.com> <488e459a.0108062211.7d7c2587@posting.google.com> <69517bb8.0108070431.7d4da755@posting.google.com> <3b6fed89$0$25324$6e365a64@newsreader02.highway.telekom.at> <3b7196d6$0$9872$6e365a64@newsreader02.highway.telekom.at> <69517bb8.0108090409.417bd8dd@posting.google.com> "Scott Klein" wrote in message news:69517bb8.0108090409.417bd8dd@posting.google.com... > Werner, > > Thanks, but management has already been sold on the idea that both > recording & playback can be done without the middle-ware to re-process > the data. > > Scott This is of course bad news for you :.-/ Probably you could write the few block necessary to make the thing look like a FAT drive yourself. Given what I know of your app so far (predefined block size, etc.) this should not be a big deal. There should be sufficient documentation on how a FAT file system should look like on the net. Using FAT32 (if that is an option) you could scale down to a single file, one partition layout. hth Werner --------------------------- Newsgroups: comp.os.vxworks Subject: ethernet link reset problem Date: 9 Aug 2001 06:58:58 -0700 From: zzesu@yahoo.com (Kim, Junghyun) Organization: http://groups.google.com/ Message-ID: Hi, I'm using vxWorks with Motorola N750 board which is connected to a switching hub. And I want to check the link status using snmp agent in the hub. When I disconnect lan cable from the board, the hub reports link-down, and when I reconncet the cable to the board, the hub reports link-up. At this time every thing is ok. But, just after the link-up report, a pair(or pairs) of link-down report and link-up report comes from the hub. From same tests with non-vxWorks systems, I got to know that this is not by hub. I think vxWorks OS do some initialization after the reconnction. (** With Motorola 750(not N750) boards, this didn't happend. I know that N750 is system board ) Any comment will be welcomed. Thanks. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: MicroMonitor (Tiny File System) Date: Thu, 09 Aug 2001 10:08:46 -0400 From: Doug Dotson Organization: Verio Message-ID: <3B72996E.D6A82CF0@digidata.com> References: We ported TFS and it worked fine for what it does. We decided to buy TrueFFS because we wanted directories and wear leveling, etc. Have yet to get it working though. Doug Raymond Yeung wrote: > Has anybody used/ported the freeware > MicroMonitor? What's the experience > with the porting effort required, and > its quality? My interest with it is > mainly with its Flash File System. > > Any comments would be much appreciated. > > Raymond --------------------------- Newsgroups: comp.os.vxworks Subject: Re: a problem when build VxWorks_rom Date: 09 Aug 2001 11:23:45 -0400 From: Bill Pringlemeir Organization: Bell Sympatico Message-ID: References: <937a20a8.0108081747.6bfb791b@posting.google.com> <9ktsmp$l2m$1@news1.xs4all.nl> Sender: bpringlemeir@DeadDuck >>> "Jade" wrote: Jade> I build vxworks_rom in the tornado enviroment ,but get the Jade> following message .(part of). ldppc -X -N -e _romInit -Ttext [snip] >>>>> "Rui" == Rui writes: Rui> Greatings: binArrayStart and binArrayEnd are defined in the Rui> Bootinit.c. They being defined depends on the the fact of the Rui> CPU_FAMILY being defined as either MIPS or PPC . Since you're Rui> using ldppc I assume you are using a PPC target environment. In Rui> my opinion you should make sure that you have CPU_FAMILY defined Rui> with the correct value at compile time by for example adding the Rui> flag '-DCPUFAMILY = PPC' to the compiler flags This is good advice. Also, the vxWorks_rom image builds a kernel that is then compressed (deflate i think). The compressed image is run through a program called `binToAsm'. Take a look in the build output for this program execution. The temporary file that it creates should contain the symbols that you noted. This `.s' file should be assembled to a `.o' and linked with the image that is giving the errors. If it is not, there is some sort of problem and you need to get this image to link. You might check to make sure the symbols in the generated `.s' file are the EXACT symbols the linker complains; they may be named differently for some reason. hth, Bill Pringlemeir. - -- TORPEDOED BY ATOMIC BOMB. CREW UNMOTIVATED. SINKING. U-404. vxWorks FAQ, "http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html" --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Unresolved simbols in c++ Templates Date: Thu, 9 Aug 2001 11:26:14 -0400 From: "Kevin Livingston" Message-ID: References: how are you linking into the ACE library? statically or dynamically? if you are attempting static linking have you added the library to your PRJ_LIBS marco in tornado? if dynamic, is the necessary library code on the target when you attempt to download your code? Kevin "santiago vazquez" wrote in message news:c9e00b4a.0108090150.62bd6329@posting.google.com... > Hello. > > I'm a user of Tornado 2.0.2, developing SW cross-compiling on a NT > host and a PPC target. The problem I have is that when I use c++, I > can compile code to generate a .out without any problem, and download > it to the target. But, when I use Templates (trying to use ACE), > during download (while linking) unresolved symbols appear. I think all > paths are OK, cause the problem arise only with templates, even if > defined on the .h files, so the problem isn't a path one. > Tornado 2.0.2 uses a cygnus gcc version 2.7.2, and we think this is > the problem, cause it's and old compiler version and may be it doesn't > manage correctly templates, but we still don't have a newer gcc > version, to verify this point. > I wish to know if compiler is really our problem and in that case, > where could I download a Tornado patch or a newer cross compiler > version (NT-PPC) to fix the problem. I would appreciate a place to > find the final executable for cross-compiling in order not to have to > fight with gcc cross-compiler building. > > Thanks in advance. --------------------------- Newsgroups: comp.os.vxworks Subject: Send Trame Date: Thu, 9 Aug 2001 17:34:53 +0200 From: "News" Message-ID: <3b72acb6_1@news.vo.lu> Hi, I have already the same pb. I want to send an ethernet trame. To do it, i'm using this following function : etherOutput( ) - send a packet on an Ethernet interface SYNOPSIS STATUS etherOutput ( struct ifnet * pIf, /* interface on which to send */ struct ether_header * pEtherHeader, /* Ethernet header to send */ char * pData, /* data to send */ int dataLength /* # of bytes of data to send */ ) Librairie : etherLib My bigger pb it's about struct ifnet * pIf . I use ifunit to get this pointeur : I do : pIf = ifunit ("ln0"); But I don't know what king of variable i have to give to this function : If I use ln0, the function returns pointeur NULL If I use lo0, I have this error message : 0x1690f60 (tShell): lo0: can't handle af0 I don't know what I have to do to resolve this pb. I also listen ethernet network with hook, like that : etherInputHookAdd (etHandle1,"rx",0) I have no pb, I can see IP packets. FYI : I know Vxworks since 2 weeks ago. Regards, Christophe --------------------------- Newsgroups: comp.os.vxworks Subject: support for explicit instantiation Date: Thu, 09 Aug 2001 18:29:43 GMT From: langston.reek@grassvalleygroup.com (Langston Reek) Organization: SBC Internet Services Message-ID: <3b72d625.89808421@news.pacbell.net> Hello, I'm trying to explicitly instantiate templates and I can't for the life of me get this to compile and load. I'm using Tornado2 (GNU 2.7.2 compiler), and I'm compiling with the compile flag, -fno-implicit-templates. { template class vector; // explicit instantiation vector list; } Is it the case that the compiler doesn't support explicit instantiation, or do I just have the wrong syntax. --------------------------- Newsgroups: comp.os.vxworks Subject: STL and Implicit Instantiation Date: Thu, 09 Aug 2001 18:41:41 GMT From: langston.reek@grassvalleygroup.com (Langston Reek) Organization: SBC Internet Services Message-ID: <3b72d94f.90617843@news.pacbell.net> I'm building x86 C++ software for VxWorks (5.4) and I've found that I need to use the -fno-implicit-templates compile flag. I was getting internal compiler errors when I used STL vector operations, specifically the push_back() operator. The manual talks about implicit and explicit instantiation. I'm still coming up to speed on STL, what is an implicit instantiation? Also, are there any pitfalls in using this compile flag? --------------------------- Newsgroups: comp.os.vxworks Subject: STL and Implicit Instantiation Date: Thu, 09 Aug 2001 18:43:33 GMT From: langston.reek@grassvalleygroup.com (Langston Reek) Organization: SBC Internet Services Message-ID: <3b72d7d3.90238187@news.pacbell.net> I'm building x86 C++ software for VxWorks (5.4) and I've found that I need to use the -fno-implicit-templates compile flag. I was getting internal compiler errors when I used STL vector operations, specifically the push_back() operator. The manual talks about implicit and explicit instantiation. I'm still coming up to speed on STL, what is an implicit instantiation? Also, are there any pitfalls in using this compile flag? --------------------------- Newsgroups: comp.os.vxworks Subject: STL and Implicit Instantiation Date: Thu, 09 Aug 2001 18:43:33 GMT From: langston.reek@grassvalleygroup.com (Langston Reek) Organization: SBC Internet Services Message-ID: <3b72d855.90368078@news.pacbell.net> I'm building x86 C++ software for VxWorks (5.4) and I've found that I need to use the -fno-implicit-templates compile flag. I was getting internal compiler errors when I used STL vector operations, specifically the push_back() operator. The manual talks about implicit and explicit instantiation. I'm still coming up to speed on STL, what is an implicit instantiation? Also, are there any pitfalls in using this compile flag? --------------------------- Newsgroups: comp.os.vxworks Subject: STL and Implicit Instantiation Date: Thu, 09 Aug 2001 18:43:33 GMT From: langston.reek@grassvalleygroup.com (Langston Reek) Organization: SBC Internet Services Message-ID: <3b72d8b2.90461406@news.pacbell.net> I'm building x86 C++ software for VxWorks (5.4) and I've found that I need to use the -fno-implicit-templates compile flag. I was getting internal compiler errors when I used STL vector operations, specifically the push_back() operator. The manual talks about implicit and explicit instantiation. I'm still coming up to speed on STL, what is an implicit instantiation? Also, are there any pitfalls in using this compile flag? --------------------------- Newsgroups: comp.os.vxworks Subject: Re: STL and Implicit Instantiation Date: Thu, 09 Aug 2001 20:13:03 +0100 From: molochai Message-ID: References: <3b72d7d3.90238187@news.pacbell.net> In message <3b72d7d3.90238187@news.pacbell.net> langston.reek@grassvalleygroup.com (Langston Reek) wrote: > I'm building x86 C++ software for VxWorks (5.4) and I've > found that I need to use the -fno-implicit-templates compile > flag. I was getting internal compiler errors when I used STL vector > operations, specifically the push_back() operator. > > The manual talks about implicit and explicit instantiation. I'm still > coming up to speed on STL, what is an implicit instantiation? I'd recomment reading the GNU tools manual. There's a section titled something like 'where's the template?' > > Also, are there any pitfalls in using this compile flag? > Yep.. IIRC you are reponsible for supplying an explicit instantiation for each type the template is used with in the entire linked program when - -fno-implicit-templates is used. By instantiation, I'm refering to the generation of template member functions and statics for the types needed. Anything inline, including RTTI will still be generated upon each apparent use. This approach is very difficult to manage IMHO, but gives ultimate control Otherwise, the compiler will generate this code whenever an instantiation is implied (provided it has been seen in the compilation unit by the compiler. This is why template headers often include a .cpp file at the end to unsure tis). This comes with a code-bloat penalty unless the linker is smart enough to spot it. Rumour has it that newer versions of ld are capable of doing exactly this. Also is the template repository (-frepo option). I've never got this to work. FWIW, I use the 'Borland' model (eg. the default). I believe this is the one prefered in later GCCs. If you're not already doing so, I'd recommend getting upgrading to the version GCC2.95.4 available from http://www.lysergic.u-net.com/. Amongst other things, this fixes a bug concerning static constructors which is present in GCC 2.7.2. M --------------------------- Newsgroups: comp.os.vxworks Subject: Memory Protection (x86) and vxWorks? Date: 9 Aug 2001 12:38:53 -0700 From: saustin_deja@my-deja.com (S Austin) Organization: http://groups.google.com/ Message-ID: Getting some memory corruption on a Pentium-II/vxWorks system, wondering where to begin to understand x86's limit checking capabilities with vxWorks? Thanks S. Austin --------------------------- Newsgroups: comp.os.vxworks Subject: IP over Atm in Vxworks Date: Thu, 9 Aug 2001 10:31:38 +0800 From: "Bruce" Organization: The news server of Changzhou TeleCom Message-ID: <9kstju$2jia$1@news.cz.js.cn> Hi,all: I am doing a END driver to implement IP over ATM in VxWorks 5.4. I will use PVC(permanent virtual circuit). According to RFC1577, PVC(VPI,VCI) is correspondind to IP address, so if I have several PVC, we need IP address of the same number. but we only have an END driver, does END can be binded to several IP address? Thank you for any answer! best regards bruce --------------------------- Newsgroups: comp.os.vxworks Subject: Questions on VxWorks boot loading Date: 9 Aug 2001 14:02:22 -0700 From: ryeung@earthlink.net (Raymond Yeung) Organization: http://groups.google.com/ Message-ID: I've a PowerPC board with boot flash and I'd like to use that for boot code and kernel/applications loading. My questions are 1. if I only have flash read/write/erase drivers but have no flash file system, would I still be able to use VxWorks loader(if yes, anything special I need to do) to load the image? 2. If not, what's my next cost-effective alternative (e.g. write my own loader, install a flash file system etc)? Thanks, Raymond --------------------------- Newsgroups: comp.os.vxworks Subject: high level question on menus Date: Thu, 09 Aug 2001 16:04:58 -0400 From: Pick a user name <> Organization: http://extra.newsguy.com Message-ID: <41r5ntk8frtsjosht8mk615hjjijek4gll@4ax.com> can someone give me a direction on where to start making my own menus for our project. maybe a document or an api... thx in advance, pete --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Questions on VxWorks boot loading Date: Thu, 09 Aug 2001 16:23:52 -0500 From: Dinker Charak Organization: Fermi National Accelerator Laboratory Message-ID: <3B72FF68.BE472A18@fnal.gov> References: > > 1. if I only have flash read/write/erase drivers > but have no flash file system, would I still be > able to use VxWorks loader(if yes, anything special > I need to do) to load the image? try this ... open romInit.c ... take all the chip select code and make your own boot loader ... it is easy to do ... burn your ram-downloadable kernel at some location into flash and burn this boot loader at your reset vector address (which should point in flash of course) ... let your boot loader configure chip selects and do UPM table setting ... and copy the ram-downloadable kernel image from flash to RAM and set PC to start address in RAM ... bingo .. off you go ... or burn your vxWorks_rom at some location in flash and burn this boot loader at your reset vector address let your boot loader configure chip selects jump the PC to start of image and let it proceed on it's own ... the boot loader does not do much here .. but it would be good to follow this design as if u want to add pre-OS diag code ... you could add it in the loader ... hope that helps ... dinker --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Questions on VxWorks boot loading Date: Thu, 09 Aug 2001 16:25:50 -0500 From: Dinker Charak Organization: Fermi National Accelerator Laboratory Message-ID: <3B72FFDE.9B89875E@fnal.gov> References: <3B72FF68.BE472A18@fnal.gov> > try this ... open romInit.c ... take all the chip select code and make oops! romInit.s dinker --------------------------- Newsgroups: comp.os.vxworks Subject: MCP750 Date: 9 Aug 2001 15:44:31 -0700 From: pete_jankus@quintum.com (Pete Jankus) Organization: http://groups.google.com/ Message-ID: <46095d81.0108091444.45fcd3a6@posting.google.com> Has anyone been able to use the Watchdog timer in the Raven ASIC on a Motorola MCP750 cPCI board? I am trying to enable WDT2 to reset the board on a Watchdog timeout, however I am unable to get the WDT2 to begin counting down. Does anyone have an example code snippet which would enable and "kick" the watchdog timer on this board? Below is the code that I am trying unsucessfully. After this code executes the timer value stays at 0xffff (as per the WDT2Status register) and does not decrement, nor is a reset generated. It behaves as though the watchdog is still disabled. I tried this with WDT1 as well, with the same results. Thanks KickWatchdog: lis r3,HI(RAVEN_BASE_ADRS) ori r3, r3, LO(RAVEN_BASE_ADRS) isync /* synchronize */ addis r4,r0,0x0000 /* load PATTERN_1 */ ori r4,r4,0x0055 isync /* synchronize */ stb r4,RAVEN_MPC_WDT2CNTL(r3) /* arm Watchdog Timer 2 */ eieio /* synchronize */ sync /* synchronize */ addis r4,r0,0xaa1f /* load PATTERN_2 */ ori r4,r4,0xffff /* Max resolution, Max reload value, Enable Watchdog timer */ isync /* synchronize */ stw r4,RAVEN_MPC_WDT2CNTL(r3) /* enable Timer2 */ eieio /* synchronize */ sync /* synchronize */ KickRavenWatchdogExit: bclr 20,0 --------------------------- Newsgroups: comp.os.vxworks Subject: System level debugging Date: Fri, 10 Aug 2001 03:05:36 +0200 From: "Zvika Vered" Organization: Internet Gold, ISRAEL Message-ID: <9kv8in$n1t$1@news.inter.net.il> Has anyone used the Tornado's debugger or Single step for system level debugging ? Thanks, Zvika Vered --------------------------- Newsgroups: comp.os.vxworks Subject: Re: support for explicit instantiation Date: Fri, 10 Aug 2001 08:59:42 +0200 From: "Albert Wijnja" Message-ID: <9l00p1$6r8nf$1@ID-67650.news.dfncis.de> References: <3b72d625.89808421@news.pacbell.net> Hi Langston, The 2.7.2 compiler does support explicit template instantiation. I use it and it works fine for me. From your example it seems to me that you are trying to instantiate the template from a local scopy (between the brackets). I instantiate the templates on global scope (thus, outside the brackets). Furthermore, I ran into some problems with unresolved externals regarding template functions (no template classes!!!) that are used from within the templates. These template functions must also be instantiated explicitly. In the newsgroup archives I found a convenient solution for this provided by Lothar Werzinger. He made some macros to make life easier. They are listed below. So, if you are using vector < string > you should add the following to your code (on global scope): // Since string is also a template, you need to instantiate it also !!! template class dm_string; // Now instantiate the vector and the related template functions template class vector < dm_string >; template dm_fill( dm_string ); template dm_fill_n( dm_string ); Hope this works for you. Good luck. Albert #if defined (ACE_HAS_EXPLICIT_TEMPLATE_INSTANTIATION) /////////////////////////////////////////////////////////////////////// // Some helper macros to define template instatiations of STL easier // Sample instantiation of vector < int > // template class vector < int >; // template dm_fill( int ); // template dm_fill_n( int ); // #if defined (__GNUC__) // template dm_uninitialized_copy_aux( int ); // template dm_uninitialized_copy_aux_const( int ); // template dm_uninitialized_fill_n_aux( int ); // #endif /* defined (__GNUC__) */ // Sample instantiation of list < MyClass > // template class list < MyClass >; // template dm_insert( MyClass ); // Macro to define the string #define dm_string \ basic_string < char > // Macro's to define some template functions for a vector #define dm_fill( _X_ ) \ void fill < _X_ *, _X_ > ( _X_ *, _X_ *, _X_ const & ) #define dm_fill_n( _X_ ) \ _X_ * fill_n < _X_ *, unsigned int, _X_ > ( _X_ *, unsigned int, _X_ const & ) #define dm_rb_tree(_X_, _Y_) \ rb_tree < _X_, pair < _X_ const, _Y_ >, select1st < pair < _X_ const, _Y_ > >, less < _X_ > > // Macro's to define the insert template function for a list #define list_par( _X_ ) \ _X_, __default_alloc_template #define l_iter( _X_ ) \ __list_iterator<_X_, _X_ &, _X_ *> #define const_l_iter( _X_ ) \ __list_iterator<_X_, _X_ const &, _X_ const *> #define dm_insert( _X_ ) \ void list< list_par( _X_ ) >::insert< const_l_iter( _X_ ) >(l_iter( _X_ ), const_l_iter( _X_ ), const_l_iter( _X_ )); // GNU STL needs that (other STLs maybe also, but we do not know) #if defined (__GNUC__) #define dm_uninitialized_copy_aux( _X_ ) \ _X_ * __uninitialized_copy_aux <_X_ *, _X_ *> ( _X_ *, _X_ *, _X_ *, __false_type ) #define dm_uninitialized_copy_aux_const( _X_ ) \ _X_ * __uninitialized_copy_aux <_X_ const *, _X_ *> ( _X_ const *, _X_ const *, _X_ *, __false_type ) #define dm_uninitialized_fill_n_aux( _X_ ) \ _X_ * __uninitialized_fill_n_aux <_X_ *, unsigned int, _X_ > ( _X_ *, unsigned int, _X_ const &, __false_type ) #endif /* defined (__GNUC__) */ #endif /* ACE_HAS_EXPLICIT_TEMPLATE_INSTANTIATION */ "Langston Reek" wrote in message news:3b72d625.89808421@news.pacbell.net... > Hello, > I'm trying to explicitly instantiate templates and I can't for the > life of me get this to compile and > load. I'm using Tornado2 (GNU 2.7.2 compiler), and I'm compiling with > the compile flag, > -fno-implicit-templates. > { > template class vector; // explicit instantiation > vector list; > } > Is it the case that the compiler doesn't support explicit > instantiation, > or do I just have the wrong syntax. > --------------------------- Newsgroups: comp.os.vxworks Subject: what's wrong with this? please help Date: 10 Aug 2001 00:44:21 -0700 From: liuzh@founder.com.cn (xing) Organization: http://groups.google.com/ Message-ID: hi,vxworks experts, i'm now using tornado 2.0 for X86, after i build a bootdisk and a vxworks from WRS's BSP,there is following promble occured,can anybody tell me why? any help or hint will be great appreicated! i use a serial line to connect my target and host,and have made some changes to WRS's BSP (config.h and vxworks's configure),then get a bootdisk and a vxworks image,this bootdisk and vxworks really works,but still there are following prombles: 1)when i start tgtsvr,the tgtsvr console alawys report a warning "core file checksums do not much ", but the booting process still going normally. 2)even my vxworks's configure has include "symbol table" ,when i config a target server's "core file and symbol table" option ,i can only use "no symbol" option,if i set to "global symbol" or "all symbol",the tgtsvr console will report " rpccore backend client time out" ,and boot process stoped. 3)after vxworks booted , i start a shell,but when i type a "i" command ,the shell report "the target system is not initialized yet .Can't processed,vaule =-1=0xfffffff" 4)when i download a demo program i built from Tornado demo ,after download completed ,it report a error list box "unresolved symbols encountered ",all vxwoks's function (such as _printf,_free .. ) are listed, and when i use browser to check the downloaded project's status, i found these function (_printf,_free...)marks as "unresolved Symbols". can anybody help me? best regards --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Unresolved simbols in c++ Templates Date: 10 Aug 2001 01:58:24 -0700 From: Santiago.Vazquez@casa.eads.net (santiago vazquez) Organization: http://groups.google.com/ Message-ID: References: Yes, we, are linking with an static ACE lib and we have included it in the PRJ_LIBS. That's not the problem, because, if we link a code that uses ACE, we can do it fine only if it uses no templates. We think the problem is that our gcc c++ compiler (version 2.7.2 '96) is quite old to fully support templates. We found that problem compiling ACE, but we've found out that using one of the c++ examples that comes with Tornado, we have a similar problem, and of course, it has templates. A simple C++ example with no templates is compiled fine. Santiago "Kevin Livingston" wrote in message news:... > how are you linking into the ACE library? > statically or dynamically? > if you are attempting static linking have you added the library to your > PRJ_LIBS marco in tornado? > if dynamic, is the necessary library code on the target when you attempt to > download your code? > Kevin > > > "santiago vazquez" wrote in message > news:c9e00b4a.0108090150.62bd6329@posting.google.com... > > Hello. > > > > I'm a user of Tornado 2.0.2, developing SW cross-compiling on a NT > > host and a PPC target. The problem I have is that when I use c++, I > > can compile code to generate a .out without any problem, and download > > it to the target. But, when I use Templates (trying to use ACE), > > during download (while linking) unresolved symbols appear. I think all > > paths are OK, cause the problem arise only with templates, even if > > defined on the .h files, so the problem isn't a path one. > > Tornado 2.0.2 uses a cygnus gcc version 2.7.2, and we think this is > > the problem, cause it's and old compiler version and may be it doesn't > > manage correctly templates, but we still don't have a newer gcc > > version, to verify this point. > > I wish to know if compiler is really our problem and in that case, > > where could I download a Tornado patch or a newer cross compiler > > version (NT-PPC) to fix the problem. I would appreciate a place to > > find the final executable for cross-compiling in order not to have to > > fight with gcc cross-compiler building. > > > > Thanks in advance. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: ata flash disk versus ide flash disk Date: Fri, 10 Aug 2001 10:10:22 +0100 From: James Marshall Organization: Agilent Technologies UK Ltd Message-ID: <3B73A4FE.5D20578D@agilent.com> References: <3B727FEB.CC10FAF2@optonline.net> I'm interested in what you say about the driver, but my recent experience is that there are large differences in the performance of different manufacturer's flash products. For 128MB CompactFlash disks, the SanDisk part defaults to operation at PIO1 and the Smart Modular Technology part defaults to PIO3, even though both of them will work at other rates. However the sustained throughput of the Smart Modular part is 4 times better than the SanDisk. This is the bottleneck rather than the driver. Make sure you benchmark the parts you'll use. James Marshall. Gururaj patkar wrote: > Hi all, > > I need pointers to the performance of using ata flash disk versus ide > flash disk with vxworks. Looking at the ide and ata driver code supplied > by wrs, the writes are done in polling mode (by checking the status > register values on the flash controller). The usage of intConnect() in > the ideDrv.c and ataDrv.c code is further intriguing as to why you poll > the status register values and when STATUS_READY is TRUE then only > provide interrupt driven functionality. > One uses interrupt driven process to enhance the read/write performance > to flash disk, then why use polling mode concurrent with it? > Some benchmark figures in terms of read/write access time for files of > varying sizes will be appreciated. > > gururaj --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Unresolved simbols in c++ Templates Date: Fri, 10 Aug 2001 12:28:16 +0200 From: "Geurt Vos" Organization: Rohill Technologies BV Message-ID: <10d0l9.s1s.ln@svr004.rohill.nl> References: Sender: rohill2@rohill.nl For one I know of no compiler that fully supports templates, but anyways gcc2.7.2 supports them actually quite well (considering it's an ancient version from early '96), so I'm certain any (simple) template example that comes from WRS should work... What are the compiler flags you are using? Geurt > Yes, we, are linking with an static ACE lib and we have included it in > the PRJ_LIBS. That's not the problem, because, if we link a code that > uses ACE, we can do it fine only if it uses no templates. We think the > problem is that our gcc c++ compiler (version 2.7.2 '96) is quite old > to fully support templates. > We found that problem compiling ACE, but we've found out that using > one of the c++ examples that comes with Tornado, we have a similar > problem, and of course, it has templates. A simple C++ example with no > templates is compiled fine. > > Santiago > > > "Kevin Livingston" wrote in message news:... > > how are you linking into the ACE library? > > statically or dynamically? > > if you are attempting static linking have you added the library to your > > PRJ_LIBS marco in tornado? > > if dynamic, is the necessary library code on the target when you attempt to > > download your code? > > Kevin > > > > > > "santiago vazquez" wrote in message > > news:c9e00b4a.0108090150.62bd6329@posting.google.com... > > > Hello. > > > > > > I'm a user of Tornado 2.0.2, developing SW cross-compiling on a NT > > > host and a PPC target. The problem I have is that when I use c++, I > > > can compile code to generate a .out without any problem, and download > > > it to the target. But, when I use Templates (trying to use ACE), > > > during download (while linking) unresolved symbols appear. I think all > > > paths are OK, cause the problem arise only with templates, even if > > > defined on the .h files, so the problem isn't a path one. > > > Tornado 2.0.2 uses a cygnus gcc version 2.7.2, and we think this is > > > the problem, cause it's and old compiler version and may be it doesn't > > > manage correctly templates, but we still don't have a newer gcc > > > version, to verify this point. > > > I wish to know if compiler is really our problem and in that case, > > > where could I download a Tornado patch or a newer cross compiler > > > version (NT-PPC) to fix the problem. I would appreciate a place to > > > find the final executable for cross-compiling in order not to have to > > > fight with gcc cross-compiler building. > > > > > > Thanks in advance. --------------------------- End of New-News digest ********************** From vxwexplo-errs@csg.lbl.gov Fri Aug 10 06:53:32 2001 From: fscheurer@manz-acs.de Date: Fri Aug 10 06:53:34 PDT 2001 Subject: AUTOREPLY comp.os.vxworks newsdigest 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_000_00017662.3B74036E Content-Type: text/plain; charset=iso-8859-1 Content-Transfer-Encoding: 7Bit This is the Mailbox of Friedrich Scheurer. R&D - ACS GmbH - D-72768 Reutlingen Sorry, I am out of office until August 20th 2001! In urgent cases please contact one of the following persons at ACS GmbH: Otto Angerhofer (Managing director) e-mail: oangerhofer@manz-acs.de Eduard Ams (Technical manager) e-mail: eams@manz-acs.de Yvonne Anhorn (Marketing) e-mail: yanhorn@manz-acs.de Margit Arnold (Sales) e-mail: marnold@manz-acs.de Christina Sixel (Sales) e-mail: csixel@manz-acs.de MfG / Kind Regards F. Scheurer ---------------------------------------------------------------- Friedrich Scheurer, Dipl.-Ing. phone +49 7121 9019-23(0) ACS GmbH - R&D fax +49 7121 9019-98 Steigaeckerstrasse 13 web www.manz-acs.de D-72768 Reutlingen (Germany) e-mail: fscheurer@manz-acs.de ---------------------------------------------------------------- ------_=_NextPart_000_00017662.3B74036E Content-Type: text/html; charset=iso-8859-1 Content-Transfer-Encoding: quoted-printable

This is the Mailbox of Friedrich Scheurer.
R&D - ACS GmbH - D-72768 Reutlingen
 
Sorry, I am out of office until August 20th 2001!
 
In urgent cases please contact one of the
following persons at ACS GmbH:
 
Otto Angerhofer (Managing director)
Eduard Ams (Technical manager)
Yvonne Anhorn (Marketing)
Margit Arnold (Sales)
e-mail: marnold@manz-acs.de
Christina Sixel (Sales)

 
MfG / Kind Regards
 
F. Scheurer
 
----------------------------------------------------------------
Friedrich Scheurer, Dipl.-Ing.    phone    +49 7121 9019-23(0)
ACS GmbH  -  R&D                  fax      +49 7121 9019-98
Steigaeckerstrasse 13             web      www.manz-acs.de
D-72768 Reutlingen (Germany)      e-mail:  fscheurer@manz-acs.de
----------------------------------------------------------------
------_=_NextPart_000_00017662.3B74036E-- From vxwexplo-errs@csg.lbl.gov Fri Aug 10 07:02:57 2001 From: Wu@mosaid.com Date: Fri Aug 10 07:02:59 PDT 2001 Subject: Re: why do I lose network connection after muxDevLoad and MuxDevStart in Hello vxWorks world, This is regarding the porting of SSP in TMS. I have noticed if I include muxDevLoad and muxDevStart for sw,ag,ap,hw END devices (for instance in usrNetwork.c). I will lose my network connection (target server can not be set up, tWdbTask/tNetTask get exception et al)). I have tried several ways but I have not solved this problem so far. It seems to me so long the software for in-band chip is loaded, then the out-of band ethernet card software won't functoin properly. The mux binding is messed up. This problem is blocking me and I cannot go any further since I'm planning to use NFS as the file system for now. In this case the debugger cannot be used either. I have tried to start wdbConfig after TMSAppInit started but it won't work either. I would guess this is a common problem when porting a SSP. Any ideas to solve it? Huang From vxwexplo-errs@csg.lbl.gov Fri Aug 10 07:55:32 2001 From: "David Laight" Date: Fri Aug 10 07:55:34 PDT 2001 Subject: Re: ata flash disk versus ide flash disk My recollections are that the ATA and IDE interfaces are really one and the same. I did look at the WRS supplied drivers when I wanted a driver for pcmcia ATA cards and compact flash (running in 'true IDE' mode), however IMHO they were not well written - large, inefficient and buggy (and only designed for x86 PC hardware). The ATA/IDE interface does require you to poll for certain events, using interrupts for others. Polling is used for 'fast' operations (eg waiting for the 'disk' to say it is ready to accept the first buffer of a write), interrupts are used for slower actions (read data available, write complete). If you can GUARANTEE you have flash media, it may be worth polling on reads - avoiding the interrupt latency. If you have an appropriate memory architecture (and other hw) one of the DMA modes may be faster - but on an embedded system I wouldn't guarantee it! I never did find out how ATA flash devices (eg compact flash) handle 512 byte sector writes using (what I assume are) standard flash memory parts giving 128kb erase sectors. Doing sequential writes (restoring a fs archive using a raw device overlaying the whole disk) you can achieve write times that look as though they are limited by the erase time (about 0.5 seconds for a 64k sector). The cards may be using a sector map (TFFS does something like that), I have seen a corrupt CF card that MIGHT have suffered a corruption of the sector map table (FAT all over sector 0-15, chunks of file in the FAT, FAT sectors replicated - a right mess, don't see how my driver or the vxworks dosfs code could do that!). Apart from the erase time, the transer rate will probably be limited by the bus cycle time. For PCMCIA and compact flash 'true ide' i/o cycles this is a 165ns command pulse - leading to a 330ns bus cycle. PCMCIA memory cycles can be slightly faster - if you can sort out the appropriate bus interface, and guarantee to obtain flash parts that support it. None of the card data sheets I saw ever gave timings for the card - they always repeated the timing in the standard! Some of the modern cards probably support 100ns bus cycles - but a different batch of cards from the same supplier may vary. Using the 'drive information' buffer doesn't help! The 'PIO modes supported' don't really apply to these devices. We had some cards (from a well known major manufacturer) that reported with a corrupt buffer (one byte missing leading to incorrect LBA size), and others that didn't seem to support the command. The device serial number (that out to be unique) was also constant - so you can't tell if the same card is re-inserted... (Try getting a distributer to agree to these sort of problems! 'Cards work in a PC' we sent tham back and they were returned a few weeks later 'fixed' (I suspect they had put the dos FAT table and filesystem back - we would have deleted those.) David From vxwexplo-errs@csg.lbl.gov Fri Aug 10 10:32:30 2001 From: "Guna Sekhar" Date: Fri Aug 10 10:32:32 PDT 2001 Subject: relation between tick and clock frequency This is a multi-part message in MIME format. ------=_NextPart_000_0027_01C121F2.C0059C80 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable Hello VxWorkers, We have VxWorks 5.4 Beta Version running on MIPS Malta Board. The = Processor is MIPS 4KC. In this context how is one tick related to the = clock frequency(233MHz). Kindly answer at the earliest. Thank you, kgs ------=_NextPart_000_0027_01C121F2.C0059C80 Content-Type: text/html; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable
Hello VxWorkers,
 
We have VxWorks 5.4 Beta Version = running on MIPS=20 Malta Board. The Processor is MIPS 4KC. In this context how is one tick = related=20 to the clock frequency(233MHz). Kindly answer at the = earliest.
 
Thank you,
 
kgs
 
 
------=_NextPart_000_0027_01C121F2.C0059C80-- From vxwexplo-errs@csg.lbl.gov Fri Aug 10 15:55:27 2001 From: Anuradha Pani Date: Fri Aug 10 15:55:29 PDT 2001 Subject: special function in VxWorks Hi, I am analyzing a PCI driver routine in VxWorks. I came across this function, special(SET_INTERRUPT, 0); I could not find any references to this function in the index. Does anybody know if this is a system call, or what this function does? Thanks Anuradha __________________________________________________ Do You Yahoo!? Send instant messages & get email alerts with Yahoo! Messenger. http://im.yahoo.com/ From vxwexplo-errs@csg.lbl.gov Fri Aug 10 17:55:52 2001 From: fscheurer@manz-acs.de Date: Fri Aug 10 17:55:54 PDT 2001 Subject: AUTOREPLY special function in VxWorks This message is in MIME format. Since your mail reader does not understand this format, some or all of this message may not be legible. ------_=_NextPart_000_00017698.3B749EB4 Content-Type: text/plain; charset=iso-8859-1 Content-Transfer-Encoding: 7Bit This is the Mailbox of Friedrich Scheurer. R&D - ACS GmbH - D-72768 Reutlingen Sorry, I am out of office until August 20th 2001! In urgent cases please contact one of the following persons at ACS GmbH: Otto Angerhofer (Managing director) e-mail: oangerhofer@manz-acs.de Eduard Ams (Technical manager) e-mail: eams@manz-acs.de Yvonne Anhorn (Marketing) e-mail: yanhorn@manz-acs.de Margit Arnold (Sales) e-mail: marnold@manz-acs.de Christina Sixel (Sales) e-mail: csixel@manz-acs.de MfG / Kind Regards F. Scheurer ---------------------------------------------------------------- Friedrich Scheurer, Dipl.-Ing. phone +49 7121 9019-23(0) ACS GmbH - R&D fax +49 7121 9019-98 Steigaeckerstrasse 13 web www.manz-acs.de D-72768 Reutlingen (Germany) e-mail: fscheurer@manz-acs.de ---------------------------------------------------------------- ------_=_NextPart_000_00017698.3B749EB4 Content-Type: text/html; charset=iso-8859-1 Content-Transfer-Encoding: quoted-printable
This is the Mailbox of Friedrich Scheurer.
R&D - ACS GmbH - D-72768 Reutlingen
 
Sorry, I am out of office until August 20th 2001!
 
In urgent cases please contact one of the
following persons at ACS GmbH:
 
Otto Angerhofer (Managing director)
Eduard Ams (Technical manager)
Yvonne Anhorn (Marketing)
Margit Arnold (Sales)
e-mail: marnold@manz-acs.de
Christina Sixel (Sales)

 
MfG / Kind Regards
 
F. Scheurer
 
----------------------------------------------------------------
Friedrich Scheurer, Dipl.-Ing.    phone    +49 7121 9019-23(0)
ACS GmbH  -  R&D                  fax      +49 7121 9019-98
Steigaeckerstrasse 13             web      www.manz-acs.de
D-72768 Reutlingen (Germany)      e-mail:  fscheurer@manz-acs.de
----------------------------------------------------------------
------_=_NextPart_000_00017698.3B749EB4-- From vxwexplo-errs@csg.lbl.gov Sat Aug 11 04:03:15 2001 From: Vxworks Exploder Date: Sat Aug 11 04:03:18 PDT 2001 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Sat Aug 11 04:03:10 PDT 2001 Subject: Re: Unresolved simbols in c++ Templates Subject: Re: what's wrong with this? please help Subject: Re: Unresolved simbols in c++ Templates Subject: Re: Anyone any experiences building gcc 2.95.X Cygwin cross x86 Subject: Re: what's wrong with this? please help Subject: protocol stack design under vxworks Subject: Re: MCP750 Subject: Re: Unresolved simbols in c++ Templates Subject: How to list files Subject: question on semaphore Subject: Re: question on semaphore Subject: Replacing the TCP/IP stack Subject: Re: support for explicit instantiation Subject: Re: How to list files Subject: Re: Write to disk files ASAP Subject: semBCreate() - Semaphore Creation is failing Subject: Re: Replacing the TCP/IP stack Subject: Re: semBCreate() - Semaphore Creation is failing Subject: Re: Replacing the TCP/IP stack Subject: Re: How to use the Intel 82559ER Lan Controller in the vxworks system? Subject: Re: How to use the Intel 82559ER Lan Controller in the vxworks system? ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: Unresolved simbols in c++ Templates Date: 10 Aug 2001 04:35:38 -0700 From: Santiago.Vazquez@casa.eads.net (santiago vazquez) Organization: http://groups.google.com/ Message-ID: References: Hello again Thank you for your answer, but that's not the problem. We've included de lib in the PRJ_LIBS macro, and this works, because the problem doesn't appear when we compile and download a program with no templates that uses that library (ACE). In that case, it works fine. On the other hand we've found out that a similar unresolved_symbols message appears while trying to download a c++ executable with templates that comes with Tornado, and that is completely independent from ACE. So we think the problem is that the compiler Tornado uses is quite old (gcc 2.7.2 '96) and doesn't fully support templates. We're trying to find a newer NT-PPC gcc cross-compiler to see if the problem is solved, but we can't find it. "Kevin Livingston" wrote in message news:... > how are you linking into the ACE library? > statically or dynamically? > if you are attempting static linking have you added the library to your > PRJ_LIBS marco in tornado? > if dynamic, is the necessary library code on the target when you attempt to > download your code? > Kevin > > > "santiago vazquez" wrote in message > news:c9e00b4a.0108090150.62bd6329@posting.google.com... > > Hello. > > > > I'm a user of Tornado 2.0.2, developing SW cross-compiling on a NT > > host and a PPC target. The problem I have is that when I use c++, I > > can compile code to generate a .out without any problem, and download > > it to the target. But, when I use Templates (trying to use ACE), > > during download (while linking) unresolved symbols appear. I think all > > paths are OK, cause the problem arise only with templates, even if > > defined on the .h files, so the problem isn't a path one. > > Tornado 2.0.2 uses a cygnus gcc version 2.7.2, and we think this is > > the problem, cause it's and old compiler version and may be it doesn't > > manage correctly templates, but we still don't have a newer gcc > > version, to verify this point. > > I wish to know if compiler is really our problem and in that case, > > where could I download a Tornado patch or a newer cross compiler > > version (NT-PPC) to fix the problem. I would appreciate a place to > > find the final executable for cross-compiling in order not to have to > > fight with gcc cross-compiler building. > > > > Thanks in advance. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: what's wrong with this? please help Date: Fri, 10 Aug 2001 20:12:08 -0700 From: "Xiao Boyu" Organization: nanwang Message-ID: <9l0j0s$5au$1@mail.cn99.com> References: SSB0aGluayBhbGwgeW91ciBxdWVzdGlvbiBvbmx5IHJlbGF0ZWQgdG8gb25lIHByb2JsZW0tIHBy b2JsZW0gMS4NCndoZW4gdG9ybmFkbyByZXBvcnQiY29yZSBmaWxlIGNoZWNrc3VtcyBkbyBub3Qg bXVjaCIsIHRoYXQgbWVhbnMNCnRoZSB2eFdvcmtzIHJ1bm5pbmcgaW4gdGFyZ2V0IGJvYXJkIGlz IGRpZmZlcmVudCBmcm9tIHRoZSBvbmUgeW91DQpzZXQgaW4gdGFyZ2V0IHNlcnZlci4gSnVzdCBz ZXQgdGhlIGNvcnJlY3QgcGF0aCBhbmQgdnhXb3JrcyBmaWxlIGluDQp0YXJnZXQgc2VydmVyIGFu ZCBldmVyeXRoaW5nIHdpbGwgYmUgT0suDQoNClhpYW8NCg0KInhpbmciIDxsaXV6aEBmb3VuZGVy LmNvbS5jbj4gd3JvdGUgaW4gbWVzc2FnZSBuZXdzOmJiMWJjMmQxLjAxMDgwOTIzNDQuMzNmMzBi OTVAcG9zdGluZy5nb29nbGUuY29tLi4uDQo+IGhpLHZ4d29ya3MgZXhwZXJ0cywNCj4gICBpJ20g bm93IHVzaW5nIHRvcm5hZG8gMi4wIGZvciBYODYsIGFmdGVyIGkgYnVpbGQgYSBib290ZGlzayBh bmQgYQ0KPiB2eHdvcmtzIGZyb20gV1JTJ3MgQlNQLHRoZXJlIGlzIGZvbGxvd2luZyBwcm9tYmxl IG9jY3VyZWQsY2FuIGFueWJvZHkNCj4gdGVsbCBtZSB3aHk/DQo+IGFueSBoZWxwIG9yIGhpbnQg d2lsbCBiZSBncmVhdCBhcHByZWljYXRlZCENCj4gICBpIHVzZSBhIHNlcmlhbCBsaW5lIHRvIGNv bm5lY3QgbXkgdGFyZ2V0IGFuZCBob3N0LGFuZCBoYXZlIG1hZGUgc29tZQ0KPiBjaGFuZ2VzIHRv IFdSUydzIEJTUCAoY29uZmlnLmggYW5kIHZ4d29ya3MncyBjb25maWd1cmUpLHRoZW4gZ2V0IGEN Cj4gYm9vdGRpc2sgYW5kIGEgdnh3b3JrcyBpbWFnZSx0aGlzIGJvb3RkaXNrIGFuZCB2eHdvcmtz IHJlYWxseQ0KPiB3b3JrcyxidXQgc3RpbGwgdGhlcmUgYXJlIGZvbGxvd2luZyBwcm9tYmxlczoN Cj4gICAgMSl3aGVuIGkgc3RhcnQgdGd0c3ZyLHRoZSB0Z3RzdnIgY29uc29sZSBhbGF3eXMgcmVw b3J0IGEgd2FybmluZw0KPiAiY29yZSBmaWxlIGNoZWNrc3VtcyBkbyBub3QgbXVjaCAiLCBidXQg dGhlIGJvb3RpbmcgcHJvY2VzcyBzdGlsbA0KPiBnb2luZyBub3JtYWxseS4NCj4gICAgMilldmVu IG15IHZ4d29ya3MncyBjb25maWd1cmUgaGFzIGluY2x1ZGUgInN5bWJvbCB0YWJsZSIgICx3aGVu IGkNCj4gY29uZmlnIGEgdGFyZ2V0IHNlcnZlcidzICJjb3JlIGZpbGUgYW5kIHN5bWJvbCB0YWJs ZSIgb3B0aW9uICxpIGNhbg0KPiBvbmx5IHVzZSAibm8gc3ltYm9sIiBvcHRpb24saWYgaSBzZXQg dG8gImdsb2JhbCBzeW1ib2wiIG9yICJhbGwNCj4gc3ltYm9sIix0aGUgdGd0c3ZyIGNvbnNvbGUg d2lsbCByZXBvcnQgIiBycGNjb3JlIGJhY2tlbmQgY2xpZW50IHRpbWUNCj4gb3V0IiAsYW5kIGJv b3QgcHJvY2VzcyBzdG9wZWQuDQo+ICAgIDMpYWZ0ZXIgdnh3b3JrcyBib290ZWQgLCBpIHN0YXJ0 IGEgc2hlbGwsYnV0IHdoZW4gaSB0eXBlIGEgImkiDQo+IGNvbW1hbmQgLHRoZSBzaGVsbCByZXBv cnQgInRoZSB0YXJnZXQgc3lzdGVtIGlzIG5vdCBpbml0aWFsaXplZCB5ZXQNCj4gLkNhbid0IHBy b2Nlc3NlZCx2YXVsZSA9LTE9MHhmZmZmZmZmIg0KPiAgICA0KXdoZW4gaSBkb3dubG9hZCBhIGRl bW8gcHJvZ3JhbSBpIGJ1aWx0IGZyb20gVG9ybmFkbyBkZW1vICxhZnRlcg0KPiBkb3dubG9hZCBj b21wbGV0ZWQgLGl0IHJlcG9ydCBhIGVycm9yIGxpc3QgYm94ICJ1bnJlc29sdmVkIHN5bWJvbHMN Cj4gZW5jb3VudGVyZWQgIixhbGwgdnh3b2tzJ3MgZnVuY3Rpb24gKHN1Y2ggYXMgX3ByaW50Zixf ZnJlZSAuLiApIGFyZQ0KPiBsaXN0ZWQsIGFuZCB3aGVuIGkgdXNlIGJyb3dzZXIgdG8gY2hlY2sg dGhlIGRvd25sb2FkZWQgcHJvamVjdCdzDQo+IHN0YXR1cywgaSBmb3VuZCB0aGVzZSBmdW5jdGlv biAoX3ByaW50ZixfZnJlZS4uLiltYXJrcyBhcyAidW5yZXNvbHZlZA0KPiBTeW1ib2xzIi4NCj4g ICAgIGNhbiBhbnlib2R5IGhlbHAgbWU/IA0KPiAgICAgDQo+IGJlc3QgcmVnYXJkcw0K --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Unresolved simbols in c++ Templates Date: Fri, 10 Aug 2001 09:24:04 -0400 From: "Kevin Livingston" Message-ID: <6dRc7.2$3r6.125@client> References: ok, because we too are using ACE - both templates (free list) and non template code (mutex), and so far I haven't run into any problems with it. Our target is a Pentium though, but we are using the same compilier that you are. if it helps here is a copy of our compile line (out of Tornado): I cut out some of our specific include directories labeled with cc386 -g -mpentium -ansi -nostdinc -DRW_MULTI_THREAD -D_REENTRANT - -fvolatile -nostdlib -fno-builtin -fno-defer-pop - -I. -ID:\Tornado\target\h -ID:\ace_5_1_14 - -ID:\ - -DCPU=PENTIUM - -c D:\\Output.cpp With the same compilier I have also used my own home built templates, and the STL templates (the definitions that came with Tornado) - and I have never had a problem with either of these things either. Kevin "santiago vazquez" wrote in message news:c9e00b4a.0108100335.1f09b426@posting.google.com... > Hello again > > Thank you for your answer, but that's not the problem. We've included > de lib in the PRJ_LIBS macro, and this works, because the problem > doesn't appear when we compile and download a program with no > templates that uses that library (ACE). In that case, it works fine. > On the other hand we've found out that a similar unresolved_symbols > message appears while trying to download a c++ executable with > templates that comes with Tornado, and that is completely independent > from ACE. So we think the problem is that the compiler Tornado uses is > quite old (gcc 2.7.2 '96) and doesn't fully support templates. > We're trying to find a newer NT-PPC gcc cross-compiler to see if the > problem is solved, but we can't find it. > > > "Kevin Livingston" wrote in message news:... > > how are you linking into the ACE library? > > statically or dynamically? > > if you are attempting static linking have you added the library to your > > PRJ_LIBS marco in tornado? > > if dynamic, is the necessary library code on the target when you attempt to > > download your code? > > Kevin > > > > > > "santiago vazquez" wrote in message > > news:c9e00b4a.0108090150.62bd6329@posting.google.com... > > > Hello. > > > > > > I'm a user of Tornado 2.0.2, developing SW cross-compiling on a NT > > > host and a PPC target. The problem I have is that when I use c++, I > > > can compile code to generate a .out without any problem, and download > > > it to the target. But, when I use Templates (trying to use ACE), > > > during download (while linking) unresolved symbols appear. I think all > > > paths are OK, cause the problem arise only with templates, even if > > > defined on the .h files, so the problem isn't a path one. > > > Tornado 2.0.2 uses a cygnus gcc version 2.7.2, and we think this is > > > the problem, cause it's and old compiler version and may be it doesn't > > > manage correctly templates, but we still don't have a newer gcc > > > version, to verify this point. > > > I wish to know if compiler is really our problem and in that case, > > > where could I download a Tornado patch or a newer cross compiler > > > version (NT-PPC) to fix the problem. I would appreciate a place to > > > find the final executable for cross-compiling in order not to have to > > > fight with gcc cross-compiler building. > > > > > > Thanks in advance. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Anyone any experiences building gcc 2.95.X Cygwin cross x86 Date: 10 Aug 2001 08:06:06 -0700 From: olandgren@yahoo.com (Owen Landgren) Organization: http://groups.google.com/ Message-ID: <8f4729c1.0108100706.671e0589@posting.google.com> References: <9i145t$65d$1@overload.lbl.gov> <8P%27.104$os1.458451@newsr2.u-net.net> "Dave Korn" wrote in message news:<8P%27.104$os1.458451@newsr2.u-net.net>... > fscheurer@manz-acs.de wrote in message <9i145t$65d$1@overload.lbl.gov>... > >Hi VxWorks-Gurus, > > > >I ran into the problem, that the integrated target-headers > >seem to be not confor with GNU expectations. > > Did you use the --with-headers= flag when configuring to supply the target > headers, or did you copy them into $prefix/$target/sys-include, or into > $prefix/$target/include ? Sadly, these other methods don't help. I'm running into the same problem with a clean GCC-3.0 using the --with-headers= option to configure. There's mention of fixincs in the stddef.h header file that's generated, but there's still a conflict that renders the build inoperable. > The error messages you posted look a lot like fixincs hasn't run. Can you > confirm or deny this? I can confirm that fixincs is being run in my build process and the problem still exists. > I suspect that the fixes already exist in the fixincs script, and merely > haven't been applied to your target headers in the course of building the > cross-compiler. However, the VxWorks targets for Gcc haven't been well > maintained of late, and it is possible that the WRS headers have changed in > such a way that the fixincs test-pattern no longer matches the WRS header > file, causing a fix to be missed; or that new code in the headers mean that > new fixes need to be added. Give it another try, being sure to use one of > the first two methods I mentioned above to supply the headers, and tell us > if it works any better. Well, this isn't exactly the best of news, but I thought it prudent to drop a line with my difficulty so that those who are thinking of doing this know that therein lie dragons. FWIW, I'm using Tornado 2.02, the latest cygwin, and am compiling for powerpc-wrs-vxworks. I use the following configure command: ../configure --target=powerpc-wrs-vxworks - --with-headers=/Tornado/target/h/ --prefix=/cross/vxtools/bin - --exec-prefix=/cross/vxtools/bin Owen Landgren --------------------------- Newsgroups: comp.os.vxworks Subject: Re: what's wrong with this? please help Date: Fri, 10 Aug 2001 15:36:58 GMT From: Andray Kaganovsky Organization: @Home Message-ID: References: liuzh@founder.com.cn (xing) wrote in news:bb1bc2d1.0108092344.33f30b95 @posting.google.com: > hi,vxworks experts, > i'm now using tornado 2.0 for X86, after i build a bootdisk and a > vxworks from WRS's BSP,there is following promble occured,can anybody > tell me why? > any help or hint will be great appreicated! > i use a serial line to connect my target and host,and have made some > changes to WRS's BSP (config.h and vxworks's configure),then get a > bootdisk and a vxworks image,this bootdisk and vxworks really > works,but still there are following prombles: > 1)when i start tgtsvr,the tgtsvr console alawys report a warning > "core file checksums do not much ", but the booting process still > going normally. [AndrayK]. Are you sure that you have specified correct filename in target server's "Core Files and Symbols" page ? Is it the file that your target boots ? > 2)even my vxworks's configure has include "symbol table" ,when i > config a target server's "core file and symbol table" option ,i can > only use "no symbol" option,if i set to "global symbol" or "all > symbol",the tgtsvr console will report " rpccore backend client time > out" ,and boot process stoped. > 3)after vxworks booted , i start a shell,but when i type a "i" > command ,the shell report "the target system is not initialized yet > .Can't processed,vaule =-1=0xfffffff" > 4)when i download a demo program i built from Tornado demo ,after > download completed ,it report a error list box "unresolved symbols > encountered ",all vxwoks's function (such as _printf,_free .. ) are > listed, and when i use browser to check the downloaded project's > status, i found these function (_printf,_free...)marks as "unresolved > Symbols". > can anybody help me? > > best regards > --------------------------- Newsgroups: comp.os.vxworks Subject: protocol stack design under vxworks Date: 10 Aug 2001 08:56:07 -0700 From: barton@comneti.com (Jim Barton) Organization: http://groups.google.com/ Message-ID: <414a7fa0.0108100756.425b1dee@posting.google.com> We are currently trying to debug an ISDN protocol stack that was recently ported from a Intel 186 platform (with no OS) to a PPC platform that uses VxWorks. Since we are currently having major problems with the stability of the stack and since we are new to VxWorks, we suspect that we may not have done things correctly. For example, each protocol layer is now represented as a separate task (each having the same priority of 100)under vxWorks. Is this the normal way that a protocol stack should be implemented in vxWorks? If not, how serious of a design problem is this? Is it worthwhile re-designing this into one(?) task or could we make this work with our current task per layer design? Also, how is timing normally accomplished within a protocol stack under VxWorks? In our old platform (without a OS) timing was accomplished by a main loop that would service messages and events at specific intervals to send out heartbeats, determine timeouts, etc. I'm not clear what is the proper way of doing this under VxWorks. Should we create timers and put all of our time-dependent stuff in the SIGALRM handler? And, yes, I know we could purchase a ISDN stack already designed for VxWorks but we would rather avoid that expense since we already have source for the stack from our older platform. We would appreciate any tips on how we should port this protocol to vxWorks! Thanks, Jim --------------------------- Newsgroups: comp.os.vxworks Subject: Re: MCP750 Date: Fri, 10 Aug 2001 12:44:59 -0500 From: Andrew Johnson Organization: APS, Argonne National Laboratory Message-ID: <3B741D9B.250289D4@aps.anl.gov> References: <46095d81.0108091444.45fcd3a6@posting.google.com> Pete Jankus wrote: > > Has anyone been able to use the Watchdog timer in the Raven ASIC on a > Motorola MCP750 cPCI board? I am trying to enable WDT2 to reset the > board on a Watchdog timeout, however I am unable to get the WDT2 to > begin counting down. Is the Raven's Timer Frequency Register set properly? The MVME2700 also uses a RavenMpic, but the BSP doesn't initialize that register as it uses a different device for the aux clock. - - Andrew - -- The world is such a cheerful place when viewed from upside-down It makes a rise of every fall, a smile of every frown --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Unresolved simbols in c++ Templates Date: Fri, 10 Aug 2001 12:49:46 -0500 From: Andrew Johnson Organization: APS, Argonne National Laboratory Message-ID: <3B741EBA.2FA1D99E@aps.anl.gov> References: santiago vazquez wrote: > > So we think the problem is that the compiler Tornado uses is > quite old (gcc 2.7.2 '96) and doesn't fully support templates. > We're trying to find a newer NT-PPC gcc cross-compiler to see if the > problem is solved, but we can't find it. Note that the C++ compiler provided with Tornado 2.0 is *not* based on the gcc-2.7.2 release although the C compiler is. The Tornado Release Notes explain that the Tornado 2.0 C++ toolchain was based on EGCS 1.0.2, although the front-end still reports that it's 2.7.2. - - Andrew - -- The world is such a cheerful place when viewed from upside-down It makes a rise of every fall, a smile of every frown --------------------------- Newsgroups: comp.os.vxworks Subject: How to list files Date: Fri, 10 Aug 2001 10:50:29 -0500 From: "Shesha Bhushan" Organization: Arizona State University Message-ID: <9l170k$lma$1@news.asu.edu> Hi All, I am a student at ASU and i am doing a research on scsi. I have connected a SCSI hard disk to a Single Board Computer by synergy. Using tornado on an NT machine, how can i list the files of that hard disk. When i give "devs" command then, i can see "/sd0/" device. But when i give "ls "/sd0/" it gives the following error - -> ls "/sd0/" Parameter format not correct - "sd0". value = -1 = 0xffffffff can anybody please help. Do i need to mount or do any thing like that. U can also reach me at Shesha@asu.edu. --------------------------- Newsgroups: comp.os.vxworks Subject: question on semaphore Date: Fri, 10 Aug 2001 14:33:51 -0400 From: "Yong Zhang" Organization: Marconi Message-ID: <9l19c2$isb$1@newsfeed.pit.comms.marconi.com> Hi, I want to know if there is a limitation on the number of semaphores? Will it cause problem if I use over 4000 semaphores? Thanks! Yong --------------------------- Newsgroups: comp.os.vxworks Subject: Re: question on semaphore Date: Fri, 10 Aug 2001 19:42:39 +0100 From: molochai Message-ID: <5442faa74a%molochai@vapour-trail.demon.co.uk> References: <9l19c2$isb$1@newsfeed.pit.comms.marconi.com> In message <9l19c2$isb$1@newsfeed.pit.comms.marconi.com> "Yong Zhang" wrote: > Hi, > > I want to know if there is a limitation on the number of semaphores? Will > it cause problem if I use over 4000 semaphores? > Only memory AFAIK. --------------------------- Newsgroups: comp.os.vxworks Subject: Replacing the TCP/IP stack Date: 10 Aug 2001 12:00:52 -0700 From: chrismooremobile@yahoo.com (Chris Moore) Organization: http://groups.google.com/ Message-ID: Hi, I'm interested in replacing the current TCP/IP stack with a newer module from Linux. Has anyone done this sort of work? I'm fairly adept at the END/MUX drivers and so forth and was looking for some estimate of work involved. Yep, I know that the VxWorks stack (that you have to pay for source) is based on the FREE open source BSD stack... funny they charge you for it considering they didn't write it. Anyway, as a side note I have found the routing performance of VxWorks to be rather hopeless. I setup two dummy NPT drivers and injected a packet into them and basically had them IP spoof so the packet bounced back and forth (up through the IP layer, routing look up etc) a certain amount of times. I ended up (on a 150Mhz MIPS) with about 5500 packets per second. This was with NO ARP and NO reassmble (so no framented packets) and no memory copies and the CPU was going hard out (all other tasks were locked out). Very sad performance. There are a number of ways to improve this but we need xs to the source code, and I would rather utilise the Linux stuff as we can then port other apps such as NAT. Any ideas? Regards, Chris --------------------------- Newsgroups: comp.os.vxworks Subject: Re: support for explicit instantiation Date: Fri, 10 Aug 2001 19:14:38 GMT From: langston.reek@grassvalleygroup.com (Langston Reek) Organization: SBC Internet Services Message-ID: <3b743260.4448578@news.pacbell.net> References: <3b72d625.89808421@news.pacbell.net> <9l00p1$6r8nf$1@ID-67650.news.dfncis.de> Albert, Thanks. This is exactly what I was looking for. On Fri, 10 Aug 2001 08:59:42 +0200, "Albert Wijnja" wrote: >Hi Langston, > >The 2.7.2 compiler does support explicit template instantiation. I use it >and it works fine for me. From your example it seems to me that you are >trying to instantiate the template from a local scopy (between the >brackets). I instantiate the templates on global scope (thus, outside the >brackets). >Furthermore, I ran into some problems with unresolved externals regarding >template functions (no template classes!!!) that are used from within the >templates. These template functions must also be instantiated explicitly. In >the newsgroup archives I found a convenient solution for this provided by >Lothar Werzinger. He made some macros to make life easier. They are listed >below. > >So, if you are using vector < string > you should add the following to your >code (on global scope): > >// Since string is also a template, you need to instantiate it also !!! >template class dm_string; > >// Now instantiate the vector and the related template functions >template class vector < dm_string >; >template dm_fill( dm_string ); >template dm_fill_n( dm_string ); > > >Hope this works for you. Good luck. > >Albert > > > > >#if defined (ACE_HAS_EXPLICIT_TEMPLATE_INSTANTIATION) >/////////////////////////////////////////////////////////////////////// >// Some helper macros to define template instatiations of STL easier >// Sample instantiation of vector < int > >// template class vector < int >; >// template dm_fill( int ); >// template dm_fill_n( int ); >// #if defined (__GNUC__) >// template dm_uninitialized_copy_aux( int ); >// template dm_uninitialized_copy_aux_const( int ); >// template dm_uninitialized_fill_n_aux( int ); >// #endif /* defined (__GNUC__) */ >// Sample instantiation of list < MyClass > >// template class list < MyClass >; >// template dm_insert( MyClass ); > >// Macro to define the string >#define dm_string \ > basic_string < char > > >// Macro's to define some template functions for a vector >#define dm_fill( _X_ ) \ > void fill < _X_ *, _X_ > ( _X_ *, _X_ *, _X_ const & ) >#define dm_fill_n( _X_ ) \ > _X_ * fill_n < _X_ *, unsigned int, _X_ > ( _X_ *, unsigned int, >_X_ const & ) >#define dm_rb_tree(_X_, _Y_) \ > rb_tree < _X_, pair < _X_ const, _Y_ >, select1st < pair < _X_ >const, _Y_ > >, less < _X_ > > > >// Macro's to define the insert template function for a list >#define list_par( _X_ ) \ > _X_, __default_alloc_template >#define l_iter( _X_ ) \ > __list_iterator<_X_, _X_ &, _X_ *> >#define const_l_iter( _X_ ) \ > __list_iterator<_X_, _X_ const &, _X_ const *> >#define dm_insert( _X_ ) \ > void list< list_par( _X_ ) >::insert< const_l_iter( _X_ ) >(l_iter( _X_ ), >const_l_iter( _X_ ), const_l_iter( _X_ )); > >// GNU STL needs that (other STLs maybe also, but we do not know) >#if defined (__GNUC__) >#define dm_uninitialized_copy_aux( _X_ ) \ > _X_ * __uninitialized_copy_aux <_X_ *, _X_ *> ( _X_ *, _X_ *, _X_ >*, __false_type ) >#define dm_uninitialized_copy_aux_const( _X_ ) \ > _X_ * __uninitialized_copy_aux <_X_ const *, _X_ *> ( _X_ const *, >_X_ const *, _X_ *, __false_type ) >#define dm_uninitialized_fill_n_aux( _X_ ) \ > _X_ * __uninitialized_fill_n_aux <_X_ *, unsigned int, _X_ > ( _X_ >*, unsigned int, _X_ const &, __false_type ) >#endif /* defined (__GNUC__) */ > >#endif /* ACE_HAS_EXPLICIT_TEMPLATE_INSTANTIATION */ > >"Langston Reek" wrote in message >news:3b72d625.89808421@news.pacbell.net... >> Hello, >> I'm trying to explicitly instantiate templates and I can't for the >> life of me get this to compile and >> load. I'm using Tornado2 (GNU 2.7.2 compiler), and I'm compiling with >> the compile flag, >> -fno-implicit-templates. >> { >> template class vector; // explicit instantiation >> vector list; >> } >> Is it the case that the compiler doesn't support explicit >> instantiation, >> or do I just have the wrong syntax. >> > > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: How to list files Date: Fri, 10 Aug 2001 16:46:13 -0400 From: "Kevin Livingston" Message-ID: References: <9l170k$lma$1@news.asu.edu> if you are using the shell in tornado through the target server you may need to preface your commands with the @ so that the shell knows it should execute them on the target instead of the host machine. for example I have a SBC with dev "/ata0a/" in the terminal for that computer ls "/ata0a/" generates a valid and approiate listing in Tornado, using the target server the same command ls "/ata0a/" returns Parameter format not correct - "ta0a". however @ls "/ata0a/" generates the results as expected. to better illustrate look at the differeence between the following commands pwd @pwd good luck with your research Kevin "Shesha Bhushan" wrote in message news:9l170k$lma$1@news.asu.edu... > Hi All, > I am a student at ASU and i am doing a research on scsi. I have > connected a SCSI hard disk to a Single Board Computer by synergy. Using > tornado on an NT machine, how can i list the files of that hard disk. > When i give "devs" command then, i can see "/sd0/" device. But when i give > "ls "/sd0/" it gives the following error > > -> ls "/sd0/" > Parameter format not correct - "sd0". > value = -1 = 0xffffffff > > can anybody please help. Do i need to mount or do any thing like that. U can > also reach me at Shesha@asu.edu. > > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Write to disk files ASAP Date: 10 Aug 2001 14:58:42 -0700 From: h0ust0n@my-deja.com (bob h) Organization: http://groups.google.com/ Message-ID: <2dab5758.0108101358.62238d31@posting.google.com> References: <3B6F495B.9800BF9E@erols.com> <488e459a.0108062211.7d7c2587@posting.google.com> <69517bb8.0108070431.7d4da755@posting.google.com> <3b6fed89$0$25324$6e365a64@newsreader02.highway.telekom.at> <3b7196d6$0$9872$6e365a64@newsreader02.highway.telekom.at> <69517bb8.0108090409.417bd8dd@posting.google.com> <3b728707$0$44634$6e365a64@newsreader02.highway.telekom.at> > "Scott Klein" wrote in message > news:69517bb8.0108090409.417bd8dd@posting.google.com... > > Werner, > > > > Thanks, but management has already been sold on the idea that both > > recording & playback can be done without the middle-ware to re-process > > the data. > > > > Scott Scott, in acquire/log application, i was able to maintain in excess of 10 Mbytes/sec to a dosFs2 scsi drive. Writing in 512 byte chunks will likely give you poor performance. I picked 100 Mbyte output file size to make it handier to move files across the network. I set aside 64 Mbytes as uncached memory. The acquire task ran with high priority putting 256Kbytes of data into the uncached in a 'circular' fashion. the logging task ran about once per second, writing these 256 Kbyte buffers to disk. read in the dosFsLib2 text and you'll see that if you do 'large' writes, the system will not spool your data, but will do it's DMA transfer to disk directly from your buffer. good luck bob --------------------------- Newsgroups: comp.os.vxworks Subject: semBCreate() - Semaphore Creation is failing Date: Fri, 10 Aug 2001 17:53:33 -0400 From: Babu Yama Organization: Lucent Technologies Message-ID: <3B7457DD.6AB5FB87@lucent.com> When vxworks OS code runs (for example, in usrRoot, muxLibInit()), the calls to semBCreate(), semMCreate() are not creating the semaphores (NULL is being returned). The code is from est8260/sbc8260 BSP. We have included the following necessary defines in config.h file #define INCLUDE_POSIX_SEM /* POSIX semaphores */ #define INCLUDE_POSIX_MQ /* POSIX msg queues */ #define INCLUDE_SEM_BINARY /* include binary semaphores */ #define INCLUDE_SEM_MUTEX /* include mutex semaphores */ #define INCLUDE_SEM_COUNTING /* include counting semaphores */ At run time, semBLibInit(), semLibInit() are being called in that order. Any comments on why semBCreate() might be failing? What can we do it fix it? Help will be appreciated, Thanks, Babu --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Replacing the TCP/IP stack Date: Fri, 10 Aug 2001 22:54:22 +0100 From: "cecil" Message-ID: <997481373.11501.0.nnrp-10.d4e4b8d1@news.demon.co.uk> References: Are you saying VxWorks is susceptible to a Denial or Service attack? Nigel Chris Moore wrote in message ... >Hi, > >I'm interested in replacing the current TCP/IP stack with a newer >module from Linux. Has anyone done this sort of work? I'm fairly adept >at the END/MUX drivers and so forth and was looking for some estimate >of work involved. Yep, I know that the VxWorks stack (that you have to >pay for source) is based on the FREE open source BSD stack... funny >they charge you for it considering they didn't write it. > >Anyway, as a side note I have found the routing performance of VxWorks >to be rather hopeless. I setup two dummy NPT drivers and injected a >packet into them and basically had them IP spoof so the packet bounced >back and forth (up through the IP layer, routing look up etc) a >certain amount of times. I ended up (on a 150Mhz MIPS) with about 5500 >packets per second. This was with NO ARP and NO reassmble (so no >framented packets) and no memory copies and the CPU was going hard out >(all other tasks were locked out). > >Very sad performance. There are a number of ways to improve this but >we need xs to the source code, and I would rather utilise the Linux >stuff as we can then port other apps such as NAT. > >Any ideas? > >Regards, > >Chris --------------------------- Newsgroups: comp.os.vxworks Subject: Re: semBCreate() - Semaphore Creation is failing Date: Fri, 10 Aug 2001 20:25:52 -0500 From: Andrew Johnson Organization: APS, Argonne National Laboratory Message-ID: <3B7489A0.B8D3A047@aps.anl.gov> References: <3B7457DD.6AB5FB87@lucent.com> Babu Yama wrote: > > When vxworks OS code runs (for example, in usrRoot, muxLibInit()), the calls to > semBCreate(), semMCreate() are not creating the semaphores (NULL is being > returned). ... > Any comments on why semBCreate() might be failing? What can we do it fix it? The reference manual for semBCreate() says that it will return NULL if it can't allocate memory, which is just about all that a semaphore needs. Do you have a system memory pool properly set up? What does memShow() report? - - Andrew - -- The world is such a cheerful place when viewed from upside-down It makes a rise of every fall, a smile of every frown --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Replacing the TCP/IP stack Date: 10 Aug 2001 20:25:35 -0700 From: mfhaigh@myrealbox.com (Mark) Organization: http://groups.google.com/ Message-ID: <64c96d8c.0108101925.340830a3@posting.google.com> References: chrismooremobile@yahoo.com (Chris Moore) wrote in message news:... > Hi, > > I'm interested in replacing the current TCP/IP stack with a newer > module from Linux. Has anyone done this sort of work? I'm fairly adept > at the END/MUX drivers and so forth and was looking for some estimate > of work involved. Yep, I know that the VxWorks stack (that you have to > pay for source) is based on the FREE open source BSD stack... funny > they charge you for it considering they didn't write it. If I remember correctly, a while ago a company called PSO systems ported the NetBSD stack to vxWorks. There's probably some at least cursory information over at the NetBSD site. Perhaps you can dig up somebody that worked on that. (If you do find out, post the answer, I'm curious) The GPL licensed linux kernel code may be a problem for you, unless you're just working on an academic project for yourself. > certain amount of times. I ended up (on a 150Mhz MIPS) with about 5500 > packets per second. This was with NO ARP and NO reassmble (so no > framented packets) and no memory copies and the CPU was going hard out > (all other tasks were locked out). ouch > > Very sad performance. There are a number of ways to improve this but > we need xs to the source code, and I would rather utilise the Linux > stuff as we can then port other apps such as NAT. Personally if I were going to port a stack I'd probably adapt the NetBSD one, as it seems much simpler/more portable than the Linux one (although I haven't done much with either though, so take that with a grain of salt) and it has the NAT and other goodies you're looking for included. Plus you could do whatever you want with the port (ie you'd have to either GPL a Linux-based stack or not distribute it). Mark S. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: How to use the Intel 82559ER Lan Controller in the vxworks system? Date: 11 Aug 2001 02:01:37 -0700 From: srs_zh@263.net (srs_zh) Organization: http://groups.google.com/ Message-ID: References: <3b6fcc9f$0$32924$6e365a64@newsreader02.highway.telekom.at> <3b72661a$0$33474$6e365a64@newsreader02.highway.telekom.at> Thank You! But my system still work wrong!I am sure that my system can find the 82559er. In config.h: #define DEFAULT_BOOT_LINE \ "eeV(0,0)host:c:/temp/vxWorks h=10.0.0.2 e=10.0.0.254 b=255.0.0.0 u=target pw=target" when system am booting,I can see: Can't laod boot file!! Panic Init End Memory error! I am waiting your answer!Thank you again!! --------------------------- Newsgroups: comp.os.vxworks Subject: Re: How to use the Intel 82559ER Lan Controller in the vxworks system? Date: Sat, 11 Aug 2001 11:24:40 +0100 From: molochai Message-ID: References: <3b6fcc9f$0$32924$6e365a64@newsreader02.highway.telekom.at> <3b72661a$0$33474$6e365a64@newsreader02.highway.telekom.at> In message srs_zh@263.net (srs_zh) wrote: > Thank You! > But my system still work wrong!I am sure that my system can find > the 82559er. As I've said before, use the fei557 driver. I can attest from personal experience that it works with the 82559ER provided that you fix up sysfei557end.c to find the device in the bus scan. --------------------------- End of New-News digest ********************** From vxwexplo-errs@csg.lbl.gov Sat Aug 11 06:25:26 2001 From: fscheurer@manz-acs.de Date: Sat Aug 11 06:25:29 PDT 2001 Subject: AUTOREPLY comp.os.vxworks newsdigest 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_000_000176A3.3B754E62 Content-Type: text/plain; charset=iso-8859-1 Content-Transfer-Encoding: 7Bit This is the Mailbox of Friedrich Scheurer. R&D - ACS GmbH - D-72768 Reutlingen Sorry, I am out of office until August 20th 2001! In urgent cases please contact one of the following persons at ACS GmbH: Otto Angerhofer (Managing director) e-mail: oangerhofer@manz-acs.de Eduard Ams (Technical manager) e-mail: eams@manz-acs.de Yvonne Anhorn (Marketing) e-mail: yanhorn@manz-acs.de Margit Arnold (Sales) e-mail: marnold@manz-acs.de Christina Sixel (Sales) e-mail: csixel@manz-acs.de MfG / Kind Regards F. Scheurer ---------------------------------------------------------------- Friedrich Scheurer, Dipl.-Ing. phone +49 7121 9019-23(0) ACS GmbH - R&D fax +49 7121 9019-98 Steigaeckerstrasse 13 web www.manz-acs.de D-72768 Reutlingen (Germany) e-mail: fscheurer@manz-acs.de ---------------------------------------------------------------- ------_=_NextPart_000_000176A3.3B754E62 Content-Type: text/html; charset=iso-8859-1 Content-Transfer-Encoding: quoted-printable
This is the Mailbox of Friedrich Scheurer.
R&D - ACS GmbH - D-72768 Reutlingen
 
Sorry, I am out of office until August 20th 2001!
 
In urgent cases please contact one of the
following persons at ACS GmbH:
 
Otto Angerhofer (Managing director)
Eduard Ams (Technical manager)
Yvonne Anhorn (Marketing)
Margit Arnold (Sales)
e-mail: marnold@manz-acs.de
Christina Sixel (Sales)

 
MfG / Kind Regards
 
F. Scheurer
 
----------------------------------------------------------------
Friedrich Scheurer, Dipl.-Ing.    phone    +49 7121 9019-23(0)
ACS GmbH  -  R&D                  fax      +49 7121 9019-98
Steigaeckerstrasse 13             web      www.manz-acs.de
D-72768 Reutlingen (Germany)      e-mail:  fscheurer@manz-acs.de
----------------------------------------------------------------
------_=_NextPart_000_000176A3.3B754E62-- From vxwexplo-errs@csg.lbl.gov Sun Aug 12 04:03:31 2001 From: Vxworks Exploder Date: Sun Aug 12 04:03:33 PDT 2001 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Sun Aug 12 04:03:03 PDT 2001 Subject: Object File download error (Tornado) ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Object File download error (Tornado) Date: Sun, 12 Aug 2001 01:21:21 GMT From: "Ashok Ramani" Organization: Excite@Home - The Leader in Broadband http://home.com/faster Message-ID: Hi, I created a project in Tornado with a few "C" files. The files compiled with no errors and I was able to build the project and generate the object files. But, when I try to download the object file to the target VxWorks machine I get an wtx object file improper error. I was able to successfully download all projects that my project depends on before trying to download my project. Note : - I am not using any main() function in my code Thanks in advance, Krupa --------------------------- End of New-News digest ********************** From vxwexplo-errs@csg.lbl.gov Sun Aug 12 05:48:04 2001 From: Rainer Stelzer Date: Sun Aug 12 05:48:07 PDT 2001 Subject: Re: protocol stack design under vxworks Hi Jim, > Also, how is timing normally accomplished within a protocol stack > under VxWorks? In our old platform (without a OS) timing was > accomplished by a main loop that would service messages and events at > specific intervals to send out heartbeats, determine timeouts, etc. > I'm not clear what is the proper way of doing this under VxWorks. > Should we create timers and put all of our time-dependent stuff in the > SIGALRM handler? I assume You are talking of basic/primary rate isdn running one D-channel and 2/32 B channel connections. For the d-channel protocol keep the task per layer design. I 'ld suggest You to use message queue interfaces between the layer task's. The FSM of layer n uses the input queue as the only stimulus input. To create timers/timeouts use vxworks watchdog timer to trigger messages that are send to this queue. layer n-1 and n+1 send their messages into this queue as well. May sound a little strange, but this solution is in my opinion formal correct, and we made best expierence designing it this way. You'll see, that channel-interaction (e.g. dailup for PPP connections) is very straightforward. The realtime requirements of the D-channel (even if X.21 is included) can be fulfilled easily. However, this model may not fit to every kind of B-channel protocol (case they are handled by the CPU and not routet via IOM2-Bus for e.g. audio applications). > And, yes, I know we could purchase a ISDN stack already designed for > VxWorks but we would rather avoid that expense since we already have > source for the stack from our older platform. If you've someone in Your design team that knows the protocol specifcation and the old stack in detail, I think you can realize the cost advantage. Case not, you should think about purchasing a ISDN stack. We ported a generic,"os-less" ISDN stack to vxworks. The resume: Buying a vxworks stack would have been much more cost efficient ! And a further hint: case you want to implement network functionality over the B channel contact your wrs sales representive for a PPP toolkit. > We would appreciate any > tips on how we should port this protocol to vxWorks! HTH Rainer vxworks tornado ===================================================== Gebrüder Stoye GmbH department: traffic control systems Name: Rainer Stelzer Adr: Longericher Strasse 177 city: Cologne, Germany Zip: 50739 phone: ++49/221/2616-551 mailto: r.stelzer@stoye.de www: http://www.stoye.de ===================================================== From vxwexplo-errs@csg.lbl.gov Sun Aug 12 07:12:34 2001 From: fscheurer@manz-acs.de Date: Sun Aug 12 07:12:36 PDT 2001 Subject: AUTOREPLY comp.os.vxworks newsdigest 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_000_000176B5.3B76AAED Content-Type: text/plain; charset=iso-8859-1 Content-Transfer-Encoding: 7Bit This is the Mailbox of Friedrich Scheurer. R&D - ACS GmbH - D-72768 Reutlingen Sorry, I am out of office until August 20th 2001! In urgent cases please contact one of the following persons at ACS GmbH: Otto Angerhofer (Managing director) e-mail: oangerhofer@manz-acs.de Eduard Ams (Technical manager) e-mail: eams@manz-acs.de Yvonne Anhorn (Marketing) e-mail: yanhorn@manz-acs.de Margit Arnold (Sales) e-mail: marnold@manz-acs.de Christina Sixel (Sales) e-mail: csixel@manz-acs.de MfG / Kind Regards F. Scheurer ---------------------------------------------------------------- Friedrich Scheurer, Dipl.-Ing. phone +49 7121 9019-23(0) ACS GmbH - R&D fax +49 7121 9019-98 Steigaeckerstrasse 13 web www.manz-acs.de D-72768 Reutlingen (Germany) e-mail: fscheurer@manz-acs.de ---------------------------------------------------------------- ------_=_NextPart_000_000176B5.3B76AAED Content-Type: text/html; charset=iso-8859-1 Content-Transfer-Encoding: quoted-printable
This is the Mailbox of Friedrich Scheurer.
R&D - ACS GmbH - D-72768 Reutlingen
 
Sorry, I am out of office until August 20th 2001!
 
In urgent cases please contact one of the
following persons at ACS GmbH:
 
Otto Angerhofer (Managing director)
Eduard Ams (Technical manager)
Yvonne Anhorn (Marketing)
Margit Arnold (Sales)
e-mail: marnold@manz-acs.de
Christina Sixel (Sales)

 
MfG / Kind Regards
 
F. Scheurer
 
----------------------------------------------------------------
Friedrich Scheurer, Dipl.-Ing.    phone    +49 7121 9019-23(0)
ACS GmbH  -  R&D                  fax      +49 7121 9019-98
Steigaeckerstrasse 13             web      www.manz-acs.de
D-72768 Reutlingen (Germany)      e-mail:  fscheurer@manz-acs.de
----------------------------------------------------------------
------_=_NextPart_000_000176B5.3B76AAED-- From vxwexplo-errs@csg.lbl.gov Sun Aug 12 07:35:45 2001 From: fscheurer@manz-acs.de Date: Sun Aug 12 07:35:47 PDT 2001 Subject: AUTOREPLY Re: protocol stack design under vxworks This message is in MIME format. Since your mail reader does not understand this format, some or all of this message may not be legible. ------_=_NextPart_000_000176B7.3B76B05A Content-Type: text/plain; charset=iso-8859-1 Content-Transfer-Encoding: 7Bit This is the Mailbox of Friedrich Scheurer. R&D - ACS GmbH - D-72768 Reutlingen Sorry, I am out of office until August 20th 2001! In urgent cases please contact one of the following persons at ACS GmbH: Otto Angerhofer (Managing director) e-mail: oangerhofer@manz-acs.de Eduard Ams (Technical manager) e-mail: eams@manz-acs.de Yvonne Anhorn (Marketing) e-mail: yanhorn@manz-acs.de Margit Arnold (Sales) e-mail: marnold@manz-acs.de Christina Sixel (Sales) e-mail: csixel@manz-acs.de MfG / Kind Regards F. Scheurer ---------------------------------------------------------------- Friedrich Scheurer, Dipl.-Ing. phone +49 7121 9019-23(0) ACS GmbH - R&D fax +49 7121 9019-98 Steigaeckerstrasse 13 web www.manz-acs.de D-72768 Reutlingen (Germany) e-mail: fscheurer@manz-acs.de ---------------------------------------------------------------- ------_=_NextPart_000_000176B7.3B76B05A Content-Type: text/html; charset=iso-8859-1 Content-Transfer-Encoding: quoted-printable
This is the Mailbox of Friedrich Scheurer.
R&D - ACS GmbH - D-72768 Reutlingen
 
Sorry, I am out of office until August 20th 2001!
 
In urgent cases please contact one of the
following persons at ACS GmbH:
 
Otto Angerhofer (Managing director)
Eduard Ams (Technical manager)
Yvonne Anhorn (Marketing)
Margit Arnold (Sales)
e-mail: marnold@manz-acs.de
Christina Sixel (Sales)

 
MfG / Kind Regards
 
F. Scheurer
 
----------------------------------------------------------------
Friedrich Scheurer, Dipl.-Ing.    phone    +49 7121 9019-23(0)
ACS GmbH  -  R&D                  fax      +49 7121 9019-98
Steigaeckerstrasse 13             web      www.manz-acs.de
D-72768 Reutlingen (Germany)      e-mail:  fscheurer@manz-acs.de
----------------------------------------------------------------
------_=_NextPart_000_000176B7.3B76B05A-- From vxwexplo-errs@csg.lbl.gov Sun Aug 12 08:39:18 2001 From: =?iso-8859-1?q?ram=20mohan?= Date: Sun Aug 12 08:39:20 PDT 2001 Subject: Re: reg Makefile hi suresh, u need to provide the exact path of all include files. also if u use any other librries u need to supply the path of that librries also in second stage ie linking of object files. with u this might be the prob because build is failing after compiling try this i hope u'll get thru the prob :) Mohan --- the vxWorks Users Group Exploder wrote: > Submitted-by vxwexplo-errs@csg.lbl.gov Tue Aug 7 > 06:28:48 2001 > Submitted-by: "Suresh T" > > > Hi All, > > We want to run the make file generated from the > Tornado Workspace > (downloadable project) from the Command prompt. > We have made sure the Environment Variables are set > properly. > While running make command we get the following > error: > make: *** No rule to make target `linkedObjs.o', > needed by `exe'. Stop > > Please provide inputs on this. > > Advanced Thanks, > Regards, > Suresh > > > > ********** > > This is a user group mailing list for vxWorks > related topics > see http://www-vxw.lbl.gov/vxworks for posting > and subscription info From vxwexplo-errs@csg.lbl.gov Sun Aug 12 08:52:17 2001 From: =?iso-8859-1?q?ram=20mohan?= Date: Sun Aug 12 08:52:19 PDT 2001 Subject: Re: Compile windoze T2 makefiles in cygwin bash shell? hey, this is because ccppc internally uses cpp.exe to compile and cpp is architecture(also os) dependent also it uses one more os dependend application as.exe as far as i know there are different combinations of options for each but i never tried changing the options for ex using ppc cpp.exe with intel cpp.exe but appropriate option u can try that finding out what options u need try this out if it does not work there is no other go except u need to buy tornado for cygwin :( :) Mohan --- the vxWorks Users Group Exploder wrote: > Submitted-by vxwexplo-errs@csg.lbl.gov Thu Aug 9 > 14:38:49 2001 > Submitted-by: "Michael Morrison" > > > Howdy, > > I have a windoze install of tornado II, but want to > use the cygwin bash > shell to run command-line make. > Unfotunatly, when I do this I get the following > error: > > $ make TARGET_BASE=iscsi > ccppc > -B/cygdrive/c/Tornado/host/x86-win32/lib/gcc-lib/ -g > -nostdinc -f > > no-builtin -nostdlib -fno-defer-pop -W -Wall > -I/cygdrive/c/Tornado/target/h -I/c > ygdrive/c/Tornado/target/config/hsiSBC8260 -I. > -I../include -c iscsi.cc > cpp: -lang-c++: linker input file unused since > linking not done > as: unrecognized option `-u' > ccppc.exe: Internal compiler error: program as got > fatal signal 1 > make: *** [iscsi.o] Error 1 > > > The same makefile runs ok in the windoze DOS > shell.... > Anyone have any ideas on how to do this? > > Do you think the cygwin 'cpp' and 'as' be getting > run instead of the > vxworks version? > > TIA > > Mikem > > > > > ********** > > This is a user group mailing list for vxWorks > related topics > see http://www-vxw.lbl.gov/vxworks for posting > and subscription info From vxwexplo-errs@csg.lbl.gov Sun Aug 12 09:46:00 2001 From: Ilia Bosis Date: Sun Aug 12 09:46:02 PDT 2001 Subject: RE: IP over Atm in Vxworks The answer is, Yes. VxWorks 5.3 with SENS, coupled even with a BSD3.4 driver, can handle a number of IP addresses per interface. We had done it a while ago. I believe that Tornado 2 with END driver can do the same (sans any possible bugs, of course). In our system there was a small number (2..4) of IP addresses per interface. I doesn't know how it will scale up to, say, hundreds of IP addresses per interface, if Your application requires this. Hope, this helps. Ilia. > Newsgroups: comp.os.vxworks > Subject: IP over Atm in Vxworks > Date: Thu, 9 Aug 2001 10:31:38 +0800 > From: "Bruce" > Organization: The news server of Changzhou TeleCom > Message-ID: <9kstju$2jia$1@news.cz.js.cn> > > Hi,all: > I am doing a END driver to implement IP over ATM in VxWorks 5.4. I will > use PVC(permanent virtual circuit). According to RFC1577, PVC(VPI,VCI) is > correspondind to IP address, so if I have several PVC, we need IP address of > the same number. but we only have an END driver, does END can be binded to > several IP address? Thank you for any answer! > > best regards > bruce -------------------------------------------- Ilia Bosis Software Engineer PacketLight Networks E-mail: Ilia_Bosis@packetlight.com Web: www.packetlight.com From vxwexplo-errs@csg.lbl.gov Sun Aug 12 11:13:15 2001 From: fscheurer@manz-acs.de Date: Sun Aug 12 11:13:17 PDT 2001 Subject: AUTOREPLY RE: IP over Atm in Vxworks This message is in MIME format. Since your mail reader does not understand this format, some or all of this message may not be legible. ------_=_NextPart_000_000176BF.3B76E357 Content-Type: text/plain; charset=iso-8859-1 Content-Transfer-Encoding: 7Bit This is the Mailbox of Friedrich Scheurer. R&D - ACS GmbH - D-72768 Reutlingen Sorry, I am out of office until August 20th 2001! In urgent cases please contact one of the following persons at ACS GmbH: Otto Angerhofer (Managing director) e-mail: oangerhofer@manz-acs.de Eduard Ams (Technical manager) e-mail: eams@manz-acs.de Yvonne Anhorn (Marketing) e-mail: yanhorn@manz-acs.de Margit Arnold (Sales) e-mail: marnold@manz-acs.de Christina Sixel (Sales) e-mail: csixel@manz-acs.de MfG / Kind Regards F. Scheurer ---------------------------------------------------------------- Friedrich Scheurer, Dipl.-Ing. phone +49 7121 9019-23(0) ACS GmbH - R&D fax +49 7121 9019-98 Steigaeckerstrasse 13 web www.manz-acs.de D-72768 Reutlingen (Germany) e-mail: fscheurer@manz-acs.de ---------------------------------------------------------------- ------_=_NextPart_000_000176BF.3B76E357 Content-Type: text/html; charset=iso-8859-1 Content-Transfer-Encoding: quoted-printable
This is the Mailbox of Friedrich Scheurer.
R&D - ACS GmbH - D-72768 Reutlingen
 
Sorry, I am out of office until August 20th 2001!
 
In urgent cases please contact one of the
following persons at ACS GmbH:
 
Otto Angerhofer (Managing director)
Eduard Ams (Technical manager)
Yvonne Anhorn (Marketing)
Margit Arnold (Sales)
e-mail: marnold@manz-acs.de
Christina Sixel (Sales)

 
MfG / Kind Regards
 
F. Scheurer
 
----------------------------------------------------------------
Friedrich Scheurer, Dipl.-Ing.    phone    +49 7121 9019-23(0)
ACS GmbH  -  R&D                  fax      +49 7121 9019-98
Steigaeckerstrasse 13             web      www.manz-acs.de
D-72768 Reutlingen (Germany)      e-mail:  fscheurer@manz-acs.de
----------------------------------------------------------------
------_=_NextPart_000_000176BF.3B76E357-- From vxwexplo-errs@csg.lbl.gov Sun Aug 12 21:34:37 2001 From: "Krishna Kishore Reddy Sureddy" Date: Sun Aug 12 21:34:40 PDT 2001 Subject: taskDelete() in vxWorks 5.4 This is a multi-part message in MIME format. --------------InterScan_NT_MIME_Boundary Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Hi vxWorks Gurus, In vxWorks 5.4 is there any way a task gets deleted other than calling exit() or normal exit when its thread of execution is over. Thanks and Regards, S.K.K.Reddy. --------------InterScan_NT_MIME_Boundary Content-Type: text/plain; name="Wipro_Disclaimer.txt" Content-Transfer-Encoding: 7bit Content-Disposition: attachment; filename="Wipro_Disclaimer.txt" ---------------------------------------------------------------------------------------------------------------------- Information transmitted by this E-MAIL is proprietary to Wipro Limited and is intended for use only by the individual or entity to which it is addressed, and may contain information that is privileged, confidential or exempt from disclosure under applicable law. If you are not the intended recipient or it appears that this mail has been forwarded to you without proper authority, you are notified that any use or dissemination of this information in any manner is strictly prohibited. In such cases, please notify us immediately at mailto:mailadmin@wipro.com and delete this mail from your records. ---------------------------------------------------------------------------------------------------------------------- --------------InterScan_NT_MIME_Boundary-- From vxwexplo-errs@csg.lbl.gov Sun Aug 12 21:42:19 2001 From: "Krishna Kishore Reddy Sureddy" Date: Sun Aug 12 21:42:21 PDT 2001 Subject: semaphore and message queue id in vxWorks This is a multi-part message in MIME format. --------------InterScan_NT_MIME_Boundary Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Hi vxWorks Gurus, Is there any way to find the List of Semaphore Id's and List of Message Queue Id's in a system running on vxWorks 5.4?. Is there any way to know which task has taken a particular semaphore?. Thanks and Regards, S.K.K.Reddy. --------------InterScan_NT_MIME_Boundary Content-Type: text/plain; name="Wipro_Disclaimer.txt" Content-Transfer-Encoding: 7bit Content-Disposition: attachment; filename="Wipro_Disclaimer.txt" ---------------------------------------------------------------------------------------------------------------------- Information transmitted by this E-MAIL is proprietary to Wipro Limited and is intended for use only by the individual or entity to which it is addressed, and may contain information that is privileged, confidential or exempt from disclosure under applicable law. If you are not the intended recipient or it appears that this mail has been forwarded to you without proper authority, you are notified that any use or dissemination of this information in any manner is strictly prohibited. In such cases, please notify us immediately at mailto:mailadmin@wipro.com and delete this mail from your records. ---------------------------------------------------------------------------------------------------------------------- --------------InterScan_NT_MIME_Boundary-- From vxwexplo-errs@csg.lbl.gov Sun Aug 12 23:44:45 2001 From: fscheurer@manz-acs.de Date: Sun Aug 12 23:44:47 PDT 2001 Subject: AUTOREPLY semaphore and message queue id in vxWorks This message is in MIME format. Since your mail reader does not understand this format, some or all of this message may not be legible. ------_=_NextPart_000_0001770B.3B779374 Content-Type: text/plain; charset=iso-8859-1 Content-Transfer-Encoding: 7Bit This is the Mailbox of Friedrich Scheurer. R&D - ACS GmbH - D-72768 Reutlingen Sorry, I am out of office until August 20th 2001! In urgent cases please contact one of the following persons at ACS GmbH: Otto Angerhofer (Managing director) e-mail: oangerhofer@manz-acs.de Eduard Ams (Technical manager) e-mail: eams@manz-acs.de Yvonne Anhorn (Marketing) e-mail: yanhorn@manz-acs.de Margit Arnold (Sales) e-mail: marnold@manz-acs.de Christina Sixel (Sales) e-mail: csixel@manz-acs.de MfG / Kind Regards F. Scheurer ---------------------------------------------------------------- Friedrich Scheurer, Dipl.-Ing. phone +49 7121 9019-23(0) ACS GmbH - R&D fax +49 7121 9019-98 Steigaeckerstrasse 13 web www.manz-acs.de D-72768 Reutlingen (Germany) e-mail: fscheurer@manz-acs.de ---------------------------------------------------------------- ------_=_NextPart_000_0001770B.3B779374 Content-Type: text/html; charset=iso-8859-1 Content-Transfer-Encoding: quoted-printable
This is the Mailbox of Friedrich Scheurer.
R&D - ACS GmbH - D-72768 Reutlingen
 
Sorry, I am out of office until August 20th 2001!
 
In urgent cases please contact one of the
following persons at ACS GmbH:
 
Otto Angerhofer (Managing director)
Eduard Ams (Technical manager)
Yvonne Anhorn (Marketing)
Margit Arnold (Sales)
e-mail: marnold@manz-acs.de
Christina Sixel (Sales)

 
MfG / Kind Regards
 
F. Scheurer
 
----------------------------------------------------------------
Friedrich Scheurer, Dipl.-Ing.    phone    +49 7121 9019-23(0)
ACS GmbH  -  R&D                  fax      +49 7121 9019-98
Steigaeckerstrasse 13             web      www.manz-acs.de
D-72768 Reutlingen (Germany)      e-mail:  fscheurer@manz-acs.de
----------------------------------------------------------------
------_=_NextPart_000_0001770B.3B779374-- From vxwexplo-errs@csg.lbl.gov Sun Aug 12 23:53:37 2001 From: fscheurer@manz-acs.de Date: Sun Aug 12 23:53:39 PDT 2001 Subject: AUTOREPLY taskDelete() in vxWorks 5.4 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_000_0001772B.3B77958E Content-Type: text/plain; charset=iso-8859-1 Content-Transfer-Encoding: 7Bit This is the Mailbox of Friedrich Scheurer. R&D - ACS GmbH - D-72768 Reutlingen Sorry, I am out of office until August 20th 2001! In urgent cases please contact one of the following persons at ACS GmbH: Otto Angerhofer (Managing director) e-mail: oangerhofer@manz-acs.de Eduard Ams (Technical manager) e-mail: eams@manz-acs.de Yvonne Anhorn (Marketing) e-mail: yanhorn@manz-acs.de Margit Arnold (Sales) e-mail: marnold@manz-acs.de Christina Sixel (Sales) e-mail: csixel@manz-acs.de MfG / Kind Regards F. Scheurer ---------------------------------------------------------------- Friedrich Scheurer, Dipl.-Ing. phone +49 7121 9019-23(0) ACS GmbH - R&D fax +49 7121 9019-98 Steigaeckerstrasse 13 web www.manz-acs.de D-72768 Reutlingen (Germany) e-mail: fscheurer@manz-acs.de ---------------------------------------------------------------- ------_=_NextPart_000_0001772B.3B77958E Content-Type: text/html; charset=iso-8859-1 Content-Transfer-Encoding: quoted-printable
This is the Mailbox of Friedrich Scheurer.
R&D - ACS GmbH - D-72768 Reutlingen
 
Sorry, I am out of office until August 20th 2001!
 
In urgent cases please contact one of the
following persons at ACS GmbH:
 
Otto Angerhofer (Managing director)
Eduard Ams (Technical manager)
Yvonne Anhorn (Marketing)
Margit Arnold (Sales)
e-mail: marnold@manz-acs.de
Christina Sixel (Sales)

 
MfG / Kind Regards
 
F. Scheurer
 
----------------------------------------------------------------
Friedrich Scheurer, Dipl.-Ing.    phone    +49 7121 9019-23(0)
ACS GmbH  -  R&D                  fax      +49 7121 9019-98
Steigaeckerstrasse 13             web      www.manz-acs.de
D-72768 Reutlingen (Germany)      e-mail:  fscheurer@manz-acs.de
----------------------------------------------------------------
------_=_NextPart_000_0001772B.3B77958E-- From vxwexplo-errs@csg.lbl.gov Mon Aug 13 04:03:30 2001 From: Vxworks Exploder Date: Mon Aug 13 04:03:32 PDT 2001 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Mon Aug 13 04:03:09 PDT 2001 Subject: Re: tiny library for memory mapped graphic interface (MPC823) Subject: How can Vxworks call the main function of an Application. Subject: Using Shared-memory networking with BSP PcPentium Subject: Re: How to use the Intel 82559ER Lan Controller in the vxworks system? Subject: Re: How can Vxworks call the main function of an Application. Subject: Re: Replacing the TCP/IP stack Subject: Re: Using Shared-memory networking with BSP PcPentium Subject: Re: Problem in scaning user input ! Subject: Re: Using Shared-memory networking with BSP PcPentium Subject: rs232 serial port driver ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: tiny library for memory mapped graphic interface (MPC823) Date: Sun, 12 Aug 2001 17:38:26 GMT From: "jdough" Organization: Excite@Home - The Leader in Broadband http://home.com/faster Message-ID: References: <3B679E61.23F5B7A5@stoye.de> Reply-To: jdough@legomyego.com You could also use OpTIC. This graphics package was designed for settop boxes that required a small foot print. It is memory mapped so that you can use almost any device with it. Talk to your local FAE and see if you can get a copy of it. --------------------------- Newsgroups: comp.os.vxworks Subject: How can Vxworks call the main function of an Application. Date: Mon, 13 Aug 2001 01:44:40 GMT From: "¹Ù¶÷°úÇÔ²²ÇÏ´Ù" Organization: Dacom Message-ID: Could you please explain to me? I'm wondering it in day and night. --------------------------- Newsgroups: comp.os.vxworks Subject: Using Shared-memory networking with BSP PcPentium Date: Mon, 13 Aug 2001 09:56:12 +0800 From: "survey liu" Organization: Bentium Ltd. (CN99) Message-ID: <9l7c12$bnj$1@mail.cn99.com> Hi,all I am now trying to build a bootrom which loads the vxworks kernel via shared memory networking. I have done next steps: 1. I defined the INCLUDE_SM_NET and INCLUDE_SM_SEQ_ADDR and then compiled it. Tornado said I have to define something else and failed. 2. I so defined the things like #define SM_ANCHOR_ADRS ((char *) (LOCAL_MEM_LOCAL_ADRS+SM_ANCHOR_OFFSET)) #define SM_MEM_ADRS SM_ANCHOR_ADRS #define SM_MEM_SIZE 0x80000 #define SM_INT_TYPE SM_INT_MAILBOX_1 #define SM_INT_ARG1 VME_AM_SUP_SHORT_IO #define SM_INT_ARG2 (0xc000|(sysProcNum *2)) #define SM_INT_ARG3 0 Then I compiled again. Also it failed with message " sysIntEnble not found in sysLib " I searched the BSP files, there is not any function named sysIntEnable(), but a sysIntEnablePIC() instead. So, can I simply copy the sysIntEnablePIC() as the sysIntEnable() that compiler needs? I have tried, it works while compiling. But I dont know whether it could boot my board or not. Any help or hint may be appreciated. - -- Best regards, Survey Liu Email: liuxw@263.net --------------------------- Newsgroups: comp.os.vxworks Subject: Re: How to use the Intel 82559ER Lan Controller in the vxworks system? Date: 12 Aug 2001 19:21:25 -0700 From: srs_zh@263.net (srs_zh) Organization: http://groups.google.com/ Message-ID: References: <3b6fcc9f$0$32924$6e365a64@newsreader02.highway.telekom.at> <3b72661a$0$33474$6e365a64@newsreader02.highway.telekom.at> Where can I get the soucr codes of the intel82559er from? --------------------------- Newsgroups: comp.os.vxworks Subject: Re: How can Vxworks call the main function of an Application. Date: Sun, 12 Aug 2001 20:45:59 -0700 From: "Michael R. Kesti" Organization: MK Associates Message-ID: <3B774D77.2E6084DD@gv.net> References: Reply-To: mkesti@gv.net "¹Ù¶÷°úÇÔ²²ÇÏ´Ù" wrote: >Could you please explain to me? >I'm wondering it in day and night. If the vxworks image was built with USER_APPL_INIT defined, the function void usrAppInit(void) is called after the OS is completely initialized. usrAppInit() may be your "main" function or it may call your "main" function, but you may not use the name "main". I often use "appMain". - -- ======================================================================== Michael Kesti | "And like, one and one don't make | two, one and one make one." mkesti@gv.net | - The Who, Bargain --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Replacing the TCP/IP stack Date: 12 Aug 2001 21:46:02 -0700 From: nthawk@chollian.net (YongSung Lee) Organization: http://groups.google.com/ Message-ID: <9c7571e8.0108122046.31b064b5@posting.google.com> References: <64c96d8c.0108101925.340830a3@posting.google.com> You can download PSO stack below link. http://www.xs4all.nl/~borkhuis/vxworks/ftp/vxhacks/pso_stack/ mfhaigh@myrealbox.com (Mark) wrote in message news:<64c96d8c.0108101925.340830a3@posting.google.com>... > chrismooremobile@yahoo.com (Chris Moore) wrote in message news:... > > Hi, > > > > I'm interested in replacing the current TCP/IP stack with a newer > > module from Linux. Has anyone done this sort of work? I'm fairly adept > > at the END/MUX drivers and so forth and was looking for some estimate > > of work involved. Yep, I know that the VxWorks stack (that you have to > > pay for source) is based on the FREE open source BSD stack... funny > > they charge you for it considering they didn't write it. > > If I remember correctly, a while ago a company called PSO systems > ported the NetBSD stack to vxWorks. There's probably some at least > cursory information over at the NetBSD site. Perhaps you can dig up > somebody that worked on that. (If you do find out, post the answer, > I'm curious) --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Using Shared-memory networking with BSP PcPentium Date: Sun, 12 Aug 2001 21:59:21 -0700 From: DrDiags Organization: laserlink.net Message-ID: <3B775EA9.FB0DFC08@covad.net> References: <9l7c12$bnj$1@mail.cn99.com> Reply-To: drdiags@covad.net Survey, I don't think you want to use the sysIntEnablePIC(). Looks like the function sysIntEnable() exists in the $(WIND_BASE)/target/src/drv/vme directory in several VME support drivers and is based on the board's VME interface. The sysIntEnablePIC() routine I found looked like it was setting up an Intel 8259A Programmable Interrupt Controller. The sysIntEnable() routine for your board should exist in the vme directory I spoke of. HTH. survey liu wrote: > > Hi,all > > I am now trying to build a bootrom which loads the vxworks kernel via shared > memory networking. I have done next steps: > > 1. I defined the INCLUDE_SM_NET and INCLUDE_SM_SEQ_ADDR and then compiled > it. Tornado said I have to define something else and failed. > > 2. I so defined the things like > #define SM_ANCHOR_ADRS ((char *) > (LOCAL_MEM_LOCAL_ADRS+SM_ANCHOR_OFFSET)) > #define SM_MEM_ADRS SM_ANCHOR_ADRS > #define SM_MEM_SIZE 0x80000 > #define SM_INT_TYPE SM_INT_MAILBOX_1 > #define SM_INT_ARG1 VME_AM_SUP_SHORT_IO > #define SM_INT_ARG2 (0xc000|(sysProcNum *2)) > #define SM_INT_ARG3 0 > Then I compiled again. Also it failed with message " sysIntEnble not > found in sysLib " > I searched the BSP files, there is not any function named > sysIntEnable(), but a sysIntEnablePIC() instead. > > So, can I simply copy the sysIntEnablePIC() as the sysIntEnable() that > compiler needs? I have tried, it works while compiling. But I dont know > whether it could boot my board or not. > > Any help or hint may be appreciated. > > -- > > Best regards, > Survey Liu > Email: liuxw@263.net - -- http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html (vxWorks FAQ home page) --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Problem in scaning user input ! Date: 12 Aug 2001 22:02:06 -0700 From: mridulgupta1@yahoo.com (Mridul Gupta) Organization: http://groups.google.com/ Message-ID: References: <9k5q3t$8g$1@news2.kornet.net> hi all, finally i found a way to scan user input without the interference of shell :) ... the only way that worked for me is to delete shell task just before reading input and respawn shell after reading ... it works ... /*******************************************************/ shelltid = taskNameToId("tShell"); taskDelete(shelltid); for( ; ; ) { // read } shellInit(0,TRUE); /*******************************************************/ Mridul --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Using Shared-memory networking with BSP PcPentium Date: Mon, 13 Aug 2001 16:02:48 +0800 From: "survey liu" Organization: Bentium Ltd. (CN99) Message-ID: <9l81gd$p6r$1@mail.cn99.com> > > I don't think you want to use the sysIntEnablePIC(). Looks like > the function sysIntEnable() exists in the > $(WIND_BASE)/target/src/drv/vme directory in several VME support drivers > and is based on the board's VME interface. The sysIntEnablePIC() routine > I found looked like it was setting up an Intel 8259A Programmable > Interrupt Controller. The sysIntEnable() routine for your board should > exist in the > vme directory I spoke of. HTH. > Thanks for you warm-heart guy. So, you mean that I will need a VME driver anyway? Just add the VME driver which can drive my VME to work into my MAKEFILE? - -- Best regards, Survey Liu 4/F Wearnes Technology Bldg. No.10 Kefa Rd.,Science & Industry Park Nanshan Shenzhen 518057,PRC Tel: +86-755-6635333-4300 Fax: +86-755-6635550 Email: survey.liu@utstar.com Web: http://www.utstar.com --------------------------- Newsgroups: comp.os.vxworks Subject: rs232 serial port driver Date: Mon, 13 Aug 2001 12:17:55 +0200 From: "Juno Kampstra" Organization: Ticsali b.v. Message-ID: <9l897p$ei2$1@nereid.worldonline.nl> Hello, Trying to make a driver for connecting to the rs232 port. Already got the ms-windows implmentation running, but now I'm searching for some vxwork (example) code. The mswin driver opens a channel, and receives data from the channel by using a seperate receive task, which is create on open a channel, and removed on closing the channel. The same idea I want to use using vxwork: use ISR to put data in a buffer, and let the receive task know new data is present, and free the lock (which blocks the receive task). Juno Kampstra. (juno.kampstra@fcs-cs.com) FCS Control Systems --------------------------- End of New-News digest ********************** From vxwexplo-errs@csg.lbl.gov Mon Aug 13 06:18:57 2001 From: "AMASS" Date: Mon Aug 13 06:18:59 PDT 2001 Subject: File sizes This is a multi-part message in MIME format. ------=_NextPart_000_000B_01C12402.1CE3A4E0 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable Hi All host : NT, target : PPC. What is the maximum file size when using the raw filesystem? Also, does = anyone know thet maximum file size permitted under NT? Cheers Andy vxworks ------=_NextPart_000_000B_01C12402.1CE3A4E0 Content-Type: text/html; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable
Hi All
 
host : NT, target = : PPC.
 
What is the maximum file size when = using the raw=20 filesystem? Also, does anyone know thet maximum file size permitted = under=20 NT?
 
Cheers
 
Andy
 
vxworks
------=_NextPart_000_000B_01C12402.1CE3A4E0-- From vxwexplo-errs@csg.lbl.gov Mon Aug 13 06:36:34 2001 From: fscheurer@manz-acs.de Date: Mon Aug 13 06:36:36 PDT 2001 Subject: AUTOREPLY comp.os.vxworks newsdigest 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_000_000177F5.3B77F3FD Content-Type: text/plain; charset=iso-8859-1 Content-Transfer-Encoding: 7Bit This is the Mailbox of Friedrich Scheurer. R&D - ACS GmbH - D-72768 Reutlingen Sorry, I am out of office until August 20th 2001! In urgent cases please contact one of the following persons at ACS GmbH: Otto Angerhofer (Managing director) e-mail: oangerhofer@manz-acs.de Eduard Ams (Technical manager) e-mail: eams@manz-acs.de Yvonne Anhorn (Marketing) e-mail: yanhorn@manz-acs.de Margit Arnold (Sales) e-mail: marnold@manz-acs.de Christina Sixel (Sales) e-mail: csixel@manz-acs.de MfG / Kind Regards F. Scheurer ---------------------------------------------------------------- Friedrich Scheurer, Dipl.-Ing. phone +49 7121 9019-23(0) ACS GmbH - R&D fax +49 7121 9019-98 Steigaeckerstrasse 13 web www.manz-acs.de D-72768 Reutlingen (Germany) e-mail: fscheurer@manz-acs.de ---------------------------------------------------------------- ------_=_NextPart_000_000177F5.3B77F3FD Content-Type: text/html; charset=iso-8859-1 Content-Transfer-Encoding: quoted-printable
This is the Mailbox of Friedrich Scheurer.
R&D - ACS GmbH - D-72768 Reutlingen
 
Sorry, I am out of office until August 20th 2001!
 
In urgent cases please contact one of the
following persons at ACS GmbH:
 
Otto Angerhofer (Managing director)
Eduard Ams (Technical manager)
Yvonne Anhorn (Marketing)
Margit Arnold (Sales)
e-mail: marnold@manz-acs.de
Christina Sixel (Sales)

 
MfG / Kind Regards
 
F. Scheurer
 
----------------------------------------------------------------
Friedrich Scheurer, Dipl.-Ing.    phone    +49 7121 9019-23(0)
ACS GmbH  -  R&D                  fax      +49 7121 9019-98
Steigaeckerstrasse 13             web      www.manz-acs.de
D-72768 Reutlingen (Germany)      e-mail:  fscheurer@manz-acs.de
----------------------------------------------------------------
------_=_NextPart_000_000177F5.3B77F3FD-- From vxwexplo-errs@csg.lbl.gov Mon Aug 13 06:55:37 2001 From: Champak Nath Date: Mon Aug 13 06:55:39 PDT 2001 Subject: Booting VxWorks on Intel PIII board through network. hello techies, my work environment is as follows: I have: 1.Target Standard Intel PIII board, 815 chipset with on board NIC. 2.Host Sun Solaris 2.5.1 3.DHCP server on the same host. 4.TFTP server on the same host. 5.VxWorks Pentium/PentiumPro BSP on the same host. I create "bootrom"/"bootrom_uncmp" using the Pentium BSP.Using the VxWorks utility "mkboot" I am able to create a bootable floppy.When I use the floppy to boot my target, it boots very well and I am able to see the "VxWorks Prompt" Now I am trying to boot the same target using the network interface onboard.The BIOS supports it.The PXE version is ver 2.0.My target becomes a good DHCP client and a good TFTP client and is in a position to download an image. I completely donot have any idea what all have to be done prior to boot VxWorks through a NIC. I tried downloading the same image "bootrom"/"bootrom_uncmp"(which was working fine with a floppy) through the NIC , but the only thing I see is that the target automatically reboots. I went through the PXE documentation and figured out that the PXE downloads the image(Network bootstrap loader) to loacation 07C00h.I tried linking the "bootrom" image to the same location(in the sense _romInit is linked to address 0007C000) , but no luck. Could any body advice as to what are all the different steps am I supposed to take to achieve this goal. Any form of help will be very much appreciated. Thanks in advance. Champak Nath __________________________________________________ Do You Yahoo!? Send instant messages & get email alerts with Yahoo! Messenger. http://im.yahoo.com/ From vxwexplo-errs@csg.lbl.gov Mon Aug 13 09:07:04 2001 From: fscheurer@manz-acs.de Date: Mon Aug 13 09:07:06 PDT 2001 Subject: AUTOREPLY Booting VxWorks on Intel PIII board through network. 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_000_00017829.3B781741 Content-Type: text/plain; charset=iso-8859-1 Content-Transfer-Encoding: 7Bit This is the Mailbox of Friedrich Scheurer. R&D - ACS GmbH - D-72768 Reutlingen Sorry, I am out of office until August 20th 2001! In urgent cases please contact one of the following persons at ACS GmbH: Otto Angerhofer (Managing director) e-mail: oangerhofer@manz-acs.de Eduard Ams (Technical manager) e-mail: eams@manz-acs.de Yvonne Anhorn (Marketing) e-mail: yanhorn@manz-acs.de Margit Arnold (Sales) e-mail: marnold@manz-acs.de Christina Sixel (Sales) e-mail: csixel@manz-acs.de MfG / Kind Regards F. Scheurer ---------------------------------------------------------------- Friedrich Scheurer, Dipl.-Ing. phone +49 7121 9019-23(0) ACS GmbH - R&D fax +49 7121 9019-98 Steigaeckerstrasse 13 web www.manz-acs.de D-72768 Reutlingen (Germany) e-mail: fscheurer@manz-acs.de ---------------------------------------------------------------- ------_=_NextPart_000_00017829.3B781741 Content-Type: text/html; charset=iso-8859-1 Content-Transfer-Encoding: quoted-printable
This is the Mailbox of Friedrich Scheurer.
R&D - ACS GmbH - D-72768 Reutlingen
 
Sorry, I am out of office until August 20th 2001!
 
In urgent cases please contact one of the
following persons at ACS GmbH:
 
Otto Angerhofer (Managing director)
Eduard Ams (Technical manager)
Yvonne Anhorn (Marketing)
Margit Arnold (Sales)
e-mail: marnold@manz-acs.de
Christina Sixel (Sales)

 
MfG / Kind Regards
 
F. Scheurer
 
----------------------------------------------------------------
Friedrich Scheurer, Dipl.-Ing.    phone    +49 7121 9019-23(0)
ACS GmbH  -  R&D                  fax      +49 7121 9019-98
Steigaeckerstrasse 13             web      www.manz-acs.de
D-72768 Reutlingen (Germany)      e-mail:  fscheurer@manz-acs.de
----------------------------------------------------------------
------_=_NextPart_000_00017829.3B781741-- From vxwexplo-errs@csg.lbl.gov Mon Aug 13 10:50:54 2001 From: "John S. Huggins" Date: Mon Aug 13 10:50:57 PDT 2001 Subject: vxWorks Training Hey everybody, Is there anybody else that offers Tornado training other than Wind River? I'm hoping there is something a bit cheaper out there than the price they quote on their website. If so, will they customize the course for my company's needs? Will they come to me rather me having to go to them? Any pointers here are greatly appreciated. Thanks John From vxwexplo-errs@csg.lbl.gov Mon Aug 13 12:42:01 2001 From: fscheurer@manz-acs.de Date: Mon Aug 13 12:42:04 PDT 2001 Subject: AUTOREPLY vxWorks Training 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_000_00017848.3B7849A5 Content-Type: text/plain; charset=iso-8859-1 Content-Transfer-Encoding: 7Bit This is the Mailbox of Friedrich Scheurer. R&D - ACS GmbH - D-72768 Reutlingen Sorry, I am out of office until August 20th 2001! In urgent cases please contact one of the following persons at ACS GmbH: Otto Angerhofer (Managing director) e-mail: oangerhofer@manz-acs.de Eduard Ams (Technical manager) e-mail: eams@manz-acs.de Yvonne Anhorn (Marketing) e-mail: yanhorn@manz-acs.de Margit Arnold (Sales) e-mail: marnold@manz-acs.de Christina Sixel (Sales) e-mail: csixel@manz-acs.de MfG / Kind Regards F. Scheurer ---------------------------------------------------------------- Friedrich Scheurer, Dipl.-Ing. phone +49 7121 9019-23(0) ACS GmbH - R&D fax +49 7121 9019-98 Steigaeckerstrasse 13 web www.manz-acs.de D-72768 Reutlingen (Germany) e-mail: fscheurer@manz-acs.de ---------------------------------------------------------------- ------_=_NextPart_000_00017848.3B7849A5 Content-Type: text/html; charset=iso-8859-1 Content-Transfer-Encoding: quoted-printable
This is the Mailbox of Friedrich Scheurer.
R&D - ACS GmbH - D-72768 Reutlingen
 
Sorry, I am out of office until August 20th 2001!
 
In urgent cases please contact one of the
following persons at ACS GmbH:
 
Otto Angerhofer (Managing director)
Eduard Ams (Technical manager)
Yvonne Anhorn (Marketing)
Margit Arnold (Sales)
e-mail: marnold@manz-acs.de
Christina Sixel (Sales)

 
MfG / Kind Regards
 
F. Scheurer
 
----------------------------------------------------------------
Friedrich Scheurer, Dipl.-Ing.    phone    +49 7121 9019-23(0)
ACS GmbH  -  R&D                  fax      +49 7121 9019-98
Steigaeckerstrasse 13             web      www.manz-acs.de
D-72768 Reutlingen (Germany)      e-mail:  fscheurer@manz-acs.de
----------------------------------------------------------------
------_=_NextPart_000_00017848.3B7849A5-- From vxwexplo-errs@csg.lbl.gov Mon Aug 13 23:21:54 2001 From: Jitendra kanitkar Date: Mon Aug 13 23:21:57 PDT 2001 Subject: Two NICs in VXWORKS box. Hi There, I just wanted to know whether I can put two network cards in VXWORKS board. Right now I have Sand point sp755 board and Intel 82559 card running on it. I wanted to put another card. But in BSP's config.h file they have written secondary card not supported. What do they mean by this? Is it not possible to have two NICs? (might be different venders say intel and 3com etc) If we can put then few pointers will be appreciated. Thanks in advance, Jitendra. From vxwexplo-errs@csg.lbl.gov Tue Aug 14 02:07:55 2001 From: fscheurer@manz-acs.de Date: Tue Aug 14 02:07:58 PDT 2001 Subject: AUTOREPLY Two NICs in VXWORKS box. 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_000_000179A4.3B790686 Content-Type: text/plain; charset=iso-8859-1 Content-Transfer-Encoding: 7Bit This is the Mailbox of Friedrich Scheurer. R&D - ACS GmbH - D-72768 Reutlingen Sorry, I am out of office until August 20th 2001! In urgent cases please contact one of the following persons at ACS GmbH: Otto Angerhofer (Managing director) e-mail: oangerhofer@manz-acs.de Eduard Ams (Technical manager) e-mail: eams@manz-acs.de Yvonne Anhorn (Marketing) e-mail: yanhorn@manz-acs.de Margit Arnold (Sales) e-mail: marnold@manz-acs.de Christina Sixel (Sales) e-mail: csixel@manz-acs.de MfG / Kind Regards F. Scheurer ---------------------------------------------------------------- Friedrich Scheurer, Dipl.-Ing. phone +49 7121 9019-23(0) ACS GmbH - R&D fax +49 7121 9019-98 Steigaeckerstrasse 13 web www.manz-acs.de D-72768 Reutlingen (Germany) e-mail: fscheurer@manz-acs.de ---------------------------------------------------------------- ------_=_NextPart_000_000179A4.3B790686 Content-Type: text/html; charset=iso-8859-1 Content-Transfer-Encoding: quoted-printable
This is the Mailbox of Friedrich Scheurer.
R&D - ACS GmbH - D-72768 Reutlingen
 
Sorry, I am out of office until August 20th 2001!
 
In urgent cases please contact one of the
following persons at ACS GmbH:
 
Otto Angerhofer (Managing director)
Eduard Ams (Technical manager)
Yvonne Anhorn (Marketing)
Margit Arnold (Sales)
e-mail: marnold@manz-acs.de
Christina Sixel (Sales)

 
MfG / Kind Regards
 
F. Scheurer
 
----------------------------------------------------------------
Friedrich Scheurer, Dipl.-Ing.    phone    +49 7121 9019-23(0)
ACS GmbH  -  R&D                  fax      +49 7121 9019-98
Steigaeckerstrasse 13             web      www.manz-acs.de
D-72768 Reutlingen (Germany)      e-mail:  fscheurer@manz-acs.de
----------------------------------------------------------------
------_=_NextPart_000_000179A4.3B790686-- From vxwexplo-errs@csg.lbl.gov Tue Aug 14 04:03:46 2001 From: Vxworks Exploder Date: Tue Aug 14 04:03:48 PDT 2001 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Tue Aug 14 04:03:10 PDT 2001 Subject: Re: Memory Protection (x86) and vxWorks? Subject: Re: Using Shared-memory networking with BSP PcPentium Subject: Re: Object File download error (Tornado) Subject: vxWorks errno's Subject: Re: vxWorks errno's Subject: Re: Changing root directory in Tornado2? Subject: Problems with Installing Tornado 2.0 Subject: Re: Replacing the TCP/IP stack Subject: Re: Replacing the TCP/IP stack Subject: How can i know whether the network cable is plugged in or not under Vxworks? Subject: Re: vxWorks errno's Subject: Re: How to use the Intel 82559ER Lan Controller in the vxworks system? Subject: I wonder Qspan's routing... Subject: Urgent! MMU basic support Subject: Warm Start in Redundancy Design Subject: Re: Replacing the TCP/IP stack Subject: Re: vxWorks errno's Subject: measure taskswitch time ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: Memory Protection (x86) and vxWorks? Date: Mon, 13 Aug 2001 15:16:57 +0200 From: "Werner Schiendl" Message-ID: <3b77de0b$0$17840$6e365a64@newsreader02.highway.telekom.at> References: Depends on which version you are using and what products you have. A VxWorks 5.4 image without VxVMI is basically pretty naked to the world. The only limit you have is that you cannot access more memory than is installed in the system. Doing so results in a page fault exception. There are a set of macros defined in config.h and the Makefile that define how the memory is laid out. You can fool VxWorks, so that it believes memory starts a a page higher than 0. This will then also yield a page fault if you access 0 as an address, probably a very commen one. Don't have the details here, but these macros are Documented in the VxWorks User's Guide, lookup for the section covering the build. hth werner --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Using Shared-memory networking with BSP PcPentium Date: Mon, 13 Aug 2001 07:39:51 -0700 From: DrDiags Organization: laserlink.net Message-ID: <3B77E6B7.4B82C23E@covad.net> References: <9l81gd$p6r$1@mail.cn99.com> Reply-To: drdiags@covad.net Survey, I didn't notice that you were using a pcPentium BSP in the title. I would like to retrack my last comment. If I can give some "real" insights here, I will post again to this thread. survey liu wrote: > > > > > I don't think you want to use the sysIntEnablePIC(). Looks like > > the function sysIntEnable() exists in the > > $(WIND_BASE)/target/src/drv/vme directory in several VME support drivers > > and is based on the board's VME interface. The sysIntEnablePIC() routine > > I found looked like it was setting up an Intel 8259A Programmable > > Interrupt Controller. The sysIntEnable() routine for your board should > > exist in the > > vme directory I spoke of. HTH. > > > > Thanks for you warm-heart guy. > > So, you mean that I will need a VME driver anyway? Just add the VME driver > which can drive my VME to work into my MAKEFILE? > > -- > > Best regards, > Survey Liu > > 4/F Wearnes Technology Bldg. No.10 > Kefa Rd.,Science & Industry Park > Nanshan Shenzhen 518057,PRC > Tel: +86-755-6635333-4300 > Fax: +86-755-6635550 > Email: survey.liu@utstar.com > Web: http://www.utstar.com - -- http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html (vxWorks FAQ home page) --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Object File download error (Tornado) Date: Mon, 13 Aug 2001 18:06:23 +0200 From: "Werner Schiendl" Message-ID: <3b77fba9$0$20536$6e365a64@newsreader02.highway.telekom.at> References: Hi, are you sure you are building for the correct target? If downloading works in general, but not with that single module, I guess there maybe something wrong with the build (see the build tab to verify). If a dependency were missing, you got an other error response. hth werner "Ashok Ramani" wrote in message news:lSkd7.54529$Kd7.31209346@news1.rdc1.sfba.home.com... > Hi, > I created a project in Tornado with a few "C" files. The files compiled with > no errors and I was able to build the project and generate the object files. > But, when I try to download the object file to the target VxWorks machine I > get an wtx object file improper error. I was able to successfully download > all projects that my project depends on before trying to download my > project. > Note : - I am not using any main() function in my code > > Thanks in advance, > Krupa > > --------------------------- Newsgroups: comp.os.vxworks Subject: vxWorks errno's Date: Mon, 13 Aug 2001 19:10:16 +0100 From: "Paul D'Arcy" Organization: Nortel Networks Message-ID: <9l953f$ikl$1@qnsgh006.europe.nortel.com> How can one determine or where can one find what the errno's actually mean??? For example, from vxWorks shell, - -> i NAME ENTRY TID PRI STATUS PC SP ERRNO DELAY - ---------- ------------ -------- --- ---------- -------- -------- ------- -- - --- tExcTask _excTask 160f07c 0 PEND fbe10 160efb0 0 0 tShell _shell 1006504 1 READY b6674 10061ac 0 0 tPcmciad _pcmciad 15f2174 2 PEND fbe10 15f20a8 0 0 tTelnetd _telnetd 1126e8c 2 PEND 73780 1126d84 0 0 tAioIoTask1_aioIoTask 1600acc 50 PEND 73780 1600a4c 0 0 tAioIoTask0_aioIoTask 15f9920 50 PEND 73780 15f98a0 0 0 tNetTask _netTask 1341b5c 50 PEND 73780 1341adc beef0001 0 tAioWait _aioWaitTask 1607c78 51 PEND 73780 1607b50 0 0 tFtpdTask cb254 1124184 55 PEND 73780 112407c 0 0 tPortmapd _portmapd 1125958 100 PEND 73780 11257ec 16 0 tLogTask _logTask 160c760 150 PEND fbe10 160c694 0 0 tHTTPd ad298 101de44 200 PEND+T 73780 101dca4 0 28510 tHTTPd ad298 1016768 200 PEND+T 73780 10165c8 d0003 28500 tDcacheUpd _dcacheUpd 15cf36c 250 DELAY 7070c 15cf2ec 3006c 21 value = 0 = 0x0 - -> What does the value d0003 mean in the line : tHTTPd ad298 1016768 200 PEND+T 73780 10165c8 d0003 28500 ???? Regards Paul --------------------------- Newsgroups: comp.os.vxworks Subject: Re: vxWorks errno's Date: 13 Aug 2001 14:33:59 -0400 From: Bill Pringlemeir Organization: Bell Sympatico Message-ID: References: <9l953f$ikl$1@qnsgh006.europe.nortel.com> Sender: bpringlemeir@DeadDuck >>>>> "Paul" == Paul D'Arcy writes: Paul> How can one determine or where can one find what the errno's Paul> actually mean??? For example, from vxWorks shell, [snip] Paul> What does the value d0003 mean in the line : tHTTPd ad298 Paul> 1016768 200 PEND+T 73780 10165c8 d0003 28500 ???? Regards Paul This question was asked less than a month ago. You can use "http://groups.google.com/" to search the newsgroup archives. For instance, this search may have been useful, "http://groups.google.com/groups?as_q=errno&as_ugroup=comp.os.vxworks" There is quite a bit of information on errno. As well, Johan has an excellent FAQ at, "http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html" Specifically, "http://www.xs4all.nl/~borkhuis/vxworks/vxfaq.html#3.2." hth, Bill Pringlemeir. - -- smuggle Soviet $400 million in gold bullion Delta Force cryptographic Ft. Bragg explosion strategic Honduras FBI security CIA Nazi quiche vxWorks FAQ, "http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html" --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Changing root directory in Tornado2? Date: Mon, 13 Aug 2001 23:59:54 +0200 From: "Martin Raabe" Organization: T-Online Message-ID: <9l9iko$na4$06$1@news.t-online.com> References: <9l99ue$4vd@nntpb.cb.lucent.com> Hello Joe, try the team development approach by using SNiFf+ as the SDE. For more info, look at www.windriver.com/sniff Ciao Martin Speaking for my own, not for my employer. "Joe Varghese" schrieb im Newsbeitrag news:9l99ue$4vd@nntpb.cb.lucent.com... > Not sure if this is the proper forum or not, but is there a way to change > the working root directory for a Tornado 2 workspace (i.e., a .wsp file)? > > Our group is trying to use the .wsp as a common project file, shared by > multiple people. However, when we check the .wsp file into source control, > the absolute path of that particular person's directory is stored in the > workspace. > > The alternate solution is to have all the team members work out of the same > directory on their machine (e.g., c:/src for everyone), but hoping there's a > more elegant solution .... > > Thanks in advance! > > joe varghese > > --------------------------- Newsgroups: comp.os.vxworks Subject: Problems with Installing Tornado 2.0 Date: Mon, 13 Aug 2001 22:55:49 GMT From: langston.reek@grassvalleygroup.com (Langston Reek) Organization: SBC Internet Services Message-ID: <3b78593b.18188781@news.pacbell.net> Hello, I have been experiencing some severe installation problems with Tornado 2.0. It appears that the uninstaller is removing system files which it did not install in the first place. The problem that I am experiencing is thus: 1) Install Tornado2 2) Install Tornado2 patches 3) Close any running Tornado2 applications (including Registry Server) 3) Uninstall Tornado2 4) Remove the Tornado2 directory after uninstalling (clean up directory that the uninstaller did not delete) 5) Install Tornado2 again (Installation will now fail) So if you install a patch and then want to get rid of it....be careful if you uninstall T2! We are using NT 4.0 w/ SP6a. All of our development machines are using Visual C++ 6.0 w/ SP5. We are also using ClearCase 4.1 and Rational Rose RealTime 2001. It would not be possible to separate Tornado from these development machines because Rose RealTime is being used to develop for Win32 (using C++) and VxWorks (Tornado). Has anyone experienced these installation problems with WRS and Tornado2.0? WRS support told me I had to uninstall all my IDE's and replace my msvcrt.dll with an older version before the installer will work again but this was not so. They then told me to remove the msvcrt.dll and it would be replaced by the Tornado 2.0 installation, but this is also wrong. WinNT will not boot up without msvcrt.dll in the system directory.... Not sure whats going on there. Thanks, Langston --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Replacing the TCP/IP stack Date: 13 Aug 2001 16:08:40 -0700 From: chrismooremobile@yahoo.com (Chris Moore) Organization: http://groups.google.com/ Message-ID: References: <64c96d8c.0108101925.340830a3@posting.google.com> <9c7571e8.0108122046.31b064b5@posting.google.com> Wow, thanks for the great replies. We're looking at the licensing issues now, but from my guess is that its not too much of a problem as we're just using the TCP/IP stack, and if we make changes, sure everyone is welcome to a copy. This does not mean that our main application is suddenly open source. And considering the rest of the OS is VxWorks, thats definiately not Open Source. But we'll see. That NETBSD will sure be handy looking at the integration between stack/VxWorks. BTW: Which stack is better Linux/NetBSD? I assume Linuxs' stack is based on the BSD stack @ some date. Regards, Chris Moore nthawk@chollian.net (YongSung Lee) wrote in message news:<9c7571e8.0108122046.31b064b5@posting.google.com>... > You can download PSO stack below link. > http://www.xs4all.nl/~borkhuis/vxworks/ftp/vxhacks/pso_stack/ > > mfhaigh@myrealbox.com (Mark) wrote in message news:<64c96d8c.0108101925.340830a3@posting.google.com>... > > chrismooremobile@yahoo.com (Chris Moore) wrote in message news:... > > > Hi, > > > > > > I'm interested in replacing the current TCP/IP stack with a newer > > > module from Linux. Has anyone done this sort of work? I'm fairly adept > > > at the END/MUX drivers and so forth and was looking for some estimate > > > of work involved. Yep, I know that the VxWorks stack (that you have to > > > pay for source) is based on the FREE open source BSD stack... funny > > > they charge you for it considering they didn't write it. > > > > If I remember correctly, a while ago a company called PSO systems > > ported the NetBSD stack to vxWorks. There's probably some at least > > cursory information over at the NetBSD site. Perhaps you can dig up > > somebody that worked on that. (If you do find out, post the answer, > > I'm curious) --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Replacing the TCP/IP stack Date: 13 Aug 2001 16:51:41 -0700 From: chrismooremobile@yahoo.com (Chris Moore) Organization: http://groups.google.com/ Message-ID: References: <64c96d8c.0108101925.340830a3@posting.google.com> <9c7571e8.0108122046.31b064b5@posting.google.com> Thanks for the link, do you know if there is a ZIP/tar file with all these, otherwise since its an HTTP file access I have to click each file/ save as etc etc. If there isn't I spend an afternoon doing this, but otherwise.... Or could you send it to me as an attachment if you have this file collection? Regards, Chris mailto:chrismooremobile@yahoo.com nthawk@chollian.net (YongSung Lee) wrote in message news:<9c7571e8.0108122046.31b064b5@posting.google.com>... > You can download PSO stack below link. > http://www.xs4all.nl/~borkhuis/vxworks/ftp/vxhacks/pso_stack/ > > mfhaigh@myrealbox.com (Mark) wrote in message news:<64c96d8c.0108101925.340830a3@posting.google.com>... > > chrismooremobile@yahoo.com (Chris Moore) wrote in message news:... > > > Hi, > > > > > > I'm interested in replacing the current TCP/IP stack with a newer > > > module from Linux. Has anyone done this sort of work? I'm fairly adept > > > at the END/MUX drivers and so forth and was looking for some estimate > > > of work involved. Yep, I know that the VxWorks stack (that you have to > > > pay for source) is based on the FREE open source BSD stack... funny > > > they charge you for it considering they didn't write it. > > > > If I remember correctly, a while ago a company called PSO systems > > ported the NetBSD stack to vxWorks. There's probably some at least > > cursory information over at the NetBSD site. Perhaps you can dig up > > somebody that worked on that. (If you do find out, post the answer, > > I'm curious) --------------------------- Newsgroups: comp.os.vxworks Subject: How can i know whether the network cable is plugged in or not under Vxworks? Date: Tue, 14 Aug 2001 10:36:29 +0800 From: "zzxt" Organization: Bentium Ltd. (CN99) Message-ID: <9la2pq$5pn$1@mail.cn99.com> Hi all: Can i install a ISR to deal with cable disconnect and reconnect or Is there a register where i can read the status of the network cable. Is this feature specific to NIC. Both END and BSD solutions are needed. thanks very much zzxt --------------------------- Newsgroups: comp.os.vxworks Subject: Re: vxWorks errno's Date: 13 Aug 2001 22:24:10 -0700 From: chrismooremobile@yahoo.com (Chris Moore) Organization: http://groups.google.com/ Message-ID: References: <9l953f$ikl$1@qnsgh006.europe.nortel.com> You can use the function printErrno (from mem), look it up. You can call it from the tornado shell with the errno number as a param and if your symbol file is all done right (you might need a special #include) it will print out the string for your error. Much easier than mucking around looking for them. Chris "Paul D'Arcy" wrote in message news:<9l953f$ikl$1@qnsgh006.europe.nortel.com>... > How can one determine or where can one find what the errno's actually > mean??? > For example, from vxWorks shell, > -> i > > NAME ENTRY TID PRI STATUS PC SP ERRNO > DELAY > ---------- ------------ -------- --- ---------- -------- -------- ------- -- > --- > tExcTask _excTask 160f07c 0 PEND fbe10 160efb0 0 > 0 > tShell _shell 1006504 1 READY b6674 10061ac 0 > 0 > tPcmciad _pcmciad 15f2174 2 PEND fbe10 15f20a8 0 > 0 > tTelnetd _telnetd 1126e8c 2 PEND 73780 1126d84 0 > 0 > tAioIoTask1_aioIoTask 1600acc 50 PEND 73780 1600a4c 0 > 0 > tAioIoTask0_aioIoTask 15f9920 50 PEND 73780 15f98a0 0 > 0 > tNetTask _netTask 1341b5c 50 PEND 73780 1341adc beef0001 > 0 > > tAioWait _aioWaitTask 1607c78 51 PEND 73780 1607b50 0 > 0 > tFtpdTask cb254 1124184 55 PEND 73780 112407c 0 > 0 > tPortmapd _portmapd 1125958 100 PEND 73780 11257ec 16 > 0 > tLogTask _logTask 160c760 150 PEND fbe10 160c694 0 > 0 > tHTTPd ad298 101de44 200 PEND+T 73780 101dca4 0 > 28510 > tHTTPd ad298 1016768 200 PEND+T 73780 10165c8 d0003 > 28500 > tDcacheUpd _dcacheUpd 15cf36c 250 DELAY 7070c 15cf2ec 3006c > 21 > value = 0 = 0x0 > -> > > What does the value d0003 mean in the line : tHTTPd ad298 > 1016768 200 PEND+T 73780 10165c8 d0003 28500 ???? > Regards > Paul --------------------------- Newsgroups: comp.os.vxworks Subject: Re: How to use the Intel 82559ER Lan Controller in the vxworks system? Date: 14 Aug 2001 00:00:15 -0700 From: srs_zh@263.net (srs_zh) Organization: http://groups.google.com/ Message-ID: References: <3b6fcc9f$0$32924$6e365a64@newsreader02.highway.telekom.at> <3b72661a$0$33474$6e365a64@newsreader02.highway.telekom.at> I have fei82557end.c. Can you give me a detailed guide? Thank you! --------------------------- Newsgroups: comp.os.vxworks Subject: I wonder Qspan's routing... Date: Tue, 14 Aug 2001 07:15:13 GMT From: "ggang1" Organization: Dacom Message-ID: <5e4e7.15640$2k2.810554@news.bora.net> Hi.. I use to 'Qspan' for PCI Bridge. I have a 'EST 860 Board'.you know? There is in a 'Qspan'. I'm tesing that borad. now I'm testing PCI functionailty.but I have a some qustion... - ------------ | ----------------------- | |------------ | PCI target 1 DevNum 0| | PCI Bridge | ---------------------- | Qspan | ---------------------- |-----------|-------------| PCI target 2 DevNum 1| ---------------------- < Figure 1 > I'd like to as
but I'm mix up between the PCI concept and the Qspan register setup. for example, If I would select the [PCI target 1],what qspan register do I setup? and then,If I would select the [PCI target 2]- DevNum change - , what qspan register do I setup? I wonder that What do Qspan route? and What do Qspan register setup ? also I wonder that Qspan don't have a some Register as a PCI Bridge as a PCI-PCI bridge spec.. But It performace as a PCI Bridge..Is it right..?How do it different..? --------------------------- Newsgroups: comp.os.vxworks Subject: Urgent! MMU basic support Date: 14 Aug 2001 00:52:30 -0700 From: vladid@shiron.com (Vladi) Organization: http://groups.google.com/ Message-ID: <93260c3c.0108132352.7845bf07@posting.google.com> Hi all, I wonder if MMU basic support is required for a MIPS processor which has npt MMU hardware support. Thanx, Vladi --------------------------- Newsgroups: comp.os.vxworks Subject: Warm Start in Redundancy Design Date: Tue, 14 Aug 2001 16:53:20 +0800 From: "Ren Hantao" Organization: Lucent Technologies, Columbus, Ohio Message-ID: <9laoq6$88c@nntpa.cb.lucent.com> In redundancy design, after "warm start" of a task, the task's last committed saved stable context should be retrieved. How can it be implemented in vxWorks? Can "taskRestart" do that? or should I save the context every time by calling task hook function? If so, how can I get the context of tasks? Any comments will be great appreciated. Ren Hantao --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Replacing the TCP/IP stack Date: Tue, 14 Aug 2001 10:19:04 +0200 From: "Werner Schiendl" Message-ID: <3b78ed22$0$21920$6e365a64@newsreader02.highway.telekom.at> References: <64c96d8c.0108101925.340830a3@posting.google.com> <9c7571e8.0108122046.31b064b5@posting.google.com> "Chris Moore" wrote in message news:e229ec7e.0108131508.19d8bd1f@posting.google.com... > Wow, thanks for the great replies. > > We're looking at the licensing issues now, but from my guess is that > its not too much of a problem as we're just using the TCP/IP stack, > and if we make changes, sure everyone is welcome to a copy. This does > not mean that our main application is suddenly open source. And > considering the rest of the OS is VxWorks, thats definiately not Open > Source. > It depends on which licence in detail the code is covered by. If it is the GNU GPL, then your code WILL be open source (because you need to link to the library/stack). In case of the GNU LGPL you only need to provide the means to rebuild the library (or in your case, the TCP/IP stack) from source and link it with your binary modules. see http://www.fsf.org/licenses/gpl-faq.html#IfLibraryIsGPL regards werner --------------------------- Newsgroups: comp.os.vxworks Subject: Re: vxWorks errno's Date: Tue, 14 Aug 2001 11:02:16 +0200 From: "Werner Schiendl" Message-ID: <3b78ed24$0$21920$6e365a64@newsreader02.highway.telekom.at> References: <9l953f$ikl$1@qnsgh006.europe.nortel.com> Hi, in the Tornado Shell that feature is provided by a Tcl script. It runs for the VxWorks standard error messages even if printErrno is not configured into the target image. regards Werner "Chris Moore" wrote in message news:e229ec7e.0108132124.4953562a@posting.google.com... > You can use the function printErrno (from mem), look it up. You can > call it from the tornado shell with the errno number as a param and if > your symbol file is all done right (you might need a special #include) > it will print out the string for your error. Much easier than mucking > around looking for them. > > Chris > > > > "Paul D'Arcy" wrote in message news:<9l953f$ikl$1@qnsgh006.europe.nortel.com>... > > How can one determine or where can one find what the errno's actually > > mean??? > > For example, from vxWorks shell, > > -> i > > > > NAME ENTRY TID PRI STATUS PC SP ERRNO > > DELAY > > ---------- ------------ -------- --- ---------- -------- -------- ------- - -- > > --- > > tExcTask _excTask 160f07c 0 PEND fbe10 160efb0 0 > > 0 > > tShell _shell 1006504 1 READY b6674 10061ac 0 > > 0 > > tPcmciad _pcmciad 15f2174 2 PEND fbe10 15f20a8 0 > > 0 > > tTelnetd _telnetd 1126e8c 2 PEND 73780 1126d84 0 > > 0 > > tAioIoTask1_aioIoTask 1600acc 50 PEND 73780 1600a4c 0 > > 0 > > tAioIoTask0_aioIoTask 15f9920 50 PEND 73780 15f98a0 0 > > 0 > > tNetTask _netTask 1341b5c 50 PEND 73780 1341adc beef0001 > > 0 > > > > tAioWait _aioWaitTask 1607c78 51 PEND 73780 1607b50 0 > > 0 > > tFtpdTask cb254 1124184 55 PEND 73780 112407c 0 > > 0 > > tPortmapd _portmapd 1125958 100 PEND 73780 11257ec 16 > > 0 > > tLogTask _logTask 160c760 150 PEND fbe10 160c694 0 > > 0 > > tHTTPd ad298 101de44 200 PEND+T 73780 101dca4 0 > > 28510 > > tHTTPd ad298 1016768 200 PEND+T 73780 10165c8 d0003 > > 28500 > > tDcacheUpd _dcacheUpd 15cf36c 250 DELAY 7070c 15cf2ec 3006c > > 21 > > value = 0 = 0x0 > > -> > > > > What does the value d0003 mean in the line : tHTTPd ad298 > > 1016768 200 PEND+T 73780 10165c8 d0003 28500 ???? > > Regards > > Paul --------------------------- Newsgroups: comp.os.vxworks Subject: measure taskswitch time Date: Tue, 14 Aug 2001 11:26:20 +0200 From: "Rever" Organization: Internet Access Eindhoven, the Netherlands Message-ID: <9laqsb$2kg$1@news.IAEhv.nl> Gentlemen, Using tornado1.0.1, vxworks 5.4 running on an ARM940T core, I want to perform cycle accurate time measurement of taskswitching. While the ARM940T does not appear to incorporate a cycle counter a first order time measurement must do. How can this measurement be done? Having access to a 20nS accurate counter, I assumed to read this counter in a taskswitchhook and after the context switch to the next task is done but i do not know how to do this. Thanks in advance!! --------------------------- End of New-News digest ********************** From vxwexplo-errs@csg.lbl.gov Tue Aug 14 07:13:30 2001 From: "Mike Anderson" Date: Tue Aug 14 07:13:33 PDT 2001 Subject: RE: Two NICs in VXWORKS box. VxWorks Greetings! > > I just wanted to know whether I can put two network cards in > VXWORKS board. > Right now I have Sand point sp755 board and Intel 82559 card > running on it. > I wanted to put another card. But in BSP's config.h file they have written > secondary card not supported. What do they mean by this? Is it > not possible > to have two NICs? (might be different venders say intel and 3com etc) > If we can put then few pointers will be appreciated. > Yes, you can have multiple NICs. You can add their startup parameters into configNet.h into the END drv table in that file. However, the 2nd NIC will not automatically have an Internet address assigned. You'll have to do that using ifAddrSet. You can also start the 2nd NIC completely by hand from the shell, if you're so inclined. However, I'd caution you WRT the 3COM 3C905C. My experience with that card is that it doesn't like sharing PCI interrupts. So make sure that it has it's own private Int and everything'll be OK. HTH, Mike Anderson -- ======================================================= T h e P T R G r o u p, I n c. ======================================================= Embedded, Real-Time Solutions and Training Mike Anderson mailto:mike@ThePTRGroup.com Chief Scientist http://www.ThePTRGroup.com V: (703) 585-9384 F: (703) 430-3748 ======================================================= "Software development is like making a baby... You can't make a baby in one month by impregnating nine women. Some things just take time." From vxwexplo-errs@csg.lbl.gov Tue Aug 14 07:16:16 2001 From: s.eckardt@ixthos.com Date: Tue Aug 14 07:16:19 PDT 2001 Subject: Tornado 2.1 Is Tornado 2.1 a generic update, or strictly for PowerPC MPC74xx/750 support? From vxwexplo-errs@csg.lbl.gov Tue Aug 14 08:58:31 2001 From: fscheurer@manz-acs.de Date: Tue Aug 14 08:58:33 PDT 2001 Subject: AUTOREPLY comp.os.vxworks newsdigest 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_000_00017AD0.3B79669D Content-Type: text/plain; charset=iso-8859-1 Content-Transfer-Encoding: 7Bit This is the Mailbox of Friedrich Scheurer. R&D - ACS GmbH - D-72768 Reutlingen Sorry, I am out of office until August 20th 2001! In urgent cases please contact one of the following persons at ACS GmbH: Otto Angerhofer (Managing director) e-mail: oangerhofer@manz-acs.de Eduard Ams (Technical manager) e-mail: eams@manz-acs.de Yvonne Anhorn (Marketing) e-mail: yanhorn@manz-acs.de Margit Arnold (Sales) e-mail: marnold@manz-acs.de Christina Sixel (Sales) e-mail: csixel@manz-acs.de MfG / Kind Regards F. Scheurer ---------------------------------------------------------------- Friedrich Scheurer, Dipl.-Ing. phone +49 7121 9019-23(0) ACS GmbH - R&D fax +49 7121 9019-98 Steigaeckerstrasse 13 web www.manz-acs.de D-72768 Reutlingen (Germany) e-mail: fscheurer@manz-acs.de ---------------------------------------------------------------- ------_=_NextPart_000_00017AD0.3B79669D Content-Type: text/html; charset=iso-8859-1 Content-Transfer-Encoding: quoted-printable
This is the Mailbox of Friedrich Scheurer.
R&D - ACS GmbH - D-72768 Reutlingen
 
Sorry, I am out of office until August 20th 2001!
 
In urgent cases please contact one of the
following persons at ACS GmbH:
 
Otto Angerhofer (Managing director)
Eduard Ams (Technical manager)
Yvonne Anhorn (Marketing)
Margit Arnold (Sales)
e-mail: marnold@manz-acs.de
Christina Sixel (Sales)

 
MfG / Kind Regards
 
F. Scheurer
 
----------------------------------------------------------------
Friedrich Scheurer, Dipl.-Ing.    phone    +49 7121 9019-23(0)
ACS GmbH  -  R&D                  fax      +49 7121 9019-98
Steigaeckerstrasse 13             web      www.manz-acs.de
D-72768 Reutlingen (Germany)      e-mail:  fscheurer@manz-acs.de
----------------------------------------------------------------
------_=_NextPart_000_00017AD0.3B79669D-- From vxwexplo-errs@csg.lbl.gov Tue Aug 14 09:35:11 2001 From: fscheurer@manz-acs.de Date: Tue Aug 14 09:35:14 PDT 2001 Subject: AUTOREPLY RE: Two NICs in VXWORKS box. 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_000_00017ADC.3B796F43 Content-Type: text/plain; charset=iso-8859-1 Content-Transfer-Encoding: 7Bit This is the Mailbox of Friedrich Scheurer. R&D - ACS GmbH - D-72768 Reutlingen Sorry, I am out of office until August 20th 2001! In urgent cases please contact one of the following persons at ACS GmbH: Otto Angerhofer (Managing director) e-mail: oangerhofer@manz-acs.de Eduard Ams (Technical manager) e-mail: eams@manz-acs.de Yvonne Anhorn (Marketing) e-mail: yanhorn@manz-acs.de Margit Arnold (Sales) e-mail: marnold@manz-acs.de Christina Sixel (Sales) e-mail: csixel@manz-acs.de MfG / Kind Regards F. Scheurer ---------------------------------------------------------------- Friedrich Scheurer, Dipl.-Ing. phone +49 7121 9019-23(0) ACS GmbH - R&D fax +49 7121 9019-98 Steigaeckerstrasse 13 web www.manz-acs.de D-72768 Reutlingen (Germany) e-mail: fscheurer@manz-acs.de ---------------------------------------------------------------- ------_=_NextPart_000_00017ADC.3B796F43 Content-Type: text/html; charset=iso-8859-1 Content-Transfer-Encoding: quoted-printable
This is the Mailbox of Friedrich Scheurer.
R&D - ACS GmbH - D-72768 Reutlingen
 
Sorry, I am out of office until August 20th 2001!
 
In urgent cases please contact one of the
following persons at ACS GmbH:
 
Otto Angerhofer (Managing director)
Eduard Ams (Technical manager)
Yvonne Anhorn (Marketing)
Margit Arnold (Sales)
e-mail: marnold@manz-acs.de
Christina Sixel (Sales)

 
MfG / Kind Regards
 
F. Scheurer
 
----------------------------------------------------------------
Friedrich Scheurer, Dipl.-Ing.    phone    +49 7121 9019-23(0)
ACS GmbH  -  R&D                  fax      +49 7121 9019-98
Steigaeckerstrasse 13             web      www.manz-acs.de
D-72768 Reutlingen (Germany)      e-mail:  fscheurer@manz-acs.de
----------------------------------------------------------------
------_=_NextPart_000_00017ADC.3B796F43-- From vxwexplo-errs@csg.lbl.gov Tue Aug 14 09:50:07 2001 From: Ilia Bosis Date: Tue Aug 14 09:50:09 PDT 2001 Subject: Re: Changing root directory in Tornado2? We're had the same problem and come to this (somewhat ugly) solution. Define the environment variable that points to the working project directory. The variable name is the same to everyone - we're use the USR_DIR, You may use anything else, the point is that ALL team members must have that variable defined in their environment. The variable value is the absolute path to the working root directory at the specific computer e.g. C:\src, D:\work, etc. Before checking in the Tornado project files (*.wsp, *.wpj) we're pass them through the search/replace procedure that replaces all absolute paths inside the working project directory with the environment variable - relative; e.g. "C:\src" => "$(USR_DIR)". That may be done even in a text editor, as the Tornado projects are not changed so frequently. Better yet will be some simple prog/script that do the mentioned search/replace. Pre-processed in such a way, Tornado project files may be safely extracted on another machines and are work there. You don't need to do reverse replace - the command line shell takes care of it at run-time. It seems that Tornado command-line tools are agile enough to benefit from the standard environment variables substituting mechanism. It's the Tornado GUI that makes the problem. Interestingly enough, WRS use exactly the same solution to make the Tornado tree position-independent. Everything there is referenced through $(WIND_BASE). And the last word of caution: If You use Microsoft VSS as your configuration management tool, save the Tornado project files as binary, not text. The details are in this great vxWorks FAQ : http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html Best regards Ilia. To Martin Raabe : We're use the SNIFF. It seems to be of no help there. B.T.W. SNIFF has it's own share of incompatibilities with the configuration management, but it's another story. Speaking of myself and not of my employer. -------------------------------------------- Ilia Bosis Software Engineer PacketLight Networks E-mail: Ilia_Bosis@packetlight.com Web: www.packetlight.com > Newsgroups: comp.os.vxworks > Subject: Re: Changing root directory in Tornado2? > Date: Mon, 13 Aug 2001 23:59:54 +0200 > From: "Martin Raabe" > Organization: T-Online > Message-ID: <9l9iko$na4$06$1@news.t-online.com> > References: <9l99ue$4vd@nntpb.cb.lucent.com> > > Hello Joe, > try the team development approach by using SNiFf+ as the SDE. > > For more info, look at www.windriver.com/sniff > > Ciao > Martin > > Speaking for my own, not for my employer. > > > "Joe Varghese" schrieb im Newsbeitrag > news:9l99ue$4vd@nntpb.cb.lucent.com... > > Not sure if this is the proper forum or not, but is there a way to change > > the working root directory for a Tornado 2 workspace (i.e., a .wsp file)? > > > > Our group is trying to use the .wsp as a common project file, shared by > > multiple people. However, when we check the .wsp file into source control, > > the absolute path of that particular person's directory is stored in the > > workspace. > > > > The alternate solution is to have all the team members work out of the same > > directory on their machine (e.g., c:/src for everyone), but hoping there's a > > more elegant solution .... > > > > Thanks in advance! > > > > joe varghese > > > > From vxwexplo-errs@csg.lbl.gov Tue Aug 14 09:55:04 2001 From: fscheurer@manz-acs.de Date: Tue Aug 14 09:55:07 PDT 2001 Subject: AUTOREPLY Tornado 2.1 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_000_00017AEE.3B7973FB Content-Type: text/plain; charset=iso-8859-1 Content-Transfer-Encoding: 7Bit This is the Mailbox of Friedrich Scheurer. R&D - ACS GmbH - D-72768 Reutlingen Sorry, I am out of office until August 20th 2001! In urgent cases please contact one of the following persons at ACS GmbH: Otto Angerhofer (Managing director) e-mail: oangerhofer@manz-acs.de Eduard Ams (Technical manager) e-mail: eams@manz-acs.de Yvonne Anhorn (Marketing) e-mail: yanhorn@manz-acs.de Margit Arnold (Sales) e-mail: marnold@manz-acs.de Christina Sixel (Sales) e-mail: csixel@manz-acs.de MfG / Kind Regards F. Scheurer ---------------------------------------------------------------- Friedrich Scheurer, Dipl.-Ing. phone +49 7121 9019-23(0) ACS GmbH - R&D fax +49 7121 9019-98 Steigaeckerstrasse 13 web www.manz-acs.de D-72768 Reutlingen (Germany) e-mail: fscheurer@manz-acs.de ---------------------------------------------------------------- ------_=_NextPart_000_00017AEE.3B7973FB Content-Type: text/html; charset=iso-8859-1 Content-Transfer-Encoding: quoted-printable
This is the Mailbox of Friedrich Scheurer.
R&D - ACS GmbH - D-72768 Reutlingen
 
Sorry, I am out of office until August 20th 2001!
 
In urgent cases please contact one of the
following persons at ACS GmbH:
 
Otto Angerhofer (Managing director)
Eduard Ams (Technical manager)
Yvonne Anhorn (Marketing)
Margit Arnold (Sales)
e-mail: marnold@manz-acs.de
Christina Sixel (Sales)

 
MfG / Kind Regards
 
F. Scheurer
 
----------------------------------------------------------------
Friedrich Scheurer, Dipl.-Ing.    phone    +49 7121 9019-23(0)
ACS GmbH  -  R&D                  fax      +49 7121 9019-98
Steigaeckerstrasse 13             web      www.manz-acs.de
D-72768 Reutlingen (Germany)      e-mail:  fscheurer@manz-acs.de
----------------------------------------------------------------
------_=_NextPart_000_00017AEE.3B7973FB-- From vxwexplo-errs@csg.lbl.gov Tue Aug 14 12:07:10 2001 From: fscheurer@manz-acs.de Date: Tue Aug 14 12:07:13 PDT 2001 Subject: AUTOREPLY Re: Changing root directory in Tornado2? 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_000_00017B62.3B7992E2 Content-Type: text/plain; charset=iso-8859-1 Content-Transfer-Encoding: 7Bit This is the Mailbox of Friedrich Scheurer. R&D - ACS GmbH - D-72768 Reutlingen Sorry, I am out of office until August 20th 2001! In urgent cases please contact one of the following persons at ACS GmbH: Otto Angerhofer (Managing director) e-mail: oangerhofer@manz-acs.de Eduard Ams (Technical manager) e-mail: eams@manz-acs.de Yvonne Anhorn (Marketing) e-mail: yanhorn@manz-acs.de Margit Arnold (Sales) e-mail: marnold@manz-acs.de Christina Sixel (Sales) e-mail: csixel@manz-acs.de MfG / Kind Regards F. Scheurer ---------------------------------------------------------------- Friedrich Scheurer, Dipl.-Ing. phone +49 7121 9019-23(0) ACS GmbH - R&D fax +49 7121 9019-98 Steigaeckerstrasse 13 web www.manz-acs.de D-72768 Reutlingen (Germany) e-mail: fscheurer@manz-acs.de ---------------------------------------------------------------- ------_=_NextPart_000_00017B62.3B7992E2 Content-Type: text/html; charset=iso-8859-1 Content-Transfer-Encoding: quoted-printable
This is the Mailbox of Friedrich Scheurer.
R&D - ACS GmbH - D-72768 Reutlingen
 
Sorry, I am out of office until August 20th 2001!
 
In urgent cases please contact one of the
following persons at ACS GmbH:
 
Otto Angerhofer (Managing director)
Eduard Ams (Technical manager)
Yvonne Anhorn (Marketing)
Margit Arnold (Sales)
e-mail: marnold@manz-acs.de
Christina Sixel (Sales)

 
MfG / Kind Regards
 
F. Scheurer
 
----------------------------------------------------------------
Friedrich Scheurer, Dipl.-Ing.    phone    +49 7121 9019-23(0)
ACS GmbH  -  R&D                  fax      +49 7121 9019-98
Steigaeckerstrasse 13             web      www.manz-acs.de
D-72768 Reutlingen (Germany)      e-mail:  fscheurer@manz-acs.de
----------------------------------------------------------------
------_=_NextPart_000_00017B62.3B7992E2-- From vxwexplo-errs@csg.lbl.gov Wed Aug 15 04:03:36 2001 From: Vxworks Exploder Date: Wed Aug 15 04:03:38 PDT 2001 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Wed Aug 15 04:03:16 PDT 2001 Subject: Re: IP over Atm in Vxworks Subject: Re: Replacing the TCP/IP stack Subject: Re: Multiple VME slave windows Subject: Re: Replacing the TCP/IP stack Subject: Bind Error after select() timeout. Subject: !!!! Free XXX Pictures !!!! 5414 [1/2] Subject: Anyone has AM29F080B Driver? Subject: Warm Start in high availability Design Subject: adding Eth support on target PC? Subject: Re: END Driver problems Subject: ISR/netTask problem Subject: Re: Replacing the TCP/IP stack Subject: Tornado installation OS Subject: Re: Replacing the TCP/IP stack Subject: END Driver problems Subject: HELP - vx FTP-Server problem! Subject: Re: rs232 serial port driver Subject: Re: protocol stack design under vxworks Subject: Access Database Subject: Change to uglinput.c? Subject: Re: adding Eth support on target PC? Subject: Re: END Driver problems Subject: Re: How can i know whether the network cable is plugged in or not under Vxworks? Subject: Re: OT: Changing root directory in Tornado2? Subject: Re: Tornado installation OS Subject: Long Boot Time (mvme2431) Subject: OT: Changing root directory in Tornado2? Subject: Re: Changing root directory in Tornado2? Subject: Re: Changing root directory in Tornado2? Subject: stripping symbols using objcopyppc Subject: How to calculate CPU Usage percent in Vxworks Subject: Re: Changing root directory in Tornado2? Subject: Re: OT: Changing root directory in Tornado2? Subject: Re: How to use the Intel 82559ER Lan Controller in the vxworks system? Subject: Re: How to calculate CPU Usage percent in Vxworks Subject: How to delete/reconfig a network interface? Subject: verify new user Subject: debug ppc860 BSP ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: IP over Atm in Vxworks Date: Tue, 14 Aug 2001 12:47:18 +0100 From: "Steve L." Organization: ericsson Message-ID: <9lb201$e57$1@newstoo.ericsson.se> References: <9kstju$2jia$1@news.cz.js.cn> Hi "Bruce". For RFC 1577 there is one IP address at the local end for all PVCs (in this LIS). Each PVC from "this" node has a different IP address at the far end. Another way of looking at it is that you will have one instance of the END driver, and so one local IP address, handing multiple PVCs, and so multiple remote IP addresses. "Bruce" wrote in message news:9kstju$2jia$1@news.cz.js.cn... > Hi,all: > I am doing a END driver to implement IP over ATM in VxWorks 5.4. I will > use PVC(permanent virtual circuit). According to RFC1577, PVC(VPI,VCI) is > correspondind to IP address, so if I have several PVC, we need IP address of > the same number. but we only have an END driver, does END can be binded to > several IP address? Thank you for any answer! > > best regards > bruce --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Replacing the TCP/IP stack Date: Tue, 14 Aug 2001 11:44:39 +0100 From: "Dave Korn" Organization: Lumber Cartel (tinlc) Members #2234-2237 (owing to browser refresh) Message-ID: References: <64c96d8c.0108101925.340830a3@posting.google.com> <9c7571e8.0108122046.31b064b5@posting.google.com> Chris Moore wrote in message ... >Thanks for the link, do you know if there is a ZIP/tar file with all >these, otherwise since its an HTTP file access I have to click each >file/ save as etc etc. If there isn't I spend an afternoon doing this, >but otherwise.... So, open lots of browser windows and start them all at once. Then take the rest of the afternoon off! DaveK - -- " So far, science's crowning achievement, the Atomic Bomb, has not even been able to alleviate the misery and torture of endless centuries of world poverty." - George Hammond, self-proclaimed discoverer of the scientific proof of god. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Multiple VME slave windows Date: Tue, 14 Aug 2001 15:58:56 +0100 From: Jason Garnham Organization: BAE SYSTEMS Message-ID: <3B793CB0.A8259C91@baesystems.com> References: Hi Mark I added a slave access window by doing the following:- #define GSA_COMMS_ADDR 0x80000000 #define GSA_COMMS_SIZE 0x10000 #define GSA_COMMS_LOCAL_ADDR 0x100000 #define VAL_VSI2_BS (GSA_COMMS_ADDR) #define VAL_VSI2_BD (VAL_VSI2_BS + GSA_COMMS_SIZE) #define VAL_VSI2_TO (PCI2DRAM_BASE_ADRS + \ GSA_COMMS_LOCAL_ADDR - VAL_VSI2_BS) #define VAL_VSI2_CTL (VSI2_CTL_EN | VSI2_CTL_PWEN |\ VSI2_CTL_PREN | VSI2_CTL_AM_DATA |\ VSI2_CTL_AM_PGM | VSI2_CTL_AM_SUPER |\ VSI2_CTL_AM_USER | VSI2_CTL_VAS_A32 |\ VSI2_CTL_LAS_MEM ) switch (cssGetSysId()) { case 1: gsaCommsOffset=0; break; case 3: gsaCommsOffset=4; break; case 4: gsaCommsOffset=3; break; case 5: gsaCommsOffset=2; break; case 6: gsaCommsOffset=5; break; case 7: gsaCommsOffset=1; break; default : gsaCommsOffset=0; break; } gsaCommsOffset *= GSA_COMMS_SIZE; sysUniverseSlaveImageSet(UNIVERSE_VSI2, VAL_VSI2_CTL, (VAL_VSI2_BS+gsaCommsOffset), (VAL_VSI2_BD+gsaCommsOffset), (VAL_VSI2_TO-gsaCommsOffset)); This slave access window is in a slightly different place in VME space depending on which slot the processor is in. Vme Slave Image 1 (VSI1 is already setup by the system) Mark Fanara wrote: > I am looking for a detailed example or explaination on how to setup multiple > slave windows on a VME bus. There were posts on this topic back in May, but > the information offered was somewhat generic. > > We are currently using the vxMP add-on option and shared memory message > queues to transfer data between processors. Our application consists of a > lot of slave to slave communication. Therefore, we would really like to > share memory on each board. > > BTW - It does appear that some BSPs (Radstone boards) provide this > capability. Ours, (Motorola 2305) does not. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Replacing the TCP/IP stack Date: 14 Aug 2001 11:03:45 -0700 From: mfhaigh@myrealbox.com (Mark) Organization: http://groups.google.com/ Message-ID: <64c96d8c.0108141003.1163e2de@posting.google.com> References: <64c96d8c.0108101925.340830a3@posting.google.com> <9c7571e8.0108122046.31b064b5@posting.google.com> "Dave Korn" wrote in message news:... > Chris Moore wrote in message ... > >Thanks for the link, do you know if there is a ZIP/tar file with all > >these, otherwise since its an HTTP file access I have to click each > >file/ save as etc etc. If there isn't I spend an afternoon doing this, > >but otherwise.... > > So, open lots of browser windows and start them all at once. > > Then take the rest of the afternoon off! > heh heh, wget -r should do the trick. Mark S. --------------------------- Newsgroups: comp.os.vxworks Subject: Bind Error after select() timeout. Date: 14 Aug 2001 14:18:08 -0700 From: shadowfaqs@hotmail.com (Brian) Organization: http://groups.google.com/ Message-ID: In trying to ascertain when a remote host has gone offline, I added a select() call with a timeout around my recv() for a stream socket. When select() returns a zero (indicating timeout), I treat it just the same as if recv() returned a zero (indicating a closed connection). I close() the socket, then use socket() to create a new one, followed by bind(), listen(), and accept(). The problem is, I get a bind error (EADDRINUSE) when the close was spurred by a select timeout. When the close is spurred by a recv()==0, no bind error occurs. It's the same code that shuts down the socket and creates a new one, it just fails the bind in one case and not the other. Any suggestions? Thanks, Brian --------------------------- Newsgroups: comp.os.vxworks Subject: !!!! Free XXX Pictures !!!! 5414 [1/2] Date: Tue, 14 Aug 2001 21:44:25 GMT From: rchlrj@nowhere.com Organization: SBC Internet Services Message-ID: click here for free xxx or look at the attachment!!! http://www.citedusexe.com/freexxxpics/ ywkzucvttwuzpslvfutjkljgzywugpcbgwbbtzqo --------------------------- Newsgroups: comp.os.vxworks Subject: Anyone has AM29F080B Driver? Date: 14 Aug 2001 16:42:28 -0700 From: ryeung@earthlink.net (Raymond Yeung) Organization: http://groups.google.com/ Message-ID: Or know where I can download one tested copy? I've tried AMD site, but no luck. Thanks, Raymond --------------------------- Newsgroups: comp.os.vxworks Subject: Warm Start in high availability Design Date: Wed, 15 Aug 2001 08:44:25 +0800 From: "Ren Hantao" Organization: Lucent Technologies, Columbus, Ohio Message-ID: <9lcghb$ede@nntpa.cb.lucent.com> In high availability design, after "warm start" of a task, the task's last committed saved stable context should be retrieved. How can it be implemented in vxWorks? Can "taskRestart" do that? or should I save the context every time by calling task hook function? If so, how can I get the context of tasks? Any comments will be great appreciated. Ren Hantao --------------------------- Newsgroups: comp.os.vxworks Subject: adding Eth support on target PC? Date: Tue, 14 Aug 2001 10:44:30 -0400 From: Bruce Chrustie Organization: Nortel Message-ID: <3B79394E.D43D1A17@somewhereinhell.com> Hi folks, I had this problem about 6 months ago and never seemed to get it resolved to I abandoned my prject and now want to restart. Starting from scratch I have tornado 2.0 running under NT. My target is a 486 with a 3c509 ISA nic. I confogured the ISA nic with the DOS program. Now I want to build a load with support for the NIC and connect it to my host via a hub. Without adding any of my own code to the load I have done: - -used the default vxworks_uncmp and built a boot floppy - -from the vxworks project facility I have enabled support for the END driver, no support for the BSD version of the driver, is this OK? (I thought I need either BSD or END?) - -where do I define that I want the END 509 driver to be used? So I build my load and load it. I get the following error on the console: muxDevLoad failed for device entry 0! adding interface lo0...done Adding 2927 symbols for standalone. wdbConfig: error configuring WDB communication interfa VxWorks Copyright..... CPU version BSP WDB: Agent congfiguration failed. muxLoad failed (tRootTask) which leaves me frustrated and lost in space....I long to be working on my code as opposed to config issues :( Bruce, note my email address is mailto:chrustie@nortelnetworks.com --------------------------- Newsgroups: comp.os.vxworks Subject: Re: END Driver problems Date: Tue, 14 Aug 2001 17:34:08 -0700 From: Carlos Villalpando Organization: Jet Propulsion Laboratory - Pasadena CA Message-ID: <3B79C380.2C0003B3@brain.jpl.nasa.gov> References: <3B798E58.7D35A701@brain.jpl.nasa.gov> Carlos Villalpando wrote: > > Hey all. > > I'm having problems with Intel 82559er ethernet adapter and END drivers > under Tornado 2.02 on an Advanced Digital Logic P3 PC104+ board. Ok, still having problems, but as a follow on with more info: tcpdump -e (decode hardware address) shows the END drivers not sending to the destination hardware MAC address, but to the broadcast MAC address for all packets, while the BSD version correctly using the destination MAC address. The END drivers are building the ARP table ok, doing ARP who-has requests and filling the table. It just not using them. - --Carlos V. --------------------------- Newsgroups: comp.os.vxworks Subject: ISR/netTask problem Date: 14 Aug 2001 15:08:30 GMT From: Johan Borkhuis Organization: Agere Systems Message-ID: Hello, I am having a problem with an interrupt from an END driver. When I try pinging the target I see response times of up to 200 msec. When examining using Windview I can see that the interrupt from the card is executed when netTask starts. NetTask is running once every 200 msec, so that explains the delay of up to 200 msec in the ping. But what I am not getting is why the ISR is only executed when NetTask is being executed. There is no relation between these 2, I even use my own nettask-link task to deal with the data, but this makes no difference. Does this look familiar? If so, why is this happening, and how can I improve this? My target is an IDT MIPS processor. 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: Replacing the TCP/IP stack Date: Tue, 14 Aug 2001 15:22:01 +0200 From: "Werner Schiendl" Message-ID: <3b7926a5$0$26754$6e365a64@newsreader02.highway.telekom.at> References: <64c96d8c.0108101925.340830a3@posting.google.com> <9c7571e8.0108122046.31b064b5@posting.google.com> Hi, you could also use a download manager like GetRight (http://www.getright.com/). This particular one has a browser that presents all linked files of a page in a list and can download all of them or selected ones. hth werner "Dave Korn" wrote in message news:Pk8e7.14$ni3.32476@newsr2.u-net.net... > Chris Moore wrote in message ... > >Thanks for the link, do you know if there is a ZIP/tar file with all > >these, otherwise since its an HTTP file access I have to click each > >file/ save as etc etc. If there isn't I spend an afternoon doing this, > >but otherwise.... > > So, open lots of browser windows and start them all at once. > > Then take the rest of the afternoon off! > > DaveK > -- > " So far, science's crowning achievement, the Atomic Bomb, has not even > been able to alleviate the misery and torture of endless centuries of > world poverty." - George Hammond, self-proclaimed discoverer of the > scientific proof of god. > > --------------------------- Newsgroups: comp.os.vxworks Subject: Tornado installation OS Date: 14 Aug 2001 20:07:14 -0700 From: fengwk@bhnec.nec.co.jp (bhnecvxworks) Organization: http://groups.google.com/ Message-ID: Hi, Is there anyone who had compared the Tornado performance between Windows 2000 and Windows NT, which is more available? For in Windows 98, unexpected result happended frequently. So we want to reinstall Tornado in another OS, windows 2000 or windows NT? Regards Shen --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Replacing the TCP/IP stack Date: 13 Aug 2001 19:18:04 -0400 From: Bill Pringlemeir Organization: Bell Sympatico Message-ID: References: <64c96d8c.0108101925.340830a3@posting.google.com> <9c7571e8.0108122046.31b064b5@posting.google.com> Sender: bpringlemeir@DeadDuck Chris> BTW: Which stack is better Linux/NetBSD? I assume Linuxs' Chris> stack is based on the BSD stack @ some date. Perhaps at some point (2.0.x?), but the latest stack (2.4.x) is a complete zero-buffer re-write. I think that WRS also did some work on the BSD stack to create a zero buffer version. Also, the RouterWare code is quite good. RouterWare was bought out by WRS; it is much more than just a TCP/IP stack. It was possible to run the RouterWare code in a polled mode without an OS; which is quite quick and deterministic. hth, Bill Pringlemeir. - -- Have you ever wondered if it was possible to have your car make annoying phone calls to someone you saw at a party? Or tried to make your telephone generate a miasma? Or joined a church where everyone but you can become one with your telephone? You will. And the company that will bring it to you: AT&T. vxWorks FAQ, "http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html" --------------------------- Newsgroups: comp.os.vxworks Subject: END Driver problems Date: Tue, 14 Aug 2001 13:47:20 -0700 From: Carlos Villalpando Organization: Jet Propulsion Laboratory - Pasadena CA Message-ID: <3B798E58.7D35A701@brain.jpl.nasa.gov> Hey all. I'm having problems with Intel 82559er ethernet adapter and END drivers under Tornado 2.02 on an Advanced Digital Logic P3 PC104+ board. I've tried both BSD and END versions of the drivers downloaded from Intel (eeE) and the WindRiver '557 drivers (fei) (with a modified PCI_DEVICE_ID of 0x1209). The BSD drivers work great. However we need the END drivers. The END drivers refuse to talk to PCs. I can ping/telnet to and from Suns, SGIs, and Macs. However, I cannot ping or telnet to/from PCs running either Win2000 or Linux. I can sit on the W2k machine and run tcpdump and ping the target running the END driver. Tcpdump shows the echo request going out, the echo and the echo reply coming in from the target. However, ping doesn't receive it. icmpstatShow on the target shows that it is getting and sending ping requests/replies. I'm stumped. Why do only the PCs get the cold shoulders? - --Carlos V. --------------------------- Newsgroups: comp.os.vxworks Subject: HELP - vx FTP-Server problem! Date: Mon, 13 Aug 2001 16:59:04 +0200 From: Joachim Pluschke Organization: Stoye GmbH, =?iso-8859-1?Q?K=F6ln?= Message-ID: <3B77EB38.DBD20CFA@stoye.de> Hi all, we have a big problem with FTP-Servers coming with VxWorks on the TornadoII CDs (Host: WinNT, Target: PowerPC). DOES ANYBODY KNOW ABOUT A RELIABLE FTP-SERVER for VxWorks THAT really SUPPORTS DIRECTORIES??? Detail: ======= (We're using T2 / VxWorks 5.4 / PPC motorola 860/821 BSP / dosFS-2.) We get in trouble with access via FTP to a ata/ide device e.g. like ramdisk,flashdisk. (We tried two FTP Servers, one of them from the /unsupported area of T2 CD.) We loose a lot chains on disk overwriting contents of directories as we can see calling chkdsk() from dosFs2Lib after the overwrite-actions. We switched to use only a ramdisk, but the problem ws the same. We had a close look on the dosFS-2 / FTP-Servers behaviour. It seems to be a problem of the FTP-Server ... Seen effects: - ------------- Everytime we overwrite files somewhere in sub-directories (///files) the lost chains occur. Doing overwrites in the root always works fine! At first we started using wrs-supported ftp-server running it without this effect, but this one doesn't support directories (we need dirs!). We switched to the /unsupported section of code from the Tornado CD and found one FTP server that (seems to) support directories... This was fine for the first moment; but as we found out, this one is not really usable due to the lost chains!!!! This was reproduced with : WinNT/Win2000 hosts using Tools like WinCommander,WsFtp,MS-Ftp from the Prompt connected to : VxWorks Target, VxWorks 5.4, Kernel: WIND version 2.5, PPC860/821-BSP, DosFS-2. DOES ANYBODY KNOW ABOUT A RELIABLE FTP-SERVER for VxWorks THAT really SUPPORTS DIRECTORIES??? Any ideas would be appreciated. regards joachim - -- ===================================================== Gebrüder Stoye GmbH department: traffic control systems name: Joachim Pluschke adr: Longericher Strasse 177 city: Cologne, Germany zip: 50739 phone: ++49/221/1707013 fax: ++49/221/1707010 mailto: j.pluschke@stoye.de www: http://www.stoye.de ===================================================== --------------------------- Newsgroups: comp.os.vxworks Subject: Re: rs232 serial port driver Date: Tue, 14 Aug 2001 10:44:02 -0400 From: david lindauer Message-ID: <3B793932.69BAA76F@notifier-is.net> References: <9l897p$ei2$1@nereid.worldonline.nl> the easy way is to use the built-in RS232 drivers supplied with VXworks. You may have to modify your BSP to get this working, if using nonstandard hardware. But once you do you can do lots of high-level stuff with select and pipes and life is fairly simple. David Juno Kampstra wrote: > Hello, > Trying to make a driver for connecting to the rs232 port. > Already got the ms-windows implmentation running, but now > I'm searching for some vxwork (example) code. > The mswin driver opens a channel, and receives data from > the channel by using a seperate receive task, which is > create on open a channel, and removed on closing the > channel. > The same idea I want to use using vxwork: use ISR to put > data in a buffer, and let the receive task know new data > is present, and free the lock (which blocks the receive > task). > > Juno Kampstra. (juno.kampstra@fcs-cs.com) > FCS Control Systems --------------------------- Newsgroups: comp.os.vxworks Subject: Re: protocol stack design under vxworks Date: Tue, 14 Aug 2001 10:41:55 -0400 From: david lindauer Message-ID: <3B7938B3.35C1A15D@notifier-is.net> References: <414a7fa0.0108100756.425b1dee@posting.google.com> Jim Barton wrote: > We are currently trying to debug an ISDN protocol stack that was > recently ported from a Intel 186 platform (with no OS) to a PPC > platform that uses VxWorks. > > Since we are currently having major problems with the stability of the > stack and since we are new to VxWorks, we suspect that we may not have > done things correctly. For example, each protocol layer is now > represented as a separate task (each having the same priority of > 100)under vxWorks. Is this the normal way that a protocol stack should > be implemented in vxWorks? If not, how serious of a design problem is > this? Is it worthwhile re-designing this into one(?) task or could we > make this work with our current task per layer design? much ado is made about multitasking, but you really have to be careful with it or you get into all sorts of trouble with timing and such. For a protocol stack, I would be tempted to mostly write straight code , with each layer calling the next... the only problme with is if you stack is going to use lots of stack space. At the bottom you may need a task to service the hardware, so that you don't put unnecessary delays into whatever task called the stack to begin with. If you are going to stay with the multitasked scheme, make sure you are using message queues and or pipes for intertask communication, and not relying on in-memory buffers (you can also put semaphores around communication segments if you are determined to rewrite as little as possible, but I don't recommend that approach for design reasons). > > > Also, how is timing normally accomplished within a protocol stack > under VxWorks? In our old platform (without a OS) timing was > accomplished by a main loop that would service messages and events at > specific intervals to send out heartbeats, determine timeouts, etc. > I'm not clear what is the proper way of doing this under VxWorks. > Should we create timers and put all of our time-dependent stuff in the > SIGALRM handler? there are three methods that I know of: 1) watchdog. This is a vxworks construct that allows you to fire a service routine after a specific interval. I'm not sure I recommend it as I think the service routine is run in an interrupt context, so what you can do is limited... 2) make a task and use sleep(). For example for heartbeats, you could make a simple task that sleeps for 20 seconds, then fires off a heartbeat to whatever task is managing the protocol stack and goes back to sleep. 3) make a task and use select(). This can be used for simple timeouts, and has the added advantage that it will return prematurely if some type of I/O (which you specify) has occurred. David > > > And, yes, I know we could purchase a ISDN stack already designed for > VxWorks but we would rather avoid that expense since we already have > source for the stack from our older platform. We would appreciate any > tips on how we should port this protocol to vxWorks! > > Thanks, > > Jim --------------------------- Newsgroups: comp.os.vxworks Subject: Access Database Date: Tue, 14 Aug 2001 10:46:56 -0400 From: david lindauer Message-ID: <3B7939E0.149C6382@notifier-is.net> hi! is there any way to query a microsoft access database from VXWorks? Perhaps through ODBC or something? We are trying to decide whether it is worth the effort to access a database directly from an embedded target (to get login information) or whether we should build code into our PC hosts to do authorization indirectly. Thanks, David --------------------------- Newsgroups: comp.os.vxworks Subject: Change to uglinput.c? Date: 13 Aug 2001 11:27:32 -0400 From: Bill Pringlemeir Organization: Bell Sympatico Message-ID: Sender: bpringlemeir@DeadDuck Does this change make sense in "$(WIND_BASE)/target/src/ugl/event/uglevent.c"? 85c84 < (pEvent2->type.pointer.buttonChange == 0)) - --- > (pEvent1->type.pointer.buttonChange == 0)) I have used this patch in a system with a pointing device and it seems to behave well. However, I didn't see anything behaving strangely prior to the change. `pEvent2' doesn't make logical sense to me however. regards, Bill Pringlemeir. - -- vxWorks FAQ, "http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html" --------------------------- Newsgroups: comp.os.vxworks Subject: Re: adding Eth support on target PC? Date: Tue, 14 Aug 2001 21:47:27 -0700 From: DrDiags Organization: laserlink.net Message-ID: <3B79FEDF.BAACA574@covad.net> References: <3B79394E.D43D1A17@somewhereinhell.com> Reply-To: drdiags@covad.net Bruce, I just looked in the config.h for an pc486 BSP and if you #define INCLUDE_ELT, you will pick up the END driver for the 3c509 (Though I am not sure whether it is ISA or PCI), see the following: # ifdef INCLUDE_ELT # undef INCLUDE_ELT /* don't use BSD driver */ # define INCLUDE_ELT_3C509_END /* Use END driver instead */ # endif The other item of interest is the default vxworks_uncmp file you said you used. I imagine it is a vxworks.st image, but the real question is whether when you stop the boot countdown and type the "h"elp command, what type of boot devices are listed at the bottom of the screen. If you see the elt0 device, then your image has built in support for your NIC. Look in your BSP for the configNet.h file and determine the name of the END driver load routine. You could instrument it to determine what is going on, but the muxDevLoad failure of the first END driver found is one of your problems. Check the config.h file (or the listing at the bottom of your help screen) because you may be trying to load several NICs that are not present in the system and thus your network init falls apart. The vxWorks FAQ should have some other insights on using STANDALONE and the tricks you have to use to make sure the network is initialize, but first you need to prove to yourself that the bootup code supports the NIC you are using and that it is the first listed in the endDevTbl[] found in the configNet.h file at the bottom. I would edit the config.h file and make sure this is the only NIC #defined. I would then rebuild the vxWorks.st or vxWorks_uncmp (you must be building from the IDE for this image). If all else fails, try submitting a TSR to Wind River. You might as well pass on your frustration to someone paid to handle this type of problem. Since you put this off for 6 months, it is probably not something you need resolved right now, so relax and let's see what you can find if any of these suggestions help in getting you started. BTW, I am sure you know the link to the vxWorks FAQ, but it is listed at the bottom..... HTH. Bruce Chrustie wrote: > > Hi folks, > > I had this problem about 6 months ago and never seemed to get it > resolved to I abandoned my prject and now want to restart. > > Starting from scratch I have tornado 2.0 running under NT. My target is > a 486 with a 3c509 ISA nic. I confogured the ISA nic with the DOS > program. Now I want to build a load with support for the NIC and connect > it to my host via a hub. > > Without adding any of my own code to the load I have done: > > -used the default vxworks_uncmp and built a boot floppy > -from the vxworks project facility I have enabled support for the END > driver, no support for the BSD version of the driver, is this OK? (I > thought I need either BSD or END?) > -where do I define that I want the END 509 driver to be used? > > So I build my load and load it. I get the following error on the > console: > > muxDevLoad failed for device entry 0! > adding interface lo0...done > > Adding 2927 symbols for standalone. > wdbConfig: error configuring WDB communication interfa > > VxWorks > > Copyright..... > > CPU > version > BSP > WDB: Agent congfiguration failed. > > muxLoad failed (tRootTask) > > which leaves me frustrated and lost in space....I long to be working on > my code as opposed to config issues :( > > Bruce, > > note my email address is mailto:chrustie@nortelnetworks.com - -- http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html (vxWorks FAQ home page) --------------------------- Newsgroups: comp.os.vxworks Subject: Re: END Driver problems Date: Tue, 14 Aug 2001 21:59:44 -0700 From: DrDiags Organization: laserlink.net Message-ID: <3B7A01C0.EC212C94@covad.net> References: <3B798E58.7D35A701@brain.jpl.nasa.gov> <3B79C380.2C0003B3@brain.jpl.nasa.gov> Reply-To: drdiags@covad.net Carlos, Don't have an answer, but I understand that T2.0.2 is a major change in the WRS network stack. They reworked and changed the API for their DHCP code picking up support for a couple of RFCs and other changes. They also have moved to the NPT model (NPT Drivers are enhanced END drivers) to allow non-ethernet (?) interfaces to use the MUX. Maybe this could be a side affect of these changes that should be reported to WRS to see if this has occurred with others developers. If not, you are the first to report it and hopefully the test case will lead to a fix for your problem, if it is a bug. Carlos Villalpando wrote: > > Carlos Villalpando wrote: > > > > Hey all. > > > > I'm having problems with Intel 82559er ethernet adapter and END drivers > > under Tornado 2.02 on an Advanced Digital Logic P3 PC104+ board. > > Ok, still having problems, but as a follow on with more info: > > tcpdump -e (decode hardware address) shows the END drivers not sending > to the destination hardware MAC address, but to the broadcast MAC > address for all packets, while the BSD version correctly using the > destination MAC address. > > The END drivers are building the ARP table ok, doing ARP who-has > requests and filling the table. It just not using them. > > --Carlos V. - -- http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html (vxWorks FAQ home page) --------------------------- Newsgroups: comp.os.vxworks Subject: Re: How can i know whether the network cable is plugged in or not under Vxworks? Date: Tue, 14 Aug 2001 09:06:36 -0700 From: "Dan Gold" Message-ID: <3b794d05$0$90634$e2e8da3@nntp.cts.com> References: <9la2pq$5pn$1@mail.cn99.com> "zzxt" wrote in message news:9la2pq$5pn$1@mail.cn99.com... > Hi all: > Can i install a ISR to deal with cable disconnect and reconnect or Is > there a register where i can read the status of the network cable. Is this > feature specific to NIC. Both END and BSD solutions are needed. thanks very > much This is totally dependent on the PHY device you are using. Most modern PHYs have a set of status and control registers you can read/write via the MDIO interface. Most of them are also capable of generating an interrupt based on various conditions (link establishment or not, auto-negotiation complete, etc...). You need to find out what PHY device you are using, if you have an MDIO interface (2 wires, clock and data) to the part, and if you can configure it to generate interrupts. hth, Dan Gold gold@ensemble.com --------------------------- Newsgroups: comp.os.vxworks Subject: Re: OT: Changing root directory in Tornado2? Date: 14 Aug 2001 16:09:26 GMT From: Johan Borkhuis Organization: Agere Systems Message-ID: References: <9l99ue$4vd@nntpb.cb.lucent.com> "Joe Varghese" wrote: > Not sure if this is the proper forum or not, but is there a way to > change the working root directory for a Tornado 2 workspace (i.e., a > .wsp file)? > > Our group is trying to use the .wsp as a common project file, shared by > multiple people. However, when we check the .wsp file into source > control, the absolute path of that particular person's directory is > stored in the workspace. Yes, if you look in the WSP file you see a absolute path. This can be replaced by a variable, for example $WIND_BASE. You can do this automatically, but you have to change the TCL files for this. Groeten, Johan - -- o o o o o o o . . . _____________________________ o _____ || Johan Borkhuis | .][__n_n_|DD[ ====_____ | borkhuis@agere.com | >(________|__|_[_________]_|__________________________| _/oo OOOOO oo` ooo ooo 'o!o!o o!o!o` === VxWorks FAQ: http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html === --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Tornado installation OS Date: Tue, 14 Aug 2001 22:10:40 -0700 From: DrDiags Organization: laserlink.net Message-ID: <3B7A0450.352774B7@covad.net> References: Reply-To: drdiags@covad.net Shen, The comments I would give is that Windows 2000 support is just starting to settle in, since it was not officially supported until earlier this year. There are some Tech Tips on WindSurf showing what Service Pack on Windows 2000 align with installation of WRS patches and products. One particular problem is the requirement of running the target server in realtime mode, otherwise some of the tools behave erratically, like the debugger. Most folks I talk to have high marks for Windows 2000, but I mainly use WinNT. Of course, others I know would like to see T2.x products backported to run on Linux machines, since compilation times are greatly reduced, but I guess you didn't ask for that input.... Just my 1/2 cents worth.. bhnecvxworks wrote: > > Hi, > Is there anyone who had compared the Tornado performance between > Windows 2000 and Windows NT, > which is more available? > For in Windows 98, unexpected result happended frequently. > So we want to reinstall Tornado in another OS, windows 2000 or windows NT? > > Regards > Shen - -- http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html (vxWorks FAQ home page) --------------------------- Newsgroups: comp.os.vxworks Subject: Long Boot Time (mvme2431) Date: Mon, 13 Aug 2001 18:04:09 GMT From: "Tim Ellis" Organization: Verio Message-ID: VxWorks, Tornado2.02 Has anyone encountered lengthy boot up time using the mv2400 bsp? (network boot on the dc interface) I'm currently migrating an application from the mv2300 cpu to the mv2400 cpu. On the mv2300 it takes approx 30 seconds to network boot. (timed from reset to vxworks prompt) On the mv2400 it takes approx 2.5 minutes to network boot. (if feels like forever) I believe that the extra delay is happening in muxDevStart(). A delay of approx 1 minute occurs at the muxDevStart() in the boot_rom code and at the muxDevStart() in usrEndLib.c. (timed by sandwiching muxDevStart with printf's) That's as far as I can go, without the source for the dec21x40 start routine. Any idea on what could be causing this delay? Thanks, Tim --------------------------- Newsgroups: comp.os.vxworks Subject: OT: Changing root directory in Tornado2? Date: Mon, 13 Aug 2001 14:31:38 -0500 From: "Joe Varghese" Organization: Lucent Technologies, Columbus, Ohio Message-ID: <9l99ue$4vd@nntpb.cb.lucent.com> Not sure if this is the proper forum or not, but is there a way to change the working root directory for a Tornado 2 workspace (i.e., a .wsp file)? Our group is trying to use the .wsp as a common project file, shared by multiple people. However, when we check the .wsp file into source control, the absolute path of that particular person's directory is stored in the workspace. The alternate solution is to have all the team members work out of the same directory on their machine (e.g., c:/src for everyone), but hoping there's a more elegant solution .... Thanks in advance! joe varghese --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Changing root directory in Tornado2? Date: Wed, 15 Aug 2001 00:48:24 -0700 From: "Denis" Organization: University of California, Los Angeles Message-ID: <9ld9d9$ec1$1@siamese.noc.ucla.edu> References: <9l99ue$4vd@nntpb.cb.lucent.com> <9l9iko$na4$06$1@news.t-online.com> "Martin Raabe" wrote in message news:9l9iko$na4$06$1@news.t-online.com... > Hello Joe, > try the team development approach by using SNiFf+ as the SDE. > > For more info, look at www.windriver.com/sniff > > Ciao > Martin > > Speaking for my own, not for my employer. that was pretty funny :) considering ... Martin Raabe Project Manager Integration of SNiFF+ in Tornado II and pRISM+ TakeFive Software GmbH , a Wind River Company that really did not sound like you were speaking for yourself there. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Changing root directory in Tornado2? Date: Tue, 14 Aug 2001 07:28:34 GMT From: "Pär Sjöholm" Organization: Telia Internet Message-ID: References: <9l99ue$4vd@nntpb.cb.lucent.com> We use the same approach as you describe. But in our .wsp there is no absolute path stored on check in. There is a $(PRJ_LIB), that seems to do the job. I haven't actively thied to figure out how it works, but it does. We get into trouble with SlickEdit, which does store a absolute path to the source. We use SourceSafe. Can't win them all... /Paer Sjoeholm "Joe Varghese" wrote in message news:9l99ue$4vd@nntpb.cb.lucent.com... > Not sure if this is the proper forum or not, but is there a way to change > the working root directory for a Tornado 2 workspace (i.e., a .wsp file)? > > Our group is trying to use the .wsp as a common project file, shared by > multiple people. However, when we check the .wsp file into source control, > the absolute path of that particular person's directory is stored in the > workspace. > > The alternate solution is to have all the team members work out of the same > directory on their machine (e.g., c:/src for everyone), but hoping there's a > more elegant solution .... > > Thanks in advance! > > joe varghese > > --------------------------- Newsgroups: comp.os.vxworks Subject: stripping symbols using objcopyppc Date: 13 Aug 2001 11:52:18 -0700 From: asriniv1@email.mot.com (Arun Srinivasan) Organization: http://groups.google.com/ Message-ID: Hi, I am trying to use objcopyppc to strip symbols off objects. These objects work internally, i.e other objects do not invoke functions in them. So I tried just leaving the undefined symbols in the object and taskMain, so that when the object is loaded up, the undefined symbols will get resolved ( as other objects loaded before this one have those symbols defined in them). But this did not work right. Apparently, there are more symbols that I need to leave behind for the object to execute without crashing on the target. Is there anyway I can find out whether a symbol is necessary or not for the object to execute successfully on the target ?- which in my case is a elf32-powerpc target. I first used nmppc to find out the symbols in the object, and then used objcopyppc to strip symbols off. Since leaving the undefined symbols alone in the object did not help, I tried leaving text symbols (denoted by type T in the nmppc output) also - in addition to undefined symbols but even that did not work. Apparently, there were symbols that needed to be in the object which I was missing. I have heard that stripping symbols off objects can result in a lot of space saving, but atleast in my case it does not seem to be trivial. Any pointers to find out symbols that NEED to be in the object for it to execute successfully will be greatly appreciated. Is there any other Windriver commands that I can use which might simplify my problem? Hope I was clear on this. Thanks! Arun Srinivasan Software Engineer Motorola Inc. --------------------------- Newsgroups: comp.os.vxworks Subject: How to calculate CPU Usage percent in Vxworks Date: 15 Aug 2001 01:29:20 -0700 From: srdswj@sbell.com.cn (Sun Wenjian) Organization: http://groups.google.com/ Message-ID: <14c57c4f.0108150029.7e29b0e0@posting.google.com> Is any one has any idea to calculate CPU utilization in vxworks OS? thanks a lot. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Changing root directory in Tornado2? Date: Wed, 15 Aug 2001 10:27:23 +0200 From: "Martin Raabe" Organization: T-Online Message-ID: <9ldbp8$nj3$07$1@news.t-online.com> References: <9l99ue$4vd@nntpb.cb.lucent.com> <9l9iko$na4$06$1@news.t-online.com> <9ld9d9$ec1$1@siamese.noc.ucla.edu> Hello Denis, since it does not sound like, I explicitely express it here! I am speaking for my own, despite my position within WindRiver. So the signature you found is older than one year isn't it? Ciao Martin Raabe Speaking for my own, not for my employer! "Denis" schrieb im Newsbeitrag news:9ld9d9$ec1$1@siamese.noc.ucla.edu... > > "Martin Raabe" wrote in message > news:9l9iko$na4$06$1@news.t-online.com... > > Hello Joe, > > try the team development approach by using SNiFf+ as the SDE. > > > > For more info, look at www.windriver.com/sniff > > > > Ciao > > Martin > > > > Speaking for my own, not for my employer. > > that was pretty funny :) considering ... > > Martin Raabe > Project Manager > Integration of SNiFF+ in Tornado II and pRISM+ > TakeFive Software GmbH , a Wind River Company > > that really did not sound like you were speaking for yourself there. > > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: OT: Changing root directory in Tornado2? Date: Mon, 13 Aug 2001 20:47:57 +0100 From: molochai Message-ID: References: <9l99ue$4vd@nntpb.cb.lucent.com> In message <9l99ue$4vd@nntpb.cb.lucent.com> "Joe Varghese" wrote: > > The alternate solution is to have all the team members work out of the same > directory on their machine (e.g., c:/src for everyone), but hoping there's > a more elegant solution .... This *is* IMHO an elegent approach. We have user accounts mapped onto the same place on each developer's workstation and this is the working area, This also ensures that user-work-space gets reliably backed up. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: How to use the Intel 82559ER Lan Controller in the vxworks system? Date: Mon, 13 Aug 2001 20:41:20 +0100 From: molochai Message-ID: References: <3b6fcc9f$0$32924$6e365a64@newsreader02.highway.telekom.at> <3b72661a$0$33474$6e365a64@newsreader02.highway.telekom.at> In message srs_zh@263.net (srs_zh) wrote: > Where can I get the soucr codes of the intel82559er from? You don't get source code to either Intel's or WRS's 82557/8/9 driver. The WRS varient is supplied in object code form with Tornado/VxWorks. You might be able to get source code from an WRS FAE if you ask nicely and keep your support paid-up. However, I doubt the source code will help you much in this case. You might have a sysFei557End.c in your BSP which is a small amount of glue needed to find the device on the PCI bus and start the driver with the correct parameters. I suspect it is this that you need. M --------------------------- Newsgroups: comp.os.vxworks Subject: Re: How to calculate CPU Usage percent in Vxworks Date: Wed, 15 Aug 2001 16:46:47 +0800 From: "tanghaoyu" Organization: Bentium Ltd. (CN99) Message-ID: <9ldcun$hr5$1@mail.cn99.com> References: <14c57c4f.0108150029.7e29b0e0@posting.google.com> you can call some functions in spylib. "Sun Wenjian" wrote in message news:14c57c4f.0108150029.7e29b0e0@posting.google.com... > Is any one has any idea to calculate CPU utilization in vxworks OS? > thanks a lot. > --------------------------- Newsgroups: comp.os.vxworks Subject: How to delete/reconfig a network interface? Date: Wed, 15 Aug 2001 17:14:22 +0800 From: "survey liu" Organization: Bentium Ltd. (CN99) Message-ID: <9ldeek$im3$1@mail.cn99.com> Hi,all I have configed one of my target board on the backplane to support Shared-Memory-Network.It works well. But, how could I delete/reconfig the "sm0" network interface while I dont need it anymore or while I want to attach the "sm0" to another Shared-Memory block? I searched all the manual about vxWorks, but not any hints are mentioned. Anyway, my question is just how to delete/reconfig a network interface, not only sm, but also any others like Ethernet,PPP,etc.. Thanks in advance - -- Best regards, Survey Liu 4/F Wearnes Technology Bldg. No.10 Kefa Rd.,Science & Industry Park Nanshan Shenzhen 518057,PRC Tel: +86-755-6635333-4300 Fax: +86-755-6635550 Email: liuxw@263.net --------------------------- Newsgroups: comp.os.vxworks Subject: verify new user Date: 15 Aug 2001 02:18:18 -0700 From: kevinatnj@yahoo.com.cn (kewin) Organization: http://groups.google.com/ Message-ID: <221caa76.0108150118.44e13575@posting.google.com> hi all --------------------------- Newsgroups: comp.os.vxworks Subject: debug ppc860 BSP Date: 15 Aug 2001 02:30:17 -0700 From: kevinatnj@yahoo.com.cn (kewin) Organization: http://groups.google.com/ Message-ID: <221caa76.0108150130.27295133@posting.google.com> hi all i need your generous help. The following problem have bothered me for a long time. i am debuging my vxWorks BSP for ppc960 using the software, singlestep.But i can't see the source programs of my BSP,only assembly language appearing in the singlestep environment.How can i added debugging information to my BSP? And I have added "-g" in the file ,\tornado\target\h\make\defs.bsp,to the corresponding FLAGS,CFLAGS,CASFLAGS and CFLAGS_AS. But I still can't see the source program. Will anybody tell me what the reason is and how i can make it. thanks a lot for your generous help. --------------------------- End of New-News digest ********************** From vxwexplo-errs@csg.lbl.gov Wed Aug 15 08:24:34 2001 From: fscheurer@manz-acs.de Date: Wed Aug 15 08:24:36 PDT 2001 Subject: AUTOREPLY comp.os.vxworks newsdigest 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_000_00017D29.3B7AB028 Content-Type: text/plain; charset=iso-8859-1 Content-Transfer-Encoding: 7Bit This is the Mailbox of Friedrich Scheurer. R&D - ACS GmbH - D-72768 Reutlingen Sorry, I am out of office until August 20th 2001! In urgent cases please contact one of the following persons at ACS GmbH: Otto Angerhofer (Managing director) e-mail: oangerhofer@manz-acs.de Eduard Ams (Technical manager) e-mail: eams@manz-acs.de Yvonne Anhorn (Marketing) e-mail: yanhorn@manz-acs.de Margit Arnold (Sales) e-mail: marnold@manz-acs.de Christina Sixel (Sales) e-mail: csixel@manz-acs.de MfG / Kind Regards F. Scheurer ---------------------------------------------------------------- Friedrich Scheurer, Dipl.-Ing. phone +49 7121 9019-23(0) ACS GmbH - R&D fax +49 7121 9019-98 Steigaeckerstrasse 13 web www.manz-acs.de D-72768 Reutlingen (Germany) e-mail: fscheurer@manz-acs.de ---------------------------------------------------------------- ------_=_NextPart_000_00017D29.3B7AB028 Content-Type: text/html; charset=iso-8859-1 Content-Transfer-Encoding: quoted-printable
This is the Mailbox of Friedrich Scheurer.
R&D - ACS GmbH - D-72768 Reutlingen
 
Sorry, I am out of office until August 20th 2001!
 
In urgent cases please contact one of the
following persons at ACS GmbH:
 
Otto Angerhofer (Managing director)
Eduard Ams (Technical manager)
Yvonne Anhorn (Marketing)
Margit Arnold (Sales)
e-mail: marnold@manz-acs.de
Christina Sixel (Sales)

 
MfG / Kind Regards
 
F. Scheurer
 
----------------------------------------------------------------
Friedrich Scheurer, Dipl.-Ing.    phone    +49 7121 9019-23(0)
ACS GmbH  -  R&D                  fax      +49 7121 9019-98
Steigaeckerstrasse 13             web      www.manz-acs.de
D-72768 Reutlingen (Germany)      e-mail:  fscheurer@manz-acs.de
----------------------------------------------------------------
------_=_NextPart_000_00017D29.3B7AB028-- From vxwexplo-errs@csg.lbl.gov Wed Aug 15 08:58:18 2001 From: "Michael Morrison" Date: Wed Aug 15 08:58:21 PDT 2001 Subject: fcntl() for VxWorks? Hi folks What is the equivalent for fcntl() in vxworks. I need to set a socket to non-blocking mode. Thanks, MM From vxwexplo-errs@csg.lbl.gov Wed Aug 15 10:24:21 2001 From: "Clark, Mark" Date: Wed Aug 15 10:24:29 PDT 2001 Subject: RE: comp.os.vxworks newsdigest VXWorks Sure looking forward to August 20th. :) From vxwexplo-errs@csg.lbl.gov Wed Aug 15 11:15:22 2001 From: fscheurer@manz-acs.de Date: Wed Aug 15 11:15:25 PDT 2001 Subject: AUTOREPLY fcntl() for VxWorks? This message is in MIME format. Since your mail reader does not understand this format, some or all of this message may not be legible. ------_=_NextPart_000_00017D79.3B7AD81B Content-Type: text/plain; charset=iso-8859-1 Content-Transfer-Encoding: 7Bit This is the Mailbox of Friedrich Scheurer. R&D - ACS GmbH - D-72768 Reutlingen Sorry, I am out of office until August 20th 2001! In urgent cases please contact one of the following persons at ACS GmbH: Otto Angerhofer (Managing director) e-mail: oangerhofer@manz-acs.de Eduard Ams (Technical manager) e-mail: eams@manz-acs.de Yvonne Anhorn (Marketing) e-mail: yanhorn@manz-acs.de Margit Arnold (Sales) e-mail: marnold@manz-acs.de Christina Sixel (Sales) e-mail: csixel@manz-acs.de MfG / Kind Regards F. Scheurer ---------------------------------------------------------------- Friedrich Scheurer, Dipl.-Ing. phone +49 7121 9019-23(0) ACS GmbH - R&D fax +49 7121 9019-98 Steigaeckerstrasse 13 web www.manz-acs.de D-72768 Reutlingen (Germany) e-mail: fscheurer@manz-acs.de ---------------------------------------------------------------- ------_=_NextPart_000_00017D79.3B7AD81B Content-Type: text/html; charset=iso-8859-1 Content-Transfer-Encoding: quoted-printable
This is the Mailbox of Friedrich Scheurer.
R&D - ACS GmbH - D-72768 Reutlingen
 
Sorry, I am out of office until August 20th 2001!
 
In urgent cases please contact one of the
following persons at ACS GmbH:
 
Otto Angerhofer (Managing director)
Eduard Ams (Technical manager)
Yvonne Anhorn (Marketing)
Margit Arnold (Sales)
e-mail: marnold@manz-acs.de
Christina Sixel (Sales)

 
MfG / Kind Regards
 
F. Scheurer
 
----------------------------------------------------------------
Friedrich Scheurer, Dipl.-Ing.    phone    +49 7121 9019-23(0)
ACS GmbH  -  R&D                  fax      +49 7121 9019-98
Steigaeckerstrasse 13             web      www.manz-acs.de
D-72768 Reutlingen (Germany)      e-mail:  fscheurer@manz-acs.de
----------------------------------------------------------------
------_=_NextPart_000_00017D79.3B7AD81B-- From vxwexplo-errs@csg.lbl.gov Wed Aug 15 12:12:08 2001 From: fscheurer@manz-acs.de Date: Wed Aug 15 12:12:10 PDT 2001 Subject: AUTOREPLY RE: comp.os.vxworks newsdigest 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_000_00017D81.3B7AE56D Content-Type: text/plain; charset=iso-8859-1 Content-Transfer-Encoding: 7Bit This is the Mailbox of Friedrich Scheurer. R&D - ACS GmbH - D-72768 Reutlingen Sorry, I am out of office until August 20th 2001! In urgent cases please contact one of the following persons at ACS GmbH: Otto Angerhofer (Managing director) e-mail: oangerhofer@manz-acs.de Eduard Ams (Technical manager) e-mail: eams@manz-acs.de Yvonne Anhorn (Marketing) e-mail: yanhorn@manz-acs.de Margit Arnold (Sales) e-mail: marnold@manz-acs.de Christina Sixel (Sales) e-mail: csixel@manz-acs.de MfG / Kind Regards F. Scheurer ---------------------------------------------------------------- Friedrich Scheurer, Dipl.-Ing. phone +49 7121 9019-23(0) ACS GmbH - R&D fax +49 7121 9019-98 Steigaeckerstrasse 13 web www.manz-acs.de D-72768 Reutlingen (Germany) e-mail: fscheurer@manz-acs.de ---------------------------------------------------------------- ------_=_NextPart_000_00017D81.3B7AE56D Content-Type: text/html; charset=iso-8859-1 Content-Transfer-Encoding: quoted-printable
This is the Mailbox of Friedrich Scheurer.
R&D - ACS GmbH - D-72768 Reutlingen
 
Sorry, I am out of office until August 20th 2001!
 
In urgent cases please contact one of the
following persons at ACS GmbH:
 
Otto Angerhofer (Managing director)
Eduard Ams (Technical manager)
Yvonne Anhorn (Marketing)
Margit Arnold (Sales)
e-mail: marnold@manz-acs.de
Christina Sixel (Sales)

 
MfG / Kind Regards
 
F. Scheurer
 
----------------------------------------------------------------
Friedrich Scheurer, Dipl.-Ing.    phone    +49 7121 9019-23(0)
ACS GmbH  -  R&D                  fax      +49 7121 9019-98
Steigaeckerstrasse 13             web      www.manz-acs.de
D-72768 Reutlingen (Germany)      e-mail:  fscheurer@manz-acs.de
----------------------------------------------------------------
------_=_NextPart_000_00017D81.3B7AE56D-- From vxwexplo-errs@csg.lbl.gov Wed Aug 15 13:06:44 2001 From: "lee" Date: Wed Aug 15 13:06:47 PDT 2001 Subject: Re: comp.os.vxworks newsdigest VXWorks > Sure looking forward to August 20th. :) Yup-per-dee-doo. > VXWorks > From vxwexplo-errs@csg.lbl.gov Wed Aug 15 14:18:43 2001 From: "lee" Date: Wed Aug 15 14:18:45 PDT 2001 Subject: RE: comp.os.vxworks newsdigest > Sure looking forward to August 20th. :) Yup. From vxwexplo-errs@csg.lbl.gov Wed Aug 15 15:13:26 2001 From: fscheurer@manz-acs.de Date: Wed Aug 15 15:13:28 PDT 2001 Subject: AUTOREPLY Re: comp.os.vxworks newsdigest 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_000_00017D87.3B7B0FFC Content-Type: text/plain; charset=iso-8859-1 Content-Transfer-Encoding: 7Bit This is the Mailbox of Friedrich Scheurer. R&D - ACS GmbH - D-72768 Reutlingen Sorry, I am out of office until August 20th 2001! In urgent cases please contact one of the following persons at ACS GmbH: Otto Angerhofer (Managing director) e-mail: oangerhofer@manz-acs.de Eduard Ams (Technical manager) e-mail: eams@manz-acs.de Yvonne Anhorn (Marketing) e-mail: yanhorn@manz-acs.de Margit Arnold (Sales) e-mail: marnold@manz-acs.de Christina Sixel (Sales) e-mail: csixel@manz-acs.de MfG / Kind Regards F. Scheurer ---------------------------------------------------------------- Friedrich Scheurer, Dipl.-Ing. phone +49 7121 9019-23(0) ACS GmbH - R&D fax +49 7121 9019-98 Steigaeckerstrasse 13 web www.manz-acs.de D-72768 Reutlingen (Germany) e-mail: fscheurer@manz-acs.de ---------------------------------------------------------------- ------_=_NextPart_000_00017D87.3B7B0FFC Content-Type: text/html; charset=iso-8859-1 Content-Transfer-Encoding: quoted-printable
This is the Mailbox of Friedrich Scheurer.
R&D - ACS GmbH - D-72768 Reutlingen
 
Sorry, I am out of office until August 20th 2001!
 
In urgent cases please contact one of the
following persons at ACS GmbH:
 
Otto Angerhofer (Managing director)
Eduard Ams (Technical manager)
Yvonne Anhorn (Marketing)
Margit Arnold (Sales)
e-mail: marnold@manz-acs.de
Christina Sixel (Sales)

 
MfG / Kind Regards
 
F. Scheurer
 
----------------------------------------------------------------
Friedrich Scheurer, Dipl.-Ing.    phone    +49 7121 9019-23(0)
ACS GmbH  -  R&D                  fax      +49 7121 9019-98
Steigaeckerstrasse 13             web      www.manz-acs.de
D-72768 Reutlingen (Germany)      e-mail:  fscheurer@manz-acs.de
----------------------------------------------------------------
------_=_NextPart_000_00017D87.3B7B0FFC-- From vxwexplo-errs@csg.lbl.gov Wed Aug 15 16:21:18 2001 From: fscheurer@manz-acs.de Date: Wed Aug 15 16:21:21 PDT 2001 Subject: AUTOREPLY RE: comp.os.vxworks newsdigest 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_000_00017DAF.3B7B1FEB Content-Type: text/plain; charset=iso-8859-1 Content-Transfer-Encoding: 7Bit This is the Mailbox of Friedrich Scheurer. R&D - ACS GmbH - D-72768 Reutlingen Sorry, I am out of office until August 20th 2001! In urgent cases please contact one of the following persons at ACS GmbH: Otto Angerhofer (Managing director) e-mail: oangerhofer@manz-acs.de Eduard Ams (Technical manager) e-mail: eams@manz-acs.de Yvonne Anhorn (Marketing) e-mail: yanhorn@manz-acs.de Margit Arnold (Sales) e-mail: marnold@manz-acs.de Christina Sixel (Sales) e-mail: csixel@manz-acs.de MfG / Kind Regards F. Scheurer ---------------------------------------------------------------- Friedrich Scheurer, Dipl.-Ing. phone +49 7121 9019-23(0) ACS GmbH - R&D fax +49 7121 9019-98 Steigaeckerstrasse 13 web www.manz-acs.de D-72768 Reutlingen (Germany) e-mail: fscheurer@manz-acs.de ---------------------------------------------------------------- ------_=_NextPart_000_00017DAF.3B7B1FEB Content-Type: text/html; charset=iso-8859-1 Content-Transfer-Encoding: quoted-printable
This is the Mailbox of Friedrich Scheurer.
R&D - ACS GmbH - D-72768 Reutlingen
 
Sorry, I am out of office until August 20th 2001!
 
In urgent cases please contact one of the
following persons at ACS GmbH:
 
Otto Angerhofer (Managing director)
Eduard Ams (Technical manager)
Yvonne Anhorn (Marketing)
Margit Arnold (Sales)
e-mail: marnold@manz-acs.de
Christina Sixel (Sales)

 
MfG / Kind Regards
 
F. Scheurer
 
----------------------------------------------------------------
Friedrich Scheurer, Dipl.-Ing.    phone    +49 7121 9019-23(0)
ACS GmbH  -  R&D                  fax      +49 7121 9019-98
Steigaeckerstrasse 13             web      www.manz-acs.de
D-72768 Reutlingen (Germany)      e-mail:  fscheurer@manz-acs.de
----------------------------------------------------------------
------_=_NextPart_000_00017DAF.3B7B1FEB-- From vxwexplo-errs@csg.lbl.gov Thu Aug 16 04:03:36 2001 From: Vxworks Exploder Date: Thu Aug 16 04:03:39 PDT 2001 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Thu Aug 16 04:03:15 PDT 2001 Subject: Personal Java or CORBA on VxWorks? Subject: Re: networking issues Subject: Re: debug ppc860 BSP Subject: Re: adding Eth support on target PC? Subject: Re: stripping symbols using objcopyppc Subject: Re: debug ppc860 BSP Subject: Re: How to calculate CPU Usage percent in Vxworks Subject: Re: Changing root directory in Tornado2? Subject: System V Packaging and VxWorks Subject: WindML 2.0/Zinc 6.0, key loss. Subject: Re: How to calculate CPU Usage percent in Vxworks Subject: Re: WindML 2.0/Zinc 6.0, key loss. Subject: Re: semBCreate() - Semaphore Creation is failing Subject: Any command to find how much of the STACK is full/empty?? Subject: how to use objdump386 Subject: Re: how to use objdump386 Subject: Re: Any command to find how much of the STACK is full/empty?? Subject: Re: debug ppc860 BSP Subject: Re: how to use objdump386 Subject: pagefaults during startup if network is plugged in Subject: Re: adding Eth support on target PC? Subject: Isn't there IP_ONESBCAST. Subject: Multiple Initiator and one HDD Subject: Re: Any command to find how much of the STACK is full/empty?? Subject: can timer_create() and timer_connect() be called in an ISR? Subject: Re: How can i know whether the network cable is plugged in or not under Vxworks? Subject: Re: adding Eth support on target PC? Subject: Re: adding Eth support on target PC? Subject: Isn't there "IP_ONESBCAST"? Subject: Serial IO and RTS/CTS Subject: Re: Any command to find how much of the STACK is full/empty?? Subject: Re: RPC File Descriptor Count Question ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Personal Java or CORBA on VxWorks? Date: Wed, 15 Aug 2001 14:17:17 +0200 From: "Anne B. Knudtzon" Organization: Schlumberger,Geco-Prakla Message-ID: <3B7A684D.DB7C8115@oslo.geco-prakla.slb.com> Hei! We are developing a new generation of our seismic network, i.e. a high throughput, time critical, low power network for field usage. Our network nodes will most likely still be based on a PPC with VxWorks. We have not yet decided whether to use Java (Personal Jworks) or C++ for the not so time critical functionality. Has any of you tried WindRiver's Jworks Personal Java VM? Is it nice? We're also considering using CORBA for the network management. Any experience with embedded CORBA? Any ORBs to recommend? Has anyone experience with e*ORB? The ORB that looks best suited for us is e*ORB from Vertel - at least it's the only I've found that runs on WRS Jworks. Looking forward to hearing from you! Regards, Anne --------------------------- Newsgroups: comp.os.vxworks Subject: Re: networking issues Date: Wed, 15 Aug 2001 09:11:46 -0400 From: Jacque Madore Organization: Raytheon Company Message-ID: <3B7A7512.676E4A60@a.com> References: <3B69B897.BBA85C91@improvsys.com> <488e459a.0108022140.149341fc@posting.google.com> Try holding the network i/f in reset first thing once booted. Perhaps the frames are arriving and unitialized handlers are being invoked. IfLib also has a routine to change the IP address at runtime.... - ------ John wrote: > Hello Jeremy, > > > The issue at hand is that on my MIPS Malta card, usrNetInit will > > sparadically cause the system to hang when it is called from my code. > > Basically, the system waits for the user to choose an IP address, then > > using usrNetInit, initializes the system with the specified IP. > > Unfortunately, this sometimes causes the system to seemingly lock up > > immediately after the following lines: > > > > Attached TCP/IP interface to lnPci unit 0 > > Attaching network interface lo0... done. > > > > So, in testing it I've tried a number of things. First off, I noticed > > that lockups typically occur when I attempt to change the IP to > > something different than what it was when I was doing my initial > > development on the board. > > This sounds like ARP related problems... often if you use network > booting, but use a different IP address in the kernel startup (or the > MAC/ethernet address for that matter) than the bootrom caused these > delays can come in. What should the target be doing next? Is it going > to be accessing the network for anything? > > Check the ARP table on your host (for unix arp -a shows you the info, > for windoze I don't have a clue, but there must be a way ;-). > > HTH, > > John... --------------------------- Newsgroups: comp.os.vxworks Subject: Re: debug ppc860 BSP Date: Wed, 15 Aug 2001 15:09:24 GMT From: Jaz Organization: Bell Sympatico Message-ID: References: <221caa76.0108150130.27295133@posting.google.com> kevinatnj@yahoo.com.cn (kewin) wrote in news:221caa76.0108150130.27295133@posting.google.com: > hi all > i need your generous help. > The following problem have bothered me for a long time. > i am debuging my vxWorks BSP for ppc960 using the software, > singlestep.But i can't see the source programs of my BSP,only assembly > language appearing in the singlestep environment.How can i added > debugging information to my BSP? And I have added "-g" in the file > ,\tornado\target\h\make\defs.bsp,to the corresponding > FLAGS,CFLAGS,CASFLAGS and CFLAGS_AS. But I still can't see the source > program. > Will anybody tell me what the reason is and how i can make it. > thanks a lot for your generous help. > You have to create a symbol file and ensure singlestep can find it and/or tell it the directories where it can find the source code. - -- Regards Jaz PS: Please remove 'X' in my email address to obtain the correct address. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: adding Eth support on target PC? Date: Wed, 15 Aug 2001 10:59:34 -0400 From: Bruce Chrustie Organization: Nortel Message-ID: <3B7A8E56.4DC58F59@somewhereinhell.com> References: <3B79394E.D43D1A17@somewhereinhell.com> <3B79FEDF.BAACA574@covad.net> DrDiags wrote: > > Bruce, > > I just looked in the config.h for an pc486 BSP and if you > #define INCLUDE_ELT, you will pick up the END driver for the > 3c509 (Though I am not sure whether it is ISA or PCI), see the > following: > > # ifdef INCLUDE_ELT > # undef INCLUDE_ELT /* don't use BSD driver */ > # define INCLUDE_ELT_3C509_END /* Use END driver instead */ > # endif Fine Dr!.... I realized that after posting my message. > The other item of interest is the default vxworks_uncmp file > you said you used. I imagine it is a vxworks.st image, but the > real question is whether when you stop the boot countdown and > type the "h"elp command, what type of boot devices are listed > at the bottom of the screen. If you see the elt0 device, then > your image has built in support for your NIC. Geeze I just had a synapse fire. I have stopped the booting to look at the config and if I type a '?' I see I have support for FEI and ELT. Now I do not have an FEI card in my machine, only the ELT so I wonder if it is trying to config the FEI and crapping out as it cannot find it. The other thing, it seems that the driver is initializing the ELT as I do get an active status on my hub on the port it is connected to. > Look in your BSP for > the configNet.h file and determine the name of the END driver load > routine. You could instrument it to determine what is going on, but > the muxDevLoad failure of the first END driver found is one of > your problems. Check the config.h file (or the listing at the bottom of > your help screen) because you may be trying to load several NICs that > are not present in the system and thus your network init falls apart. Amen... Thx for the insights Dr, and I will report back this weekend. I am also going to install an FEI NIC in the system to see what happens. Plus with the default vxworks image I should be able to use the FEI without modifying any more code. Then I can continue on my merry way debugging my own real (or not so real) driver for a POTS card! > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: stripping symbols using objcopyppc Date: Wed, 15 Aug 2001 15:18:14 +0100 From: "Dave Korn" Organization: Lumber Cartel (tinlc) Members #2234-2237 (owing to browser refresh) Message-ID: References: Arun Srinivasan wrote in message ... >Hi, > I am trying to use objcopyppc to strip symbols off objects. These >objects work internally, i.e other objects do not invoke functions in >them. So I tried just leaving the undefined symbols in the object and >taskMain, so that when the object is loaded up, the undefined symbols >will get resolved ( as other objects loaded before this one have those >symbols defined in them). But this did not work right. Apparently, >there are more symbols that I need to leave behind for the object to >execute without crashing on the target. > >Is there anyway I can find out whether a symbol is necessary or not >for the object to execute successfully on the target ?- which in my >case is a elf32-powerpc target. > >I first used nmppc to find out the symbols in the object, and then >used objcopyppc to strip symbols off. > > >Since leaving the undefined symbols alone in the object did not help, >I tried leaving text symbols (denoted by type T in the nmppc output) >also - in addition to undefined symbols but even that did not work. >Apparently, there were symbols that needed to be in the object which I >was missing. > >I have heard that stripping symbols off objects can result in a lot of >space saving, but atleast in my case it does not seem to be trivial. >Any pointers to find out symbols that NEED to be in the object for it >to execute successfully will be greatly appreciated. Is there any >other Windriver commands that I can use which might simplify my >problem? This is just an educated guess, but the symbols "_ctors" and "_dtors" (which point to the tables generated by the munch stage of the build) are absolutely vital if you have any statically-constructed class objects. The dynamic loader won't complain if they're missing, but it won't call any of the constructors for your static objects if it can't find this table - it will just assume there are no static class objects. Other than that, I would have thought that the loader would give you an error message if it tries to resolve a symbol during loading and can't do so. DaveK - -- " So far, science's crowning achievement, the Atomic Bomb, has not even been able to alleviate the misery and torture of endless centuries of world poverty." - George Hammond, self-proclaimed discoverer of the scientific proof of god. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: debug ppc860 BSP Date: Wed, 15 Aug 2001 12:44:54 -0400 From: david lindauer Message-ID: <3B7AA706.DDADC3E5@notifier-is.net> References: <221caa76.0108150130.27295133@posting.google.com> you would have to enable it with the linker as well (is there an LFLAGS?) David kewin wrote: > hi all > i need your generous help. > The following problem have bothered me for a long time. > i am debuging my vxWorks BSP for ppc960 using the software, > singlestep.But i can't see the source programs of my BSP,only assembly > language appearing in the singlestep environment.How can i added > debugging information to my BSP? And I have added "-g" in the file > ,\tornado\target\h\make\defs.bsp,to the corresponding > FLAGS,CFLAGS,CASFLAGS and CFLAGS_AS. But I still can't see the source > program. > Will anybody tell me what the reason is and how i can make it. > thanks a lot for your generous help. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: How to calculate CPU Usage percent in Vxworks Date: Wed, 15 Aug 2001 12:11:11 -0500 From: Dinker Charak Organization: Fermi National Accelerator Laboratory Message-ID: <3B7AAD2F.1CB4B9D0@fnal.gov> References: <14c57c4f.0108150029.7e29b0e0@posting.google.com> Sun Wenjian wrote: > > Is any one has any idea to calculate CPU utilization in vxworks OS? > thanks a lot. try using spyTask. If you are using Tornado II tools ... try Browser tool ... - -- Computing Division/Online and Database Systems 630-840-8604 (office) Fermi National Accelerator Laboratory http://www.fnal.gov --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Changing root directory in Tornado2? Date: Wed, 15 Aug 2001 10:33:53 -0700 From: "Denis" Message-ID: <9leblv$90cor$1@ID-44494.news.dfncis.de> References: <9l99ue$4vd@nntpb.cb.lucent.com> <9l9iko$na4$06$1@news.t-online.com> <9ld9d9$ec1$1@siamese.noc.ucla.edu> <9ldbp8$nj3$07$1@news.t-online.com> Martin, "Martin Raabe" wrote in message news:9ldbp8$nj3$07$1@news.t-online.com... | Hello Denis, | since it does not sound like, I explicitely express it here! | I am speaking for my own, despite my position within WindRiver. | So the signature you found is older than one year isn't it? the only reason I even looked up a signature, was because I saw you trying to pitch sniff in another thread not a long time ago "How common is SNIFF usage". or, if not pitch, then strongly advocate it, to a degree where i got a feeling that you were affiliated with them. perhaps you are speaking for yourself, and you are not affilitated with sniff development anymore. it just sounded funny, but perhaps i should've held off on posting that message. my apologies. denis --------------------------- Newsgroups: comp.os.vxworks Subject: System V Packaging and VxWorks Date: Wed, 15 Aug 2001 12:56:25 -0500 From: Kamal Hussain Organization: Lucent Technologies, Columbus, Ohio Message-ID: <3B7AB7C9.E0775177@lucent.com> Hi, Does VxWorks support System V Packaging APIs? - -Kamal --------------------------- Newsgroups: comp.os.vxworks Subject: WindML 2.0/Zinc 6.0, key loss. Date: 15 Aug 2001 14:21:49 -0400 From: Bill Pringlemeir Organization: Bell Sympatico Message-ID: Sender: bpringlemeir@DeadDuck In WindML 2.0 $(WIND_BASE)/src/ugl/win/winroute.c, [start patch] - --- winroute2.c Wed Aug 15 14:16:12 2001 +++ winroute.c Wed Aug 15 14:15:51 2001 @@ -137,6 +137,9 @@ { UGL_WINDOW_ID keyboardGrabId = winKeyboardGrabGet (eventRouterId); + if (keyboardGrabId == UGL_NULL_ID) + keyboardGrabId = winRootGet (eventRouterId); + if (keyboardGrabId != UGL_NULL_ID) { pInputEvent->header.objectId = keyboardGrabId; [end patch] If keyboardGrabId is NULL, then use the winRootGet() member to send the keys to. While using Zinc 6.0, a combobox on the screen would cause edit boxes to stop receiving keyboard input. If someone from WRS/Zinc is listening, I would be interested in better patches or comments on this. If you are using Zinc 6.0 or WindML, you might consider using this. regards, Bill Pringlemeir. - -- Truth decays into beauty, while beauty soon becomes merely charm. Charm ends up as strangeness, and even that doesn't last, but up and down are forever. - Anonymous vxWorks FAQ, "http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html" --------------------------- Newsgroups: comp.os.vxworks Subject: Re: How to calculate CPU Usage percent in Vxworks Date: Wed, 15 Aug 2001 18:07:26 GMT From: Lee DeRaud Organization: None to speak of Message-ID: <38elntk81kho36srs7aoe5bqv22jt6f4jv@4ax.com> References: <14c57c4f.0108150029.7e29b0e0@posting.google.com> <3B7AAD2F.1CB4B9D0@fnal.gov> Sender: nntp@news.boeing.com (Boeing NNTP News Access) On Wed, 15 Aug 2001 12:11:11 -0500, Dinker Charak wrote: >Sun Wenjian wrote: >> >> Is any one has any idea to calculate CPU utilization in vxworks OS? >> thanks a lot. > >try using spyTask. A helpful hint from someone who's been-there-done-that: Use the highest sampling frequency your BSP auxClock supports, and set the global 'spyTaskPriority' to a rather high value (low priority) *before* turning spy on. The default settings will crash most time- critical apps without yielding a useful measurement. DAMHIKT. Lee --------------------------- Newsgroups: comp.os.vxworks Subject: Re: WindML 2.0/Zinc 6.0, key loss. Date: 15 Aug 2001 14:38:07 -0400 From: Bill Pringlemeir Organization: Bell Sympatico Message-ID: References: Sender: bpringlemeir@DeadDuck >>>>> "Bill" == Bill Pringlemeir writes: Bill> In WindML 2.0 $(WIND_BASE)/src/ugl/win/winroute.c, [snip] Bill> If keyboardGrabId is NULL, then use the winRootGet() member to Bill> send the keys to. While using Zinc 6.0, a combobox on the Bill> screen would cause edit boxes to stop receiving keyboard input. Bill> If someone from WRS/Zinc is listening, I would be interested in Bill> better patches or comments on this. If you are using Zinc 6.0 Bill> or WindML, you might consider using this. Sorry. I should add that the only call to winKeyboardUngrab() is done in the window.c code under winUnmanage(). This gets called {through some Byzantine call hierarchy} by Zinc. No window will receive any key input as the current eventRouter will have the keyboard window set to NULL. regards, Bill Pringlemeir. - -- NSA class struggle munitions Peking World Trade Center [Hello to all my fans in domestic surveillance] supercomputer Noriega arrangements FBI Marxist genetic SEAL Team 6 Waco, Texas Semtex vxWorks FAQ, "http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html" --------------------------- Newsgroups: comp.os.vxworks Subject: Re: semBCreate() - Semaphore Creation is failing Date: Wed, 15 Aug 2001 14:15:42 -0400 From: Pick a user name <> Organization: http://extra.newsguy.com Message-ID: References: <3B7457DD.6AB5FB87@lucent.com> i too was having a problem with semBCreate failing. i did a test and took out the includes of the posix stuff and it worked. --------------------------- Newsgroups: comp.os.vxworks Subject: Any command to find how much of the STACK is full/empty?? Date: Wed, 15 Aug 2001 14:41:14 -0400 From: Ajay Oka +1 973 448 8137 Organization: Lucent Technologies Message-ID: <3B7AC24A.76FEE2A4@lucent.com> Folks, Anyone knows about how to find out how much STACK space is filled out and/or how much is empty? Thanks Ajay --------------------------- Newsgroups: comp.os.vxworks Subject: how to use objdump386 Date: Wed, 15 Aug 2001 16:29:54 -0400 From: "Wang,Xiaoguang(Freeman)" Organization: Marconi Message-ID: <9lem1m$qr0$1@newsfeed.pit.comms.marconi.com> Hi, I've been using objdumparm for a long time, and it works pretty well. Recently I was assigned a job debugging some pentium code, then I turned to use objdump386 when I want to deassemble the image. But, to my supprise, I found the information it gave me was wrong --- maybe I was wrong:( Say, if FUNC_A calls FUNC_B, with objdumparm, I can see the name of FUNC_B in the deassembed code of FUNC_A. But there is no calling of FUNC_B in the deassembled code with objdump386, and I saw some other strange function calls in the deassembled code. Is there any problem with the objdump386 provided with vxWorks 5.4? Or did I misuse it? What I did is: objdump386 -d --start-address=0x... --stop-address=0x.... image Thanks --------------------------- Newsgroups: comp.os.vxworks Subject: Re: how to use objdump386 Date: 15 Aug 2001 16:47:47 -0400 From: Bill Pringlemeir Organization: Bell Sympatico Message-ID: References: <9lem1m$qr0$1@newsfeed.pit.comms.marconi.com> Sender: bpringlemeir@DeadDuck >>>>> "Freeman" == writes: [snip] Freeman> Is there any problem with the objdump386 Freeman> provided with vxWorks 5.4? Or did I misuse Freeman> it? type `objdump386 --help'. What does it say at the bottom? no try `objdumpsimpc --help'. What does it say? I get, objdumpsimpc: supported targets: pe-i386 pei-i386 ... pe-i386 is a modified coff format that Window executables use. Are you compiling with `cc386' or `ccsimpc' (or just reverse-engineering something)? You will need to use one or the other depending on the format of the executable. You shouldn't need to supply the start and end addresses, unless it is raw assembly. If it is raw assembly, it may not decode the function names as you would like as there is no symbol table. regards, Bill Pringlemeir. - -- CHUBBY CHECKER owns my BUILDING! vxWorks FAQ, "http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html" --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Any command to find how much of the STACK is full/empty?? Date: Wed, 15 Aug 2001 23:25:29 GMT From: "Peter Marqui" Organization: Verio Message-ID: References: <3B7AC24A.76FEE2A4@lucent.com> checkStack "Ajay Oka +1 973 448 8137" wrote in message news:3B7AC24A.76FEE2A4@lucent.com... > Folks, > > Anyone knows about how to find out how much STACK > space is filled out and/or how much is empty? > > Thanks > > Ajay --------------------------- Newsgroups: comp.os.vxworks Subject: Re: debug ppc860 BSP Date: 15 Aug 2001 17:39:00 -0700 From: kevinatnj@yahoo.com.cn (kewin) Organization: http://groups.google.com/ Message-ID: <221caa76.0108151639.1cfe1eef@posting.google.com> References: <221caa76.0108150130.27295133@posting.google.com> <3B7AA706.DDADC3E5@notifier-is.net> david lindauer wrote in message news:<3B7AA706.DDADC3E5@notifier-is.net>... > you would have to enable it with the linker as well (is there an LFLAGS?) > > David > > kewin wrote: > > > hi all > > i need your generous help. > > The following problem have bothered me for a long time. > > i am debuging my vxWorks BSP for ppc960 using the software, > > singlestep.But i can't see the source programs of my BSP,only assembly > > language appearing in the singlestep environment.How can i added > > debugging information to my BSP? And I have added "-g" in the file > > ,\tornado\target\h\make\defs.bsp,to the corresponding > > FLAGS,CFLAGS,CASFLAGS and CFLAGS_AS. But I still can't see the source > > program. > > Will anybody tell me what the reason is and how i can make it. > > thanks a lot for your generous help. thanks to David.I have tried. i didn't find LFLAGS,but i find LD=ldppc.i make the following change,LD=ldppc -g.The underlines are what i see, ####### bootConfig.c: at unknown line number 002029A4 blr ####### bootConfig.c: at unknown line number 002029A8 usrBpInit: 002029A8 stwu r1,-8(r1) 002029AC mfspr r0,LR 002029B0 stw r0,0xC(r1) 002029B4 lis r3,0x00250000@HI 002029B8 addi r3,r3,0x75A0 002029BC bl printf 002029C0 li r3,-1 002029C4 lwz r0,0xC(r1) 002029C8 mtspr LR,r0 002029CC addi r1,r1,8 ####### bootConfig.c: at unknown line number 002029D0 blr ####### bootConfig.c: at unknown line number 002029D4 usrSlipInit: 002029D4 stwu r1,-8(r1) 002029D8 mfspr r0,LR 002029DC stw r0,0xC(r1) 002029E0 lis r3,0x00250000@HI 002029E4 addi r3,r3,0x75D8 002029E8 bl printf 002029EC li r3,-1 002029F0 lwz r0,0xC(r1) 002029F4 mtspr LR,r0 002029F8 addi r1,r1,8 --------------------------- Newsgroups: comp.os.vxworks Subject: Re: how to use objdump386 Date: Wed, 15 Aug 2001 17:55:55 -0700 From: "Denis" Message-ID: <9lf5im$9d5ou$1@ID-44494.news.dfncis.de> References: <9lem1m$qr0$1@newsfeed.pit.comms.marconi.com> "Wang,Xiaoguang(Freeman)" wrote in message news:9lem1m$qr0$1@newsfeed.pit.comms.marconi.com... [snip] | Say, if FUNC_A calls FUNC_B, with objdumparm, I can see the name of FUNC_B | in the deassembed code of FUNC_A. | But there is no calling of FUNC_B in the deassembled code with objdump386, | and I saw some other strange function calls in the deassembled code. [snip] are you saying that there are things like call 00000000 ? in any case, try objdump386 -d --reloc object.o denis --------------------------- Newsgroups: comp.os.vxworks Subject: pagefaults during startup if network is plugged in Date: Wed, 15 Aug 2001 18:02:19 -0700 From: "Denis" Message-ID: <9lf5uu$93qqc$1@ID-44494.news.dfncis.de> hello, we are running T 1.01. the ethernet interfaces are intel 82557. we are running into bizarre problem, where when we download an image to the box upon bootup, a crash happens, if there are any packets entering ethernet interface. in particular, the box comes up, starts the bootloader process, downloads an image, and then we must unplug the network interface, or ensure that there are no packets entering it. we can plug the cable back in once the box is up and running. if we do not do so, the box will crash. we tried debugging it at some point, and it crashes in places which seem to not make much sense. has anyone seen this ? please let me know if you'd like to see more information. thanks, denis --------------------------- Newsgroups: comp.os.vxworks Subject: Re: adding Eth support on target PC? Date: Wed, 15 Aug 2001 20:55:41 -0400 From: Bruce Chrustie Organization: Nortel Message-ID: <3B7B1A0D.53FC3BC4@somewhereinhell.com> References: <3B79394E.D43D1A17@somewhereinhell.com> <3B79FEDF.BAACA574@covad.net> <3B7A8E56.4DC58F59@somewhereinhell.com> Bruce Chrustie wrote: > Geeze I just had a synapse fire. I have stopped the booting to look at > the config and if I type a '?' I see I have support for FEI and ELT. Now > I do not have an FEI card in my machine, only the ELT so I wonder if it > is trying to config the FEI and crapping out as it cannot find it. The > other thing, it seems that the driver is initializing the ELT as I do > get an active status on my hub on the port it is connected to. Finally! Seems as though somewhere INCLUDE_FEI (by default) is set as well as INCLUDE_ELT. So it craps out trying to find the FEI NIC. I #undef'd INCLUDE_FEI everywhere it was used and then some. Needless to say it booted right up and got rid of the error. I still did not have tcp/ip running for some reason (TBD) so I tried to load VXW off of the network as opposed to the disk and it worked! I got "attached TCP/IP inerface to elt0" from the network load, but do not get it on the disk load. DOn't know why yet. TBD unless you have the answer for me :) Thx for the help and once I figure out the culprit #define INCLUDE_FEI I will write it up nicely for the FAQ. I am still going to buy an Intel 82557 NIC to see if the default loads and config are automagically supported to save in others frustrations! I have seen a few posts on this in the past. The other thing is when I showElt (or whatever it was) I get about half of the stats on the screen and then it locks up requiring a reboot. Off I go to PCI land for a while..... thx all! --------------------------- Newsgroups: comp.os.vxworks Subject: Isn't there IP_ONESBCAST. Date: 15 Aug 2001 18:35:45 -0700 From: nfysh@hanmail.net (SungHwanYun) Organization: http://groups.google.com/ Message-ID: I want to set socket option of BSD/OS 3.0 "IP_ONESBCAST". Don't VxWorks support "IP_ONESBCAST"? If VxWorks don't support "IP_ONESBCAST", I want to know another way of doing same effect. Best Regards. Thanks. --------------------------- Newsgroups: comp.os.vxworks Subject: Multiple Initiator and one HDD Date: Thu, 16 Aug 2001 01:36:56 GMT From: "Andy Beard" Organization: RoadRunner - Central Florida Message-ID: I have a system comprised of two MVME 172 processor and one SCSI hard drive with a DOS filesystem. I'm trying to share the hard drive between the two processors and am running into problems with corrupt files and locked up SCSI interface. I've tried some of the setting on the DOS filesystem and that didn't seem to help. The SCSI interface is using an NCR 710 chip. If the driver supports multiple initiators, I'm assuming I need to change some of the SCSI parameters. Two questions: Does anyone know if the driver supports multiple initiators and what should the SCSI setting be? Thanks in advance Andy --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Any command to find how much of the STACK is full/empty?? Date: 16 Aug 2001 01:51:14 GMT From: Rush Manbert Organization: a2i network Message-ID: <9lf8uj$3r5$1@samba.rahul.net> References: <3B7AC24A.76FEE2A4@lucent.com> In article <3B7AC24A.76FEE2A4@lucent.com>, Ajay Oka +1 973 448 8137 wrote: >Folks, > >Anyone knows about how to find out how much STACK >space is filled out and/or how much is empty? > >Thanks > >Ajay Just use checkStack at the VxWorks shell. Gives you this info for all tasks. - - Rush --------------------------- Newsgroups: comp.os.vxworks Subject: can timer_create() and timer_connect() be called in an ISR? Date: 15 Aug 2001 18:56:33 -0700 From: xgdong@linkair.com.cn (Docular Xu) Organization: http://groups.google.com/ Message-ID: <72a8983.0108151756.50676655@posting.google.com> When implement a driver, I need to start a timer to show if time-out comes before ACK. The exact time for sending out a frame is the device's transmit completion interrupt. So, may I call timer_create() and timer_connect() in ISR ? Thanks a lot. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: How can i know whether the network cable is plugged in or not under Vxworks? Date: Thu, 16 Aug 2001 10:38:49 +0800 From: "zzxt" Organization: Bentium Ltd. (CN99) Message-ID: <9lfbm1$nsj$1@mail.cn99.com> References: <9la2pq$5pn$1@mail.cn99.com> <3b794d05$0$90634$e2e8da3@nntp.cts.com> Thanks for your information, but what do you mean by "MDIO"? "Dan Gold" wrote in message news:3b794d05$0$90634$e2e8da3@nntp.cts.com... > > "zzxt" wrote in message news:9la2pq$5pn$1@mail.cn99.com... > > Hi all: > > Can i install a ISR to deal with cable disconnect and reconnect or Is > > there a register where i can read the status of the network cable. Is this > > feature specific to NIC. Both END and BSD solutions are needed. thanks > very > > much > > This is totally dependent on the PHY device you are using. Most modern PHYs > have a set of status and control registers you can read/write via the MDIO > interface. Most of them are also capable of generating an interrupt based > on various conditions (link establishment or not, auto-negotiation complete, > etc...). You need to find out what PHY device you are using, if you have an > MDIO interface (2 wires, clock and data) to the part, and if you can > configure it to generate interrupts. > > hth, > Dan Gold gold@ensemble.com > > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: adding Eth support on target PC? Date: Wed, 15 Aug 2001 19:44:09 -0700 From: DrDiags Organization: laserlink.net Message-ID: <3B7B3379.CF6F589@covad.net> References: <3B79394E.D43D1A17@somewhereinhell.com> <3B79FEDF.BAACA574@covad.net> <3B7A8E56.4DC58F59@somewhereinhell.com> <3B7B1A0D.53FC3BC4@somewhereinhell.com> Reply-To: drdiags@covad.net Bruce, The failure to init NIC when loading from disk can be cured by putting the name of your NIC in the (o)ther: field in the boot parameters. I believe Johan has a nice write-up in the FAQ about using a standalone vxWorks_unmcp(.st) image. What happens from what I know, is since the NIC was not used to download the image, during the kernel startup, the code looks at the type of device used and if it matches certain type of devices, the code will scan the (o)ther field to see if there is a network device which needs initialization. There are other factors that can come into play, like #define STANDALONE_NETWORK, but look on WindSurf if you have an account. There are some Tech Tips for the configuration you have. Then of course, you have to remember the saving of your uncompressed kernel image to be used by the target server, so you won't get the "target server not initialized" message when you start your debugging or tool usage. Thank the (politically correct) great deity above that Johan put together the FAQ and we have this group. Anyway, what else you gonna do when the project is sliding... Good Luck. Bruce Chrustie wrote: > > Bruce Chrustie wrote: > > > Geeze I just had a synapse fire. I have stopped the booting to look at > > the config and if I type a '?' I see I have support for FEI and ELT. Now > > I do not have an FEI card in my machine, only the ELT so I wonder if it > > is trying to config the FEI and crapping out as it cannot find it. The > > other thing, it seems that the driver is initializing the ELT as I do > > get an active status on my hub on the port it is connected to. > > Finally! Seems as though somewhere INCLUDE_FEI (by default) is set as > well as INCLUDE_ELT. So it craps out trying to find the FEI NIC. I > #undef'd INCLUDE_FEI everywhere it was used and then some. Needless to > say it booted right up and got rid of the error. > > I still did not have tcp/ip running for some reason (TBD) so I tried to > load VXW off of the network as opposed to the disk and it worked! > > I got "attached TCP/IP inerface to elt0" from the network load, but do > not get it on the disk load. DOn't know why yet. TBD unless you have the > answer for me :) > > Thx for the help and once I figure out the culprit #define INCLUDE_FEI I > will write it up nicely for the FAQ. I am still going to buy an Intel > 82557 NIC to see if the default loads and config are automagically > supported to save in others frustrations! I have seen a few posts on > this in the past. > > The other thing is when I showElt (or whatever it was) I get about half > of the stats on the screen and then it locks up requiring a reboot. > > Off I go to PCI land for a while..... > > thx all! - -- http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html (vxWorks FAQ home page) --------------------------- Newsgroups: comp.os.vxworks Subject: Re: adding Eth support on target PC? Date: Wed, 15 Aug 2001 20:04:08 -0700 From: DrDiags Organization: laserlink.net Message-ID: <3B7B3828.D18BE1@covad.net> References: <3B79394E.D43D1A17@somewhereinhell.com> <3B79FEDF.BAACA574@covad.net> <3B7A8E56.4DC58F59@somewhereinhell.com> <3B7B1A0D.53FC3BC4@somewhereinhell.com> Reply-To: drdiags@covad.net Bruce, One more item, make sure the DEFAULT_BOOTLINE (sp) has the network information you want setup on the NIC when you build the vxWorks_uncmp(.st) image. I am not totally sure, but the STANDALONE kernel may use the one embedded in the image instead of using the information provided in the bootstrap. (Probably not correct info here, but just another item to add to the confusion). > I still did not have tcp/ip running for some reason (TBD) so I tried to > load VXW off of the network as opposed to the disk and it worked! > > I got "attached TCP/IP inerface to elt0" from the network load, but do > not get it on the disk load. DOn't know why yet. TBD unless you have the > answer for me :) > > The other thing is when I showElt (or whatever it was) I get about half > of the stats on the screen and then it locks up requiring a reboot. Darn!! Not sure what to say about this one, is the interrupt being shared or something? (PCI) Maybe you can control this with your BIOS, but again this is a total WAG. - -- http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html (vxWorks FAQ home page) --------------------------- Newsgroups: comp.os.vxworks Subject: Isn't there "IP_ONESBCAST"? Date: 15 Aug 2001 20:57:37 -0700 From: nfysh@hanmail.net (SungHwanYun) Organization: http://groups.google.com/ Message-ID: I want to set socket option of BSD/OS 3.0 "IP_ONESBCAST". Don't VxWorks support "IP_ONESBCAST"? If VxWorks don't support "IP_ONESBCAST", I want to know another way of doing same effect. Best Regards. Thanks. --------------------------- Newsgroups: comp.os.vxworks Subject: Serial IO and RTS/CTS Date: 16 Aug 2001 00:01:36 -0700 From: pritamganguly@hotmail.com (Pritam) Organization: http://groups.google.com/ Message-ID: <166770ed.0108152301.58d85049@posting.google.com> Hi all I am writing a serial io module for communicating with a NT host. But I am unable to figure out how to enable RTS/CTS in VxWorks. The 'ioctl' function does not have any option to do that. Any help are welcome. Thanking in advance Pritam --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Any command to find how much of the STACK is full/empty?? Date: 16 Aug 2001 00:41:35 -0700 From: john_94501@yahoo.com (John) Organization: http://groups.google.com/ Message-ID: <488e459a.0108152341.565a32a6@posting.google.com> References: <3B7AC24A.76FEE2A4@lucent.com> Hello, You could use checkStack from a shell to get the information. I seem to remember that there is a graphical version in the IDE's browser too... If you need it programmatically, take a look at taskInfoGet() - the TASK_DESC structure should return information about stack usage for a given task (except the calling task). For that one, you can get the current stack address easily (take the address of an automatic variable) and use that along with the stack size info (from the TASK_DESC structure) to calculate the current usage. HTH, John... Ajay Oka +1 973 448 8137 wrote in message news:<3B7AC24A.76FEE2A4@lucent.com>... > Folks, > > Anyone knows about how to find out how much STACK > space is filled out and/or how much is empty? > > Thanks > > Ajay --------------------------- Newsgroups: comp.os.vxworks Subject: Re: RPC File Descriptor Count Question Date: 16 Aug 2001 00:53:24 -0700 From: john_94501@yahoo.com (John) Organization: http://groups.google.com/ Message-ID: <488e459a.0108152353.6d71e86e@posting.google.com> References: <9lelh1$7q$1@samba.rahul.net> Hello, Sounds like a classic network server strategy. The server must listen for incoming connections on a 'well known' port, otherwise the clients couldn't contact it. If it is to handle more than one client simultaneously, it could not use that port for the actual communication or other incoming connections would not be seen. So, a new task is spawned to handle the client's request(s). The server task goes back to listening for new connections on the well known port. That way multiple clients can be handled without problems. Each has its own server task, and its own network port/socket. I can't say if your RPC server is working that way, but the behaviour does not sound unusual for network client/server apps so it wouldn't surprise me to see it. HTH, John... Rush Manbert wrote in message news:<9lelh1$7q$1@samba.rahul.net>... > I hope this is a stupid, easily answered question... > > We are using RPC on VxWorks and WinNT 4.0. The server is running under > VxWOrks on our embedded processor. The client is a program running under > Windows NT on a PC. The general configuration places multiple servers > and multiple clients on a LAN, but the simplest configuration is one > server and one client. > > We observe that each client who makes RPC calls to the server opens > a new file descriptor on the server side. All communication with that > client seems to occur on the same socket/file descriptor. If we exit > the client program, then the file descriptor gets closed. This means > that if we have five client programs running, they all talk to our > one server, and the server ends up with five sockets open in order to > support the communications. > > From reading the server side code, it appears to open one UDP socket > and one TCP socket and register both of them with the portmapper. The > client side code appears to ask the server portmapper which server > port it should use (they always use port 0). I would expect that all > of the communications would use the two sockets that were initially > opened and registered by the server, but that is not the case. > > Can anyone tell me why each client opens a new file descriptor on > the server? Is this normal and expected for RPC? (I have never done any > network programming before, so am a babe in the woods.) > > Thanks, > Rush --------------------------- End of New-News digest ********************** From vxwexplo-errs@csg.lbl.gov Thu Aug 16 07:09:25 2001 From: fscheurer@manz-acs.de Date: Thu Aug 16 07:09:27 PDT 2001 Subject: AUTOREPLY comp.os.vxworks newsdigest 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_000_00017EC0.3B7BF017 Content-Type: text/plain; charset=iso-8859-1 Content-Transfer-Encoding: 7Bit This is the Mailbox of Friedrich Scheurer. R&D - ACS GmbH - D-72768 Reutlingen Sorry, I am out of office until August 20th 2001! In urgent cases please contact one of the following persons at ACS GmbH: Otto Angerhofer (Managing director) e-mail: oangerhofer@manz-acs.de Eduard Ams (Technical manager) e-mail: eams@manz-acs.de Yvonne Anhorn (Marketing) e-mail: yanhorn@manz-acs.de Margit Arnold (Sales) e-mail: marnold@manz-acs.de Christina Sixel (Sales) e-mail: csixel@manz-acs.de MfG / Kind Regards F. Scheurer ---------------------------------------------------------------- Friedrich Scheurer, Dipl.-Ing. phone +49 7121 9019-23(0) ACS GmbH - R&D fax +49 7121 9019-98 Steigaeckerstrasse 13 web www.manz-acs.de D-72768 Reutlingen (Germany) e-mail: fscheurer@manz-acs.de ---------------------------------------------------------------- ------_=_NextPart_000_00017EC0.3B7BF017 Content-Type: text/html; charset=iso-8859-1 Content-Transfer-Encoding: quoted-printable
This is the Mailbox of Friedrich Scheurer.
R&D - ACS GmbH - D-72768 Reutlingen
 
Sorry, I am out of office until August 20th 2001!
 
In urgent cases please contact one of the
following persons at ACS GmbH:
 
Otto Angerhofer (Managing director)
Eduard Ams (Technical manager)
Yvonne Anhorn (Marketing)
Margit Arnold (Sales)
e-mail: marnold@manz-acs.de
Christina Sixel (Sales)

 
MfG / Kind Regards
 
F. Scheurer
 
----------------------------------------------------------------
Friedrich Scheurer, Dipl.-Ing.    phone    +49 7121 9019-23(0)
ACS GmbH  -  R&D                  fax      +49 7121 9019-98
Steigaeckerstrasse 13             web      www.manz-acs.de
D-72768 Reutlingen (Germany)      e-mail:  fscheurer@manz-acs.de
----------------------------------------------------------------
------_=_NextPart_000_00017EC0.3B7BF017-- From vxwexplo-errs@csg.lbl.gov Thu Aug 16 07:36:38 2001 From: Jean.Sarkis@axcelis.com Date: Thu Aug 16 07:36:40 PDT 2001 Subject: Re: Bind Error after select() timeout. VxWorks >In trying to ascertain when a remote host has gone offline, I added a >select() call with a timeout around my recv() for a stream socket. >When select() returns a zero (indicating timeout), I treat it just the >same as if recv() returned a zero (indicating a closed connection). >I close() the socket, then use socket() to create a new one, followed > by bind(), listen(), and accept(). >The problem is, I get a bind error (EADDRINUSE) when the close was >spurred by a select timeout. When the close is spurred by a >recv()==0, no bind error occurs. It's the same code that shuts down >the socket and creates a new one, it just fails the bind in one case >and not the other. Any suggestions? Brian, Did you try setting the socket option, SO_REUSEADDR? int off=0; setsockopt (sock, SOL_SOCKET, SO_REUSEADDR, (void *)&off, sizeof(off)); Jean From vxwexplo-errs@csg.lbl.gov Thu Aug 16 09:42:30 2001 From: "thien dinh" Date: Thu Aug 16 09:42:33 PDT 2001 Subject: DMA register info for the MCP750 board Hi VxWorks Readers, I have been looking through the Motorola BSP documentation for the information on DMA registers and how to program them, but so far I have no luck. Could anyone have or know where these info may be and let me know. TVD From vxwexplo-errs@csg.lbl.gov Thu Aug 16 21:37:38 2001 From: "Thomas Keith Buchanan" Date: Thu Aug 16 21:37:40 PDT 2001 Subject: RE: vxWorks Training > Submitted-by vxwexplo-errs@csg.lbl.gov Mon Aug 13 10:50:54 2001 > Submitted-by: "John S. Huggins" > > Hey everybody, > > Is there anybody else that offers Tornado training other than Wind > River? I'm hoping there is something a bit cheaper out there than the > price they quote on their website. If so, will they customize the course > for my company's needs? Will they come to me rather me having to go to > them? > > Any pointers here are greatly appreciated. > Try The PTR Group (http://www.ThePTRGroup.com). They take a different approach. The on-site versions are more hands-on than those provided by Wind River focusing on multiple aspects of real-time embedded development. The material uses Tornado as the toolset of choice but is not solely about the nuances of the Tornado toolset. Many Exploder readers are aware that SPARTA used to provide VxWorks training courses. For the past couple years, I've referred many former customers to The PTR Group. I can highly recommend them to you... the feedback has been great. HTH, ------------------------------------ /\ Thomas Keith Buchanan /**\ Program Manager /****\ SPARTA, Inc. /****/.. 205 East IH 30 /****/.... Rockwall, TX 75087 .\****\__... ...\**\ /.... Voice: 972 772 4487 ....\**\/\.... FAX: 972 772 0739 ....\****\.. PCS: 972 672 9657 ....\****/ WWW: www.sparta.com .../***/ ./***/ SPARTA \**/ ====== \/ Pride In Performance From vxwexplo-errs@csg.lbl.gov Fri Aug 17 04:03:37 2001 From: Vxworks Exploder Date: Fri Aug 17 04:03:40 PDT 2001 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Fri Aug 17 04:03:10 PDT 2001 Subject: Re: Serial IO and RTS/CTS Subject: VxWorks v/s Crosscompiler based development Subject: Re: Bind Error after select() timeout. Subject: Re: how to use objdump386 Subject: Re: how to use objdump386 Subject: Re: HELP - vx FTP-Server problem! Subject: Re: how to use objdump386 Subject: Debugging multiple tasks Subject: Question on using FOREVER in tasks, please answer if you can.... Subject: Re: Question on using FOREVER in tasks, please answer if you can.... Subject: MCP750 Boot Flash Subject: How to Format Disk Subject: Re: Debugging multiple tasks Subject: Re: MCP750 Boot Flash Subject: Re: Access Database Subject: Re: can timer_create() and timer_connect() be called in an ISR? Subject: Re: MCP750 Boot Flash Subject: Re: VxWorks v/s Crosscompiler based development Subject: Re: Debugging multiple tasks ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: Serial IO and RTS/CTS Date: Thu, 16 Aug 2001 12:52:17 +0100 From: "David Laight" Organization: BT Internet Message-ID: <9lgc3l$8n6$1@plutonium.btinternet.com> References: <166770ed.0108152301.58d85049@posting.google.com> None of the vxWorks serial port drivers have support for monitoring or changing and of the modem control lines. If you need to control these, you will have to modify the driver. Additionaly the 16x50 UARTs (as used on PC motherboards) do not have internal support for flow control (software XON/XOFF or CTS/RFS) so you will have to implement that as well. On the transmit side, you cannot stop bytes already in the tx fifo being sent, so you will need to know how much space is left in the remote systems RX buffer when it de-asserts RFS. This, together with knowledge of your system's interrupt latentcy, will tell you how many bytes you can put into your TX fifo. A CBU (Crystal Ball Unit) may be of service here. RFS -> Ready For Sending (I and ready for YOU to send ME data) RTS -> Ready To Send (I want to send) CTS -> Clear To Send (input - allowing tx) Asserting RTS should cause CTS to be asserted (at the same end). The 'standard' serial interface does not have a wire for RFS, the one designated RTS is often used in its place. Even the 25 way D connector (best defined in ISO-2110) misses RFS, it may be on the second connector - - but I think that only carries the parallel autodial signals. This leads to a significant number of ways of wiring a 'null modem' cable. David Pritam wrote in message news:166770ed.0108152301.58d85049@posting.google.com... > Hi all > I am writing a serial io module for communicating with a NT host. But > I am unable to figure out how to enable RTS/CTS in VxWorks. The > 'ioctl' function does not have any option to do that. > Any help are welcome. > Thanking in advance > Pritam --------------------------- Newsgroups: comp.os.vxworks Subject: VxWorks v/s Crosscompiler based development Date: 16 Aug 2001 07:24:43 -0700 From: ishwara_a@yahoo.com (Ishwar) Organization: http://groups.google.com/ Message-ID: <5286f56c.0108160624.26c14fb3@posting.google.com> Hi, In emdebbed development, when we program the hardware directly(i.e. without any RTOS), we create the HEX image on PC and burn it on ROM on the board. I am not sure about the equivalent when we go for production of vxworks based systems. Does the OS "runs" the .exe created using tarnedo or the whole thing running(in the field) is one single HEX image ? I'm new to VxWorks. Thanks, Ishwar --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Bind Error after select() timeout. Date: 16 Aug 2001 07:39:08 -0700 From: shadowfaqs@hotmail.com (Brian) Organization: http://groups.google.com/ Message-ID: References: Thanks to MM, I got the answer I needed (REUSEADDR socket option) at http://www.lcg.org/sock-faq , item 11. It works fine now. shadowfaqs@hotmail.com (Brian) wrote in message news:... > In trying to ascertain when a remote host has gone offline, I added a > select() call with a timeout around my recv() for a stream socket. > When select() returns a zero (indicating timeout), I treat it just the > same as if recv() returned a zero (indicating a closed connection). I > close() the socket, then use socket() to create a new one, followed by > bind(), listen(), and accept(). > > The problem is, I get a bind error (EADDRINUSE) when the close was > spurred by a select timeout. When the close is spurred by a > recv()==0, no bind error occurs. It's the same code that shuts down > the socket and creates a new one, it just fails the bind in one case > and not the other. Any suggestions? > > Thanks, > Brian --------------------------- Newsgroups: comp.os.vxworks Subject: Re: how to use objdump386 Date: Thu, 16 Aug 2001 10:47:32 -0400 From: "Wang,Xiaoguang(Freeman)" Organization: Marconi Message-ID: <9lgmbp$cdc$1@newsfeed.pit.comms.marconi.com> References: <9lem1m$qr0$1@newsfeed.pit.comms.marconi.com> Bill Thanks. Sorry I didn't give some detailed information last time. The file I wanted to investigate is a.out-i386 obj file. The reason I supplied start and stop addresses is because the file is too large and I's just interested in one function. The compiler is gcc 2.95.2, arch=pentium, CPU=PENTIUM, gcc-lib is i386-wrs-vxwork5.4/2.95.2. The other options are -fno-defer-pop -funsigned-char -fdollars-in-identifiers -nostdinc -fno-b uiltin -fvolatile -O2 -pipe This obj file will be processed by objcopy into an image file then gziped and packed into a tar file, which is downloaded to the target by a loader. When I examined the disassembly code I got, I found in the function 1. The 3rd instruction is 'ret' 2. All calls point to a In my opinion, without comparation, there is not going to be a ret in the beginning of a function. And all calls should point to the exactly entry point of a function, not an entry point + offset. So I doubt the correctness of the disassembly, and that's why I ask the group for help. Do I need to specify the CPU or arch when I do objdump? Are there any alternate tools that I can use to disassemble the code? Thanks a lot Xiaoguang Bill Pringlemeir wrote in message news:uzo91yrrg.fsf@yahoo.com... > >>>>> "Freeman" == writes: > [snip] > Freeman> Is there any problem with the objdump386 > Freeman> provided with vxWorks 5.4? Or did I misuse > Freeman> it? > > type `objdump386 --help'. What does it say at the bottom? > no try `objdumpsimpc --help'. What does it say? > > I get, > objdumpsimpc: supported targets: pe-i386 pei-i386 ... > > pe-i386 is a modified coff format that Window executables use. Are > you compiling with `cc386' or `ccsimpc' (or just reverse-engineering > something)? You will need to use one or the other depending on the > format of the executable. You shouldn't need to supply the start and > end addresses, unless it is raw assembly. If it is raw assembly, it > may not decode the function names as you would like as there is no > symbol table. > > regards, > Bill Pringlemeir. > > -- > CHUBBY CHECKER owns my BUILDING! > > vxWorks FAQ, > "http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html" --------------------------- Newsgroups: comp.os.vxworks Subject: Re: how to use objdump386 Date: Thu, 16 Aug 2001 11:13:29 -0400 From: "Wang,Xiaoguang(Freeman)" Organization: Marconi Message-ID: <9lgnsd$cpm$1@newsfeed.pit.comms.marconi.com> References: <9lem1m$qr0$1@newsfeed.pit.comms.marconi.com> <9lf5im$9d5ou$1@ID-44494.news.dfncis.de> Hi, Denis No, it gives some non-zero values. I just tried --reloc and almost all combinations, I still can't understand the disassembly. Thanks Xiaoguang > are you saying that there are things like call 00000000 ? > in any case, try objdump386 -d --reloc object.o > > denis > > > > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: HELP - vx FTP-Server problem! Date: 16 Aug 2001 15:22:45 GMT From: Johan Borkhuis Organization: Agere Systems Message-ID: References: <3B77EB38.DBD20CFA@stoye.de> Joachim Pluschke wrote: > Hi all, > > we have a big problem with FTP-Servers coming with VxWorks on the > TornadoII CDs (Host: WinNT, Target: PowerPC). Did you try the FTP server that comes with DosFS2? It is in sourcecode form and can be found in the unsupported directory. Groeten, Johan - -- o o o o o o o . . . _____________________________ o _____ || Johan Borkhuis | .][__n_n_|DD[ ====_____ | borkhuis@agere.com | >(________|__|_[_________]_|__________________________| _/oo OOOOO oo` ooo ooo 'o!o!o o!o!o` === VxWorks FAQ: http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html === --------------------------- Newsgroups: comp.os.vxworks Subject: Re: how to use objdump386 Date: 16 Aug 2001 11:34:06 -0400 From: Bill Pringlemeir Organization: Bell Sympatico Message-ID: References: <9lem1m$qr0$1@newsfeed.pit.comms.marconi.com> <9lgmbp$cdc$1@newsfeed.pit.comms.marconi.com> Sender: bpringlemeir@DeadDuck >>>>> "Freeman" == Wang,Xiaoguang(Freeman) writes: Freeman> The file I wanted to investigate is a.out-i386 obj file. The Freeman> reason I supplied start and stop addresses is because the Freeman> file is too large and I's just interested in one function. [snip] Freeman> When I examined the disassembly code I got, I found in the Freeman> function 1. The 3rd instruction is 'ret' 2. All calls point Freeman> to a Freeman> In my opinion, without comparation, there is not going to be Freeman> a ret in the beginning of a function. And all calls should Freeman> point to the exactly entry point of a function, not an entry Freeman> point + offset. This is quite common. If the file has not been fully relocated (and it probably hasn't in your case), then you will see expressions like foo+12: jsr 000000 Or something like that (long time since I used an x86). This is generally a sign that your code hasn't been fully relocated. Freeman> Do I need to specify the CPU or arch when I do objdump? Are Freeman> there any alternate tools that I can use to disassemble the Freeman> code? Generate a map file and take a look at it. Also, you can extract the module using `cut' or some other command and run `objdump' on the selected bit of binary code. I have taken the BFD code, modified it and then put a disassembler on a target as well. This is pretty easy. Also, you should really look at `objdump -all-headers' and look for the "SYMBOL TABLE" section. An entry like this is external, [ 15](sec 0)(fl 0x00)(ty 0)(scl 2) (nx 0) 0x00000000 _CompEntryGet... ^^^^^^ It has a section of zero, which means it isn't in the current file. If you have any symbols like this, they you don't have a totally relocated file. You shouldn't be passing the `-r' or `-Ur' options to the linker. If you are then you have a relocatable object (default for vxWorks, I believe). This will have to be processed by a loader to resolve the external symbols. hth, Bill Pringlemeir. - -- Don't forget to dye your hair. Today is not a good time for buying a new computer. Today is not a good time for making love. Many people will ask you for advice about making love. You'll once in a while be tired. vxWorks FAQ, "http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html" --------------------------- Newsgroups: comp.os.vxworks Subject: Debugging multiple tasks Date: 16 Aug 2001 09:15:58 -0700 From: andy.moskoff@strixsys.com (Andy) Organization: http://groups.google.com/ Message-ID: There was a previous post about trying to set a breakpoint at the entry point of a new task that is spawned by another task. The response was to set a "global breakpoint". Thats fine, but how do you set a global breakpoint? Using Tornado II/vxWorks 5.4.2. If there is another way I'd sure like to hear it. Finally, is there a FAQ for this newsgroup and if so where?????? Andy --------------------------- Newsgroups: comp.os.vxworks Subject: Question on using FOREVER in tasks, please answer if you can.... Date: 16 Aug 2001 16:25:24 GMT From: PC@User.com (PC User) Organization: ExecPC Internet - Milwaukee, WI Message-ID: <3b7bf3f3$0$12825$272ea4a1@news.execpc.com> Hi, I have two tasks that were using FOREVER and I changed them to a for loop with a deterministic index such as 15, 20, 25, ..... The problem is that the tasks keep running passed the loop index by 3 or 4 times each time. I thought that they would stop right at the end of the loop index. Why does VxWorks keeps running entering this loop 3 or 4 time regardless of my loop index reaching its end? Thanks for your answer in advance. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Question on using FOREVER in tasks, please answer if you can.... Date: Thu, 16 Aug 2001 20:49:55 +0200 From: "michi" Organization: [neXgo] the next generation online Message-ID: <9lh49i$cj6$1@newsread2.nexgo.de> References: <3b7bf3f3$0$12825$272ea4a1@news.execpc.com> VxWorks dosn't know something about your for-loop in your task. it sheduls the task as its real-time-sheduler rules force it to run. look in the manual how the sheduler work and you can see that task run dosn't depend on task's code. "PC User" schrieb im Newsbeitrag news:3b7bf3f3$0$12825$272ea4a1@news.execpc.com... > Hi, > > I have two tasks that were using FOREVER and I changed them to a for loop with > a deterministic index such as 15, 20, 25, ..... > > The problem is that the tasks keep running passed the loop index by 3 or 4 > times each time. I thought that they would stop right at the end of the loop > index. Why does VxWorks keeps running entering this loop 3 or 4 time > regardless of my loop index reaching its end? > > Thanks for your answer in advance. > --------------------------- Newsgroups: comp.os.vxworks Subject: MCP750 Boot Flash Date: 16 Aug 2001 22:46:57 GMT From: username@qualcomm.com (Username) Organization: QUALCOMM Inc. Message-ID: <9lhih1$h2h$1@coset.qualcomm.com> I tried to use the PPC1-Bug pflash command to copy the vxworks boot kernel to the bootable flash and typed in the wrong address (fff001000 instead of ff0001000). Now I can't even get to the PPC1-Bug prompt. I probably wrote over the PPC1-Bug code in flash. Does anybody know how I can recover from this? --------------------------- Newsgroups: comp.os.vxworks Subject: How to Format Disk Date: Thu, 16 Aug 2001 17:49:44 -0500 From: "Shesha Bhushan" Organization: Arizona State University Message-ID: <9lhpr6$2ss$1@news.asu.edu> Hi Gurus, I have connected a scsi hardisk to a Single board Computer that runs vxWorks OS. The hard disk was previously formated as NTFS. I want to format this hard disk with dosFs. How to format this previously formated disk. Tx a Bunch Shesha Bhushan --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Debugging multiple tasks Date: Thu, 16 Aug 2001 18:42:09 -0700 From: DrDiags Organization: laserlink.net Message-ID: <3B7C7671.9702E888@covad.net> References: Reply-To: drdiags@covad.net Andy, The link at the bottom contains the FAQ for this newsgroup. Andy wrote: > <......SNIP.....> > another way I'd sure like to hear it. Finally, is there a FAQ for this > newsgroup and if so where?????? - -- http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html (vxWorks FAQ home page) --------------------------- Newsgroups: comp.os.vxworks Subject: Re: MCP750 Boot Flash Date: Fri, 17 Aug 2001 13:29:00 +0800 From: "tanghaoyu" Organization: Bentium Ltd. (CN99) Message-ID: <9lia3u$mit$1@mail.cn99.com> References: <9lhih1$h2h$1@coset.qualcomm.com> There are two Flash ROM on MCP750. BANK A is 8M, start at ff000000,end at ff7fffff. BANK B is 1M start at ff800000. Default Jumper map BANK B at fff00000 too,you can change it to BANK A. PPC1-Bug is inside of BANK B that can be pull out of MCP750. In order to recover the wrong pflash on BANK B,you must have another good MCP750. then to do by this step: BANK B(good) -> BANK A -> BANK B(wrong) good luck to you! Tang Haoyu "Username" wrote in message news:9lhih1$h2h$1@coset.qualcomm.com... > I tried to use the PPC1-Bug pflash command to copy the vxworks boot kernel to > the bootable flash and typed in the wrong address (fff001000 instead of > ff0001000). > > Now I can't even get to the PPC1-Bug prompt. I probably wrote over the > PPC1-Bug code in flash. > > Does anybody know how I can recover from this? > > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Access Database Date: 16 Aug 2001 22:32:44 -0700 From: chrismooremobile@yahoo.com (Chris Moore) Organization: http://groups.google.com/ Message-ID: References: <3B7939E0.149C6382@notifier-is.net> david lindauer wrote in message news:<3B7939E0.149C6382@notifier-is.net>... > hi! > is there any way to query a microsoft access database from VXWorks? > Perhaps through ODBC or something? We are trying to decide whether it > is worth the effort to access a database directly from an embedded > target (to get login information) or whether we should build code into > our PC hosts to do authorization indirectly. > > Thanks, > > David Hi, I don't know of any ODBC drivers for VxWorks. Perhaps you could do it running Java and JDBC? But perhaps an easier soloution is a small program running on your Database server (or any other PC on that network) that listens on a UDP port and you pass it your SQL requests via UDP, it then executes those requests (via ODBC to the SQL/Access database) and transmits the results back as UDP data. Pretty easy, think of it as a database proxy! Rgds, Chris --------------------------- Newsgroups: comp.os.vxworks Subject: Re: can timer_create() and timer_connect() be called in an ISR? Date: 16 Aug 2001 23:21:50 -0700 From: john_94501@yahoo.com (John) Organization: http://groups.google.com/ Message-ID: <488e459a.0108162221.78aee19e@posting.google.com> References: <72a8983.0108151756.50676655@posting.google.com> Hello all, Never ceases to amaze me how many people ask questions about calling routines in an ISR... general rule of thumb, the answer is almost always no you can't. The routines that are safe are listed in the manual (and have been quoted on this newsgroup many times). Check section 2.5.3 of the VxWorks 5.4 Programmer's Guide for the list of functions you can call from an ISR. If it's not in the list, don't call it. Instead, push the interrupt processing to task level... have a high priority task waiting on a semaphore, when the interrupt arrives, have the ISR give the semaphore and exit (if there is something that needs to be done very early on that doesn't involve restricted OS calls, then you can do that in the ISR too of course). Your high priority task can then create timers etc, etc. Also, there might be a better way of dealing with your timeouts... check out the watchdog library (wdLib) - but be careful; the routine that gets executed by the watchdog timer is a type of ISR and should follow the rules for an ISR function. HTH, John... xgdong@linkair.com.cn (Docular Xu) wrote in message news:<72a8983.0108151756.50676655@posting.google.com>... > When implement a driver, I need to start a timer to show if time-out > comes before ACK. The exact time for sending out a frame is the > device's transmit completion interrupt. > > So, may I call timer_create() and timer_connect() in ISR ? > > Thanks a lot. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: MCP750 Boot Flash Date: 17 Aug 2001 00:13:09 -0700 From: john_94501@yahoo.com (John) Organization: http://groups.google.com/ Message-ID: <488e459a.0108162313.7823dd1e@posting.google.com> References: <9lhih1$h2h$1@coset.qualcomm.com> Hello, Don't know about the 750, but older MCG boards had dual flash, one bank soldered the other in a socket. The PPCBug s/w is normally in the socketed part. The solution to your predicament is one of the following: 1) If you have another board, try swapping the socketed flash (good idea to get a copy of the flash image on disk before going here in case you repeat the mistake ;-) 2) Download a new image from the motorola web site or ftp site and reprogram your socketed device in an external programmer. HTH, John... username@qualcomm.com (Username) wrote in message news:<9lhih1$h2h$1@coset.qualcomm.com>... > I tried to use the PPC1-Bug pflash command to copy the vxworks boot kernel to > the bootable flash and typed in the wrong address (fff001000 instead of > ff0001000). > > Now I can't even get to the PPC1-Bug prompt. I probably wrote over the > PPC1-Bug code in flash. > > Does anybody know how I can recover from this? --------------------------- Newsgroups: comp.os.vxworks Subject: Re: VxWorks v/s Crosscompiler based development Date: 17 Aug 2001 00:25:36 -0700 From: john_94501@yahoo.com (John) Organization: http://groups.google.com/ Message-ID: <488e459a.0108162325.344fdf8f@posting.google.com> References: <5286f56c.0108160624.26c14fb3@posting.google.com> Hello, Read the VxWorks Programmer's Guide, at least the first few chapters if not all of it. That will explain a lot of the background to developing with VxWorks, and the features available in the OS itself. You should also take a look at sections 2.4 to 2.6 in the Tornado User's Guide which explain how to set up your target and get it booted and running the operating system. HTH, John... ishwara_a@yahoo.com (Ishwar) wrote in message news:<5286f56c.0108160624.26c14fb3@posting.google.com>... > Hi, > > In emdebbed development, when we program the hardware directly(i.e. > without any RTOS), we create the HEX image on PC and burn it on ROM on > the board. > I am not sure about the equivalent when we go for production of > vxworks based systems. > Does the OS "runs" the .exe created using tarnedo or the whole thing > running(in the field) is one single HEX image ? > > I'm new to VxWorks. > > Thanks, > Ishwar --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Debugging multiple tasks Date: Fri, 17 Aug 2001 01:35:39 -0700 From: "Albert H Chen" Message-ID: References: I think all you have to do is in the Tornado shell type "b " to set a global breakpoint... Albert "Andy" wrote in message news:e397b57.0108160815.76bffc15@posting.google.com... > There was a previous post about trying to set a breakpoint at the > entry point of a new task that is spawned by another task. The > response was to set a "global breakpoint". Thats fine, but how do you > set a global breakpoint? Using Tornado II/vxWorks 5.4.2. If there is > another way I'd sure like to hear it. Finally, is there a FAQ for this > newsgroup and if so where?????? > > Andy --------------------------- End of New-News digest ********************** From vxwexplo-errs@csg.lbl.gov Sat Aug 18 04:03:46 2001 From: Vxworks Exploder Date: Sat Aug 18 04:03:49 PDT 2001 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Sat Aug 18 04:03:10 PDT 2001 Subject: Re: Question on using FOREVER in tasks, please answer if you can.... Subject: Re: VxWorks v/s Crosscompiler based development Subject: Debugging when the debugger crashes Subject: Re: Using Sym8xx SCSI adapter on PPC Subject: Re: Debugging when the debugger crashes Subject: DiffServ Subject: VxWorks - SA1110 and the Sophia ICE Subject: Re: pagefaults during startup if network is plugged in Subject: Re: How can i know whether the network cable is plugged in or not under Vxworks? Subject: Re: Debugging multiple tasks Subject: watchdog info Subject: Re: pagefaults during startup if network is plugged in Subject: Tornado 2.0.1 dependency generation Subject: Re: Tornado 2.0.1 dependency generation Subject: Re: Tornado 2.0.1 dependency generation Subject: Re: pagefaults during startup if network is plugged in Subject: Inherited Priority Problems? Subject: the problem of debugging ppc860 BSP has been solved Subject: Re: Debugging when the debugger crashes Subject: memory 'free' suspends task... Subject: Re: Tornado 2.0.1 dependency generation Subject: Re: memory 'free' suspends task... Subject: how to send broadcast messages with dest ip "255.255.255.255" ? Subject: Re: memory 'free' suspends task... Subject: Re: Inherited Priority Problems? Subject: Re: memory 'free' suspends task... ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: Question on using FOREVER in tasks, please answer if you can.... Date: Fri, 17 Aug 2001 11:08:50 +0100 From: "Dave Korn" Organization: Lumber Cartel (tinlc) Members #2234-2237 (owing to browser refresh) Message-ID: References: <3b7bf3f3$0$12825$272ea4a1@news.execpc.com> PC User wrote in message <3b7bf3f3$0$12825$272ea4a1@news.execpc.com>... >Hi, > >I have two tasks that were using FOREVER and I changed them to a for loop with >a deterministic index such as 15, 20, 25, ..... > >The problem is that the tasks keep running passed the loop index by 3 or 4 >times each time. I thought that they would stop right at the end of the loop >index. Why does VxWorks keeps running entering this loop 3 or 4 time >regardless of my loop index reaching its end? I'd guess that you're running multiple tasks through the same function code and using a global variable as a counter or something similarly non-thread-safe. If that isn't it you're going to have to show us your code before anyone can guess what's wrong with it. DaveK - -- " So far, science's crowning achievement, the Atomic Bomb, has not even been able to alleviate the misery and torture of endless centuries of world poverty." - George Hammond, self-proclaimed discoverer of the scientific proof of god. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: VxWorks v/s Crosscompiler based development Date: Fri, 17 Aug 2001 11:43:26 +0100 From: vxworksintent@tao-group.com (David Given) Organization: I do not speak for anyone but myself, and barely that. Message-ID: References: <5286f56c.0108160624.26c14fb3@posting.google.com> Sender: dg@pearl.tao.co.uk In article <5286f56c.0108160624.26c14fb3@posting.google.com>, ishwara_a@yahoo.com (Ishwar) writes: > In emdebbed development, when we program the hardware directly(i.e. > without any RTOS), we create the HEX image on PC and burn it on ROM on > the board. > I am not sure about the equivalent when we go for production of > vxworks based systems. > Does the OS "runs" the .exe created using tarnedo or the whole thing > running(in the field) is one single HEX image ? It's easier to think of VxWorks as less of an OS and more of an OS support library. What you do is to compile your application, statically link it against VxWorks, and that gives you a stand alone image that you can download onto the target and run there. (In fact, for development you tend to have the VxWorks image on ROM and dynamically link your application against it when you download it onto the board; this saves a considerable amount of time because the bulk of the VxWorks code is already there. But for production systems, the above still applies.) - -- David Given Tao Group Ltd. --------------------------- Newsgroups: comp.os.vxworks Subject: Debugging when the debugger crashes Date: Fri, 17 Aug 2001 11:48:02 +0100 From: vxworksintent@tao-group.com (David Given) Organization: I do not speak for anyone but myself, and barely that. Message-ID: <2psil9.q9e.ln@127.0.0.1> Sender: dg@pearl.tao.co.uk I've got a particularly annoying problem on my MIPS-based device. I have Tornado attached to it via ethernet; unfortunately, my application crashes so hard that it takes the VxWorks ethernet driver down with it. (I think it's locking up with interrupts locked.) This means, of course, that the debugger doesn't work any more. I suspect it's jumping into the wild blue yonder and stomping all over memory. I can tell this because I see gibberish appear on the screen once it's locked. But I can't tell any more about it, because I can't debug it. What I'd really like is ICE or a JTAG debugger, but unfortunately my hardware doesn't support it; all I can do is use the ethernet debugger. Does anyone have any suggestions as to how I might be able to tell what's going on here? The main thing I'd like to know is the last thing it was doing before it died... - -- David Given Tao Group Ltd. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Using Sym8xx SCSI adapter on PPC Date: Fri, 17 Aug 2001 14:02:43 +0200 From: =?iso-8859-1?Q?Torbj=F6rn?= Frostemark Organization: FOI Message-ID: <3B7D07E3.B77514A8@foi.se> References: This is a multi-part message in MIME format. - --------------A137F16189B74501458D9732 Content-Type: text/plain; charset=iso-8859-1 Content-Transfer-Encoding: 8bit Koen Van Hulle wrote: > Has any one been able to use a Symbis Logic SCSI controller with VxWorks on > the PPC? I have used sym825 and sym875 on a MPC750 (MVME2700 processor board has the sym825 built in and the sym875 is on a PMC module) and a sym895 on a MPC7400 (MVME5100 with a IPMC712 module). I had no problems with the default values, but I had to patch T2.1 with the sym895 driver. Now you know it works, but I cannot help you more... /Torbjörn - --------------A137F16189B74501458D9732 Content-Type: text/x-vcard; charset=us-ascii; name="torfr.vcf" Content-Transfer-Encoding: 7bit Content-Description: Card for Torbjörn Frostemark Content-Disposition: attachment; filename="torfr.vcf" begin:vcard n:Frostemark;Torbjörn tel;cell:+46-70-249 76 27 tel;fax:+46-(13-37 80 49 tel;work:+46-13-37 80 14 x-mozilla-html:TRUE org:FOI - Swedish Defence Research Agency;Electronic Warfare Systems adr:;;Box 1165;SE-581 11 Linköping;;;SWEDEN version:2.1 email;internet:torfr@foi.se title:Research Engineer fn:Torbjörn Frostemark end:vcard - --------------A137F16189B74501458D9732-- --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Debugging when the debugger crashes Date: Fri, 17 Aug 2001 11:05:09 -0700 From: Joe Durusau Organization: Lockheed Martin Corporation Message-ID: <3B7D5CD5.43C79D44@lmco.com> References: <2psil9.q9e.ln@127.0.0.1> If your board has non-volatile RAM, and it does not get damaged by the error, you could have selected functions store something there when they are entered, and narrow down from there. Otherwise, you will have to delete functions (or tasks) until ti no longer fails, then add back, etc. Another approach might be to use the simulator if you and emulate any special hardware you have. Speaking only for myself, Joe Durusau David Given wrote: > > I've got a particularly annoying problem on my MIPS-based device. I have > Tornado attached to it via ethernet; unfortunately, my application crashes > so hard that it takes the VxWorks ethernet driver down with it. (I think > it's locking up with interrupts locked.) This means, of course, that the > debugger doesn't work any more. > > I suspect it's jumping into the wild blue yonder and stomping all over > memory. I can tell this because I see gibberish appear on the screen once > it's locked. But I can't tell any more about it, because I can't debug it. > > What I'd really like is ICE or a JTAG debugger, but unfortunately my > hardware doesn't support it; all I can do is use the ethernet debugger. > Does anyone have any suggestions as to how I might be able to tell what's > going on here? The main thing I'd like to know is the last thing it was > doing before it died... > > -- > David Given > Tao Group Ltd. --------------------------- Newsgroups: comp.os.vxworks Subject: DiffServ Date: 17 Aug 2001 09:30:45 -0700 From: lfoucu@pacbell.net (Thierry Foucu) Organization: http://groups.google.com/ Message-ID: HI, Does anyone know an implementation of Diffserv (Differentiated Services) for VxWorks? If not, maybe you know some good link, where I could find some information. Thanks for your help --------------------------- Newsgroups: comp.os.vxworks Subject: VxWorks - SA1110 and the Sophia ICE Date: Fri, 17 Aug 2001 10:26:32 -0600 From: "Myrl Tanton" Organization: CADVision/PSINet Message-ID: <9ljh0u$qft$1@news3.cadvision.com> Hi all, I'm trying to debug a VxWorks Image cross compiled for the Strong Arm 1110 In-Circuit Emulator. This requires the use of the '-gdwarf' flag during the compile, however the SA1110 cross compiler provided by VxWorks does not support the -gdwarf compile flag. Does anyone have experince getting the Sophia ICE and a VxWorks SA1110 image to work together? If so I can use some guidance. Specifically, is there a utility that can convert to the same output as produce with the -gdwarf flag? Is there a different cross-compiler variant I can use? Any other suggestions? Thank-you in advance Myrl --------------------------- Newsgroups: comp.os.vxworks Subject: Re: pagefaults during startup if network is plugged in Date: Fri, 17 Aug 2001 09:37:34 -0700 From: "Dan Gold" Message-ID: <3b7d48e2$0$279$e2e8da3@nntp.cts.com> References: <9lf5uu$93qqc$1@ID-44494.news.dfncis.de> "Denis" wrote in message news:9lf5uu$93qqc$1@ID-44494.news.dfncis.de... > hello, > > we are running T 1.01. the ethernet interfaces are intel 82557. > > we are running into bizarre problem, where when we download an image > to the box upon bootup, a crash happens, if there are any packets > entering ethernet interface. > > in particular, the box comes up, starts the bootloader process, > downloads an image, and then we must unplug the network interface, > or ensure that there are no packets entering it. we can plug the > cable back in once the box is up and running. if we do not do so, > the box will crash. we tried debugging it at some point, and it > crashes in places which seem to not make much sense. Once you've downloaded the image, do you disable the network interface ? You don't need it anymore since you've already got your image. Is it possible that while you're decompressing your downloaded image and putting it into RAM, you're colliding with receive buffers for the network interface which is still receiving packets ? hth, Dan Gold gold@ensemble.com --------------------------- Newsgroups: comp.os.vxworks Subject: Re: How can i know whether the network cable is plugged in or not under Vxworks? Date: Fri, 17 Aug 2001 09:42:33 -0700 From: "Dan Gold" Message-ID: <3b7d4a0d$0$90634$e2e8da3@nntp.cts.com> References: <9la2pq$5pn$1@mail.cn99.com> <3b794d05$0$90634$e2e8da3@nntp.cts.com> <9lfbm1$nsj$1@mail.cn99.com> "zzxt" wrote in message news:9lfbm1$nsj$1@mail.cn99.com... > Thanks for your information, but what do you mean by "MDIO"? The MDIO interface is a 2-wire interface to read/write registers on the PHY device. Most PHY devices today (that do 10/100) support this. I think it stands for Management Data I/O or something like that. One wire is the clock (that you toggle), and the other wire is the data. You need 2 general purpose lines for this. One is strictly output (the clock, called MDC), and one is bi-directional - output on writes, input on reads. Essentially you 'bit-bang' the commands and data to/from the chip. What PHY part are you using ? Why not look at the data sheet for the part ? Is this a custom design or did you buy someone's board ? hth, Dan Gold gold@ensemble.com > "Dan Gold" wrote in message > news:3b794d05$0$90634$e2e8da3@nntp.cts.com... > > > > "zzxt" wrote in message news:9la2pq$5pn$1@mail.cn99.com... > > > Hi all: > > > Can i install a ISR to deal with cable disconnect and reconnect or Is > > > there a register where i can read the status of the network cable. Is > this > > > feature specific to NIC. Both END and BSD solutions are needed. thanks > > very > > > much > > > > This is totally dependent on the PHY device you are using. Most modern > PHYs > > have a set of status and control registers you can read/write via the MDIO > > interface. Most of them are also capable of generating an interrupt based > > on various conditions (link establishment or not, auto-negotiation > complete, > > etc...). You need to find out what PHY device you are using, if you have > an > > MDIO interface (2 wires, clock and data) to the part, and if you can > > configure it to generate interrupts. > > > > hth, > > Dan Gold gold@ensemble.com > > > > > > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Debugging multiple tasks Date: 17 Aug 2001 10:10:12 -0700 From: andy.moskoff@strixsys.com (Andy) Organization: http://groups.google.com/ Message-ID: References: Thanks. Yes, I've found that "b " works too. This stuff is not obvious in Tornado User's Guide! Who wrote that crap?!?!?! Andy "Albert H Chen" wrote in message news:... > I think all you have to do is in the Tornado shell type "b " > to set a global breakpoint... > > Albert > > > "Andy" wrote in message > news:e397b57.0108160815.76bffc15@posting.google.com... > > There was a previous post about trying to set a breakpoint at the > > entry point of a new task that is spawned by another task. The > > response was to set a "global breakpoint". Thats fine, but how do you > > set a global breakpoint? Using Tornado II/vxWorks 5.4.2. If there is > > another way I'd sure like to hear it. Finally, is there a FAQ for this > > newsgroup and if so where?????? > > > > Andy --------------------------- Newsgroups: comp.os.vxworks Subject: watchdog info Date: Fri, 17 Aug 2001 18:52:29 GMT From: "joe zhao" Organization: Magma Communications Ltd. Message-ID: Hi, Does anybody know how to check total number of watchdog timers currently being started in a system running vxWorks 5.4? Your help is highly appreciated. Thanks. Joe --------------------------- Newsgroups: comp.os.vxworks Subject: Re: pagefaults during startup if network is plugged in Date: Fri, 17 Aug 2001 13:55:56 -0500 From: Andrew Johnson Organization: APS, Argonne National Laboratory Message-ID: <3B7D68BC.105EAB46@aps.anl.gov> References: <9lf5uu$93qqc$1@ID-44494.news.dfncis.de> Denis wrote: > > we are running T 1.01. the ethernet interfaces are intel 82557. > > we are running into bizarre problem, where when we download an image > to the box upon bootup, a crash happens, if there are any packets > entering ethernet interface. > > in particular, the box comes up, starts the bootloader process, > downloads an image, and then we must unplug the network interface, > or ensure that there are no packets entering it. we can plug the > cable back in once the box is up and running. if we do not do so, > the box will crash. we tried debugging it at some point, and it > crashes in places which seem to not make much sense. It sounds like the network interrupt routine isn't being disabled properly before the new image is run. If an interrupt comes in before the new image is ready for it, then you can expect strange things to happen. Check your bootrom code, this is where the error seems likely to be. - - Andrew - -- The world is such a cheerful place when viewed from upside-down It makes a rise of every fall, a smile of every frown --------------------------- Newsgroups: comp.os.vxworks Subject: Tornado 2.0.1 dependency generation Date: Fri, 17 Aug 2001 15:01:50 -0400 From: Ben Kuhn Organization: Improv Systems, Inc. Message-ID: <3B7D6A1E.3D57949F@improvsys.com> Reply-To: benk@improvsys.com This is a multi-part message in MIME format. - --------------8513B90E54A949DF80957CC5 Content-Type: text/html; charset=us-ascii Content-Transfer-Encoding: 7bit  
I have this recurring problem in Tornado 2.0.1 where when I add my source files to the project, then modify the build line to also use my include directories, it consistently seems to find some files and not find others, even when they're in the same directory.  Other times it will list a dependency twice, once finding in the right location, and a second time not finding it.

Can anyone give me a brief explanation of how Tornado decides where include files reside, so I can stop having to constantly futz with the internals of the project files?  I feel like I must be missing something pretty basic.

--

Ben Kuhn
  - --------------8513B90E54A949DF80957CC5 Content-Type: text/x-vcard; charset=us-ascii; name="benk.vcf" Content-Transfer-Encoding: 7bit Content-Description: Card for Ben Kuhn Content-Disposition: attachment; filename="benk.vcf" begin:vcard n:Kuhn;Benjamin tel;cell:716 739 2171 tel;work:(716) 756 - 5411 x25 x-mozilla-html:FALSE adr:;;;;;; version:2.1 email;internet:benk@improvsys.com fn:Benjamin Kuhn end:vcard - --------------8513B90E54A949DF80957CC5-- --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Tornado 2.0.1 dependency generation Date: 17 Aug 2001 21:26:36 GMT From: Johan Borkhuis Organization: Agere Systems Message-ID: References: <3B7D6A1E.3D57949F@improvsys.com> Ben Kuhn wrote: > >  
I have this recurring problem in Tornado 2.0.1 where when I > add my > source files to the project, then modify the build line to also use my > include directories, it consistently seems to find some files and not > find others, even when they're in the same directory.  Other times > it will list a dependency twice, once finding in the right location, > and a second time not finding it. >

Can anyone give me a brief explanation of how Tornado decides where > include files reside, so I can stop having to constantly futz with the > internals of the project files?  I feel like I must be missing > something pretty basic. >

-- >

Ben Kuhn >
  Some advise for you. The first on is on posting: please do not post using HTML, it looks very strange, as you might see above. Now your problem. Some things that might help are: - - Use forward slashes instead of backward, the dependencies generator is easily confused by the backward slashes - - Do not use spaces between the # and the include - - If you examine the dependencies for the source files you will see that sometimes you see "../../" and other times only "". Take a look at your source files, and see if you can get these include statements equal. 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: Tornado 2.0.1 dependency generation Date: Fri, 17 Aug 2001 17:55:02 -0400 From: Ben Kuhn Organization: Improv Systems, Inc. Message-ID: <3B7D92B6.FCD3A009@improvsys.com> References: <3B7D6A1E.3D57949F@improvsys.com> Reply-To: benk@improvsys.com This is a multi-part message in MIME format. - --------------4C62EB32F59296063F5FDB3D Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Johan Borkhuis wrote: > Some advise for you. The first on is on posting: please do not post using > HTML, it looks very strange, as you might see above. > My apologies about the HTML, thought I had that turned off. I suppose that's what I get for using Netscape and not double checking. > Now your problem. Some things that might help are: > - Use forward slashes instead of backward, the dependencies generator is > easily confused by the backward slashes > - Do not use spaces between the # and the include > - If you examine the dependencies for the source files you will see that > sometimes you see "../../" and other times only " file>". Take a look at your source files, and see if you can get these > include statements equal. Thanks for the tips. I experimented with all of these. Changing the '\'s to '/'s seems to have helped, though it seems counterintuitive, since tornado still shows '\'s in the build window and window titles. I'm confused about the previously noted strange behavior though, and it hasn't completely gone away. Ben Kuhn - --------------4C62EB32F59296063F5FDB3D Content-Type: text/x-vcard; charset=us-ascii; name="benk.vcf" Content-Transfer-Encoding: 7bit Content-Description: Card for Ben Kuhn Content-Disposition: attachment; filename="benk.vcf" begin:vcard n:Kuhn;Benjamin tel;cell:716 739 2171 tel;work:(716) 756 - 5411 x25 x-mozilla-html:FALSE adr:;;;;;; version:2.1 email;internet:benk@improvsys.com fn:Benjamin Kuhn end:vcard - --------------4C62EB32F59296063F5FDB3D-- --------------------------- Newsgroups: comp.os.vxworks Subject: Re: pagefaults during startup if network is plugged in Date: Fri, 17 Aug 2001 18:10:21 -0700 From: "Denis" Message-ID: <9lkf5p$9v54l$1@ID-44494.news.dfncis.de> References: <9lf5uu$93qqc$1@ID-44494.news.dfncis.de> thanks a bunch for the suggestions. i will look in the areas thought to cause a problem. denis --------------------------- Newsgroups: comp.os.vxworks Subject: Inherited Priority Problems? Date: 17 Aug 2001 20:06:56 -0700 From: chwin@china.com (Adom) Organization: http://groups.google.com/ Message-ID: <92dbccc1.0108171906.1f47a6cb@posting.google.com> Thanks for reading these: I am developing a project on Vxworks 5.4 using Tornado II. Now i found a strang thing. I start a task with priority 110 when system boot . It creates a mutex to protect a container used by net task and itself . At beginning,when i use commond 'i' in shell, the task status is "PEND+T" with priority 110 and the net task priority is 50. After some time , i do this again , the task status changes to "PEND+T+I" with priority 50 . It puzzles me . Can anyone help me ? Any advice or hint is appreciated very much! Adom 2001/8/18 --------------------------- Newsgroups: comp.os.vxworks Subject: the problem of debugging ppc860 BSP has been solved Date: 17 Aug 2001 21:52:45 -0700 From: kevinatnj@yahoo.com.cn (kewin) Organization: http://groups.google.com/ Message-ID: <221caa76.0108172052.7fdc8b75@posting.google.com> problem has been solved! single step 7.4 does not support this function when debugging vxWorks BSP.I have update single step 7.4 to version 7.6.Now it is working very well. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Debugging when the debugger crashes Date: Fri, 17 Aug 2001 22:50:24 -0700 From: DrDiags Organization: laserlink.net Message-ID: <3B7E0220.9ACC02DE@covad.net> References: <2psil9.q9e.ln@127.0.0.1> Reply-To: drdiags@covad.net David, How about setting up a post-mortem windview trace? If you can set aside a section of memory and can reboot the target without having to power-cycle (reset switch) this may help. David Given wrote: > > I've got a particularly annoying problem on my MIPS-based device. I have > Tornado attached to it via ethernet; unfortunately, my application crashes > so hard that it takes the VxWorks ethernet driver down with it. (I think > it's locking up with interrupts locked.) This means, of course, that the > debugger doesn't work any more. > > I suspect it's jumping into the wild blue yonder and stomping all over > memory. I can tell this because I see gibberish appear on the screen once > it's locked. But I can't tell any more about it, because I can't debug it. > > What I'd really like is ICE or a JTAG debugger, but unfortunately my > hardware doesn't support it; all I can do is use the ethernet debugger. > Does anyone have any suggestions as to how I might be able to tell what's > going on here? The main thing I'd like to know is the last thing it was > doing before it died... > > -- > David Given > Tao Group Ltd. - -- http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html (vxWorks FAQ home page) --------------------------- Newsgroups: comp.os.vxworks Subject: memory 'free' suspends task... Date: Sat, 18 Aug 2001 06:04:39 GMT From: "Changhyun Oe" Organization: Dacom Message-ID: When I tried 'free'ing the memory, It always suspends the task which calls 'free'. Please tell me why... I need your help.... Previous Thanks to you~ - - chang - --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Tornado 2.0.1 dependency generation Date: Fri, 17 Aug 2001 23:13:25 -0700 From: DrDiags Organization: laserlink.net Message-ID: <3B7E0785.C276A3A5@covad.net> References: <3B7D6A1E.3D57949F@improvsys.com> <3B7D92B6.FCD3A009@improvsys.com> Reply-To: drdiags@covad.net Ben, Something I found in this area is if this is occurs on a Windows Host, make sure the directory name used in the -I statement is capatilized exactly as if this was a UNIX machine. If your directory is MyDirectory and you put -ID:/MYDiRECtory, the project tool may not be able to find the header files (at least on WinNT). I also was told to put my -I. as the last -I flag, but I am not too sure about that one. Things I do now when creating test applications or building new projects, is that I put the -Ipath in my build properties before I even start to add source file, because I have seen dependency files that had no path associated with them, then of course you get the no rules to make error. Here is some other info I got, but once again you would have to take it with a grain of salt because it seems somewhat simplistic: Dependencies Problem - - First change all #include to #include "xxx.h" If your includes are for headers that appear in multiple places, changing from #include to #include "xxx.h" may be a good idea. Problem with source search path for included files may be related to double quotes vs. pointed brackets. This should work as follows: source file that includes header file which happens to be in multiple locations, if included with quotes, should look in same directory as the main source file, then start using -I included paths. - - Specify VPATH for as described in the next document, "How to generate external dependencies in Tornado 2.0." VPATH is for all file dependencies, -i is for header files only. ================================================================== TITLE: How to generate external dependencies in Tornado 2.0? DESCRIPTION: How can I generate external dependencies in Tornado 2.0? I have two directories, source_code (where all the source codes are) and header (where all the header files are). I add all the source and header files into the downloadable project, then select "Dependencies" from the workspace context menu. All the header files are shown in the "external Dependencies" folder in the "Files" view of the workspace window. Then I click the "Build Project" from the tool bar, I got the following error message: make: *** No rule to make target `bee.h', needed by `bee.o'. Stop. Why does the build fail? SUGGESTION: The following procedures will generate a correct dependencies in order to build the object modules. Step 1: Add C/C++ compiler option "-I" in the build specification To display the build specification, double-click on the build name in the "Build" view of the workspace to display the property sheet for the build. Then, select the "C/C++ compiler" tab of the build specification property sheet to view the compiler options. Add another option in the text box: -Idirectory_path_of_the_header_files For example: -IE:\header -IE:\header2 Each directory path requires a separated "-I" option. Be sure to include a -I.. as the last of the I options so that the compiler will look in the local directory as well. It should be the last option. Step 2: Add makefile macro "VPATH" in the build specification Select the "Macros" tab of the build specification property sheet to view the makefile macros associated with the current project. Add the new macro VPATH by changing the name of an existing macro. Replace the value of the existing macro by the search path(s) of the header files, separated by spaces. For example: E:\header1 E:\header2 Then click the "Add/Set" button and the "OK" button. Step 3: Rebuild the dependencies Select "Dependencies" from the workspace context menu and click the OK button to rebuild the dependencies. Step 4: Rebuild the object modules/application You can now click "ReBuild All" button in the tool bar to rebuild the object modules/application. The new compiler option "-I" will be pulled into the dependencies and the compiler will search the specified directories for the header files. - -- http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html (vxWorks FAQ home page) --------------------------- Newsgroups: comp.os.vxworks Subject: Re: memory 'free' suspends task... Date: Fri, 17 Aug 2001 23:31:28 -0700 From: DrDiags Organization: laserlink.net Message-ID: <3B7E0BC0.B9EC83A3@covad.net> References: Reply-To: drdiags@covad.net Changhyun, This is how the free() will work if you attempt to free a pointer that is no longer pointing to malloc'd space from a task. This should be documented in the online manuals for vxWorks. What you should do is check to see if the pointer is NULL before freeing and after successfully freeing memory, set the pointer to NULL. I have seen bugs in the past where global pointers are set to malloc spaced, then some task frees it, but another task in an cleanup role, attempts to free all pointers which are non-NULL. The other technique you can use to stop taskSuspend() from being called is to spawn your task with the VX_UNBREAKABLE option set. Do that, and you can call free without having the task suspend. Changhyun Oe wrote: > > When I tried 'free'ing the memory, > > It always suspends the task which calls 'free'. > > Please tell me why... > > I need your help.... > > Previous Thanks to you~ > > - chang - - -- http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html (vxWorks FAQ home page) --------------------------- Newsgroups: comp.os.vxworks Subject: how to send broadcast messages with dest ip "255.255.255.255" ? Date: Sat, 18 Aug 2001 16:21:03 +0900 From: "Â÷Çö¿í" Organization: Korea Telecom Message-ID: <9ll4t2$h6i$1@news1.kornet.net> Reply-To: "Â÷Çö¿í" Hi, I want to know how to send broadcast messages which are destined to "255.255.255.255". I tried "routeAdd "255.255.255.255" ..." on shell prompt. Anyone knows? --------------------------- Newsgroups: comp.os.vxworks Subject: Re: memory 'free' suspends task... Date: Sat, 18 Aug 2001 03:34:54 -0700 From: "Albert H Chen" Message-ID: References: You might be freeing a piece of memory that can't be freed either because it is a stack var. or it has been freed already... therefore vxWorks calls the default exception handler which then suspends your task... try running tasktrace on your task and see what you come up with.... Most of the time it is mempartfree error... HTH Albert "Changhyun Oe" wrote in message news:Xznf7.16684$2k2.904864@news.bora.net... > When I tried 'free'ing the memory, > > It always suspends the task which calls 'free'. > > Please tell me why... > > I need your help.... > > Previous Thanks to you~ > > - chang - > > > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Inherited Priority Problems? Date: Sat, 18 Aug 2001 03:46:14 -0700 From: "Albert H Chen" Message-ID: References: <92dbccc1.0108171906.1f47a6cb@posting.google.com> When you created the MSemaphore, you probably enabled the priority inversion protection by passing in SEM_INVERSION_SAFE... thus when your task is holding on the semaphore and the net task is waiting for it to release... your task temporary inherited the net task priority. It will actually inherit the highest priority of all tasks that is queued to the semaphore... in your case 50. This is to solve the problem of lower priority task preventing a higher priority task from running if lower priority task has the semaphore and is been pre-empted by a medium priority task... therefore starving the higher priority task. HTH Albert "Adom" wrote in message news:92dbccc1.0108171906.1f47a6cb@posting.google.com... > Thanks for reading these: > > I am developing a project on Vxworks 5.4 using Tornado II. > Now i found a strang thing. > I start a task with priority 110 when system boot . > It creates a mutex to protect a container used by net task and > itself . > At beginning,when i use commond 'i' in shell, > the task status is "PEND+T" with priority 110 and the net task > priority is 50. > After some time , i do this again , the task status changes to > "PEND+T+I" with priority 50 . > It puzzles me . > Can anyone help me ? > Any advice or hint is appreciated very much! > > Adom > 2001/8/18 --------------------------- Newsgroups: comp.os.vxworks Subject: Re: memory 'free' suspends task... Date: Sat, 18 Aug 2001 11:47:06 +0100 From: molochai Message-ID: <7568edab4a%molochai@vapour-trail.demon.co.uk> References: In message "Albert H Chen" wrote: > You might be freeing a piece of memory that can't be freed either because > it is a stack var. or it has been freed already... therefore vxWorks calls > the default exception handler which then suspends your task... try running > tasktrace on your task and see what you come up with.... Most of the time > it is mempartfree error... Alternatively your application has trampled part of the free list (this ISTR is consistency checked). A Common fault is exceeding array-bounds on memory allocated on the heap. --------------------------- End of New-News digest ********************** From vxwexplo-errs@csg.lbl.gov Sun Aug 19 04:03:19 2001 From: Vxworks Exploder Date: Sun Aug 19 04:03:22 PDT 2001 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Sun Aug 19 04:03:03 PDT 2001 Subject: Re: vxWorks errno's Subject: GDB problems with Tornado 2 for Mips (libR4000gnuvx.a) Subject: Re: GDB problems with Tornado 2 for Mips (libR4000gnuvx.a) Subject: Re: GDB problems with Tornado 2 for Mips (libR4000gnuvx.a) ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: vxWorks errno's Date: Wed, 15 Aug 2001 12:10:04 +0200 From: "Anne B. Knudtzon" Organization: Schlumberger,Geco-Prakla Message-ID: <3B7A4A7C.14BC1BC8@oslo.geco-prakla.slb.com> References: <9l953f$ikl$1@qnsgh006.europe.nortel.com> From the vxWorks file vwModNum.h you can find which module the 'd' is: #define M_iosLib (13 << 16) and in iosLib.h you then find the error #define S_iosLib_INVALID_FILE_DESCRIPTOR (M_iosLib | 3) and then remains the quest for the meaning behind this! I never found the problems behind the errors we get in our tasks... Good luck! anne Paul D'Arcy wrote: > > How can one determine or where can one find what the errno's actually > mean??? > For example, from vxWorks shell, > -> i > > NAME ENTRY TID PRI STATUS PC SP ERRNO > DELAY > ---------- ------------ -------- --- ---------- -------- -------- ------- -- > --- > tHTTPd ad298 101de44 200 PEND+T 73780 101dca4 0 > 28510 > tHTTPd ad298 1016768 200 PEND+T 73780 10165c8 d0003 > 28500 > tDcacheUpd _dcacheUpd 15cf36c 250 DELAY 7070c 15cf2ec 3006c > 21 > value = 0 = 0x0 > -> > > What does the value d0003 mean in the line : tHTTPd ad298 > 1016768 200 PEND+T 73780 10165c8 d0003 28500 ???? > Regards > Paul --------------------------- Newsgroups: comp.os.vxworks Subject: GDB problems with Tornado 2 for Mips (libR4000gnuvx.a) Date: Sat, 18 Aug 2001 21:20:58 GMT From: "Nihad Alibegovic" Organization: Road Runner Message-ID: <__Af7.2724$Qh2.545742@typhoon.san.rr.com> Hello everyone, we ported our BSP from vxWorks 5.3 to vxWorks5.4 on Tornado 2. When it was all said and done everything works except GDB. Target attaches fine we can put break points but as soon as we try to single step we get tWdb exception. It is worth noting that if I put a break point in from the host shell but single step in target resident shell things work well. So, I guess this is not just a GDB problem. Additionally, I have same problem when trying to attach to the task. It is as if the host and target memories are completely out of sync and host is accessing memory it is not supposed to access on the target. We use Network to debug which means we are in task mode debugging only. The platform we run on is Mips 4310 CPU so we use libR4000gnuvx.a. We have all the WDB includes in our BSP that we need as certified by Wind River. Wind River has helped very little in this issue in the last week. While, at this does anyone get quick support from these guys. In 2 years working with vxWorks they have the slowest customer support of any company I ever worked with. Nihad Alibegovic MTS3 nalibegovic@hns.com --------------------------- Newsgroups: comp.os.vxworks Subject: Re: GDB problems with Tornado 2 for Mips (libR4000gnuvx.a) Date: Sat, 18 Aug 2001 16:10:23 -0700 From: DrDiags Organization: laserlink.net Message-ID: <3B7EF5DF.D4509E9@covad.net> References: <__Af7.2724$Qh2.545742@typhoon.san.rr.com> Reply-To: drdiags@covad.net Nihad, Was your END driver created by Wind River or did you do your own? I know that vxWorks 5.3 used SENS 1.0 while vxWorks 5.4 uses SENS 2.0. Does all your other networking work as expected? Are you even using END instead of the BSD network drivers (if_xxx) type drivers. If you have written your own network driver, maybe you could download the Network Protocol Toolkit Users guide. There is a section in it that talks about porting your vxWorks 5.3 driver to work in vxWorks 5.4. Another painful test might be to enable your kernel to use the serial port for WDB. If you test works under this condition, it might help you identify the area the disconnect is occuring. Just some random thoughts...... - -- http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html (vxWorks FAQ home page) --------------------------- Newsgroups: comp.os.vxworks Subject: Re: GDB problems with Tornado 2 for Mips (libR4000gnuvx.a) Date: Sat, 18 Aug 2001 23:19:25 GMT From: "Nihad Alibegovic" Organization: Road Runner Message-ID: <1KCf7.2744$Qh2.589102@typhoon.san.rr.com> References: <__Af7.2724$Qh2.545742@typhoon.san.rr.com> <3B7EF5DF.D4509E9@covad.net> This is not END driver. Sorry I forgot to mention that. The driver is BSD style driver and we simply ported it from 5.3 to 5.4 according to Network Protocol Toolkit guide. The driver actually works fine as far as I can tell. The throughput is the same and all of the connectivity is there. I can ping in and out, use the box for ftp, telnet, rlogin, etc. The driver is if_lnPci.c for AMD Lance controller. I will try the painful test with serial port as you suggested but thus far it all points to somewhere else but the LAN driver. I say this because we can connect Tornado and do symbol loads and synhronization. As a matter of fact we can do everything except debug. "DrDiags" wrote in message news:3B7EF5DF.D4509E9@covad.net... > Nihad, > > Was your END driver created by Wind River or did you do your > own? I know that vxWorks 5.3 used SENS 1.0 while vxWorks 5.4 uses > SENS 2.0. Does all your other networking work as expected? Are you > even using END instead of the BSD network drivers (if_xxx) type > drivers. If you have written your own network driver, maybe you > could download the Network Protocol Toolkit Users guide. There is > a section in it that talks about porting your vxWorks 5.3 driver > to work in vxWorks 5.4. Another painful test might be to enable > your kernel to use the serial port for WDB. If you test works under > this condition, it might help you identify the area the disconnect > is occuring. Just some random thoughts...... > > > -- > http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html (vxWorks FAQ home > page) > --------------------------- End of New-News digest ********************** From vxwexplo-errs@csg.lbl.gov Mon Aug 20 01:11:46 2001 From: "Rashid Raj" Date: Mon Aug 20 01:11:48 PDT 2001 Subject: Dynamic PCI Bus Management in VxWorks Hello Everyone, I want to implement Hot-Swap on VxWorks 5.4. The cards are on a cPCI chassis. I know that during PCI Auto-configuration we can dynamicaly allocate the resources to the PCI devices. But how to implement it under Hot-Swap. Do we have to re-configure all the devices whenever a particular device is inserted or removed ? Can anybody suggest the solution ? Thanking you all in anticipation. Rashid Logic Eastern (I) Pvt. Ltd. Noida From vxwexplo-errs@csg.lbl.gov Mon Aug 20 04:04:07 2001 From: Vxworks Exploder Date: Mon Aug 20 04:04:10 PDT 2001 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Mon Aug 20 04:03:06 PDT 2001 Subject: Re: check this out Subject: netJobAdd Subject: About memory snapshot for watchdog Subject: test Subject: Re: Using Sym8xx SCSI adapter on PPC Subject: Re: How can i know whether the network cable is plugged in or not under Subject: Watchdog timer problems ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: check this out Date: Sun, 19 Aug 2001 16:33:24 +0100 From: molochai Message-ID: <97748bac4a%molochai@vapour-trail.demon.co.uk> References: In message tvnmfb@zwallet.com wrote: > DID YOU KNOW IF YOU DONT KNOW JESUS AS YOUR PERSONAL SAVIOR YOU WILL GO TO HELL. REPENT YOURSELF! > http://www.radiovie.qc.ca > killed already.. kewl :-) --------------------------- Newsgroups: comp.os.vxworks Subject: netJobAdd Date: 19 Aug 2001 17:14:26 -0700 From: pepsi_000@hotmail.com (vxJunkie) Organization: http://groups.google.com/ Message-ID: <676eb679.0108191614.64b8383d@posting.google.com> hi, We are using Motorola's PCC-860 processor. I have an ISR which does some amount of processing and calls netJobAdd. void ISR(void) { ***** ***** ctr1++; netJobAdd(func1,....); } void func1(void) { *** ctr2++; } After a while i find that "ctr2" stops incrementing, even though "ctr1" is continously incrementing. I checked all the tasks giving an "i" command. None of the tasks were suspended, nor did i see any errno. I did not get any work Q panic errors either. I also checked the return value of netJobAdd and that was fine too. I feel the interrupt rate is high. But why didn't this cause a work Q panic error? I am still able to ping the processor. Thanks in advance.. --------------------------- Newsgroups: comp.os.vxworks Subject: About memory snapshot for watchdog Date: Mon, 20 Aug 2001 10:22:12 +0800 From: "survey liu" Organization: Bentium Ltd. (CN99) Message-ID: <9lps5p$4i6$1@mail.cn99.com> Hi,Pals I want to add somes features on my system. It is to dump out the snapshot of memory after system crash/reset. How could I do? I am not willing to affect the system performance. Maybe the "Core dump" tech. of gcc/gdb can help? Do you have some good ideas? - -- Best regards, Survey Liu 4/F Wearnes Technology Bldg. No.10 Kefa Rd.,Science & Industry Park Nanshan Shenzhen 518057,PRC Tel: +86-755-6635333-4300 Fax: +86-755-6635550 Email: survey.liu@utstar.com Web: http://www.utstar.com --------------------------- Newsgroups: comp.os.vxworks Subject: test Date: Mon, 20 Aug 2001 11:33:40 +0900 From: "Â÷Çö¿í" Organization: A poorly-installed InterNetNews site Message-ID: <9lpsf3$bs9$1@news2.kornet.net> Reply-To: "Â÷Çö¿í" --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Using Sym8xx SCSI adapter on PPC Date: Mon, 20 Aug 2001 07:04:16 GMT From: "Koen Van Hulle" Organization: Xeikon Message-ID: References: <3B7D07E3.B77514A8@foi.se> Torbjörn, Thanks for letting me know. In the mean time, I have managed to get the local Wind River FAE interested in looking into my problem. Let's hope he can shed some light on this. Koen "Torbjörn Frostemark" wrote in message news:3B7D07E3.B77514A8@foi.se... > Koen Van Hulle wrote: > > > Has any one been able to use a Symbis Logic SCSI controller with VxWorks on > > the PPC? > > I have used sym825 and sym875 on a MPC750 (MVME2700 processor board has the > sym825 built in and the sym875 is on a PMC module) and a sym895 on a MPC7400 > (MVME5100 with a IPMC712 module). I had no problems with the default values, but > I had to patch T2.1 with the sym895 driver. > > Now you know it works, but I cannot help you more... > > /Torbjörn > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: How can i know whether the network cable is plugged in or not under Vxworks? Date: Mon, 20 Aug 2001 10:29:52 +0100 From: James Marshall Organization: Agilent Technologies UK Ltd Message-ID: <3B80D890.506A005F@agilent.com> References: <9la2pq$5pn$1@mail.cn99.com> <3b794d05$0$90634$e2e8da3@nntp.cts.com> <9lfbm1$nsj$1@mail.cn99.com> <3b7d4a0d$0$90634$e2e8da3@nntp.cts.com> > One wire is the > clock (that you toggle), and the other wire is the data. You need 2 general > purpose lines for this. Some controllers (for example, Intel 82557/8/9/9ER) have their own MDIO interface through a register so you don't need to provide your own clock - you just write a register and wait for status to indicate when its done. I think you can also enable an interrupt to indicate "MDIO operation complete", although an operation only takes about 32-33us. James Marshall. --------------------------- Newsgroups: comp.os.vxworks Subject: Watchdog timer problems Date: Mon, 20 Aug 2001 12:21:58 +0200 From: Glenn Ivar Klausen Organization: Nextra Public Access Message-ID: <3B80E4C6.6CA7164D@oslo.online.no> I've written som watchdog timer example code, but it dosent seem to be working. If anyoun could tell med the problem here. #include #include WDOG_ID wid; int wdTimerid; void wdExpired(void *msg); void wdTest(void) { wid = wdCreate(); if(wdStart(wid, (sysClkRateGet()*5), wdExpired, "timer expired") == OK) { fprintf(stdout, "Timer created ok. (wid = %lx) \n", wid); } else { fprintf(stdout, "Timer failed."); } } void wdExpired(void *msg) { fprintf(stdout, " msg = %s \n", (char *)msg); } As far as I can see, this is supposed to work. But it dont. - -- Glenn --------------------------- End of New-News digest ********************** From vxwexplo-errs@csg.lbl.gov Mon Aug 20 05:02:55 2001 From: "Zhu Zhiqiang" Date: Mon Aug 20 05:02:57 PDT 2001 Subject: I want to join Vxworks mailing list This is a multi-part message in MIME format. ------=_NextPart_000_0021_01C129B3.32580A80 Content-Type: text/plain; charset="gb2312" Content-Transfer-Encoding: base64 SSB3YW50IHRvIGpvaW4gVnh3b3JrcyBtYWlsaW5nIGxpc3QNCg== ------=_NextPart_000_0021_01C129B3.32580A80 Content-Type: text/html; charset="gb2312" Content-Transfer-Encoding: base64 PCFET0NUWVBFIEhUTUwgUFVCTElDICItLy9XM0MvL0RURCBIVE1MIDQuMCBUcmFuc2l0aW9uYWwv L0VOIj4NCjxIVE1MPjxIRUFEPg0KPE1FVEEgY29udGVudD0idGV4dC9odG1sOyBjaGFyc2V0PWdi MjMxMiIgaHR0cC1lcXVpdj1Db250ZW50LVR5cGU+DQo8TUVUQSBjb250ZW50PSJNU0hUTUwgNS4w MC4yNjE0LjM1MDAiIG5hbWU9R0VORVJBVE9SPg0KPFNUWUxFPjwvU1RZTEU+DQo8L0hFQUQ+DQo8 Qk9EWSBiZ0NvbG9yPSNmZmZmZmY+DQo8RElWPjxGT05UIHNpemU9Mj5JIHdhbnQgdG8gam9pbiBW eHdvcmtzIG1haWxpbmcgbGlzdDwvRk9OVD48L0RJVj48L0JPRFk+PC9IVE1MPg0K ------=_NextPart_000_0021_01C129B3.32580A80-- From vxwexplo-errs@csg.lbl.gov Mon Aug 20 05:03:22 2001 From: "Zhu Zhiqiang" Date: Mon Aug 20 05:03:25 PDT 2001 Subject: I want to join Vxworks mailing list This is a multi-part message in MIME format. ------=_NextPart_000_002A_01C129B3.42C86A40 Content-Type: text/plain; charset="gb2312" Content-Transfer-Encoding: base64 SSB3YW50IHRvIGpvaW4gVnh3b3JrcyBtYWlsaW5nIGxpc3QNCg== ------=_NextPart_000_002A_01C129B3.42C86A40 Content-Type: text/html; charset="gb2312" Content-Transfer-Encoding: base64 PCFET0NUWVBFIEhUTUwgUFVCTElDICItLy9XM0MvL0RURCBIVE1MIDQuMCBUcmFuc2l0aW9uYWwv L0VOIj4NCjxIVE1MPjxIRUFEPg0KPE1FVEEgY29udGVudD0idGV4dC9odG1sOyBjaGFyc2V0PWdi MjMxMiIgaHR0cC1lcXVpdj1Db250ZW50LVR5cGU+DQo8TUVUQSBjb250ZW50PSJNU0hUTUwgNS4w MC4yNjE0LjM1MDAiIG5hbWU9R0VORVJBVE9SPg0KPFNUWUxFPjwvU1RZTEU+DQo8L0hFQUQ+DQo8 Qk9EWSBiZ0NvbG9yPSNmZmZmZmY+DQo8RElWPjxGT05UIHNpemU9Mj5JIHdhbnQgdG8gam9pbiBW eHdvcmtzIG1haWxpbmcgbGlzdDwvRk9OVD48L0RJVj48L0JPRFk+PC9IVE1MPg0K ------=_NextPart_000_002A_01C129B3.42C86A40-- From vxwexplo-errs@csg.lbl.gov Mon Aug 20 05:03:53 2001 From: "Zhu Zhiqiang" Date: Mon Aug 20 05:03:55 PDT 2001 Subject: I want to join Vxworks mailing list This is a multi-part message in MIME format. ------=_NextPart_000_0033_01C129B3.555FA1A0 Content-Type: text/plain; charset="gb2312" Content-Transfer-Encoding: base64 SSB3YW50IHRvIGpvaW4gVnh3b3JrcyBtYWlsaW5nIGxpc3QNCg== ------=_NextPart_000_0033_01C129B3.555FA1A0 Content-Type: text/html; charset="gb2312" Content-Transfer-Encoding: base64 PCFET0NUWVBFIEhUTUwgUFVCTElDICItLy9XM0MvL0RURCBIVE1MIDQuMCBUcmFuc2l0aW9uYWwv L0VOIj4NCjxIVE1MPjxIRUFEPg0KPE1FVEEgY29udGVudD0idGV4dC9odG1sOyBjaGFyc2V0PWdi MjMxMiIgaHR0cC1lcXVpdj1Db250ZW50LVR5cGU+DQo8TUVUQSBjb250ZW50PSJNU0hUTUwgNS4w MC4yNjE0LjM1MDAiIG5hbWU9R0VORVJBVE9SPg0KPFNUWUxFPjwvU1RZTEU+DQo8L0hFQUQ+DQo8 Qk9EWSBiZ0NvbG9yPSNmZmZmZmY+DQo8RElWPjxGT05UIHNpemU9Mj5JIHdhbnQgdG8gam9pbiBW eHdvcmtzIG1haWxpbmcgbGlzdDwvRk9OVD48L0RJVj48L0JPRFk+PC9IVE1MPg0K ------=_NextPart_000_0033_01C129B3.555FA1A0-- From vxwexplo-errs@csg.lbl.gov Mon Aug 20 05:04:18 2001 From: "Zhu Zhiqiang" Date: Mon Aug 20 05:04:20 PDT 2001 Subject: I want to join Vxworks mailing list This is a multi-part message in MIME format. ------=_NextPart_000_0041_01C129B3.638DB280 Content-Type: text/plain; charset="gb2312" Content-Transfer-Encoding: base64 DQo= ------=_NextPart_000_0041_01C129B3.638DB280 Content-Type: text/html; charset="gb2312" Content-Transfer-Encoding: base64 PCFET0NUWVBFIEhUTUwgUFVCTElDICItLy9XM0MvL0RURCBIVE1MIDQuMCBUcmFuc2l0aW9uYWwv L0VOIj4NCjxIVE1MPjxIRUFEPg0KPE1FVEEgY29udGVudD0idGV4dC9odG1sOyBjaGFyc2V0PWdi MjMxMiIgaHR0cC1lcXVpdj1Db250ZW50LVR5cGU+DQo8TUVUQSBjb250ZW50PSJNU0hUTUwgNS4w MC4yNjE0LjM1MDAiIG5hbWU9R0VORVJBVE9SPg0KPFNUWUxFPjwvU1RZTEU+DQo8L0hFQUQ+DQo8 Qk9EWSBiZ0NvbG9yPSNmZmZmZmY+DQo8RElWPiZuYnNwOzwvRElWPjwvQk9EWT48L0hUTUw+DQo= ------=_NextPart_000_0041_01C129B3.638DB280-- From vxwexplo-errs@csg.lbl.gov Mon Aug 20 10:01:21 2001 From: "David Laight" Date: Mon Aug 20 10:01:23 PDT 2001 Subject: Re: PCI hot swap > I want to implement Hot-Swap on VxWorks 5.4. The cards are on a cPCI chassis. > I know that during PCI Auto-configuration we can dynamicaly allocate the > resources to the PCI devices. But how to implement it under Hot-Swap. > Do we have to re-configure all the devices whenever a particular device is > inserted or removed ? I did some work on this a few years ago.... got as far as demostrating that you could remove (and replace) a working card from a working system! 'Hot Swap' (replacing a broken card with a direct replacement) is relatively easy - given you have hardware that can isolate all the bus signals and ensure that the PCI bus isn't 'parked' on the card you isolate. 'Hot insertion' (add a card to an empty slot) is an entirely different ball game. Reconfiguring the entire PCI bus is almost certainly not possible - you would have to have special drivers for all the hardware that allowed the I/O (etc) addresses of the device ports to be dynamically changed. This would be a significant cost to the entire system. This means that you have to give a 'new' card resources that are already assigned to the relevant part of the PCI bus. In practise this means that the resources have to be reserved when the initial system grope is done. Ok, you may get away with using the resources 'just above' those already used - but this wont work in the most general cases. The big problem is with cards that contain PCI-PCI (or equivalent) bridges, particularly when they sit behind other PCI bridges. A PCI bus will forward a dense range of each resource to its subsiduary bus. As well as I/O addresses and memory addresses, PCI bus numbers have to be allocated densly. One really horrid case are CARDBUS PCI extenders (these let you connect a PCI chassis to a notebook's Cardbus (c/f pcmcia) socket). Theoretically these are hot-pluggable, in practise (rtfm) you find that they have to be inserted before the system is booted (we nearly got hot insertion working under solaris 8). The problem is the number of PCI-PCI bridges and the unknown amount of I/O and memory space that has to be configured in the system bridges. In any case, after the 'insertion' you only need to grope the slot that has the new card and allocate it resources that are 'reserved' for that slot. Beware - some graphics cards don't seem to like a second PCI grope - even trying to read the board-id can lock the system solid. If you want to remove broken boards, you will need to ensure the relevant device driver is hardened to PCI bus errors and random data reads! None ov the standard vxWorks code will be... David From vxwexplo-errs@csg.lbl.gov Mon Aug 20 10:14:04 2001 From: fscheurer@manz-acs.de Date: Mon Aug 20 10:14:06 PDT 2001 Subject: Re-2: Compile windoze T2 makefiles in cygwin bash shell? Hi, as I recently tried to use the cygwin-tool-chain as well, I assume that it is a PATH-problem. If you use the bash from Cygwin, the bash environment variable PATH sets the Cygwin paths in front of the standard environment of Windows. So the Cygwin binaries are found before your Tornado tools. Think about it! MfG / Kind Regards F. Scheurer ---------------------------------------------------------------- Friedrich Scheurer, Dipl.-Ing. phone +49 7121 9019-23(0) ACS GmbH - R&D fax +49 7121 9019-98 Steigaeckerstrasse 13 web www.manz-acs.de D-72768 Reutlingen (Germany) e-mail: fscheurer@manz-acs.de ---------------------------------------------------------------- -------- Original Message -------- Subject: Re: Compile windoze T2 makefiles in cygwin bash shell? (12-Aug-2001 19:48) From: vxwexplo@lbl.gov To: FScheurer@manz-acs.de > Submitted-by vxwexplo-errs@csg.lbl.gov Sun Aug 12 08:52:17 2001 > Submitted-by: =?iso-8859-1?q?ram=20mohan?= > > hey, > this is because ccppc internally uses cpp.exe to > compile > and cpp is architecture(also os) dependent also it > uses one more os dependend application as.exe > as far as i know there are different combinations of > options for each > but i never tried changing the options for ex using > ppc cpp.exe with intel cpp.exe but appropriate option > u can try that finding out what options u need > try this out > if it does not work there is no other go except u > need to buy tornado for cygwin :( > :) > Mohan > > --- the vxWorks Users Group Exploder > wrote: > Submitted-by > vxwexplo-errs@csg.lbl.gov Thu Aug 9 > > 14:38:49 2001 > > Submitted-by: "Michael Morrison" > > > > > > Howdy, > > > > I have a windoze install of tornado II, but want to > > use the cygwin bash > > shell to run command-line make. > > Unfotunatly, when I do this I get the following > > error: > > > > $ make TARGET_BASE=iscsi > > ccppc > > -B/cygdrive/c/Tornado/host/x86-win32/lib/gcc-lib/ -g > > -nostdinc -f > > > > no-builtin -nostdlib -fno-defer-pop -W -Wall > > -I/cygdrive/c/Tornado/target/h -I/c > > ygdrive/c/Tornado/target/config/hsiSBC8260 -I. > > -I../include -c iscsi.cc > > cpp: -lang-c++: linker input file unused since > > linking not done > > as: unrecognized option `-u' > > ccppc.exe: Internal compiler error: program as got > > fatal signal 1 > > make: *** [iscsi.o] Error 1 > > > > > > The same makefile runs ok in the windoze DOS > > shell.... > > Anyone have any ideas on how to do this? > > > > Do you think the cygwin 'cpp' and 'as' be getting > > run instead of the > > vxworks version? > > > > TIA > > > > Mikem > > > > > > > > From vxwexplo-errs@csg.lbl.gov Mon Aug 20 21:10:02 2001 From: "Vivek Soni" Date: Mon Aug 20 21:10:05 PDT 2001 Subject: Watchdog timer problems Hi Glenn Ur example code works fine for me. The problem is not with watchdog timer. Only problem is with ur fprintf(). Probably u have not included header file for fprintf(). That's why it give GPF while executing fprintf(). - Vivek. Newsgroups: comp.os.vxworks Subject: Watchdog timer problems Date: Mon, 20 Aug 2001 12:21:58 +0200 From: Glenn Ivar Klausen Organization: Nextra Public Access Message-ID: <3B80E4C6.6CA7164D@oslo.online.no> I've written som watchdog timer example code, but it dosent seem to be working. If anyoun could tell med the problem here. #include #include WDOG_ID wid; int wdTimerid; void wdExpired(void *msg); void wdTest(void) { wid = wdCreate(); if(wdStart(wid, (sysClkRateGet()*5), wdExpired, "timer expired") == OK) { fprintf(stdout, "Timer created ok. (wid = %lx) \n", wid); } else { fprintf(stdout, "Timer failed."); } } void wdExpired(void *msg) { fprintf(stdout, " msg = %s \n", (char *)msg); } As far as I can see, this is supposed to work. But it dont. - -- Glenn From vxwexplo-errs@csg.lbl.gov Tue Aug 21 03:06:41 2001 From: "Rashid Raj" Date: Tue Aug 21 03:06:43 PDT 2001 Subject: Re: PCI hot swap Hi David, Thanks a lot for the information. Each slot is having a different resource requirement. Don't you think that if we reserve the resources for each slot before groping than won't it be inefficient use of resources. For example there might be some memory left unsed. Or, should the resources allocated to a slot during the initial grope be reserved for that slot ? Thanks, Rashid Logic Eastern (I) Pvt. Ltd. Noida ---------- Original Message ---------------------------------- From: vxwexplo@lbl.gov (the vxWorks Users Group Exploder) Date: Mon, 20 Aug 2001 10:12:01 -0700 (PDT) >Submitted-by vxwexplo-errs@csg.lbl.gov Mon Aug 20 10:01:21 2001 >Submitted-by: "David Laight" > > >I did some work on this a few years ago.... got as far as demostrating that >you could remove (and replace) a working card from a working system! > >'Hot Swap' (replacing a broken card with a direct replacement) is relatively >easy - given you have hardware that can isolate all the bus signals and ensure >that the PCI bus isn't 'parked' on the card you isolate. > >'Hot insertion' (add a card to an empty slot) is an entirely different ball >game. > >Reconfiguring the entire PCI bus is almost certainly not possible - you would >have to have special drivers for all the hardware that allowed the I/O (etc) >addresses of the device ports to be dynamically changed. This would be a >significant cost to the entire system. > >This means that you have to give a 'new' card resources that are already >assigned to the relevant part of the PCI bus. In practise this means that >the resources have to be reserved when the initial system grope is done. >Ok, you may get away with using the resources 'just above' those already >used - but this wont work in the most general cases. > >The big problem is with cards that contain PCI-PCI (or equivalent) bridges, >particularly when they sit behind other PCI bridges. > >A PCI bus will forward a dense range of each resource to its subsiduary bus. >As well as I/O addresses and memory addresses, PCI bus numbers have to be >allocated densly. > >One really horrid case are CARDBUS PCI extenders (these let you connect a PCI >chassis to a notebook's Cardbus (c/f pcmcia) socket). Theoretically these >are hot-pluggable, in practise (rtfm) you find that they have to be inserted >before the system is booted (we nearly got hot insertion working under solaris >8). The problem is the number of PCI-PCI bridges and the unknown amount of >I/O and memory space that has to be configured in the system bridges. > >In any case, after the 'insertion' you only need to grope the slot that has >the new card and allocate it resources that are 'reserved' for that slot. >Beware - some graphics cards don't seem to like a second PCI grope - even trying >to read the board-id can lock the system solid. > >If you want to remove broken boards, you will need to ensure the relevant device >driver is hardened to PCI bus errors and random data reads! None ov the standard >vxWorks code will be... > > David From vxwexplo-errs@csg.lbl.gov Tue Aug 21 04:03:44 2001 From: Vxworks Exploder Date: Tue Aug 21 04:03:47 PDT 2001 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Tue Aug 21 04:03:16 PDT 2001 Subject: Re: GDB problems with Tornado 2 for Mips (libR4000gnuvx.a) Subject: Re: change sector size? HELP! Subject: Re: Multiple Initiator and one HDD Subject: Re: HELP - vx FTP-Server problem! Subject: MPC8260 BSP net driver problem Subject: Re: protocol stack design under vxworks Subject: Re: ethernet link reset problem Subject: Re: ata flash disk versus ide flash disk Subject: tNetTask does not starts Subject: Re: Tornado 2.0.1 dependency generation Subject: Half duplex network settings in MVME5100 Subject: Re: Half duplex network settings in MVME5100 Subject: Re: Watchdog timer problems Subject: Re: Debugging when the debugger crashes Subject: Anyone get VxWorks to boot quickly Subject: Tornado Prototyper Plus license key Subject: Re: END Driver problems Subject: timebase/decrementer register and vxworks Subject: S_iosLib_INVALID_FILE_DESCRIPTOR error Subject: Re: S_iosLib_INVALID_FILE_DESCRIPTOR error Subject: Re: tNetTask does not starts Subject: Re: Debugging multiple tasks Subject: Re: Debugging multiple tasks Subject: Re: Anyone get VxWorks to boot quickly Subject: Re: S_iosLib_INVALID_FILE_DESCRIPTOR error Subject: Re: Tornado 2.0.1 dependency generation Subject: Extern's in header files Subject: Re: Extern's in header files Subject: How to rebuild libppc860gnuvx.a Subject: Re: MPC8260 BSP net driver problem Subject: Re: how to send broadcast messages with dest ip "255.255.255.255" ? Subject: Re: Half duplex network settings in MVME5100 Subject: Printfs with two SMCs on PPC860 Subject: Re: Debugging when the debugger crashes Subject: Re: Printfs with two SMCs on PPC860 Subject: MPC860 rev D4 & Fast Ethernet Subject: Re: Anyone get VxWorks to boot quickly Subject: LINKAGE PROBLEM Subject: Re: How to rebuild libppc860gnuvx.a Subject: Re: Half duplex network settings in MVME5100 Subject: Re: Debugging when the debugger crashes Subject: Re: Debugging when the debugger crashes Subject: Re: Debugging when the debugger crashes ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: GDB problems with Tornado 2 for Mips (libR4000gnuvx.a) Date: Mon, 20 Aug 2001 11:52:44 GMT From: "Girish V. Gulawani" Organization: Global Online Japan // Exodus Communications KK. Message-ID: References: <__Af7.2724$Qh2.545742@typhoon.san.rr.com> perhaps i have come across similar problem earlier. that was related to SPR 26387 (unpublished). the problem is in getting mips registers while attaching task through gdb. & have already communicated to wrs. but ofcourse no reply. to this spr 26387 details are - in the WRS distributed source file /target/config/comps/src/usrWdbTask.c the function vxTaskRegsGet() fails. vxTaskRegGet () { TCB *tcb; ... tcb = taskTcb() switch(regSetType) { case WDB_REG_SET_IU: { REG_SET dummy; taskRegsGet(context, &dummy); **** problem *** *ppRegSet = (char*)tcb->regs; break; } ... } if you comment of the problem statement it works. then there is one more place you have to change that is delete hook in the same file. however i dont have the code right now. if you feel it is the same problem do post here. i'll respond asap. my debugger is working properly now. but still wrs support has not replied and its over 3/4 months. "Nihad Alibegovic" wrote in message news:__Af7.2724$Qh2.545742@typhoon.san.rr.com... > Hello everyone, > > we ported our BSP from vxWorks 5.3 to vxWorks5.4 on Tornado 2. When it was > all said and done everything works except GDB. Target attaches fine we can > put break points but as soon as we try to single step we get tWdb exception. > It is worth noting that if I put a break point in from the host shell but > single step in target resident shell things work well. So, I guess this is > not just a GDB problem. Additionally, I have same problem when trying to > attach to the task. It is as if the host and target memories are completely > out of sync and host is accessing memory it is not supposed to access on the > target. We use Network to debug which means we are in task mode debugging > only. > > The platform we run on is Mips 4310 CPU so we use libR4000gnuvx.a. We have > all the WDB includes in our BSP that we need as certified by Wind River. > Wind River has helped very little in this issue in the last week. While, at > this does anyone get quick support from these guys. In 2 years working with > vxWorks they have the slowest customer support of any company I ever worked > with. > > Nihad Alibegovic > MTS3 > nalibegovic@hns.com > > > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: change sector size? HELP! Date: Mon, 20 Aug 2001 14:28:58 +0200 From: "Leonid Rosenboim" Organization: Verio Message-ID: References: <52ccd904a%molochai@vapour-trail.demon.co.uk> <_Gl_6.69653$%a.3391129@news1.rdc1.sdca.home.com> George, The sector size is one of the parameters of the Block Device, set by the block driver. You mentioned that you use an FC controller. I am not familiar with such a driver, but if that was a SCSI device, then the SCSI "devCreate" call has a sector size parameter, which you can set when creating an instance. Your driver should have this too. Once the block device has the sector set at 2048 (or whatever you fancy), DosFs (I assume you use DosFs 2.0 cause you mentioned FAT32) will use that value all the way up, through dcache etc. Let me know how that gues. (hope my help isn't too late). Leonid "George" wrote in message news:_Gl_6.69653$%a.3391129@news1.rdc1.sdca.home.com... > We have a number of Motorola 750 hosts hooked up to 300Gbyte RAIDs (through > a Fibre Channel switch). These RAIDs are also accessed by Win32 PCs > (through the switch). We need the highest possible throughput in and out of > the RAIDs (looking for 40 to 80 Mbytes/second). Having a RAID (disk) sector > size of 2048 bytes significantly increases the efficiency of the RAIDs > allowing each drive of the RAID to have a sector size of 512. We're > formatting the RAIDs with FAT32 so that both VxWorks and Win32 can access > them. We have no trouble formatting the RAIDs with a sector size of 2048 > (we've found it's easiest to use Linux for this), now if we can just get > VxWorks to read them! > > Setting the sector size is such a trivial thing to do in Windows and Linux, > so far this seems to be impossible in VxWorks. :-( > > "Bill Pringlemeir" wrote in message > news:uhex22in6.fsf@yahoo.com... > > >>>>> "George" == George writes: > > >>>>> "molochai" == writes... > > > > George> How do you change the sector size from 512 bytes to 2048? > > George> WindRiver has > not been much help :-( > > > > > molochai> Of what? > > > > George> Of a disk drive. > > > > Usually you don't set the sector size, you set the cluster size. With > > DosFs, I think the default cluster size is 1024 or 2048 bytes. I > > don't think there is an overhead to use a small cluster size with a > > solid state devices such as flash. You might gain a little disk space > > if you could format a drive such that it had larger sector sizes. > > > > If you really want to increase the sector size, then I would suggest > > doing this under DOS and then see if DosFS will read the drive. > > However, I have had trouble (and so have others) changing the cluster > > sizes. The odds that DosFs would handle larger sector sizes seems > > rather dubious to me. > > > > hth, > > Bill Pringlemeir. > > > > ps, you could be a little more specific on what it is you wish to do. > > > > Useful vxWorks URLS, > > "http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html" > > "http://www.xs4all.nl/~borkhuis/vxworks/vxw_pt9.html" > > > > -- > > Truth decays into beauty, while beauty soon becomes merely > > charm. Charm ends up as strangeness, and even that doesn't last, but > > up and down are forever. - Anonymous > > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Multiple Initiator and one HDD Date: Mon, 20 Aug 2001 14:42:47 +0200 From: "Leonid Rosenboim" Organization: Verio Message-ID: References: Andy, as far as I can remember, nothing in VxWorks supports multiple initiators. The driver is one of many issues you might be facing with this setup. Disk cache and file system are important too. But here is how I would go about figuring out this: First set up th einitiators such that only ONE of them has a R/W access, while all others are st to R/O access. Thence you by large dealing with driver issues. Making everything deal with this setup properly is a large project, so I suggest you reconsider, to see if there isn't a simplier solution out there somewhere. Leonid "Andy Beard" wrote in message news:YsFe7.20284$G9.2173395@typhoon.tampabay.rr.com... > I have a system comprised of two MVME 172 processor and one SCSI hard drive > with a DOS filesystem. I'm trying to share the hard drive between the two > processors and am running into problems with corrupt files and locked up > SCSI interface. I've tried some of the setting on the DOS filesystem and > that didn't seem to help. The SCSI interface is using an NCR 710 chip. If > the driver supports multiple initiators, I'm assuming I need to change some > of the SCSI parameters. Two questions: Does anyone know if the driver > supports multiple initiators and what should the SCSI setting be? > > Thanks in advance > Andy > > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: HELP - vx FTP-Server problem! Date: Mon, 20 Aug 2001 15:03:51 +0200 From: "Leonid Rosenboim" Organization: Verio Message-ID: References: <3B77EB38.DBD20CFA@stoye.de> Joachim, Once upon a time, a loing time ago, I made the "unsupported' FTP server that you are probably using. The supported version was missing the mkdir and rmdir commands and had also many bugs with changing directories and listing directories other then the current one. Any FTP server, is acessing the disk via the standard I/O calls, whatever is the file system, it is up to the file system to make sure the disk structure remains consistent and nothing is being overwritten. Hence, I can only deduce the bug is in the DosFs 2.0 code. But in fact, I know one such bug in DosFs 2.0, which is related to directories and causes lost cluster chains. Suppose you have a directory "MyDir1", and a file "MyFile1", you can legally do "ename ("MyFile1","MyDir1/MyFile1"), that is fine. But if you do rename("MyDir1","MyDir1/MyDir2") you will loose all subdirectories and files under "MyDir1", which will be diagnozed as lost chains by CHKDSK later. Leonid "Joachim Pluschke" wrote in message news:3B77EB38.DBD20CFA@stoye.de... > Hi all, > > we have a big problem with FTP-Servers coming with VxWorks on the > TornadoII CDs (Host: WinNT, Target: PowerPC). > > DOES ANYBODY KNOW ABOUT A RELIABLE FTP-SERVER for VxWorks THAT really > SUPPORTS DIRECTORIES??? > > Detail: > ======= > (We're using T2 / VxWorks 5.4 / PPC motorola 860/821 BSP / dosFS-2.) > > We get in trouble with access via FTP to a ata/ide device e.g. like > ramdisk,flashdisk. > (We tried two FTP Servers, one of them from the /unsupported area of T2 > CD.) > > We loose a lot chains on disk overwriting contents of directories as we > can see calling chkdsk() from dosFs2Lib after the overwrite-actions. > We switched to use only a ramdisk, but the problem ws the same. > We had a close look on the dosFS-2 / FTP-Servers behaviour. > It seems to be a problem of the FTP-Server ... > > > Seen effects: > ------------- > Everytime we overwrite files somewhere in sub-directories > (//

/files) the lost chains occur. > Doing overwrites in the root always works fine! > > > At first we started using wrs-supported ftp-server running it without > this effect, but this one doesn't support directories (we need dirs!). > > We switched to the /unsupported section of code from the Tornado CD and > found one FTP server that (seems to) support directories... > This was fine for the first moment; but as we found out, this one is not > really usable due to the lost chains!!!! > > This was reproduced with : > WinNT/Win2000 hosts using Tools like WinCommander,WsFtp,MS-Ftp from the > Prompt > connected to : > VxWorks Target, VxWorks 5.4, Kernel: WIND version 2.5, PPC860/821-BSP, > DosFS-2. > > > > DOES ANYBODY KNOW ABOUT A RELIABLE FTP-SERVER for VxWorks THAT really > SUPPORTS DIRECTORIES??? > > Any ideas would be appreciated. > > > > regards > joachim > > > -- > > ===================================================== > Gebrüder Stoye GmbH > department: traffic control systems > name: Joachim Pluschke > adr: Longericher Strasse 177 > city: Cologne, Germany > zip: 50739 > phone: ++49/221/1707013 > fax: ++49/221/1707010 > mailto: j.pluschke@stoye.de > www: http://www.stoye.de > ===================================================== --------------------------- Newsgroups: comp.os.vxworks Subject: MPC8260 BSP net driver problem Date: 20 Aug 2001 05:28:12 -0700 From: kevinatnj@yahoo.com.cn (kewin) Organization: http://groups.google.com/ Message-ID: <221caa76.0108200428.6219f046@posting.google.com> There is a BSP net problem that has teared my hare out for weeks.The following is some my evironment. 1. i have two LANS to use,my company LAN and my own LAN.My own LAN means that i take advantage of my computer,a hub,my target board and some cables to build up a small LAN.i name the company LAN to CLAN ,and name my own LAN to SLAN. 2. My operating system is VxWorks 5.4.My target board contains a MPC8260 CPU,two ethernet ports using L80223 PHY.I use FCC1 and FCC2 to serve as Ethernet ports. After i finished my BSP,i burned it to a 512K EPROM.With ths EPROM i boot my board and then download my VxWorks image through ethernet port 1.I have binded different IP address and MAC address with each port.Till this point,everything is OK.But When i ping my target board through CLAN,there are some "request time out" ."some" means not too many.When i ping is through SLAN,my BSP works very well,and there is no "request time out". Any possible advice will be welcome.Thanks! --------------------------- Newsgroups: comp.os.vxworks Subject: Re: protocol stack design under vxworks Date: Mon, 20 Aug 2001 15:36:35 +0200 From: "Leonid Rosenboim" Organization: Verio Message-ID: References: <414a7fa0.0108100756.425b1dee@posting.google.com> Hey Jim, One of my past colegues has done something quite similar, bought a "portable" Q.9xx stack and started porting onto i960 VxWorks. The main problem as it turned out, was that the original code was never tested on a Multitasking system, so it had no concurrency safety. They could go deep into the code and make it multitask safe, but since they planned to use another stack soon after (V5.2), and signalling performance as not such a big deal, they insetead did this (which I recommend you do too): All the ISDN code runs in a single task (as if it was a OS-less CPU), and the external calls are made from other tasks by sending a msgQ to the ISDN task. Yeah, performance was a bitch, but it saved a whole lot of R&D time and money. Leonid "Jim Barton" wrote in message news:414a7fa0.0108100756.425b1dee@posting.google.com... > We are currently trying to debug an ISDN protocol stack that was > recently ported from a Intel 186 platform (with no OS) to a PPC > platform that uses VxWorks. > > Since we are currently having major problems with the stability of the > stack and since we are new to VxWorks, we suspect that we may not have > done things correctly. For example, each protocol layer is now > represented as a separate task (each having the same priority of > 100)under vxWorks. Is this the normal way that a protocol stack should > be implemented in vxWorks? If not, how serious of a design problem is > this? Is it worthwhile re-designing this into one(?) task or could we > make this work with our current task per layer design? > > Also, how is timing normally accomplished within a protocol stack > under VxWorks? In our old platform (without a OS) timing was > accomplished by a main loop that would service messages and events at > specific intervals to send out heartbeats, determine timeouts, etc. > I'm not clear what is the proper way of doing this under VxWorks. > Should we create timers and put all of our time-dependent stuff in the > SIGALRM handler? > > And, yes, I know we could purchase a ISDN stack already designed for > VxWorks but we would rather avoid that expense since we already have > source for the stack from our older platform. We would appreciate any > tips on how we should port this protocol to vxWorks! > > Thanks, > > Jim --------------------------- Newsgroups: comp.os.vxworks Subject: Re: ethernet link reset problem Date: Mon, 20 Aug 2001 15:39:56 +0200 From: "Leonid Rosenboim" Organization: Verio Message-ID: References: Kim, This might be due to 100baseT auto-negotiation protocol. If your hub is a plain old 10baseT, it will get some false signals before the 100baseT NIC settles down on the lowest old standard, which did not support auto-negotiation at the time. Also, even if you want full-duplex at 10 Mbps, it still uses 100BaseT standard for negotiation. "Kim, Junghyun" wrote in message news:d572a01a.0108090558.5088025@posting.google.com... > Hi, > > I'm using vxWorks with Motorola N750 board which is > connected to a switching hub. And I want to check the > link status using snmp agent in the hub. > > When I disconnect lan cable from the board, the hub reports > link-down, and when I reconncet the cable to the board, > the hub reports link-up. > At this time every thing is ok. > > But, just after the link-up report, a pair(or pairs) of link-down report and > link-up report comes from the hub. > > From same tests with non-vxWorks systems, I got to know that this is > not by hub. I think vxWorks OS do some initialization after the > reconnction. > (** With Motorola 750(not N750) boards, this didn't happend. > I know that N750 is system board ) > > > Any comment will be welcomed. > Thanks. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: ata flash disk versus ide flash disk Date: Mon, 20 Aug 2001 15:45:31 +0200 From: "Leonid Rosenboim" Organization: Verio Message-ID: References: <3B727FEB.CC10FAF2@optonline.net> Guru, you are asking many questions at once. The reason for mixed interrupt/polling mode in the driver is because many times, an interrupt line is shared by two disks, for example, if both master and slave are connected to one IDE channel. Also, the driver will wait for interrupt for operations which it expects to take a long time, i.e. operations which require disk head movements, but uses polling when it only expects an electronic component to respond - polling is much chepaer to use when response times are quick. In your Flash case, response times of the disk are very different from magneto-mechanical disks, so you can expect that the driver will poll, when in fact your Fash disk is exhibitting a lengthy operation. Lastly, CompactFash is an ATA extention which defines new commands which are appropriate for Flash disks only, and that is yet another reason to reqrite that old driver code, from scratch. (which was what I had in mind before I left WRS on Jan 1999). Leonid "Gururaj patkar" wrote in message news:3B727FEB.CC10FAF2@optonline.net... > Hi all, > > I need pointers to the performance of using ata flash disk versus ide > flash disk with vxworks. Looking at the ide and ata driver code supplied > by wrs, the writes are done in polling mode (by checking the status > register values on the flash controller). The usage of intConnect() in > the ideDrv.c and ataDrv.c code is further intriguing as to why you poll > the status register values and when STATUS_READY is TRUE then only > provide interrupt driven functionality. > One uses interrupt driven process to enhance the read/write performance > to flash disk, then why use polling mode concurrent with it? > Some benchmark figures in terms of read/write access time for files of > varying sizes will be appreciated. > > gururaj > --------------------------- Newsgroups: comp.os.vxworks Subject: tNetTask does not starts Date: Mon, 20 Aug 2001 15:27:31 +0200 From: "elop42" Organization: NetVision Israel Message-ID: <9lr0b8$7u3$1@news.netvision.net.il> I create a standalone bootable VxWorks application VxWorks.st_rom.hex. I think, I did all required includes and defines in configuration files (INCLUDE_NETWORK and INCLUDE_NET_INIT). I compiled code with STANDALONE_NET flag in rules.bsp. I changed boot line, give target IP of my board. But tNetTask does not starts. I think that usrNetInit() starts and failed. Maybe anyone has some experience with this problem? I found this PR of WindRiver - ---------------------------------------------------------------------------- - ---- SPR #: 5800 ARCH: Motorola PowerPC BOARD: mv1600 HOST: Sun4 DESCRIPTION: Depending on which make utility used to build vxWorks.st, the standalone image may not work. WORKAROUND: Try different make utilities. - ---------------------------------------------------------------------------- - ---- so tried another make utilities, but this don't help. My board is svme177 of DY4. Thanks in advanced. Yuri. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Tornado 2.0.1 dependency generation Date: Mon, 20 Aug 2001 10:10:31 -0400 From: "Kevin Livingston" Message-ID: References: <3B7D6A1E.3D57949F@improvsys.com> <3B7D92B6.FCD3A009@improvsys.com> I am experiencing similar problems. I am glad to know that I am not alone. Although I can not yet get a simple test case to replicate the problem (I haven't tried very hard yet though) WRS won't take my problem request except with a highly over simplified test case. ( never mind the fact that it is only complex projects that exhibit the behavior ) I believe there is a bug in the Tornado dependency generator, or several of them. I will give Tornado all the information to find the dependent file (e.g. -I on the compile line etc.) However like you mentioned some of the dependencies (in the make file and in the external dependency list) are listed as being in no directory, some are listed properly. Some get listed with the $(PRJ_DIR) macro as part of their path, some as absolute paths. Also, sometime when I have bad dependencies listed not even a regenerate dependencies will change it - the only solution I have found for that is to remove all the files from your project (yes you have to do it one by one) then rebuild dependencies, the build all. then add the files back in and try again. (needless to say that is an extreme nuisance) The reason I am most annoyed with this problem is make (and dependencies) are tools to help you build faster, but due to some odd behavior, sometimes I cannot build at all - since with bad dependencies it get all confused so even a rebuild all doesn't matter because make will still try to look at the dependencies, and with out a simple case WRS refuses to care - even though it is affecting our development. I don't understand if they are already willing to bundle third party tools such as compiliers in with their tool set, why not use a proven tool for their dependency generator like imake or the equilavent. why reinvent the wheel? could you suggest any more work arounds? this problem definately goes beyond making sure all your slashes tip this way '/' Kevin "Ben Kuhn" wrote in message news:3B7D92B6.FCD3A009@improvsys.com... > > > Johan Borkhuis wrote: > > > Some advise for you. The first on is on posting: please do not post using > > HTML, it looks very strange, as you might see above. > > > > My apologies about the HTML, thought I had that turned off. I suppose that's > what I get for using Netscape and not double checking. > > > Now your problem. Some things that might help are: > > - Use forward slashes instead of backward, the dependencies generator is > > easily confused by the backward slashes > > - Do not use spaces between the # and the include > > - If you examine the dependencies for the source files you will see that > > sometimes you see "../../" and other times only " > file>". Take a look at your source files, and see if you can get these > > include statements equal. > > Thanks for the tips. I experimented with all of these. Changing the '\'s to > '/'s seems to have helped, though it seems counterintuitive, since tornado > still shows '\'s in the build window and window titles. I'm confused about > the previously noted strange behavior though, and it hasn't completely gone > away. > > Ben Kuhn > > --------------------------- Newsgroups: comp.os.vxworks Subject: Half duplex network settings in MVME5100 Date: Mon, 20 Aug 2001 17:36:41 +0200 From: =?iso-8859-1?Q?Torbj=F6rn?= Frostemark Organization: FOI Message-ID: <3B812E89.FF7F24B4@foi.se> This is a multi-part message in MIME format. - --------------198215CF7E46E70E6136C39E Content-Type: text/plain; charset=iso-8859-1 Content-Transfer-Encoding: 8bit I want to run a MVME5100 in 100BaseT half duplex mode, but I cannot find a way to tell the driver to change from full duplex mode. Suggestions? /Torbjörn - --------------198215CF7E46E70E6136C39E Content-Type: text/x-vcard; charset=us-ascii; name="torfr.vcf" Content-Transfer-Encoding: 7bit Content-Description: Card for Torbjörn Frostemark Content-Disposition: attachment; filename="torfr.vcf" begin:vcard n:Frostemark;Torbjörn tel;cell:+46-70-249 76 27 tel;fax:+46-(13-37 80 49 tel;work:+46-13-37 80 14 x-mozilla-html:TRUE org:FOI - Swedish Defence Research Agency;Electronic Warfare Systems adr:;;Box 1165;SE-581 11 Linköping;;;SWEDEN version:2.1 email;internet:torfr@foi.se title:Research Engineer fn:Torbjörn Frostemark end:vcard - --------------198215CF7E46E70E6136C39E-- --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Half duplex network settings in MVME5100 Date: Mon, 20 Aug 2001 09:23:44 -0700 From: "Dan Gold" Message-ID: <3b813a03$0$348$e2e8da3@nntp.cts.com> References: <3B812E89.FF7F24B4@foi.se> "Torbjörn Frostemark" wrote in message news:3B812E89.FF7F24B4@foi.se... > I want to run a MVME5100 in 100BaseT half duplex mode, but I cannot find > a way to tell the driver to change from full duplex mode. Suggestions? First, a better description of your environment, i.e. I assume 5.4, which ethernet chip (and more importantly) which PHY chip you're using. The ability to do 100Base-T HD must be specified in the PHY chip. You can either set it's configuration via some resistors (you need to look at the schematic and data sheet for your particular board), or by writing some of the control registers for the part, via the MDIO interface. There are a few ways to access this: 1. WRS in some of their drivers gives routines to access the PHY chip via the MDIO interface. If you don't see anything related to that in your driver, you may need to ask them about it, or just ask for the source to the driver and put your own support in. 2. If WRS doesn't have support in their driver for this, you need to look at the schematic and roll your own if that's possible. 3. You can limit the autonegotiation parameters via resistors. Again, see the schematic and data sheet for the details. hth, Dan Gold gold@ensemble.com --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Watchdog timer problems Date: 20 Aug 2001 12:29:02 -0400 From: Bill Pringlemeir Organization: Bell Sympatico Message-ID: References: <3B80E4C6.6CA7164D@oslo.online.no> Sender: bpringlemeir@DeadDuck >>>>> "Glenn" == Glenn Ivar Klausen writes: Glenn> I've written som watchdog timer example code, but it dosent Glenn> seem to be working. If anyoun could tell med the problem here. [snip] void wdExpired(void *msg) { fprintf(stdout, " msg = %s \n", (char *)msg); } Glenn> As far as I can see, this is supposed to work. But it dont. The watchdog function run at interrupt level. I don't think that you can use `printf' or `fprintf' at this level. You could try the function, logMsg or set a global variable and examine it in a regular task. See the "vxWorks Programmers guide, section 2.5.3". Especially, Table 2-23: Routines that Can Be Called by Interrupt Service Routines Library Routines ======= ======== bLib All routines errnoLib errnoGet(), errnoSet() [snip] logLib logMsg() [snip] wdLib wdStart(), wdCancel() Also see the rngLib section. rings are nice data structure to use in an interrupt routine. hth, Bill Pringlemeir. - -- Peking Treasury Clinton assassination fissionable nuclear SEAL Team 6 Khaddafi PLO colonel arrangements explosion $400 million in gold bullion supercomputer FBI vxWorks FAQ, "http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html" --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Debugging when the debugger crashes Date: Mon, 20 Aug 2001 11:12:45 +0100 From: vxworksintent@tao-group.com (David Given) Organization: I do not speak for anyone but myself, and barely that. Message-ID: References: <2psil9.q9e.ln@127.0.0.1> <3B7E0220.9ACC02DE@covad.net> Sender: dg@pearl.tao.co.uk In article <3B7E0220.9ACC02DE@covad.net>, DrDiags writes: > How about setting up a post-mortem windview trace? If you can > set aside a section of memory and can reboot the target > without having to power-cycle (reset switch) this may help. Unfortunately, it crashes so dead that the only option is to hit the reset button. And I don't think my BSP has Windview installed, anyway. And a simulator's no use to me because the entire target is special hardware... Currently I'm reduced to randomly poking things and trying to see any patterns on when it crashes. - -- David Given Tao Group Ltd. --------------------------- Newsgroups: comp.os.vxworks Subject: Anyone get VxWorks to boot quickly Date: Mon, 20 Aug 2001 12:04:35 -0500 From: "Brad Crabtree" Organization: Raytheon Company Message-ID: I have a need to get a Tornado 2.0 VxWorks ROM image (about 4MB in size) loaded and booted on a Power PC VME board in 3-4 seconds. It is currently taking about 6. Does anyone have any lessons learned on getting a VxWorks image to boot quickly? - -- - -------------------------------------------- Bradley D. Crabtree b-crabtree@raytheon.com --------------------------- Newsgroups: comp.os.vxworks Subject: Tornado Prototyper Plus license key Date: Mon, 20 Aug 2001 13:33:30 -0400 From: "Irene" Organization: Bell Sympatico Message-ID: Hi, I recently downloaded the Tornado Prototyper and installed it on my machine. However, when I tried to request a license key by using the web site "Automatic License Generation", I kept getting error message complaining: "Missing Required Field The required field version, is missing. Please go back and check your entry. " after I submit my computer's hostname and hostid. I don't know what's wrong there. Can anyone help me? Mimi --------------------------- Newsgroups: comp.os.vxworks Subject: Re: END Driver problems Date: Mon, 20 Aug 2001 19:04:28 +0100 From: "timallen" Organization: ntlworld News Service Message-ID: References: <3B798E58.7D35A701@brain.jpl.nasa.gov> <3B79C380.2C0003B3@brain.jpl.nasa.gov> <3B7A01C0.EC212C94@covad.net> Reply-To: "timallen" DrDiags wrote in message news:3B7A01C0.EC212C94@covad.net... > Carlos, > > Don't have an answer, but I understand that T2.0.2 is a major > change in the WRS network stack. They reworked and changed the API > for their DHCP code picking up support for a couple of RFCs and other > changes. They also have moved to the NPT model (NPT Drivers are enhanced > END drivers) to allow non-ethernet (?) interfaces to use the MUX. > Maybe this could be a side affect of these changes that should be > reported to WRS to see if this has occurred with others developers. > If not, you are the first to report it and hopefully the test case > will lead to a fix for your problem, if it is a bug. I have seen problems with 2 other network drivers / devices when changing from T2.0 to T202. So far, neither Wind River, nor the device manufacturer (who wrote one of the drivers) have been able to find out why the driver doesn't work. I'm aware of the change to NPT, but it is supposed to be compatible with END. However, one workaround which I've used is as follows: edit /target/src/drv/wdbEndPktDrv.c to change line 87: #define WDB_NPT_CAPABLE to: #undef WDB_NPT_CAPABLE and rebuild in /target/src/drv directory. I've discussed this with Wind River UK tech support and they don't see a problem with doing that other than losing the Network Protocol Toolkit capability. If anyone knows any more details about the T2 to T202 differences, I'd be grateful too! Cheers, Tim. --------------------------- Newsgroups: comp.os.vxworks Subject: timebase/decrementer register and vxworks Date: Mon, 20 Aug 2001 14:24:48 -0400 From: Bhavesh Patel Organization: Mailgate.ORG Server - http://www.Mailgate.ORG Message-ID: <3B8155F0.7AAC9AEC@fwion.com> Sender: bpatel@fwion.com Hi, pardon me if the answer to this question is floating around in one of the newsgroup archives. I am using BSP based on SBC8260 (ie PPC8260), vxworks.5.4/Tornado2.0. The system clock is 66Mhz. I have defined INCLUDE_TIMESTAMP in config.h to enable high resolution timestamping. questions - - --------- 1) I want to know what value is used by vxworks for the update frequency of the time base register ( the Progamming Environment Manual for PowerPC, page 2-15 says that the system software controls the update frequency of the time base), is it the system clock frequency which in my case is 66Mhz. 2) The page 2-33 of the "Programming Environment Manual for PowerPC" says the DEC (decremeter register) frequency is based on the same implementation dependent frequency that drives the time base. Is the DEC frequency and timebase frequency same for vxworks? 3) where is sysTimestampConnect() called from, does user code need to call it in usrconfig.c? 4) Do I need to call sysTimestampEnable() in usrConfig.c explicitly to enable the timestamp driver? bhavesh --------------------------- Newsgroups: comp.os.vxworks Subject: S_iosLib_INVALID_FILE_DESCRIPTOR error Date: Mon, 20 Aug 2001 14:52:33 -0400 From: Sunil Joseph Organization: Lucent Technologies Message-ID: <3B815C71.5A08DC6C@lucent.com> Folks, I am trying to send a buffer over a socket using write f (write(sFd, (char *) &myRequest, sizeof(myRequest)) == ERROR) . I get the following error S_iosLib_INVALID_FILE_DESCRIPTOR. I looked at the FD it seemed OK. What does the error really mean ? Sunil --------------------------- Newsgroups: comp.os.vxworks Subject: Re: S_iosLib_INVALID_FILE_DESCRIPTOR error Date: Mon, 20 Aug 2001 22:02:00 +0200 From: "Werner Schiendl" Message-ID: <3b816d65$0$31966$6e365a64@newsreader02.highway.telekom.at> References: <3B815C71.5A08DC6C@lucent.com> Hi, Are you sure your file descriptor is OK? Maybe you closed it before due to some error condition? Or the nice-looking value is from an earlier iteration of the software? For me this means, that the number is NOT recognised as a valid file descriptor at the moment you call write. You can get a list of all active file descriptors printed with iosFdShow (I don't actually know, if sockets are listed - but give it a try) hth werner "Sunil Joseph" wrote in message news:3B815C71.5A08DC6C@lucent.com... > Folks, > I am trying to send a buffer over a socket using write > > f (write(sFd, (char *) &myRequest, sizeof(myRequest)) == ERROR) > . I get the following error S_iosLib_INVALID_FILE_DESCRIPTOR. > I looked at the FD it seemed OK. What does the error really mean ? > Sunil --------------------------- Newsgroups: comp.os.vxworks Subject: Re: tNetTask does not starts Date: Mon, 20 Aug 2001 22:12:55 +0200 From: "Werner Schiendl" Message-ID: <3b816ff5$0$28674$6e365a64@newsreader02.highway.telekom.at> References: <9lr0b8$7u3$1@news.netvision.net.il> Hi, what build environment are you using? Tornado II project facility or command line build? If you use Tornado II, be careful when editing the config files. Lots of this stuff is overwritten by the build mechanism of the project facility and will not take effect. Otherwise, are you sure you specified which network adapter to use? On x86, you usually use the other field of the boot line to specify the network adapter if you don't boot from a network connection (I assume you don't boot over the network, based on the file name you gave). hth Werner "elop42" wrote in message news:9lr0b8$7u3$1@news.netvision.net.il... > > I create a standalone bootable VxWorks application VxWorks.st_rom.hex. > I think, I did all required includes and defines in configuration files > (INCLUDE_NETWORK and INCLUDE_NET_INIT). > I compiled code with STANDALONE_NET flag in rules.bsp. > I changed boot line, give target IP of my board. > But tNetTask does not starts. > I think that usrNetInit() starts and failed. > Maybe anyone has some experience with this problem? > > I found this PR of WindRiver > -------------------------------------------------------------------------- - -- > ---- > > SPR #: 5800 > > ARCH: Motorola PowerPC > BOARD: mv1600 > HOST: Sun4 > > DESCRIPTION: > Depending on which make utility used to build vxWorks.st, the > standalone image may not work. > > WORKAROUND: > Try different make utilities. > -------------------------------------------------------------------------- - -- > ---- > > so tried another make utilities, but this don't help. > My board is svme177 of DY4. > Thanks in advanced. > Yuri. > > > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Debugging multiple tasks Date: Mon, 20 Aug 2001 22:18:24 +0200 From: "Werner Schiendl" Message-ID: <3b81713e$0$21924$6e365a64@newsreader02.highway.telekom.at> References: Hi, at least on Win NT you can simply right-click the line where you want to have the global breakpoint set and select "Toggle Global Breakpoint". This should be easier to handle, under normal circumstances. hth Werner "Albert H Chen" wrote in message news:uD4f7.1691$P3.405648@newsrump.sjc.telocity.net... > I think all you have to do is in the Tornado shell type "b " > to set a global breakpoint... > > Albert > > > "Andy" wrote in message > news:e397b57.0108160815.76bffc15@posting.google.com... > > There was a previous post about trying to set a breakpoint at the > > entry point of a new task that is spawned by another task. The > > response was to set a "global breakpoint". Thats fine, but how do you > > set a global breakpoint? Using Tornado II/vxWorks 5.4.2. If there is > > another way I'd sure like to hear it. Finally, is there a FAQ for this > > newsgroup and if so where?????? > > > > Andy > > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Debugging multiple tasks Date: 20 Aug 2001 16:42:29 -0400 From: Bill Pringlemeir Organization: Bell Sympatico Message-ID: References: <3b81713e$0$21924$6e365a64@newsreader02.highway.telekom.at> Sender: bpringlemeir@DeadDuck >>>> "Andy" wrote: >> There was a previous post about trying to set a breakpoint at the >> entry point of a new task that is spawned by another task. The >> response was to set a "global breakpoint". Thats fine, but how do >> you set a global breakpoint? Using Tornado II/vxWorks 5.4.2. If >> there is another way I'd sure like to hear it. Finally, is there a >> FAQ for this newsgroup and if so where?????? >>>> "Albert H Chen" wrote: > I think all you have to do is in the Tornado shell type "b name>" to set a global breakpoint... >>>>> "Werner" == Werner Schiendl writes: Werner> Hi, at least on Win NT you can simply right-click the line Werner> where you want to have the global breakpoint set and select Werner> "Toggle Global Breakpoint". This should be easier to handle, Werner> under normal circumstances. These things are not the same (for some reason). In the case of "b function", you will break in the shell and use the shell commands like `s' and `c'. I think you can only debug at the assembler level here. Setting a global break point on the other hand seems to result in use of gdb/wdb backend. You must connect the debugger (and WDB) to the target. When a global break point is hit, you have to use the `debug|attach' menu item to attach to the task that has hit the global break point. You can then debug modules at the source level if your object file have debugging info. fwiw, Bill Pringlemeir. - -- nuclear Qaddafi domestic disruption Serbian SEAL Team 6 quiche cracking munitions plutonium BATF $400 million in gold bullion bomb fissionable FBI Honduras vxWorks FAQ, "http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html" --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Anyone get VxWorks to boot quickly Date: Mon, 20 Aug 2001 16:42:14 -0400 From: david lindauer Message-ID: <3B817625.E1C7AD96@notifier-is.net> References: the only trick I know is disable the memory test... David Brad Crabtree wrote: > I have a need to get a Tornado 2.0 VxWorks ROM image (about 4MB in size) > loaded and booted on a Power PC VME board in 3-4 seconds. It is currently > taking about 6. Does anyone have any lessons learned on getting a VxWorks > image to boot quickly? > > -- > -------------------------------------------- > Bradley D. Crabtree > b-crabtree@raytheon.com --------------------------- Newsgroups: comp.os.vxworks Subject: Re: S_iosLib_INVALID_FILE_DESCRIPTOR error Date: Mon, 20 Aug 2001 16:44:06 -0400 From: david lindauer Message-ID: <3B817696.A43FCEED@notifier-is.net> References: <3B815C71.5A08DC6C@lucent.com> That error comes up when a file hasn't been opened properly. Did the create socket succeed, along with all the binding, connection, and so forth? David Sunil Joseph wrote: > Folks, > I am trying to send a buffer over a socket using write > > f (write(sFd, (char *) &myRequest, sizeof(myRequest)) == ERROR) > . I get the following error S_iosLib_INVALID_FILE_DESCRIPTOR. > I looked at the FD it seemed OK. What does the error really mean ? > Sunil --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Tornado 2.0.1 dependency generation Date: Mon, 20 Aug 2001 17:01:35 -0400 From: Ben Kuhn Organization: Improv Systems, Inc. Message-ID: <3B817AAF.916DBE5E@improvsys.com> References: <3B7D6A1E.3D57949F@improvsys.com> <3B7D92B6.FCD3A009@improvsys.com> Reply-To: benk@improvsys.com This is a multi-part message in MIME format. - --------------7082CA0202F82ABBB2C9E88C Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Thanks for all of the tips, folks. With some combination of all of the stuff everyone haas mentioned, I've managed to get things up and running. Should this be added / updated in the Tornado FAQ? The list of things to check, as best I can tell from everyone's suggestions are: Make sure you use UNIX '/'s, make sure path names match the case in the file system, even on Windows, use quotes to include project files, instead of '<', '>', and set the VPATH macro to specify include directories. [I didn't actually try this one] My apologies if I've missed anything. Thanks again. - -- Ben Kuhn - --------------7082CA0202F82ABBB2C9E88C Content-Type: text/x-vcard; charset=us-ascii; name="benk.vcf" Content-Transfer-Encoding: 7bit Content-Description: Card for Ben Kuhn Content-Disposition: attachment; filename="benk.vcf" begin:vcard n:Kuhn;Benjamin tel;cell:716 739 2171 tel;work:(716) 756 - 5411 x25 x-mozilla-html:FALSE adr:;;;;;; version:2.1 email;internet:benk@improvsys.com fn:Benjamin Kuhn end:vcard - --------------7082CA0202F82ABBB2C9E88C-- --------------------------- Newsgroups: comp.os.vxworks Subject: Extern's in header files Date: 20 Aug 2001 15:57:55 -0700 From: venkat_sr@yahoo.com (Venkat S) Organization: http://groups.google.com/ Message-ID: Hi, What is the purpose of externs in the the header files such as sysLib.h? Is is only for the obvious reason that an user who #includes this file can call the extern functions or does it have other purposes too. Another purpose I can think of is that the designer of vxWorks have defined an inteface and developers of VxWorks BSP's provide the code for the functions? For example sysLib.h defines the inteface while sysLib.c written by the BSP developer provides the code which implements these functions. Am I thinking right? Regards, Venkat --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Extern's in header files Date: Mon, 20 Aug 2001 19:19:06 -0400 From: joe durusau Organization: Bellsouth.Net Message-ID: <3B819AEA.D98624CF@bellsouth.net> References: Venkat S wrote: > Hi, > What is the purpose of externs in the the header files such as > sysLib.h? Is is only for the obvious reason that an user who #includes > this file can call the extern functions or does it have other > purposes too. > <> > Regards, > Venkat It's pretty much like any C header file. The declarations tell the compiler the number and type of parameters the function takes, and the type of variable (if any) the function returns. You could still write calls without the header files, but there would be no parameter error-checking, and the functions would be compiled as if they all returned integer. Speaking only for myself, Joe Durusau --------------------------- Newsgroups: comp.os.vxworks Subject: How to rebuild libppc860gnuvx.a Date: 20 Aug 2001 17:55:01 -0700 From: vinhh@bridgewave.com (Vinh Huynh) Organization: http://groups.google.com/ Message-ID: <6031d81a.0108201655.5b6f10ad@posting.google.com> Hello, I make some changes to the bootline configuration. I like to update the bootChange function in usrLib.c to use our new bootline. 1. Is this require to rebuild to vxWorks library(libppc860gnuvx.a)? 2. If the first question is true, how do build new library for vxWorks? Thanks, Vinh --------------------------- Newsgroups: comp.os.vxworks Subject: Re: MPC8260 BSP net driver problem Date: Mon, 20 Aug 2001 18:32:28 -0700 From: DrDiags Organization: laserlink.net Message-ID: <3B81BA2C.398314B4@covad.net> References: <221caa76.0108200428.6219f046@posting.google.com> Reply-To: drdiags@covad.net Kewin, What does your routing table look like? I am also told that if both interfaces are on the same subnet, that all bets are off. I have not personally tried this, but the claim is the routing table will not be able to consistently select the correct interface. What does your mRouteShow() routine return. Maybe someone can show you where the problem may lie, if you post an abbreviated dump of these types of utilities. (arpShow(), etc....) kewin wrote: > > There is a BSP net problem that has teared my hare out for weeks.The > following is some my evironment. > 1. i have two LANS to use,my company LAN and my own LAN.My own LAN > means that i take advantage of my computer,a hub,my target board and > some cables to build up a small LAN.i name the company LAN to CLAN > ,and name my own LAN to SLAN. > 2. My operating system is VxWorks 5.4.My target board contains a > MPC8260 CPU,two > ethernet ports using L80223 PHY.I use FCC1 and FCC2 to serve as > Ethernet ports. > > After i finished my BSP,i burned it to a 512K EPROM.With ths EPROM i > boot my board and then download my VxWorks image through ethernet port > 1.I have binded different IP address and MAC address with each > port.Till this point,everything is OK.But When i ping my target board > through CLAN,there are some "request time out" ."some" means not too > many.When i ping is through SLAN,my BSP works very well,and there is > no "request time out". > > Any possible advice will be welcome.Thanks! - -- http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html (vxWorks FAQ home page) --------------------------- Newsgroups: comp.os.vxworks Subject: Re: how to send broadcast messages with dest ip "255.255.255.255" ? Date: Mon, 20 Aug 2001 19:06:23 -0700 From: Albert H Chen Organization: Newsfeeds.com http://www.newsfeeds.com 73,000+ UNCENSORED Newsgroups. Message-ID: <3B81C21F.6080301@cornell.edu> References: <9ll4t2$h6i$1@news1.kornet.net> The following code fragment will send a broadcast packet with a destination address of 255.255.255.255. It requires the creation of a routing entry which matches that destination address. The gateway for the route entry determines the interface that sends the broadcast. The following example creates the necessary entry. - -> mRouteEntryAdd (0xffffffff, 0x930b7497, 0xffffffff, 0, 0, 1, 0) value = 0 = 0x0 - -> routeShow ROUTE NET TABLE destination gateway flags Refcnt Use Interface - ---------------------------------------------------------------------------- 147.11.116.0 147.11.116.151 101 0 0 ln0 255.255.255.255 147.11.116.151 3 0 0 ln0 - ---------------------------------------------------------------------------- ROUTE HOST TABLE destination gateway flags Refcnt Use Interface - ---------------------------------------------------------------------------- 127.0.0.1 127.0.0.1 5 1 4 lo0 - ---------------------------------------------------------------------------- value = 77 = 0x4d = 'M' The entry can be removed by: mRouteEntryDelete (0xffffffff, 0x930b7497, 0xffffffff, 0, 0, 1) when the broadcasts are completed. The route entry allows the route lookup for the 255.255.255.255 destination to succeed so that the packet is sent. Otherwise, the sendto() call fails with EHOSTUNREACH (0x41). A default route may also be created instead of the explicit destination shown above. Normally, the socket layer transmission calls such as sendto() replace the all-ones broadcast address with the locally configured broadcast address of the primary (i.e - first attached) interface, which has the form , <-1>. This behavior occurs for convenience so that users do not need to know the assigned address to send broadcasts. It also enhances security since use of the all-ones broadcast is limited to prevent network flooding. That broadcast address should only be used by uninitialized interfaces using protocols like DHCP or BOOTP to obtain configuration information. In other cases, the standard subnet local broadcast which is normally produced is appropriate. The following code should only be used by customers attempting to write a DHCP or BOOTP client. It replaces the broadcast address of the primary interface with the all-ones broadcast so that the substition performed by the socket layer still produces the desired result. The actual interface which transmits the broadcast is determined by the gateway of the routing entry, but the destination address for that packet is always set to the broadcast address of the primary interface. The broadcast address must be changed by directly issuing the appropriate low-level ioctl() call because the inet_addr() conversion performed when processing the higher-level ifBroadcastSet() routine produces a value of -1 (ERROR) when the all-ones broadcast is specified. For the same reason, the mRouteEntryAdd() routine must be used when creating a route entry to the 255.255.255.255 address. If a default route is created instead, any of the routing API routines may be used. To completely mimic certain parts of a DHCP/BOOTP client, the IP address of the sending interface must be changed to 0.0.0.0 (using ifAddrSet or the corresponding ioctl call) at some point before the sendto() call. That step is not shown here, but was also tested successfully. #include "vxWorks.h" #include "stdio.h" #include "sockLib.h" #include "netinet/in.h" #include "sys/ioctl.h" #include "logLib.h" #define SOCK 7000 int client_test() { char buf[512]; int bufLen = 512; struct sockaddr_in name; int status; struct ifreq ifr; int optval = 1; int fd = socket (AF_INET, SOCK_DGRAM, 0); if (fd == ERROR) { perror("no socket "); return (ERROR); } memset (&name, 0, sizeof (name)); name.sin_family = AF_INET; name.sin_port = htons (SOCK); name.sin_addr.s_addr = 0xffffffff; /* * Replace the broadcast address to prevent substitution * of the network number (e.g. 138.11.255.255) */ strcpy (ifr.ifr_name, "dc0"); bzero ((caddr_t) &ifr.ifr_addr, sizeof (ifr.ifr_addr)); ifr.ifr_addr.sa_len = sizeof (struct sockaddr_in); ifr.ifr_addr.sa_family = AF_INET; ((struct sockaddr_in *) &ifr.ifr_addr)->sin_addr.s_addr = 0xffffffff; status = ioctl (fd, SIOCSIFBRDADDR, &ifr); if (status != 0) { if (status != ERROR) /* iosIoctl() can return ERROR */ (void)errnoSet (status); perror ("ioctl failed"); close (fd); return (ERROR); } setsockopt (fd, SOL_SOCKET, SO_BROADCAST, (char *)&optval, sizeof (optval)); if (sendto (fd, buf, bufLen, 0, (struct sockaddr *)&name, sizeof (name))==ERROR) { perror(" sendto failed "); close (fd); return (ERROR); } close (fd); return (OK); } With the route entry and the code above the packet goes out with 255.255.255.255 as the IP destination address. However, due to the route entry the destination MAC address is the target's Ethernet address, not the Ethernet broadcast address. The following code can be used to insert the Ethernet broadcast address: #include "vxWorks.h" #include "logLib.h" #include "ifLib.h" BOOL myOutputHook (struct ifnet *pIf, char *buffer, int length) { char *enet = buffer; char *ipaddress = buffer+30; /* Ethernet header is 14 bytes * Ip dest field is located 16 bytes * after that */ unsigned char etbcastaddr [6] = { 0xff, 0xff, 0xff, 0xff, 0xff, 0xff }; logMsg("Output hook called\n", 0,0,0,0,0,0); if (ipaddress[0] == 0xff && ipaddress[1] == 0xff && ipaddress[2] == 0xff && ipaddress[3] == 0xff) { logMsg(" Destination Ethernet address is %02x:%02x:%02x:%02x:%02x:%02x\n", enet [0], enet [1], enet [2], enet [3], enet [4], enet[5]); bcopy(etbcastaddr, enet, 6); logMsg(" Destination Ethernet address is %02x:%02x:%02x:%02x:%02x:%02x\n", enet [0], enet [1], enet [2], enet [3], enet [4], enet[5]); } return (FALSE); } HTH Albert Â÷Çö¿í wrote: > Hi, > I want to know how to send broadcast messages which are destined to > "255.255.255.255". > I tried "routeAdd "255.255.255.255" ..." on shell prompt. > Anyone knows? > > > > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Half duplex network settings in MVME5100 Date: Mon, 20 Aug 2001 18:46:05 -0700 From: DrDiags Organization: laserlink.net Message-ID: <3B81BD5D.145C07B3@covad.net> References: <3B812E89.FF7F24B4@foi.se> Reply-To: drdiags@covad.net Hello, Besides Dan's suggestions, you could also look in the configNet.h file in your BSP directory to see if you have something similar to the following: /* * this table may be customized by the user to force a * particular order how different technology abilities may be * negotiated by the PHY. */ INT16 motFecPhyAnOrderTbl [] = { PHY_100BASE_T4, /* 100Base-T4 */ PHY_100BASE_TX_FDX, /* 100Base-T full duplex*/ PHY_100BASE_TX, /* 100Base-T */ PHY_10BASE_T_FDX, /* 10Base-T full duplex */ PHY_10BASE_T, /* 10Base-T */ -1 /* Signal end of table */ }; The above came from an ADS860 BSP directory for a MOTFEC END driver, but you may find something similar for your NIC (if you are using an END). Torbjörn Frostemark wrote: > > I want to run a MVME5100 in 100BaseT half duplex mode, but I cannot find > a way to tell the driver to change from full duplex mode. Suggestions? > > /Torbjörn - -- http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html (vxWorks FAQ home page) --------------------------- Newsgroups: comp.os.vxworks Subject: Printfs with two SMCs on PPC860 Date: Mon, 20 Aug 2001 22:04:47 -0400 From: Angelo LaMonaca Organization: MindSpring Enterprises Message-ID: <3B81C1BF.7EBB3F66@infi.net> I am doing a BSP for a PPC860EN platform. It turns out that there is a problem with SMC1 but SMC2 is functional on our PPC860 target hardware. This creates a need to steer the printfs to SMC2. Is that possible? If so, what do I need to do in the serial initialization to accomplish this? I looked at the ADS860 BSP and I see how to initialize the two SMCs. I can not figure out how to tell VxWorks to use SMC2 instead of SMC1 for printfs and the usual serial I/O stuff. Thanks in advance, Angelo --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Debugging when the debugger crashes Date: Mon, 20 Aug 2001 19:15:27 -0700 From: DrDiags Organization: laserlink.net Message-ID: <3B81C43F.3D33081B@covad.net> References: <2psil9.q9e.ln@127.0.0.1> <3B7E0220.9ACC02DE@covad.net> Reply-To: drdiags@covad.net David, I just want to clear up any mis-impression I may have given in regards to having to hit the reset button. Since your post-mortem buffer would not be managed by vxWorks, the buffer remains intact unless you power down the target. Could you talk to your FAE/Account Manager to see if you could eval windView, because it might help? But as you say, adding this feature will add overhead and change your environment to an extent. David Given wrote: > > In article <3B7E0220.9ACC02DE@covad.net>, > DrDiags writes: > > How about setting up a post-mortem windview trace? If you can > > set aside a section of memory and can reboot the target > > without having to power-cycle (reset switch) this may help. > > Unfortunately, it crashes so dead that the only option is to hit the reset > button. And I don't think my BSP has Windview installed, anyway. And a > simulator's no use to me because the entire target is special hardware... > > Currently I'm reduced to randomly poking things and trying to see any > patterns on when it crashes. > > -- > David Given > Tao Group Ltd. - -- http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html (vxWorks FAQ home page) --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Printfs with two SMCs on PPC860 Date: Mon, 20 Aug 2001 19:34:46 -0700 From: DrDiags Organization: laserlink.net Message-ID: <3B81C8C6.AD994734@covad.net> References: <3B81C1BF.7EBB3F66@infi.net> Reply-To: drdiags@covad.net Angelo, What about the BSP config.h file or in Tornado IDE, I think you can redefine the CONSOLE_TTY in your config.h (it also exists in the configAll.h file, but you should override it in your BSP header file): #define NUM_TTY 2 /* number of tty channels */ #define CONSOLE_TTY 0 /* console channel */ Or from the online help pages (Windows version): ioGlobalStdSet( ) NAME ioGlobalStdSet( ) - set the file descriptor for global standard input/output/error SYNOPSIS void ioGlobalStdSet ( int stdFd, /* std input (0), output (1), or error (2) */ int newFd /* new underlying file descriptor */ ) DESCRIPTION This routine changes the assignment of a specified global standard file descriptor stdFd (0, 1, or, 2) to the specified underlying file descriptor newFd. newFd should be a file descriptor open to the desired device or file. All tasks will use this new assignment when doing I/O to stdFd, unless they have specified a task-specific standard file descriptor (see ioTaskStdSet( )). If stdFd is not 0, 1, or 2, this routine has no effect. Angelo LaMonaca wrote: > > I am doing a BSP for a PPC860EN platform. It turns out that there is a > problem with SMC1 but SMC2 is functional on our PPC860 target hardware. > This creates a need to steer the printfs to SMC2. Is that possible? If > so, what do I need to do in the serial initialization to accomplish > this? I looked at the ADS860 BSP and I see how to initialize the two > SMCs. I can not figure out how to tell VxWorks to use SMC2 instead of > SMC1 for printfs and the usual serial I/O stuff. > > Thanks in advance, > Angelo - -- http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html (vxWorks FAQ home page) --------------------------- Newsgroups: comp.os.vxworks Subject: MPC860 rev D4 & Fast Ethernet Date: Tue, 21 Aug 2001 07:11:55 GMT From: lourens@mecalc.co.za (Lourens Geldenhuys) Organization: Mecalc (Pty) Ltd Message-ID: <3b820962.2131915@modem.mecalc> Reply-To: lourens@mecalc.co.za Is anybody using an MPC860D4 processor with WindRiver's ads860 BSP? We replaced the MPC860B3 silicon on our (working) design with rev D4 silicon, and now our Ethernet does not work. We are using the FEC and RELEASE 1.2/1, Jul 2000 of the ads860 BSP, which does support rev D4 silicon. Also, I have modified the XFC capacitors to match the rev D4 spec, still no Ethernet. Any advice? Thank you, 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: Re: Anyone get VxWorks to boot quickly Date: 21 Aug 2001 00:22:20 -0700 From: john_94501@yahoo.com (John) Organization: http://groups.google.com/ Message-ID: <488e459a.0108202322.421fa0f7@posting.google.com> References: Hello, Are you using compressed images? Sometime they can be faster than pure copy, sometimes slower. If you have space in the ROM, try both and see. Another thing to look at is the cache setting that the ROM image is running in (the copy out/decompression code runs ROM resident). You want to be careful here, but having the I-cache on is a good way of getting a boost (I suspect that with times around 6 seconds you already have the I-cache on though - it would be probably be much, much slower without). You don't say which board you are using, but if it is something like the MVME series, be careful about which flash memory you use for the image. If I remember correctly, the socketed flash is "narrower" than the soldered flash causing accesses to the flash to take longer (the memory controller has to do several fetches from flash). You may want to check this out too. HTH, John... "Brad Crabtree" wrote in message news:... > I have a need to get a Tornado 2.0 VxWorks ROM image (about 4MB in size) > loaded and booted on a Power PC VME board in 3-4 seconds. It is currently > taking about 6. Does anyone have any lessons learned on getting a VxWorks > image to boot quickly? --------------------------- Newsgroups: comp.os.vxworks Subject: LINKAGE PROBLEM Date: Tue, 21 Aug 2001 09:38:11 +0200 From: "RK" Message-ID: <9lt368$asbte$1@ID-69140.news.dfncis.de> Hi We are using Tornado 2 to download our application on VxWorks 5.4 based bootable image. Here is the problem. When we use WindNet function in application we get unresolved symbols when try to download the application (This function are not called in bootable image, so linker could not link them into bootable image - All these function are placed in libPPC860gnuvx.a library ). Is there some efficient way how to force linking and resolve symbols in bootable image without calling those function? We need to initialize WindNet protocols not earlier than in application, because some configuration need to read before initialization. Thanks, Dejan --------------------------- Newsgroups: comp.os.vxworks Subject: Re: How to rebuild libppc860gnuvx.a Date: 21 Aug 2001 00:41:39 -0700 From: john_94501@yahoo.com (John) Organization: http://groups.google.com/ Message-ID: <488e459a.0108202341.6664dd4@posting.google.com> References: <6031d81a.0108201655.5b6f10ad@posting.google.com> Hello, You can rebuild the file you have changed in a number of ways. There should be a Makefile in the usr directory where you found. Inside there, there is a definition of OBJS that will likely include more than you have source for, so there are a couple of possibilities: 1) Edit the definition of OBJS in the Makefile to include only the files you have (or want to build), then use: make CPU=PPC860 2) Better, simply override the definition on the command line like this: make CPU=PPC860 OBJS=usrLib.o Both of these should build your new .o file and update the archive. Once that's done, rebuild your project or BSP and try the new version out. HTH, John... PS If you use the IDE normally, make sure you source the torVars.sh or torVar.csh file (Unix) or execute torVars.bat (Windoze) before using the command line... vinhh@bridgewave.com (Vinh Huynh) wrote in message news:<6031d81a.0108201655.5b6f10ad@posting.google.com>... > Hello, > > I make some changes to the bootline configuration. I like to update > the bootChange function in usrLib.c to use our new bootline. > > 1. Is this require to rebuild to vxWorks library(libppc860gnuvx.a)? > 2. If the first question is true, how do build new library for > vxWorks? > > Thanks, > Vinh --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Half duplex network settings in MVME5100 Date: 21 Aug 2001 01:01:37 -0700 From: john_94501@yahoo.com (John) Organization: http://groups.google.com/ Message-ID: <488e459a.0108210001.3eb240e1@posting.google.com> References: <3B812E89.FF7F24B4@foi.se> Hello, The MVME5100 uses the 82559ER parts, and therefore the fei driver. You can configure this driver to half duplex mode, but not as easily as some others. In the BSP there is a file called sysFei82557End.c. That file contains a functions called sys557Init(), the second parameter of which is a pointer to a structure of type FEI_BOARD_INFO. Towards the end of this function there is an ifdef for 10MBps operation that sets the phySpeed and phyDpx values. This latter value controls the duplex settings programmed into the PHY. To get half duplex, even on a link that could support full, add this line at the end of sys557Init(): pBoard->phyDpx = PHY_HALF_DPX; That should force it to half duplex mode... HTH, John... Torbjörn Frostemark wrote in message news:<3B812E89.FF7F24B4@foi.se>... > I want to run a MVME5100 in 100BaseT half duplex mode, but I cannot find > a way to tell the driver to change from full duplex mode. Suggestions? > > /Torbjörn > > -- --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Debugging when the debugger crashes Date: 21 Aug 2001 01:09:56 -0700 From: john_94501@yahoo.com (John) Organization: http://groups.google.com/ Message-ID: <488e459a.0108210009.5773c1c5@posting.google.com> References: <2psil9.q9e.ln@127.0.0.1> <3B7E0220.9ACC02DE@covad.net> <3B81C43F.3D33081B@covad.net> Hello, Have you tried configuring a serial driver for the WDB connection? These are normally used for system mode debugging rather than problems like this, but often a serial driver is more robust than a network one. If you suspect interrupt problems, then you might also want to use system mode and not task mode (this will switch the serial driver into polled mode). Another possible technique if the serial port is still working would be the target based debug facilities (target shell etc). It is only assembler level, but better than nothing. HTH, John... DrDiags wrote in message news:<3B81C43F.3D33081B@covad.net>... > David, > > I just want to clear up any mis-impression I may have > given in regards to having to hit the reset button. Since > your post-mortem buffer would not be managed by vxWorks, the > buffer remains intact unless you power down the target. Could > you talk to your FAE/Account Manager to see if you could > eval windView, because it might help? But as you say, adding this > feature will add overhead and change your environment to an extent. > > David Given wrote: > > > > In article <3B7E0220.9ACC02DE@covad.net>, > > DrDiags writes: > > > How about setting up a post-mortem windview trace? If you can > > > set aside a section of memory and can reboot the target > > > without having to power-cycle (reset switch) this may help. > > > > Unfortunately, it crashes so dead that the only option is to hit the reset > > button. And I don't think my BSP has Windview installed, anyway. And a > > simulator's no use to me because the entire target is special hardware... > > > > Currently I'm reduced to randomly poking things and trying to see any > > patterns on when it crashes. > > > > -- > > David Given > > Tao Group Ltd. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Debugging when the debugger crashes Date: Tue, 21 Aug 2001 10:51:51 +0100 From: James Marshall Organization: Agilent Technologies UK Ltd Message-ID: <3B822F37.6B80B863@agilent.com> References: <2psil9.q9e.ln@127.0.0.1> <3B7E0220.9ACC02DE@covad.net> > Currently I'm reduced to randomly poking things and trying to see any > patterns on when it crashes. > > -- > David Given > Tao Group Ltd. This isn't a helpful suggestion, just a wicked observation. Does anyone else see the delicious irony in someone from this company reduced to "randomly poking" to find a bug? Maybe you should ask "what is the essence of this crash" ;-) James Marshall. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Debugging when the debugger crashes Date: Tue, 21 Aug 2001 06:34:18 -0700 From: Joe Durusau Organization: Lockheed Martin Corporation Message-ID: <3B82635A.60C7BADC@lmco.com> References: <2psil9.q9e.ln@127.0.0.1> <3B7E0220.9ACC02DE@covad.net> <3B822F37.6B80B863@agilent.com> Maybe the original poster has already looked at this, but th emost common cause of crashes is forgetting that there are very few functions that you can call from an ISR. There's a list in the programmer's guide. Those restrictions also apply to routines called be a watchdog timer. Speaking only for myself, Joe Durusau James Marshall wrote: > > > Currently I'm reduced to randomly poking things and trying to see any > > patterns on when it crashes. > > > > -- > > David Given > > Tao Group Ltd. > > This isn't a helpful suggestion, just a wicked observation. > > Does anyone else see the delicious irony in someone from this company reduced to "randomly poking" > to find a bug? > > Maybe you should ask "what is the essence of this crash" ;-) > > James Marshall. --------------------------- End of New-News digest ********************** From vxwexplo-errs@csg.lbl.gov Tue Aug 21 07:15:41 2001 From: John Moore Date: Tue Aug 21 07:15:43 PDT 2001 Subject: Re: Watchdog timer problems Watchdog timers execute within the context of the system clock interrupt. fprintf is not good to call inside an interrupt. Try logMsg instead. jmoore > > Submitted-by: "Vivek Soni" > > Hi Glenn > Ur example code works fine for me. > The problem is not with watchdog timer. > Only problem is with ur fprintf(). Probably u have not included > header file for fprintf(). That's why it give GPF while executing > fprintf(). > > - Vivek. > > Newsgroups: comp.os.vxworks > Subject: Watchdog timer problems > Date: Mon, 20 Aug 2001 12:21:58 +0200 > From: Glenn Ivar Klausen > Organization: Nextra Public Access > Message-ID: <3B80E4C6.6CA7164D@oslo.online.no> > > I've written som watchdog timer example code, but it dosent seem to be > working. If anyoun could tell med the problem here. > > #include > #include > > WDOG_ID wid; > int wdTimerid; > > void wdExpired(void *msg); > > void wdTest(void) > { > wid = wdCreate(); > > if(wdStart(wid, (sysClkRateGet()*5), wdExpired, "timer expired") == > OK) > { > fprintf(stdout, "Timer created ok. (wid = %lx) \n", wid); > } > else > { > fprintf(stdout, "Timer failed."); > } > } > > void wdExpired(void *msg) > { > fprintf(stdout, " msg = %s \n", (char *)msg); > } > > As far as I can see, this is supposed to work. But it dont. > - -- From vxwexplo-errs@csg.lbl.gov Tue Aug 21 09:51:34 2001 From: "David Laight" Date: Tue Aug 21 09:51:36 PDT 2001 Subject: Re: PCI hot swap > Each slot is having a different resource requirement. Don't you think that > if we reserve the resources for each slot before groping than won't it be > inefficient use of resources. For example there might be some memory left > unused. These are all 'virtual' resources, so not using them isn't a big problem. (unless you are short of physical address space on a system.) > Or, should the resources allocated to a slot during the initial grope be > reserved for that slot ? Absolutely. If cards have bridges then you cannot switch resource from behind a bridge with massive reconfiguration - which will change I/O and memory addresses assigned to other cards. Most PCI grope code allocates resource 'densly' during the grope - typically because your average PC is assumed to have a fixed configuration. Unfortunately, one this has been done it is almost impossible to support hot insertion. David no vxworks here..... From vxwexplo-errs@csg.lbl.gov Tue Aug 21 10:06:06 2001 From: "David Laight" Date: Tue Aug 21 10:06:09 PDT 2001 Subject: Re: dosfs file system corruptions > But in fact, I know one such bug in DosFs 2.0, which is related to > directories and causes lost cluster chains. Suppose you have a directory > "MyDir1", and a file "MyFile1", you can legally do "ename > ("MyFile1","MyDir1/MyFile1"), that is fine. But if you do > rename("MyDir1","MyDir1/MyDir2") you will loose all subdirectories and files do you mean the 'recursive one'? > under "MyDir1", which will be diagnozed as lost chains by CHKDSK later. DosFs 1 will lose the old contents of mydir if you do: rename( "myfile", "mydir" ) this is VERY easy to do with NFS... mv myfile mydir So is dosfs 2 any better here? dosFs 1 also gets confused by empty files. I've forgotten the exact sequence, but open, close, reboot (i think) leaves a FAT block allocated to the file and free list. I wrote a file system checker in order to try and isolate some corruptions - they stopped happening if we checked the fs..... but we never could reproduce the error, just suddenly got VERY corrupt FS. I read somewhere that the vxworks CHKDSK code uses the second copy of the FAT as temporary storage. What does it do if there is only one copy - probably sensible on compact flash etc. David From vxwexplo-errs@csg.lbl.gov Tue Aug 21 10:31:06 2001 From: "David Laight" Date: Tue Aug 21 10:31:08 PDT 2001 Subject: Re: Half duplex network settings The typical problem here is that the device driver has to set a bit in one of the ethernet control registers to enable/disable full/half duplex in the state engine of the ethernet chipset. This has to match the state of that the transceiver is in - or terrible things happen to the entire network segment. Note that the ethernet controller doesn't need to know whether you are running at 10MHz or 100MHz. The transceiver will normally autonegotiate or autodetect the fastest speed that both ends of the TP link support. The device driver then has to work out whether this is FDX or not - decidedly non trivial even if you have code that can read/write the transceiver registers. There are two problems: 1) you cannot tell whether autonegotiation or autodetect was used. (the current operating mode isn't in one of the standard tranceiver registers!) 2) you are not told when the link goes down - unless you poll a register (some ethernet controllers will poll an MII register for you and interrupt if/when it changes). This all makes life hard. To test you need: - a 10M hub (10M only) - a 100M hub (autodetect 100M) - a 10/100 dual backplane hub (2 hubs and a switch) (autoneg. 10/100 HDX) - a 10M switched hub (autoneg 10M FDX) - a 10/100M switched hub (autoneg 10/100 FDX) and check all transitions! If you are playing with the xcver registers, you can set to autodetect 100M, but autonegotiate to 10M... I've always forced HDX in the xcver MII register - then I know how to configure the driver.... There is a performance penalty, but it probably isn't very big in real life. All the problems with the vxworks driver for the DEC chipset are caused by the xcver negotiating FDX and the driver not being able to check! If you have your own hardware - check the xcver datasheet, it MAY be possible to configure the 'link advertise register' using pullup/down links. Or add a hub between the board and switch! David From vxwexplo-errs@csg.lbl.gov Wed Aug 22 04:03:35 2001 From: Vxworks Exploder Date: Wed Aug 22 04:03:37 PDT 2001 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Wed Aug 22 04:03:16 PDT 2001 Subject: Re: Debugging when the debugger crashes Subject: Re: How to rebuild libppc860gnuvx.a Subject: using inflate in a downlable application Subject: Re: Anyone get VxWorks to boot quickly Subject: multiple developers & env vars in Tornado (was dependency generation) Subject: Zinc/UGL design problem? Subject: Re: MPC860 rev D4 & Fast Ethernet Subject: Re: Anyone get VxWorks to boot quickly Subject: Re: memory 'free' suspends task... Subject: Exception handling in VxWorks Subject: Re: GDB problems with Tornado 2 for Mips (libR4000gnuvx.a) Subject: Re: Debugging when the debugger crashes Subject: Re: LINKAGE PROBLEM Subject: Re: MPC860 rev D4 & Fast Ethernet Subject: Re: How to rebuild libppc860gnuvx.a Subject: What does bootLoadModule do? Subject: Re: Extern's in header files Subject: Re: Exception handling in VxWorks Subject: Re: memory 'free' suspends task... Subject: Re: MPC860 rev D4 & Fast Ethernet Subject: Re: multiple developers & env vars in Tornado (was dependency generation) Subject: Re: multiple developers & env vars in Tornado (was dependency generation) Subject: silly linker error! Subject: Re: GDB problems with Tornado 2 for Mips (libR4000gnuvx.a) Subject: Re: MPC8260 BSP net driver problem Subject: about virtual memory exhausted error Subject: Re: using inflate in a downlable application Subject: Re: Debugging when the debugger crashes Subject: Re: Debugging when the debugger crashes Subject: Re: about virtual memory exhausted error Subject: Re: using inflate in a downlable application Subject: Re: HELP - vx ... FTP-Server is innocent, it's a dosFS2 problem Subject: Re: MPC860 rev D4 & Fast Ethernet Subject: Re: Tornado 2.0.1 dependency generation Subject: indentify name in different node,what argument need? Subject: Memory allocation Subject: Newby Info Subject: Re: Anyone get VxWorks to boot quickly ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: Debugging when the debugger crashes Date: Tue, 21 Aug 2001 11:32:39 +0100 From: vxworksintent@tao-group.com (David Given) Organization: I do not speak for anyone but myself, and barely that. Message-ID: <7cdtl9.lr4.ln@127.0.0.1> References: <2psil9.q9e.ln@127.0.0.1> <3B7E0220.9ACC02DE@covad.net> <3B822F37.6B80B863@agilent.com> Sender: dg@pearl.tao.co.uk In article <3B822F37.6B80B863@agilent.com>, James Marshall writes: [...] >> David Given >> Tao Group Ltd. > > This isn't a helpful suggestion, just a wicked observation. > > Does anyone else see the delicious irony in someone from this company reduced to "randomly poking" > to find a bug? Alas, I have lost my centre. Re the replies: thanks everyone. Unfortunately I don't have the ability to rebuild the BSP, so getting Wind River support isn't really feasible; which is a pity, because post-mortem mode sounds like exactly what I'm looking for. I will get on to my vendor. And what's system mode/task mode? I can't find any references in the documentation. I suspect most of my problems are down to a dodgy exception handler. I haven't been able to find any decent documentation on how to catch processor exceptions from a VxWorks task, and then restart them; if anyone has any pointers, I would be most grateful. - -- David Given Tao Group Ltd. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: How to rebuild libppc860gnuvx.a Date: 21 Aug 2001 05:06:12 -0700 From: kevinatnj@yahoo.com.cn (kewin) Organization: http://groups.google.com/ Message-ID: <221caa76.0108210406.49c0a25f@posting.google.com> References: <6031d81a.0108201655.5b6f10ad@posting.google.com> i don't know if it is necessary.But if you want to rebuild this Lib,i can do like this. >make CPU=PPC860 CPU means CPU type you are using. good luck --------------------------- Newsgroups: comp.os.vxworks Subject: using inflate in a downlable application Date: Tue, 21 Aug 2001 14:04:13 +0000 (UTC) From: anthony.rouaux@st2e.com (Anthony Rouaux) Organization: TEMEX - ST2E Message-ID: I all, i am trying to use the inflate fonction in a downlable application. My probleme is that this fonction is juste present in a file called libPPC860gnuvx.a and not in my vxwork image running on the target. So when I download the .out i have an error message that says me that it doesn t know inflate. How can i include this function from the .a file? Thanks a lot. Anthony --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Anyone get VxWorks to boot quickly Date: 21 Aug 2001 07:12:20 -0700 From: polyester99@hotmail.com (Kelvin Lawson) Organization: http://groups.google.com/ Message-ID: <7fbcd71e.0108210612.648c3f12@posting.google.com> References: <488e459a.0108202322.421fa0f7@posting.google.com> I have found that complicated MMU configurations can add a lot of time to the boot process. This seems to be down to the call to usrMmuInit(). It completes much faster with a simple sysPhysMemDesc structure. Kelvin. > > I have a need to get a Tornado 2.0 VxWorks ROM image (about 4MB in size) > > loaded and booted on a Power PC VME board in 3-4 seconds. It is currently > > taking about 6. Does anyone have any lessons learned on getting a VxWorks > > image to boot quickly? --------------------------- Newsgroups: comp.os.vxworks Subject: multiple developers & env vars in Tornado (was dependency generation) Date: Tue, 21 Aug 2001 10:16:47 -0400 From: "Kevin Livingston" Message-ID: I believe some people have talked about having multiple developers working on the same Tornado project. We would like to do that as well. Our current "attempted" solution is to use environment variables to define where include directories are. (e.g. using -I on the compile line). Tornado has the very handy macro PRJ_DIR which can be used. however, if it is used, then the project is saved and then reopened. (or just reopened by anyone with tornado) PRJ_DIR will become expanded - and thus defeat the purpose, because when the project is saved it will have absolute directories, and make it difficult for any other developers. I thought I had a solution by using OS defined environment variables. As tornado does not expand these variables when a project is opened. However, I believe that tornado uses the information in the compile line: included directories, to help it form its dependency lists. But, when doing this tornado cannot understand foreign environment variables well, only its own (e.g. PRJ_DIR) and frequently generates faulty dependencies - either defeating the purpose of having dependencies, or corrupting the information in the make file to a point where the project cannot be built at all. it is not a requirement to use environment variables, we would just like to have multiple developers working on the same project, and currently the only major thing holding us back, is setting the compile line for the project. (as well as the other settings which need directories - e.g. project libs.) does anyone else experience this behavior? or have a workaround or suggestions? (I already use forward slashes; there are no spaces in my source between # and include; I have a VPATH etc. -- the problem now appears with tornado - and I guess technically it is not a problem, they do not have to support OS environment variables - but it is the easiest (no cost) way to get multiple developers seamlessly involved.) If it helps we are using Tornado 2.0.2 with the Host running windows NT. thanks for your help Kevin --------------------------- Newsgroups: comp.os.vxworks Subject: Zinc/UGL design problem? Date: 21 Aug 2001 10:19:26 -0400 From: Bill Pringlemeir Organization: Bell Sympatico Message-ID: Sender: bpringlemeir@DeadDuck Summary - ------- How to you synchronize data in a message queue if one task can delete/free a pointer in the message? Real problem - ------------ I have found a problem with the structure of WindML2.0 and Zinc 6.0. If a device has found an event and the uglInputTask() is running to handle an event, then the window callback functions such as "winCallbackEventRoute" will be called and a window ID/handle/pointer will be attached to the message. This message is placed in a queue and received by Zinc at a latter time. During this time, the window can be delete and the stale handle will be used by the Device::Poll() function. Does anyone have suggestions on how to mark events in a queue after they have been sent? I think that maybe attaching message information to the window might be possible. Then when the window is deleted, the messages could be cancelled. However, there is still a window between when the receiving task has taken the message from the queue and when it is finished with the message. Another approach that I thought of was to mark the UGL Window with an ID field and a magic number. The winDestroy() would clear the structure before deallocation. Then the Device::Poll functions can check that the Window ID is valid. The bad sequence goes, +--------------+ +--------------+ +----------+ +------------+ context:| Zinc |->| uglInput |->| Zinc |->| Zinc | action: |delete eminent|->|device message|->|win delete|->|message rcv.| +--------------+ +--------------+ +----------+ +------------+ regards, Bill Pringlemeir. - -- The birds of leaving call to us. Yet here we stand endowed with the fear of flight. - Dead can Dance vxWorks FAQ, "http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html" --------------------------- Newsgroups: comp.os.vxworks Subject: Re: MPC860 rev D4 & Fast Ethernet Date: Tue, 21 Aug 2001 14:19:32 GMT From: "Steeve Gagne" Organization: MetroNet Communications Group Inc. Message-ID: References: <3b820962.2131915@modem.mecalc> We use the motFecEnd Driver with MPC860D4 Revision It is working fine and we found the same problem than you did. There is a bug in the motFecEnd.c Ethernet driver. Ask your FAE if vxWorks found the same problem than we found. - -- Steeve Gagne Simpler Networks Inc. steeve.gagne@simplernetworks.com steeve.gagne@videotron.ca "Lourens Geldenhuys" wrote in message news:3b820962.2131915@modem.mecalc... > Is anybody using an MPC860D4 processor with WindRiver's ads860 BSP? We > replaced the MPC860B3 silicon on our (working) design with rev D4 > silicon, and now our Ethernet does not work. We are using the FEC and > RELEASE 1.2/1, Jul 2000 of the ads860 BSP, which does support rev D4 > silicon. Also, I have modified the XFC capacitors to match the rev D4 > spec, still no Ethernet. > > Any advice? > > Thank you, > 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: Re: Anyone get VxWorks to boot quickly Date: Tue, 21 Aug 2001 14:22:55 GMT From: "Steeve Gagne" Organization: MetroNet Communications Group Inc. Message-ID: <39ug7.686$bJ6.70166305@news1.mtl.metronet.ca> References: Verify if the instruction cache is ON. We found that this feature was disabled in our BSP and run-time image. You can remove the memory fill function at boot time. (the memory is filled with 0's at boot time) - -- Steeve Gagne steeve.gagne@videotron.ca "Brad Crabtree" wrote in message news:hsbg7.10$DK5.172@dfw-service2.ext.raytheon.com... > I have a need to get a Tornado 2.0 VxWorks ROM image (about 4MB in size) > loaded and booted on a Power PC VME board in 3-4 seconds. It is currently > taking about 6. Does anyone have any lessons learned on getting a VxWorks > image to boot quickly? > > -- > -------------------------------------------- > Bradley D. Crabtree > b-crabtree@raytheon.com > > > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: memory 'free' suspends task... Date: Tue, 21 Aug 2001 14:24:32 GMT From: "Steeve Gagne" Organization: MetroNet Communications Group Inc. Message-ID: References: memOptionsSet ( MEM_BLOCK_CHECK | /* optional check for bad blocks */ MEM_ALLOC_ERROR_LOG_FLAG | /* response to errors when allocating memory */ MEM_BLOCK_ERROR_LOG_FLAG | MEM_ALLOC_ERROR_RETURN | MEM_ALLOC_ERROR_LOG_MSG | MEM_BLOCK_ERROR_RETURN | MEM_BLOCK_ERROR_LOG_MSG /* | MEM_BLOCK_ERROR_SUSPEND_FLAG */ ) ; - -- Steeve Gagne Simpler Networks Inc. steeve.gagne@simplernetworks.com steeve.gagne@videotron.ca "Changhyun Oe" wrote in message news:Xznf7.16684$2k2.904864@news.bora.net... > When I tried 'free'ing the memory, > > It always suspends the task which calls 'free'. > > Please tell me why... > > I need your help.... > > Previous Thanks to you~ > > - chang - > > > --------------------------- Newsgroups: comp.os.vxworks Subject: Exception handling in VxWorks Date: 21 Aug 2001 08:27:07 -0700 From: vivekp@hotmail.com (pvpartha1) Organization: http://groups.google.com/ Message-ID: <623ba2de.0108210727.75b5534f@posting.google.com> Is it possible to trap data exception errors or bus errors and suspend the thread that causes that to happen? How is this done? any help would be appreciated Thanks --------------------------- Newsgroups: comp.os.vxworks Subject: Re: GDB problems with Tornado 2 for Mips (libR4000gnuvx.a) Date: Tue, 21 Aug 2001 15:27:49 GMT From: "Nihad Alibegovic" Organization: Road Runner Message-ID: References: <__Af7.2724$Qh2.545742@typhoon.san.rr.com> This could be the very same problem. Please get me some code to try. I will try what you put in below and then anything else you give me. We have had terrible time with this so far and like yours it has now been 2 weeks without reply. I will also check out your SPR. Thank you very much for your help. "Girish V. Gulawani" wrote in message news:gS6g7.3713$j04.314817@nnrp.gol.com... > perhaps i have come across similar problem earlier. that was related to SPR > 26387 (unpublished). > the problem is in getting mips registers while attaching task through gdb. & > have already communicated to wrs. but ofcourse no reply. > to this spr 26387 details are - > in the WRS distributed source file > /target/config/comps/src/usrWdbTask.c the function vxTaskRegsGet() fails. > vxTaskRegGet () > { > TCB *tcb; > ... > tcb = taskTcb() > switch(regSetType) { > case WDB_REG_SET_IU: > { > REG_SET dummy; > taskRegsGet(context, &dummy); **** problem *** > *ppRegSet = (char*)tcb->regs; > break; > } ... > } > if you comment of the problem statement it works. then there is one more > place you have to change that is delete hook in the same file. however i > dont have the code right now. > if you feel it is the same problem do post here. i'll respond asap. > my debugger is working properly now. but still wrs support has not replied > and its over 3/4 months. > > > "Nihad Alibegovic" wrote in message > news:__Af7.2724$Qh2.545742@typhoon.san.rr.com... > > Hello everyone, > > > > we ported our BSP from vxWorks 5.3 to vxWorks5.4 on Tornado 2. When it > was > > all said and done everything works except GDB. Target attaches fine we > can > > put break points but as soon as we try to single step we get tWdb > exception. > > It is worth noting that if I put a break point in from the host shell but > > single step in target resident shell things work well. So, I guess this > is > > not just a GDB problem. Additionally, I have same problem when trying to > > attach to the task. It is as if the host and target memories are > completely > > out of sync and host is accessing memory it is not supposed to access on > the > > target. We use Network to debug which means we are in task mode debugging > > only. > > > > The platform we run on is Mips 4310 CPU so we use libR4000gnuvx.a. We > have > > all the WDB includes in our BSP that we need as certified by Wind River. > > Wind River has helped very little in this issue in the last week. While, > at > > this does anyone get quick support from these guys. In 2 years working > with > > vxWorks they have the slowest customer support of any company I ever > worked > > with. > > > > Nihad Alibegovic > > MTS3 > > nalibegovic@hns.com > > > > > > > > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Debugging when the debugger crashes Date: Tue, 21 Aug 2001 16:39:10 +0100 From: James Marshall Organization: Agilent Technologies UK Ltd Message-ID: <3B82809D.775FD159@agilent.com> References: <2psil9.q9e.ln@127.0.0.1> <3B7E0220.9ACC02DE@covad.net> <3B822F37.6B80B863@agilent.com> <7cdtl9.lr4.ln@127.0.0.1> > I suspect most of my problems are down to a dodgy exception handler. I > haven't been able to find any decent documentation on how to catch > processor exceptions from a VxWorks task, and then restart them; if anyone > has any pointers, I would be most grateful. Ah now I may be able to help there. I use excHookAdd() to add my exception handler as I found out on the group archive. One problem I've had with it is the way VxWorks handles exceptions (on x86 anyway): the exception isn't reported/handled until a task runs. I don't know how MIPS processor architecture uses exceptions, but on x86 NMI is an exception rather than an interrupt. I guess that usually exceptions are synchronous, that is it's something that a running task does, so there is always a current task in which to process the exception. However NMI is asynchronous and in x86 architecture it is implemented as an exception, which seems to break VxWorks exception-handling model. I appeared to have solved it by adding a task at priority 255 that loops endlessly and is always ready to run, so my exception handler is called within 100us instead of whenever the next task happens to run. I tried to find a way of resuming an excepting task that has been suspended and came up with this: - - in the exception handler call excJobAdd(), passing it a pointer to a function that spawns a task at priority 1, and passing the current task id. [excJobAdd() works like netJobAdd(), passing a function to be called from tExcTask, and is declared in excLib.h] - - in the spawned task, wait until your excepting task is suspended (yes, I had to add a taskDelay(1)), then resume it and exit the spawned task. You can improve this by removing the taskSpawn stage (which is slow) and creating your own task, message queue and xxxJobAdd() call to process the task resume. The spawned task which does the work seems to need a high priority which is less than tExcTask so that the VxWorks exception-handling functions which execute in the context of tExcTask can complete. Although I discovered that this works, I haven't yet taken it further. Maybe others know more... James Marshall. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: LINKAGE PROBLEM Date: 21 Aug 2001 08:52:13 -0700 From: john_94501@yahoo.com (John) Organization: http://groups.google.com/ Message-ID: <488e459a.0108210752.78b4194c@posting.google.com> References: <9lt368$asbte$1@ID-69140.news.dfncis.de> Hello, You need to add the appropriate components to your kernel image. I don't know what you mean by WindNet, but what you are looking for is things of the form INCLUDE_xxx where xxx is replaced by the name of the feature you need. Now, depending on how you are building the image, the way to do this varies. If you are using the IDE, then it contains a tool for adding and removing these features. If you are building from the BSP directory (the old T1 way), then take a look in config.h and configAll.h (the latter is in config/all). If you are not building a kernel at all, but just using the default one off the CD, you need to read the manuals... At the very least, you should read the VxWorks Programmer's Guide and sections 2 and 4 of the Tornado User's Guide as well. HTH, John... "RK" wrote in message news:<9lt368$asbte$1@ID-69140.news.dfncis.de>... > Hi > > We are using Tornado 2 to download our application on VxWorks 5.4 based > bootable image. Here is the problem. When we use WindNet function in > application we get unresolved symbols when try to download the application > (This function are not called in bootable image, so linker could not link > them into bootable image - All these function are placed in libPPC860gnuvx.a > library ). Is there some efficient way how to force linking and resolve > symbols in bootable image without calling those function? We need to > initialize WindNet protocols not earlier than in application, because some > configuration need to read before initialization. > > Thanks, Dejan --------------------------- Newsgroups: comp.os.vxworks Subject: Re: MPC860 rev D4 & Fast Ethernet Date: 21 Aug 2001 09:48:38 -0700 From: chrismooremobile@yahoo.com (Chris Moore) Organization: http://groups.google.com/ Message-ID: References: <3b820962.2131915@modem.mecalc> Hi, I've written the FEC driver for the PPC860 for pSOS that we use on our old system. There are quite a few little changes for the REVD4. Basically the pin muxing on the MII port has changed, initially it was written wrong in the documents from Mot and all sorts of fun stuff. Easiest is to contact WRS as it seems there is a driver. If you want to fix it your self you can look on the mot site for erratas for the FEC. If you want your driver to work across old and new revisions you will need to check the chip revision (part of the IMMR register (high bits I think from mem) and init the pins accordingly. If you want to do it your self and can't find the info, zap me an email and I'll track down my code changes for you. Rgds, Chris Moore Ericsson lourens@mecalc.co.za (Lourens Geldenhuys) wrote in message news:<3b820962.2131915@modem.mecalc>... > Is anybody using an MPC860D4 processor with WindRiver's ads860 BSP? We > replaced the MPC860B3 silicon on our (working) design with rev D4 > silicon, and now our Ethernet does not work. We are using the FEC and > RELEASE 1.2/1, Jul 2000 of the ads860 BSP, which does support rev D4 > silicon. Also, I have modified the XFC capacitors to match the rev D4 > spec, still no Ethernet. > > Any advice? > > Thank you, > 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: Re: How to rebuild libppc860gnuvx.a Date: Tue, 21 Aug 2001 13:03:25 -0400 From: david lindauer Message-ID: <3B82945D.4D39E77D@notifier-is.net> References: <6031d81a.0108201655.5b6f10ad@posting.google.com> the easiest way to deal with this is just copy the file out of the src tree and put it in your project. Then you don't have to worry about rebuilding libs (of course if you are going to use it for dozens of different projects you do want to rebuild the libs). David Vinh Huynh wrote: > Hello, > > I make some changes to the bootline configuration. I like to update > the bootChange function in usrLib.c to use our new bootline. > > 1. Is this require to rebuild to vxWorks library(libppc860gnuvx.a)? > 2. If the first question is true, how do build new library for > vxWorks? > > Thanks, > Vinh --------------------------- Newsgroups: comp.os.vxworks Subject: What does bootLoadModule do? Date: 21 Aug 2001 10:14:42 -0700 From: ryeung@earthlink.net (Raymond Yeung) Organization: http://groups.google.com/ Message-ID: I'm customizing my system image (i.e. vxWorks + application). I need to know what the system function bootLoadModule provides? Does it take care of symbol table, code relocation? Any documentation? Thanks, Raymond --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Extern's in header files Date: 21 Aug 2001 10:18:12 -0700 From: mfhaigh@myrealbox.com (Mark) Organization: http://groups.google.com/ Message-ID: <64c96d8c.0108210918.7cdad652@posting.google.com> References: venkat_sr@yahoo.com (Venkat S) wrote in message news:... > Hi, > What is the purpose of externs in the the header files such as > sysLib.h? Is is only for the obvious reason that an user who #includes > this file can call the extern functions or does it have other > purposes too. > Another purpose I can think of is that the designer of vxWorks > have defined an inteface and developers of VxWorks BSP's provide the > code for the functions? For example sysLib.h defines the inteface > while sysLib.c written by the BSP developer provides the code which > implements these functions. Am I thinking right? The extern in the function declaration is really just a convention to show that the functions are in other files that are "far away" in libraries, modules, or object files somewhere else. See comp.lang.c FAQ 1.11. You can get rid of the extern in front of the declaration, but personally when I encounter this (in some .h): int foo(); void bar(); extern int baz(); I won't spend a whole lot of time looking in the local module for function baz. Some may debate the usefulness of this approach, but I think it can be helpful sometimes. Mark S. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Exception handling in VxWorks Date: Tue, 21 Aug 2001 17:08:54 GMT From: Lee DeRaud Organization: None to speak of Message-ID: References: <623ba2de.0108210727.75b5534f@posting.google.com> Sender: nntp@news.boeing.com (Boeing NNTP News Access) On 21 Aug 2001 08:27:07 -0700, vivekp@hotmail.com (pvpartha1) wrote: >Is it possible to trap data exception errors or bus errors and suspend >the thread that causes that to happen? How is this done? >any help would be appreciated In what architecture? Most of them already have bus error and/or data access exceptions (e.g. misaligned accesses) enabled by default, and suspension of the offending task is (part of) the default handling. (Note that, much to the surprise of everyone here, FPU exceptions (e.g. divide-by-zero) are *not* enabled in M68K.) Lee --------------------------- Newsgroups: comp.os.vxworks Subject: Re: memory 'free' suspends task... Date: 21 Aug 2001 10:35:46 -0700 From: mfhaigh@myrealbox.com (Mark) Organization: http://groups.google.com/ Message-ID: <64c96d8c.0108210935.482a13ca@posting.google.com> References: <3B7E0BC0.B9EC83A3@covad.net> DrDiags wrote in message news:<3B7E0BC0.B9EC83A3@covad.net>... > Changhyun, > > This is how the free() will work if you attempt to free a > pointer that is no longer pointing to malloc'd space from a > task. This should be documented in the online manuals for vxWorks. > What you should do is check to see if the pointer is NULL before > freeing and after successfully freeing memory, set the pointer > to NULL. I have seen bugs in the past where global pointers are > set to malloc spaced, then some task frees it, but another task > in an cleanup role, attempts to free all pointers which are > non-NULL. Macros can come in handy: #if DEBUG_OF_SOME_SORT #define FREE(x) \ do { \ printf("FREE: pointer " #x "\n"); \ free(x); \ x = NULL; \ } while(0) #else #define FREE(x) free(x) #endif This helps matters for me sometimes. Hope this is helpful to you. Mark S. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: MPC860 rev D4 & Fast Ethernet Date: Tue, 21 Aug 2001 11:20:30 -0700 From: "Michael R. Kesti" Organization: MK Associates Message-ID: <3B82A66E.F69930AF@gv.net> References: <3b820962.2131915@modem.mecalc> Reply-To: mkesti@gv.net Lourens Geldenhuys wrote: >Is anybody using an MPC860D4 processor with WindRiver's ads860 BSP? We >replaced the MPC860B3 silicon on our (working) design with rev D4 >silicon, and now our Ethernet does not work. We are using the FEC and >RELEASE 1.2/1, Jul 2000 of the ads860 BSP, which does support rev D4 >silicon. Also, I have modified the XFC capacitors to match the rev D4 >spec, still no Ethernet. > >Any advice? It turns out that Port D configuration of rev D3 and later parts differs from that of earlier revisions. Specifically, the meaning of the relevant Port D direction register bits differ. In earlier revisions, this register is used to configure I/O direction regardless of whether the pins are used for general purpose or "special" mode (such as SCC TX and RX). The D3 revision cleaned up the MII/FEC/100BaseT network interface, which is a second special function of Port D on the MPC860. The Port D direction register is now used to select between these special functions rather than the direction of a single function. Note that this change can also affect operation of SCC3 and SCC4 when they use port D for their I/O. I suspect that your BSP's sysFecEnetEnable() function in sysLib.c sets PDDIR to 0x1c58. Change this value to 0x1fff and the FEC on your D4 boards will likely come back to life. If you need to support all verisons in a single code set you can set PDDIR conditionally, depending on the value read from the IMMR. First, write a function that returns the entire contents of the IMMR. (vxImmrGet() masks off the revision data.) Then, test whether the low 16 bits of the value returned by this function are greater than or equal to 0x0501. If so, you're on a D3 or later rev, and can write the PDDIR accordingly. - -- ======================================================================== Michael Kesti | "And like, one and one don't make | two, one and one make one." mkesti@gv.net | - The Who, Bargain --------------------------- Newsgroups: comp.os.vxworks Subject: Re: multiple developers & env vars in Tornado (was dependency generation) Date: Tue, 21 Aug 2001 22:06:12 +0100 From: molochai Message-ID: <9898b1ad4a%molochai@vapour-trail.demon.co.uk> References: In message "Kevin Livingston" wrote: > I believe some people have talked about having multiple developers working > on the same Tornado project. We would like to do that as well. Our > current "attempted" solution is to use environment variables to define > where include directories are. (e.g. using -I on the compile line). > Tornado has the very handy macro PRJ_DIR which can be used. however, if it > is used, then the project is saved and then reopened. (or just reopened by > anyone with tornado) PRJ_DIR will become expanded - and thus defeat the > purpose, because when the project is saved it will have absolute > directories, and make it difficult for any other developers. The best bet is to make each developers environment appear the same to the tools. It does kind of seem as if you're not using any version-control or Configuration management tool. For any kind of multi-developer project, using one is essential, and will provide you with an evironment which is consistent across developer workstations. Any approach using environment variables - which I assume will be set differently on each developer's workstation - is asking for nasty version control and build problems. > > it is not a requirement to use environment variables, we would just like to > have multiple developers working on the same project, and currently the > only major thing holding us back, is setting the compile line for the > project. (as well as the other settings which need directories - e.g. > project libs.) Version control workspace and project files and set project options in the normal way. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: multiple developers & env vars in Tornado (was dependency generation) Date: Tue, 21 Aug 2001 23:48:36 +0200 From: "Martin Raabe" Organization: T-Online Message-ID: <9lukvk$3dm$06$1@news.t-online.com> References: <9898b1ad4a%molochai@vapour-trail.demon.co.uk> Hello folks, I recommend using SNiFf+ for Tornado, which comes with RCS Version Control Software inclusive. It has team development as the top goal, which it reaches easily. For more information look at: www.windriver.com/sniff Hope it helps. Ciao Martin Raabe "molochai" schrieb im Newsbeitrag news:9898b1ad4a%molochai@vapour-trail.demon.co.uk... > In message > "Kevin Livingston" wrote: > > > I believe some people have talked about having multiple developers working > > on the same Tornado project. We would like to do that as well. Our > > current "attempted" solution is to use environment variables to define > > where include directories are. (e.g. using -I on the compile line). > > Tornado has the very handy macro PRJ_DIR which can be used. however, if it > > is used, then the project is saved and then reopened. (or just reopened by > > anyone with tornado) PRJ_DIR will become expanded - and thus defeat the > > purpose, because when the project is saved it will have absolute > > directories, and make it difficult for any other developers. > > The best bet is to make each developers environment appear the same to the > tools. > > It does kind of seem as if you're not using any version-control or > Configuration management tool. For any kind of multi-developer project, using > one is essential, and will provide you with an evironment which is consistent > across developer workstations. > > Any approach using environment variables - which I assume will be set > differently on each developer's workstation - is asking for nasty version > control and build problems. > > > > > it is not a requirement to use environment variables, we would just like to > > have multiple developers working on the same project, and currently the > > only major thing holding us back, is setting the compile line for the > > project. (as well as the other settings which need directories - e.g. > > project libs.) > > Version control workspace and project files and set project options in the > normal way. > > --------------------------- Newsgroups: comp.os.vxworks Subject: silly linker error! Date: Tue, 21 Aug 2001 19:13:35 -0400 From: Bruce Chrustie Organization: Nortel Message-ID: <3B82EB1F.753DACBA@somewhereinhell.com> Hi folks, I am using the pci config & show routines in my code. specifically pciFindDevice(TJVENDORID & 0xFFFF, (PCIPHONEID >> 16) & 0xFFFF, i, &pciBusNo, &pciDevNo, &pciFuncNo) != ERROR) and return = pciDeviceShow(0); I have included the appropriate header defs, yet get the following error upon linking, not compiling: undefined reference to `pciDeviceShow' any thoughts? I do not get any errors from the pciFindDevice routine, and yet they are so closely related! Just 1 module away! The only think I can see is for pciDeviceShow it is conditionally included based upon INCLUDE_SHOW_ROUTINES which I think I have def'd correctly in all possible places. Thoughts always appreciated. Are the complier & linker errors documented anywhere? Bruce, --------------------------- Newsgroups: comp.os.vxworks Subject: Re: GDB problems with Tornado 2 for Mips (libR4000gnuvx.a) Date: Wed, 22 Aug 2001 00:28:32 GMT From: "Girish V. Gulawani" Organization: Global Online Japan // Exodus Communications KK. Message-ID: References: <__Af7.2724$Qh2.545742@typhoon.san.rr.com> there are infact 2 changes in the same file - /target/config/comps/src/usrWdbTask.c .1 in function vxTaskRegsGet : comment out the /* DO NOT REMOVE THIS CALL TO taskRegsGet (SPR #26387) */ taskRegsGet (pContext->contextId,&dummy); -- this one you comment out. .2 check for valid delete hook function. i dont know who changes it & makes it invalid entry, (perhaps cache?). _ _wdbTaskDeleteHook() { ..... void(*hook()); void(* myhook()); hook= TCB->WdbExitHook; myhook=wdbCtxExitNotifyHook; if (hook!= myhook) hook=myhook; ..... } i hope this should work for you too. keep us posted here. "Nihad Alibegovic" wrote in message news:V5vg7.6338$Qh2.1774693@typhoon.san.rr.com... > This could be the very same problem. Please get me some code to try. I > will try what you put in below and then anything else you give me. We have > had terrible time with this so far and like yours it has now been 2 weeks > without reply. I will also check out your SPR. Thank you very much for > your help. > > "Girish V. Gulawani" wrote in message > news:gS6g7.3713$j04.314817@nnrp.gol.com... > > perhaps i have come across similar problem earlier. that was related to > SPR > > 26387 (unpublished). > > the problem is in getting mips registers while attaching task through gdb. > & > > have already communicated to wrs. but ofcourse no reply. > > to this spr 26387 details are - > > in the WRS distributed source file > > /target/config/comps/src/usrWdbTask.c the function vxTaskRegsGet() fails. > > vxTaskRegGet () > > { > > TCB *tcb; > > ... > > tcb = taskTcb() > > switch(regSetType) { > > case WDB_REG_SET_IU: > > { > > REG_SET dummy; > > taskRegsGet(context, &dummy); **** problem *** > > *ppRegSet = (char*)tcb->regs; > > break; > > } ... > > } > > if you comment of the problem statement it works. then there is one more > > place you have to change that is delete hook in the same file. however i > > dont have the code right now. > > if you feel it is the same problem do post here. i'll respond asap. > > my debugger is working properly now. but still wrs support has not replied > > and its over 3/4 months. > > > > > > "Nihad Alibegovic" wrote in message > > news:__Af7.2724$Qh2.545742@typhoon.san.rr.com... > > > Hello everyone, > > > > > > we ported our BSP from vxWorks 5.3 to vxWorks5.4 on Tornado 2. When it > > was > > > all said and done everything works except GDB. Target attaches fine we > > can > > > put break points but as soon as we try to single step we get tWdb > > exception. > > > It is worth noting that if I put a break point in from the host shell > but > > > single step in target resident shell things work well. So, I guess this > > is > > > not just a GDB problem. Additionally, I have same problem when trying > to > > > attach to the task. It is as if the host and target memories are > > completely > > > out of sync and host is accessing memory it is not supposed to access on > > the > > > target. We use Network to debug which means we are in task mode > debugging > > > only. > > > > > > The platform we run on is Mips 4310 CPU so we use libR4000gnuvx.a. We > > have > > > all the WDB includes in our BSP that we need as certified by Wind River. > > > Wind River has helped very little in this issue in the last week. > While, > > at > > > this does anyone get quick support from these guys. In 2 years working > > with > > > vxWorks they have the slowest customer support of any company I ever > > worked > > > with. > > > > > > Nihad Alibegovic > > > MTS3 > > > nalibegovic@hns.com > > > > > > > > > > > > > > > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: MPC8260 BSP net driver problem Date: Tue, 21 Aug 2001 17:01:56 +0100 From: Graham Baxter Organization: Graham Baxter (Software) Limited Message-ID: <3B8285F4.1108B28@NOSPAM.bcs.org.uk> References: <221caa76.0108200428.6219f046@posting.google.com> If you are using both FCC1 and FCC2: Check the FCC1 and FCC2 parameter RAM. Make sure that RIPTR is different for FCC1 and FCC2. (Also TIPTR). It was hard coded in the Ethernet driver to 0x3000 (tiptr 0xB000). Regards, Graham Baxter Freelance Software Engineer gbaxter@NOSPAM.bcs.org.uk --------------------------- Newsgroups: comp.os.vxworks Subject: about virtual memory exhausted error Date: 22 Aug 2001 10:59:26 +0900 From: Yasuyuki Kino Organization: OMRON SOFTWARE Co., Ltd. Message-ID: I'm using Tornado 2.0 on Windows 2000 When I try to compile c++ source code with "-O2" compiler option, cc1plus.exe sends an error message "virtual memory exhousted". The memory is still left. Why is it? - -- Kino --------------------------- Newsgroups: comp.os.vxworks Subject: Re: using inflate in a downlable application Date: Wed, 22 Aug 2001 03:27:51 GMT From: pkockritz@home.com (Pete Kockritz) Organization: Excite@Home - The Leader in Broadband http://home.com/faster Message-ID: References: In article , anthony.rouaux@st2e.com (Anthony Rouaux) wrote: > I all, > i am trying to use the inflate fonction in a downlable application. My > probleme is that this fonction is juste present in a file called > libPPC860gnuvx.a and not in my vxwork image running on the target. So when > I download the .out i have an error message that says me that it doesn t > know inflate. > How can i include this function from the .a file? > > Thanks a lot. > > Anthony The inflateLib functions are not in your kernel because nothing in the kernel calls them. The linker will only include .o files from a library if it needs it to satisfy an unresolved reference. Inflate is typically only used in the bootrom. So you have two options: 1) Add a reference to inflate() in the kernel, so that the linker must pull in inflateLib.o to satisfy it. You don't have to actually call inflate, just add a function to an existing kernel .c file that calls inflate, but which is never called itself. 2) Extract the inflateLib.o from the .a file. You can then link this .o file with your other files. Look at the man page for arppc to see how to do this. Hope this helps. - -- Pete Kockritz --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Debugging when the debugger crashes Date: 21 Aug 2001 21:43:47 -0700 From: john_94501@yahoo.com (John) Organization: http://groups.google.com/ Message-ID: <488e459a.0108212043.29e13e3b@posting.google.com> References: <2psil9.q9e.ln@127.0.0.1> <3B7E0220.9ACC02DE@covad.net> <3B822F37.6B80B863@agilent.com> <7cdtl9.lr4.ln@127.0.0.1> Hello, > And what's system mode/task mode? I can't find any references in the > documentation. Basically, Tornado offers two modes of debugging. One where you debug the individual tasks in the system (the normal mode), and a special version called system mode debugging that disconnects the target agent from the operating system and works more like a ROM monitor. This primarily supports debugging of interrupt handlers, and as a result works in polling mode. For info, check out section 7.5 of the Tornado II User's Guide. > I suspect most of my problems are down to a dodgy exception handler. I > haven't been able to find any decent documentation on how to catch > processor exceptions from a VxWorks task, and then restart them; if anyone > has any pointers, I would be most grateful. To catch exceptions you have a few options... 1) You can use a hook (see excHookAdd() in the VxWorks reference manual). This will allow to to do some processing when a task takes an exception. If you're on x86 platforms I seem to remember that a double fault causes the CPU to halt - this might be the cause of your problem. 2) Another option for handling exceptions is to register a signal handler and catch the signal. Signals are a higher level solution though, and if you are double faulting then the signal delivery may not happen. To actually restart the task once it has crashed, see the taskRestart() function. A signal handler can call this, as can the exception hook routine I think. If your task creates anything outside of its main loop though, you may want to consider using setjmp/longjmp instead of taskRestart(). This allows you a little more control about where in the sequence of your task you resume execution from. HTH, John... --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Debugging when the debugger crashes Date: 21 Aug 2001 22:14:53 -0700 From: john_94501@yahoo.com (John) Organization: http://groups.google.com/ Message-ID: <488e459a.0108212114.6a0a2485@posting.google.com> References: <2psil9.q9e.ln@127.0.0.1> <3B7E0220.9ACC02DE@covad.net> <3B822F37.6B80B863@agilent.com> <7cdtl9.lr4.ln@127.0.0.1> <3B82809D.775FD159@agilent.com> Hello, James Marshall wrote in message news:<3B82809D.775FD159@agilent.com>... > I use excHookAdd() to add my exception handler as I found out on the group archive. One problem I've had > with it is the way VxWorks handles exceptions (on x86 anyway): the exception isn't reported/handled > until a task runs. An "exception" cannot occur unless a task is running (unless it happens in an ISR, and that is a panic situation that causes the OS to reboot). It strikes me that the NMI should in fact be hooked in using not the excConnect() function, but the intConnect() one. Since I cannot see an NMI in the standard header file for the x86 family, it will be treated as a generic exception vector (which vector number is it on?). You can change this - see solution 2 below. > I tried to find a way of resuming an excepting task that has been suspended and came up with this: > > - in the exception handler call excJobAdd(), passing it a pointer to a function that spawns a task at > priority 1, and passing the current task id. [excJobAdd() works like netJobAdd(), passing a function to > be called from tExcTask, and is declared in excLib.h] > > - in the spawned task, wait until your excepting task is suspended (yes, I had to add a taskDelay(1)), > then resume it and exit the spawned task. This is perhaps a little complicated. A couple of simpler solutions for your case might be the following: 1) Use the older hook (_func_excBaseHook) instead of the one you're using right now. This hook is called with four parameters: int vecNum, ESF0 * pESF, REG_SET * pRegs, EXC_INFO * pExcInfo The return value should be FALSE if the exception was not handled, or TRUE if it was. So, if you detect your special NMI exception, just return TRUE and the exception handler will exit without suspending the task. 2) Better still, simply replace the vector for the NMI. If you know the vector number (and I can't see it in the header files for x86 so I'm not sure where the NMI vector is), simply connect a routine to that vector using intConnect() - yes, intConnect() is used for exception handler connection on x86 platforms. Your handler routine can simply be a return, and the offending NMI exceptions will be ignored. - -- Something for others to be aware of... in the case of a normal class of exception, simply resuming the task will not do any good. When a CPU takes a synchronous exception, the program counter is normally set to the address of that exception. The intention is that the handler will fix the cause of the exception (think paging memory on big operating systems; the exception happens and the OS gets the missing page off the disk) and then retry the instruction. But all is not lost. There are a number of ways of recovering from these types of exception, such as page faults. Take a look in the reference manual at the following routines: taskRestart() or setjmp()/longjmp(). These can be called from a hook routine, or from a signal handler (all the conventional exceptions are hooked into signals too if a signal handler has been registered by the task). The way to choose? Signals are task specific; each task must register to receive them, but can easily do its own thing to handle an error. The hook routine is the same for all exception types, and all tasks. This might be what you need, but means that the hook routine has a lot more work to do... HTH, John... --------------------------- Newsgroups: comp.os.vxworks Subject: Re: about virtual memory exhausted error Date: 21 Aug 2001 22:31:50 -0700 From: john_94501@yahoo.com (John) Organization: http://groups.google.com/ Message-ID: <488e459a.0108212131.7bfebe80@posting.google.com> References: Hello, Welcome to the world of Windoze... Basically, your swap space is being exhausted by the compiler (compiling complex code can do this, especially it seems C++ code). To solve the problem you have a choice: 1) Increase the size of the swap space (paging file) using the Windows control panel. It is normally on the System tool's performance tab, but since this is Windows, it could have been moved anywhere :-( 2) Make sure all other apps are shut down, including any of those system tray ones that clutter up Windows. These all take memory, and when doing big compilations you'll want every byte you can find to keep it fast. 3) Upgrade to a real operating system ;-) I don't know how Win2K handles virtual memory, but it there is usually an option to specify the size of the page file manually, set it to be nice and large. You might also want to make sure that you have plenty of physical memory in your machine too... For complex C++ compilations I would suggest 256Mb or more (especially given the price of RAM these days). It will make things happen much more smoothly... HTH, John... Yasuyuki Kino wrote in message news:... > I'm using Tornado 2.0 on Windows 2000 > > When I try to compile c++ source code with "-O2" compiler option, > cc1plus.exe sends an error message "virtual memory exhousted". > > The memory is still left. > > Why is it? --------------------------- Newsgroups: comp.os.vxworks Subject: Re: using inflate in a downlable application Date: 21 Aug 2001 22:40:26 -0700 From: john_94501@yahoo.com (John) Organization: http://groups.google.com/ Message-ID: <488e459a.0108212140.36b359b8@posting.google.com> References: Hello, OK, three possible solutions. 1) Edit a file in your BSP, say sysLib.c - does not matter which file. Add the following line to the file, outside of any function: extern int inflate (char *, char *, int); If that alone does not work (it should for PPC), then add something like this too: FUNCPTR _func_inflate = (FUNCPTR) inflate; This should trick the build into believing that it needs the inflate() function, and it will extract it from the archive for you automatically. 2) Extract the inflateLib.o module from the archive (use "arppc x libPPC860gnuvx.a inflateLib.o" to do this). Copy the file to the BSP directory, and having done that, use MACH_EXTRA when building your BSP to add the file to the image manually: make MACH_EXTRA=inflateLib.o 3) Go to the web site given in the inflateLib manual, download the source for the compression and decompression routines and build them yourself. HTH, John... anthony.rouaux@st2e.com (Anthony Rouaux) wrote in message news:... > I all, > i am trying to use the inflate fonction in a downlable application. My > probleme is that this fonction is juste present in a file called > libPPC860gnuvx.a and not in my vxwork image running on the target. So when > I download the .out i have an error message that says me that it doesn t > know inflate. > How can i include this function from the .a file? > > Thanks a lot. > > Anthony --------------------------- Newsgroups: comp.os.vxworks Subject: Re: HELP - vx ... FTP-Server is innocent, it's a dosFS2 problem Date: Wed, 22 Aug 2001 07:56:44 +0200 From: Rainer Stelzer Organization: Stoye GmbH, =?iso-8859-1?Q?K=F6ln?= Message-ID: <3B83499C.BC841720@stoye.de> References: <3B77EB38.DBD20CFA@stoye.de> - --------------288142658BB9A4107FE702A3 Content-Type: text/plain; charset=iso-8859-1 Content-Transfer-Encoding: 8bit Hi vxworkers, the FTP Server (neither the delivered nor the unsupported one) wasn't the problem. The problem we encountered, is a known one: SPR.No. 31480. SPR# 31480 Submitted by: rodw Date: May 24 2000 Title: DosFS2.0 : Files being lost Problem_Description When a cluster chain for a directory contains a cluster where every directory entry has been deleted; if a new file is then created in that directory, the remaining entries in the cluster are changed from deleted entries to unused entries. This has the effect of preventing software from finding files beyond this point. If the check disk utility is then run the files are actually deleted and the lost fat chains are recovered. I think every one using T2 and dosFS2 should know that, cause this behavior results in complex, long term error situations that are hard to find. And that's the reason why in my opinion this kind of errors should be reported by wrs in an urgent e-mail, rather than polling windserve by the users. (Which is a problem, because of what to search for ?) We spent three day's to localize and reproduce an error that was found and corrected in 8/2000 ! rgds rainer vxworks, tornado ===================================================== Gebrüder Stoye GmbH department: traffic control systems name: Rainer Stelzer adr: Longericher Strasse 177 city: Cologne, Germany zip: 50739 mobil: 0174/8194860 phone: ++49/221/1707019 fax: ++49/221/1707010 mailto: r.stelzer@stoye.de www: http://www.stoye.de ===================================================== - --------------288142658BB9A4107FE702A3 Content-Type: text/html; charset=us-ascii Content-Transfer-Encoding: 7bit Hi vxworkers,

the FTP Server (neither the delivered nor the unsupported one) wasn't the problem.
The problem we encountered, is a known one: SPR.No. 31480.
 

SPR# 31480
Submitted by: rodw
Date: May 24 2000
Title: DosFS2.0 : Files being lost

                                          Problem_Description

 When a cluster chain for a directory contains a cluster where every
directory entry has been deleted; if a new file is then created in that
directory,
  the remaining entries in the cluster are changed from deleted entries to
unused entries.
  This has the effect of preventing software from finding files beyond this
point.
  If the check disk utility
  is then run the files are actually deleted and the lost fat chains are
recovered.
 

I think every one using T2 and dosFS2 should know that, cause this behavior
results in complex, long term error situations that are hard to find.

And that's the reason why in my opinion this kind of errors should be reported by wrs
in an urgent e-mail, rather than polling windserve by the users.
(Which is a problem, because of what to search for ?)
We spent three day's to localize and reproduce an error that was found and corrected
in 8/2000 !

rgds

rainer

vxworks, tornado
=====================================================
            Gebrüder Stoye GmbH
department: traffic control systems
      name: Rainer Stelzer
       adr: Longericher Strasse 177
      city: Cologne, Germany
       zip: 50739
     mobil: 0174/8194860
     phone: ++49/221/1707019
       fax: ++49/221/1707010
    mailto: r.stelzer@stoye.de
       www: http://www.stoye.de
=====================================================
  - --------------288142658BB9A4107FE702A3-- --------------------------- Newsgroups: comp.os.vxworks Subject: Re: MPC860 rev D4 & Fast Ethernet Date: Wed, 22 Aug 2001 07:42:44 GMT From: lourens@mecalc.co.za (Lourens Geldenhuys) Organization: Mecalc (Pty) Ltd Message-ID: <3b836242.754314@modem.mecalc> References: <3b820962.2131915@modem.mecalc> <3B82A66E.F69930AF@gv.net> Reply-To: lourens@mecalc.co.za On Tue, 21 Aug 2001 11:20:30 -0700, "Michael R. Kesti" wrote: > >I suspect that your BSP's sysFecEnetEnable() function in sysLib.c sets PDDIR >to 0x1c58. Change this value to 0x1fff and the FEC on your D4 boards will >likely come back to life. Indeed! Thank you! ================================================================ 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: Re: Tornado 2.0.1 dependency generation Date: 22 Aug 2001 07:37:21 GMT From: Johan Borkhuis Organization: Agere Systems Message-ID: References: <3B7D6A1E.3D57949F@improvsys.com> <3B7D92B6.FCD3A009@improvsys.com> <3B817AAF.916DBE5E@improvsys.com> Ben Kuhn wrote: > Thanks for all of the tips, folks. With some combination of all of the > stuff everyone haas mentioned, I've managed to get things up and > running. Should this be added / updated in the Tornado FAQ? Has been done. If there are more to add, please mention these in this thread. > set the VPATH macro to specify include directories. [I didn't actually > try this one] This one does not matter, it is only needed for building. The depoendency tool scan the CC-comandline and extracts the include paths from it. 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: indentify name in different node,what argument need? Date: Wed, 22 Aug 2001 16:36:46 +0800 From: "Kumo_Yah" Organization: SEEDNet News Service Message-ID: <9lvqut$4qr$1@news.seed.net.tw> Please help me! xxx.mib has some indentify name in different node. just like OID VS NAME 1.3.4.1.0 pp1 1.3.4.2.0 op1 1.3.5.1.0 pp1 1.3.5.2.0 op1 When I use mg-soft's mibcomp , although warning happen,but it can use also. And I want to use mibcomp in vxworks. It's broken,with the "duplicate defination" Should I change all the names? I need advice please.... ============================ Tornado 2.0 for PowerPC VxWorks 5.4 Host : Windows98 Target : PowerPC (MPC8240) ============================ - -- Kumo,Yeh. email:kumo@ms42.url.com.tw - -- Kumo,Yeh. email:kumo@ms42.url.com.tw --------------------------- Newsgroups: comp.os.vxworks Subject: Memory allocation Date: 22 Aug 2001 01:42:43 -0700 From: ike@1freeemail.com (Ike Robinson) Organization: http://groups.google.com/ Message-ID: <1a86250f.0108220042.345c2668@posting.google.com> Hello, I use VxWorks (for Motorola MCP750 - MPC 750) version 5.4. Kernel: WIND version 2.5. Here's the small example/test: void test() { FILE *f=stdout; } When I run memShow, execute test and run memShow again, then I see that I have 80 bytes less free. Does anybody know why these 80 bytes are allocated when I use stdout (even in declaration)? Also when I use 'sp test' (spawn task) memShow shows the same values before and after 'sp test'. Does anybody know why in such case there is no free memory decrease? Maybe there are some docs/faqs which I can read to find an explanation? Thanks in advance, Ike --------------------------- Newsgroups: comp.os.vxworks Subject: Newby Info Date: 22 Aug 2001 02:09:00 -0700 From: r@gors.be (WomanLover) Organization: http://groups.google.com/ Message-ID: hey, I'am an newby on the programming side and i was wondering if anybody can provide me with usefull links, tutorials or code samples. Thanks on advance Roderick --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Anyone get VxWorks to boot quickly Date: Wed, 22 Aug 2001 10:45:44 +0100 From: "timallen" Organization: ntlworld News Service Message-ID: References: Reply-To: "timallen" Brad Crabtree wrote in message news:hsbg7.10$DK5.172@dfw-service2.ext.raytheon.com... > I have a need to get a Tornado 2.0 VxWorks ROM image (about 4MB in size) > loaded and booted on a Power PC VME board in 3-4 seconds. It is currently > taking about 6. Does anyone have any lessons learned on getting a VxWorks > image to boot quickly? Other than the instruction cache enabling, as previously mentioned, if you're also initialising a network driver which auto negotiates this can be quite time consuming and can usually be fixed by fixing the network type at compile time if you know what it is. Tim. --------------------------- End of New-News digest ********************** From vxwexplo-errs@csg.lbl.gov Wed Aug 22 07:12:19 2001 From: "Mike Anderson" Date: Wed Aug 22 07:12:21 PDT 2001 Subject: MOSPF or PM-Dense for VxWorks? VxWorks Greetings! Anyone out there in netland know of implementations of MOSPF or PM-Dense multicast routing protocols for VxWorks? TIA, Mike -- ======================================================= T h e P T R G r o u p, I n c. ======================================================= Embedded, Real-Time Solutions and Training Mike Anderson mailto:mike@ThePTRGroup.com Chief Scientist http://www.ThePTRGroup.com V: (703) 585-9384 F: (703) 430-3748 ======================================================= "Software development is like making a baby... You can't make a baby in one month by impregnating nine women. Some things just take time." From vxwexplo-errs@csg.lbl.gov Wed Aug 22 07:12:24 2001 From: "Mike Anderson" Date: Wed Aug 22 07:12:27 PDT 2001 Subject: Unnumbered interfaces for VxWorks? VxWorks Greetings! Does anyone in netland know if VxWorks supports unnumbered interfaces via PPP? From my research, I'd say no. But, if someone knows differently, I'd appreciate hearing from you. TIA, Mike -- ======================================================= T h e P T R G r o u p, I n c. ======================================================= Embedded, Real-Time Solutions and Training Mike Anderson mailto:mike@ThePTRGroup.com Chief Scientist http://www.ThePTRGroup.com V: (703) 585-9384 F: (703) 430-3748 ======================================================= "Software development is like making a baby... You can't make a baby in one month by impregnating nine women. Some things just take time." From vxwexplo-errs@csg.lbl.gov Wed Aug 22 09:04:44 2001 From: Anthony Giovino Date: Wed Aug 22 09:04:47 PDT 2001 Subject: Thales / Cetia VMPC6a board usage This is a multi-part message in MIME format. --------------F2D7BC554C40B8AFAA4A5B18 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit We're wondering if anyone is using a Thales (formerly Cetia) VMPC6a (or similar) board? We want to use system mode debugging. The vendor has stated the BSP has bugs in this regard and doesn't know when (if) they will fix them. Support at WRS has stated that BSPs that don't support system mode debugging via the END connection should support it through the "WDB_SERIAL" connection. The vendor has stated system mode debugging won't work through serial either (support has been less than stellar). By the way, we're using Tornado 2.0 on Widows 2000 and VxWorks 5.4. We've put some of the 2000 patches in for Tornado, but we're waiting for the BSP update so we can run Tornado 2.0.2. The VMPC6a is a PowerPC 750 based VME CCA. Does anyone have any experience with this vendor or board family? If so, any words of wisdom with regard to system mode debugging, Tornado updates, BSP updates, etc. TIA, Tony G --------------F2D7BC554C40B8AFAA4A5B18 Content-Type: text/x-vcard; charset=us-ascii; name="agiovino.vcf" Content-Transfer-Encoding: 7bit Content-Description: Card for Anthony Giovino Content-Disposition: attachment; filename="agiovino.vcf" begin:vcard n:Giovino;Anthony tel;fax:(716) 631-7849 tel;work:(716) 631-7389 x-mozilla-html:TRUE url:www.sierra-idt.com org:Sierra Research, an Integrated Defense Technologies company version:2.1 email;internet:agiovino@sierra-idt.com title:Senior Software Engineer adr;quoted-printable:;;485 Cayuga Road=0D=0AP. O. Box 222 MS 312;Buffalo;New York;14225-0222;USA x-mozilla-cpt:;18992 fn:Anthony Giovino end:vcard --------------F2D7BC554C40B8AFAA4A5B18-- From vxwexplo-errs@csg.lbl.gov Thu Aug 23 02:46:24 2001 From: fscheurer@manz-acs.de Date: Thu Aug 23 02:46:26 PDT 2001 Subject: GCC 2.95.4 for i386-wrs-vxworks Hi Dave, (hi GCC-VxWorks gurus,) thanks a lot for your related Home-Page http://www.lysergic.u-net.com, Dave! With the comments related to the cygwin build process I succeeded to build the compiler. Here are my comments and remaining questions: 1) vxworks and target related patches ------------------------------------- Although I used "vx-ppc-gcc.diff" and "vx-ppc-binutils.diff" I build for the "i386-wrs-vxworks" target. Due to the amount of HW-related patches I assume that a special patch for "my" target x86 is necessary, isn't it? 2) gcc.diff cannot be "made" ---------------------------- Due to a not recoverable error gcc.info could not be build. .. /cygdrive/c/GNU_CrossGCC_Cygwin/DaveK_JimPisano/build/build-gcc/texinfo/makeinfo /makeinfo -I/gcc-src/gcc -o cpp.info /gcc-src/gcc/cpp.texi Making info file `cpp.info' from `/gcc-src/gcc/cpp.texi'. /cygdrive/c/GNU_CrossGCC_Cygwin/DaveK_JimPisano/build/build-gcc/texinfo/makeinfo /makeinfo -I/gcc-src/gcc -o gcc.info /gcc-src/gcc/gcc.texi /gcc-src/gcc/gcc.texi: No such file or directory make[1]: *** [gcc.info] Error 2 make[1]: Leaving directory `/cygdrive/c/GNU_CrossGCC_Cygwin/DaveK_JimPisano/buil d/build-gcc/gcc' make: *** [all-gcc] Error 2 .. I couldn't find a reason why gcc.texi shouldn't be available. All files included are there including the same attribute in that "$(srcdir)"-directory. See original build-gcc/gcc/Makefile: .. # Remake the info files. doc: info info: cpp.info gcc.info lang.info cpp.info: $(srcdir)/cpp.texi $(MAKEINFO) $(MAKEINFOFLAGS) -I$(srcdir) -o cpp.info $(srcdir)/cpp.texi gcc.info: $(srcdir)/gcc.texi $(srcdir)/extend.texi $(srcdir)/install.texi \ $(srcdir)/invoke.texi $(srcdir)/md.texi $(srcdir)/rtl.texi \ $(srcdir)/tm.texi $(srcdir)/gcov.texi $(MAKEINFO) $(MAKEINFOFLAGS) -I$(srcdir) -o gcc.info $(srcdir)/gcc.texi dvi: gcc.dvi cpp.dvi lang.dvi .. I tried to change the Makefile and tried to rename or copy the gcc.texi, with no success. So as "ultima ratio" I changed build-gcc/gcc/Makefile to prevent build and install of gcc.info: .. # Remake the info files. doc: info #info: cpp.info gcc.info lang.info info: cpp.info lang.info cpp.info: $(srcdir)/cpp.texi $(MAKEINFO) $(MAKEINFOFLAGS) -I$(srcdir) -o cpp.info $(srcdir)/cpp.texi gcc.info: $(srcdir)/gcc.texi $(srcdir)/extend.texi $(srcdir)/install.texi \ $(srcdir)/invoke.texi $(srcdir)/md.texi $(srcdir)/rtl.texi \ $(srcdir)/tm.texi $(srcdir)/gcov.texi $(MAKEINFO) $(MAKEINFOFLAGS) -I$(srcdir) -o gcc.info $(srcdir)/gcc.texi dvi: gcc.dvi cpp.dvi lang.dvi .. install-info: doc installdirs lang.install-info # -rm -f $(infodir)/cpp.info* $(infodir)/gcc.info* -rm -f $(infodir)/cpp.info* # for f in cpp.info* gcc.info*; do \ for f in cpp.info*; do \ $(INSTALL_DATA) $$f $(infodir)/$$f; \ done -if $(SHELL) -c 'install-info --version' >/dev/null 2>&1; then \ if [ -f $(infodir)/dir ] ; then \ # for f in cpp.info gcc.info; do \ for f in cpp.info; do \ install-info --dir-file=$(infodir)/dir $(infodir)/$$f; \ done; \ else true; fi; \ else true; fi; # -chmod a-x $(infodir)/cpp.info* $(infodir)/gcc.info* -chmod a-x $(infodir)/cpp.info* .. OK, it's only a doc file and not necessary for the functionality. But I cannot understand the behaviour. Is there a fix for that problem? 3) fixinc problem ------------------ With Jim Pisano's script I used a command sequence .. # first, fix includes with patches cd gcc/fixinc ./genfixes .. There I saw a problem with the fixincl.x build due to a hyphen sign `-' in the fix's name: .. autogen inclhack.def gcc -c -g -I. -I.. -I../../../gcc/fixinc -I../../../gcc/fixinc/.. -I../../../gcc/fixinc/../config -I../../../gcc/fixinc/../../incl ude ../../../gcc/fixinc/fixincl.c In file included from ../../../gcc/fixinc/fixincl.c:140: ./../../gcc/fixinc/fixincl.x:3344: warning: missing white space after `#define apzVxworks_Stdio' ./../../gcc/fixinc/fixincl.x:3352: warning: missing white space after `#define VXWORKS_STDIO' ./../../gcc/fixinc/fixincl.x:3353: warning: missing white space after `#define VXWORKS_STDIO' ./../../gcc/fixinc/fixincl.x:3353: warning: `VXWORKS_STDIO' redefined ./../../gcc/fixinc/fixincl.x:3352: warning: this is the location of the previous definition In file included from ../../../gcc/fixinc/fixincl.c:140: ./../../gcc/fixinc/fixincl.x:3334: parse error before `-' ./../../gcc/fixinc/fixincl.x:3339: parse error before `-' ./../../gcc/fixinc/fixincl.x:3349: parse error before `-' ./../../gcc/fixinc/fixincl.x:3354: parse error before `-' ./../../gcc/fixinc/fixincl.x:3360: parse error before `-' ./../../gcc/fixinc/fixincl.x:4333: `zVxworks_Stdio' undeclared here (not in a function) ./../../gcc/fixinc/fixincl.x:4333: `Lock_Cxx_CompatName' undeclared here (not in a function) ./../../gcc/fixinc/fixincl.x:4333: initializer element is not constant ./../../gcc/fixinc/fixincl.x:4333: (near initialization for `fixDescList[91].fix_name') ./../../gcc/fixinc/fixincl.x:4333: `zVxworks_Stdio' undeclared here (not in a function) ./../../gcc/fixinc/fixincl.x:4333: `Lock_Cxx_CompatList' undeclared here (not in a function) ./../../gcc/fixinc/fixincl.x:4333: initializer element is not constant ./../../gcc/fixinc/fixincl.x:4333: (near initialization for `fixDescList[91].file_list') ./../../gcc/fixinc/fixincl.x:4334: parse error before `const' ./../../gcc/fixinc/fixincl.x:4334: warning: initialization from incompatible pointer type ./../../gcc/fixinc/fixincl.x:4335: `LOCK_CXX_COMPAT_RE_CT' undeclared here (not in a function) ./../../gcc/fixinc/fixincl.x:4335: initializer element is not constant ./../../gcc/fixinc/fixincl.x:4335: (near initialization for `fixDescList[91].unused') ./../../gcc/fixinc/fixincl.x:4335: parse error before `1' ./../../gcc/fixinc/fixincl.x:4336: `aVxworks_Stdio' undeclared here (not in a function) ./../../gcc/fixinc/fixincl.x:4336: `Lock_Cxx_CompatTests' undeclared here (not in a function) ./../../gcc/fixinc/fixincl.x:4336: initializer element is not constant ./../../gcc/fixinc/fixincl.x:4336: (near initialization for `fixDescList[91].fd_flags') ./../../gcc/fixinc/fixincl.x:4336: parse error before `const' ./../../gcc/fixinc/fixincl.c: In function `initialize': ./../../gcc/fixinc/fixincl.c:349: `SIGIOT' undeclared (first use in this function) ./../../gcc/fixinc/fixincl.c:349: (Each undeclared identifier is reported only once ./../../gcc/fixinc/fixincl.c:349: for each function it appears in.) make[2]: *** [fixincl.o] Error 1 make[2]: Leaving directory `/GNU_CrossGCC_Cygwin/DaveK_JimPisano/build/gcc-2.95.3/build/gcc/fixinc' Could not install binary fixincludes. Installing shell script instead. .. In your proposal the fixinc is only done the regular way. Although the binary fixinc doesn't work: .. cd ./fixinc; /bin/sh ${srcdir}/mkfixinc.sh i386-wrs-vxworks constructing ../fixinc.sh for i386-wrs-vxworks make SHELL="/bin/sh" install make[2]: Entering directory `/cygdrive/c/GNU_CrossGCC_Cygwin/DaveK_JimPisano/bui ld/build-gcc/gcc/fixinc' autogen inclhack.def autogen -DPROGRAM=1 -b fixincl inclhack.def touch: creating `fixincl.sh': Permission denied make[2]: *** [fixincl.sh] Error 1 make[2]: Leaving directory `/cygdrive/c/GNU_CrossGCC_Cygwin/DaveK_JimPisano/buil d/build-gcc/gcc/fixinc' Could not install binary fixincludes. Installing shell script instead. rm -rf include; mkdir include TARGET_MACHINE=i386-wrs-vxworks; srcdir=`cd /gcc-src/gcc; pwd`; \ INSTALL_ASSERT_H=install-assert-h; SHELL=/bin/sh ;\ export TARGET_MACHINE srcdir INSTALL_ASSERT_H SHELL ; \ /bin/sh ./fixinc.sh `pwd`/include /usr/local/i386-wrs-vxworks/sys-include Fixing headers into /cygdrive/c/GNU_CrossGCC_Cygwin/DaveK_JimPisano/build/build- gcc/gcc/include for i386-wrs-vxworks target Finding directories and links to directories Searching /usr/local/i386-wrs-vxworks/sys-include/. All directories (including links to directories): ./arch ./arch/i86 ./arch/i86.sim ./arch/simnt ./arpa ./dhcp ./drv ./drv/dma ./dr v/end ./drv/fdisk ./drv/hdisk ./drv/intrCtl ./drv/mem ./drv/multi ./drv/netif ./ drv/parallel ./drv/pci ./drv/pcmcia ./drv/rtc ./drv/scsi ./drv/serial ./drv/sio ./drv/timer ./drv/vme ./drv/wdb ./make ./net ./netinet ./netinet/ppp ./ospf ./pr ivate ./resolv ./rip ./rpc ./std ./streams ./sys ./tffs ./types ./wdb Making symbolic directory links Fixing directory /usr/local/i386-wrs-vxworks/sys-include into /cygdrive/c/GNU_Cr ossGCC_Cygwin/DaveK_JimPisano/build/build-gcc/gcc/include Checking header files Fixed ./aio.h: machine_name .. Could this cause any further problems? 4)Policy for headers --------------------- What is WRS policy for using their headers without using their libraries? Do I need a license to call the vxWorks-Routines from the shell (which means to know the prototypes of each kernel routine), when my vxWorks system is delivered from my basic system provider? TIA MfG / Kind Regards i.A. F. Scheurer P.S.: Due to a mistake of mine I cannot post anymore to the VxW-Exploder. Unfortunately I switched on an auto-reply without filtering the vxw-explo and terrorised the whole (vxworks)-world! So probably I can't post it. I hope Alan Biocca accepts my apologies. ---------------------------------------------------------------- Friedrich Scheurer, Dipl.-Ing. phone +49 7121 9019-23(0) ACS GmbH - R&D fax +49 7121 9019-98 Steigaeckerstrasse 13 web www.manz-acs.de D-72768 Reutlingen (Germany) e-mail: fscheurer@manz-acs.de ---------------------------------------------------------------- From vxwexplo-errs@csg.lbl.gov Thu Aug 23 04:05:32 2001 From: Vxworks Exploder Date: Thu Aug 23 04:05:35 PDT 2001 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Thu Aug 23 04:04:37 PDT 2001 Subject: Re: Debugging when the debugger crashes Subject: memory allocation problems Subject: Re: MPC860 rev D4 & Fast Ethernet Subject: Test Subject: Re: using inflate in a downlable application Subject: Re: about virtual memory exhausted error Subject: Re: Memory allocation Subject: Re: memory allocation problems Subject: Newbie: inhibit / trap VME bus error? Subject: Re: Exception handling in VxWorks Subject: Re: Problem in scaning user input ! Subject: Re: Debugging when the debugger crashes Subject: Re: Newbie: inhibit / trap VME bus error? Subject: Re: Debugging when the debugger crashes Subject: Re: memory allocation problems Subject: TFFS in Tornado 2 for PPC Subject: SV: TFFS in Tornado 2 for PPC Subject: Re: Anyone get VxWorks to boot quickly Subject: TFFS in Tornado 2 for PPC Subject: porting linux kernel code to vxworks? Subject: Re: Newby Info Subject: DHCP server on multi i/faces Subject: Telnet sessions Subject: Re: about virtual memory exhausted error Subject: MIPS, breakpoints, watchpoints problem Subject: MIPS, breakpoints, watchpoints problem Subject: Re: about virtual memory exhausted error Subject: Re: Telnet sessions Subject: Re: Exception handling in VxWorks Subject: SV: TFFS in Tornado 2 for PPC Subject: Re: about virtual memory exhausted error Subject: Several Questions... Subject: Strict timing Subject: SLIP protocol under Windows 2000/NT Subject: Re: Debugging when the debugger crashes Subject: Re: Several Questions... Subject: ioctl in END Driver. Subject: any one would help me to solve the PNP problem? ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: Debugging when the debugger crashes Date: Wed, 22 Aug 2001 12:04:14 +0100 From: vxworksintent@tao-group.com (David Given) Organization: I do not speak for anyone but myself, and barely that. Message-ID: References: <2psil9.q9e.ln@127.0.0.1> <3B7E0220.9ACC02DE@covad.net> <3B822F37.6B80B863@agilent.com> <7cdtl9.lr4.ln@127.0.0.1> <488e459a.0108212043.29e13e3b@posting.google.com> Sender: dg@pearl.tao.co.uk In article <488e459a.0108212043.29e13e3b@posting.google.com>, john_94501@yahoo.com (John) writes: [...] >> And what's system mode/task mode? I can't find any references in the >> documentation. [...] > For info, check out section 7.5 of the Tornado II User's Guide. Actually, 8.6 in mine --- but no matter, the BSP seems not to support system mode (at least of Ethernet). And unfortunately I can't reconfigure it to use the serial port. [...] > 1) You can use a hook (see excHookAdd() in the VxWorks reference > manual). This will allow to to do some processing when a task takes an > exception. If you're on x86 platforms I seem to remember that a double > fault causes the CPU to halt - this might be the cause of your > problem. We're currently using excHookAdd(). Our strategy, which was developed by trial and error on another MIPS platform, is currently pretty horrible and uses a helper task; this helper task, in one place, waits for the main task to be suspended, resumes it, waits for it to be suspended *again*, and resumes it again. This did work, but only just. Our requirements are that when an exception arrives, we have to call an exception handler in that task. The exception handler needs to be passed the CPU state of the task when the exception occurred, and has to be able to either longjmp out or change the CPU state and continue executing. Ideally the solution should be portable across multiple VxWorks platforms. In another post you mention _func_excBaseHook. It looks like just what I want --- assuming it works, of course. When I call it, do I need to chain on to the old routine? - -- David Given Tao Group Ltd. --------------------------- Newsgroups: comp.os.vxworks Subject: memory allocation problems Date: Wed, 22 Aug 2001 15:16:17 +0200 From: Aleks Organization: . Message-ID: <3B83B0A1.891C8E46@yahoo.com> Reply-To: . Hi all, I use PPC603 and in the beginning some partition must be created (C code). memPartCreate() returns NULL... What additional diagnostic could I get in order to find why? Is anything in addition to printErrno()? In manual for printErrno() is said, that error-status symbol table must be included in the system. How can I check it? Thanks in advance, Aleks --------------------------- Newsgroups: comp.os.vxworks Subject: Re: MPC860 rev D4 & Fast Ethernet Date: 22 Aug 2001 13:27:17 GMT From: "Michael Gilbert" Organization: Concentric Internet Services Message-ID: <9m0bvl$nq2@dispatch.concentric.net> References: <3b820962.2131915@modem.mecalc> Speaking only for myself. Use this code at your own risk. I've used the WRS fix for the 860 D4 before. It uses a compiler switch to control the D4 changes. I made it a run time check so that we didn't have to maintain two seperate images. I added the following to sysAlib.s: /*************************************************************************** **** * * hwImmrGet - returns the PARTNUM and MASKNUM portion of the IMMR register * * The VxWorks function vxImmrGet returns the ISB potion of the IMMR register. * This function returns the portion of IMMR masked out by vxImmrGet. */ hwImmrGet: mfspr r3, IMMR andi. r3, r3, 0xffff blr /* hwImmrGet */ And added this to sysLib.c: in sysFecEnetEnable: /* rev D parts are different */ if (hwImmrGet() > 0x0500) *PDDIR (motCpmAddr) = 0x1fff; else *PDDIR (motCpmAddr) = 0x1c58; in sysHwInit: /* * If we are running on a rev D (or greater) processor set the IsRevD * flag in the load string for the FEC device in the END device table. * * This is done here so that it will be common to RAM and ROM code. */ if (hwImmrGet() > 0x0500) { ptr = endDevTbl[0].endLoadString; /* FEC is device 0 */ for (i = 0; i < 12; i++) /* "IsRevD" flag is 12th item in list */ { ptr = strchr(ptr, ':'); /* find the next colon */ ptr++; } *ptr = '1'; /* set the "IsRevD" flag */ Hope this helps. Mike "Chris Moore" wrote in message news:e229ec7e.0108210848.2108f416@posting.google.com... > Hi, > > I've written the FEC driver for the PPC860 for pSOS that we use on our > old system. There are quite a few little changes for the REVD4. > Basically the pin muxing on the MII port has changed, initially it was > written wrong in the documents from Mot and all sorts of fun stuff. > Easiest is to contact WRS as it seems there is a driver. If you want > to fix it your self you can look on the mot site for erratas for the > FEC. > > If you want your driver to work across old and new revisions you will > need to check the chip revision (part of the IMMR register (high bits > I think from mem) and init the pins accordingly. > > If you want to do it your self and can't find the info, zap me an > email and I'll track down my code changes for you. > > Rgds, > > Chris Moore > > Ericsson > > > > lourens@mecalc.co.za (Lourens Geldenhuys) wrote in message news:<3b820962.2131915@modem.mecalc>... > > Is anybody using an MPC860D4 processor with WindRiver's ads860 BSP? We > > replaced the MPC860B3 silicon on our (working) design with rev D4 > > silicon, and now our Ethernet does not work. We are using the FEC and > > RELEASE 1.2/1, Jul 2000 of the ads860 BSP, which does support rev D4 > > silicon. Also, I have modified the XFC capacitors to match the rev D4 > > spec, still no Ethernet. > > > > Any advice? > > > > Thank you, > > 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: Test Date: Wed, 22 Aug 2001 08:47:46 -0500 From: PC User Organization: ExecPC Internet - Milwaukee, WI Message-ID: <3b83b982$0$42879$272ea4a1@news.execpc.com> --------------------------- Newsgroups: comp.os.vxworks Subject: Re: using inflate in a downlable application Date: 22 Aug 2001 10:33:13 -0400 From: Bill Pringlemeir Organization: Bell Sympatico Message-ID: References: Sender: bpringlemeir@DeadDuck >>>>> Anthony Rouaux wrote: >> I all, i am trying to use the inflate fonction in a downlable >> application. My probleme is that this fonction is juste present in >> a file called libPPC860gnuvx.a and not in my vxwork image running >> on the target. So when I download the .out i have an error message >> that says me that it doesn t know inflate. How can i include this >> function from the .a file? >>>>> "Pete" == Pete Kockritz writes: [snip] Pete> So you have two options: Pete> 1) Add a reference to inflate() in the kernel, so that the Pete> 2) Extract the inflateLib.o from the .a file. You can then link 3) Get the source from "http://www.gzip.org/zlib/" or "http://www.info-zip.org/pub/infozip/zlib/". It should compile easily for vxWorks. This package is more flexible as you can handle gzip files. # makefile. CPU = ?? Your CPU ?? TOOL = gnu EXTRA_DEFINE = -Wcomment MKDIR = mkdir TGT_DIR = $(WIND_BASE)/target OBJS= adler32.o compress.o crc32.o deflate.o gzio.o infblock.o infcodes.o \ inffast.o inflate.o inftrees.o infutil.o trees.o uncompr.o zutil.o all : ($OBJS) include $(TGT_DIR)/h/make/defs.bsp include $(TGT_DIR)/h/make/make.$(CPU)$(TOOL) include $(TGT_DIR)/h/make/defs.$(WIND_HOST_TYPE) ## Allow C++ sytle comments (tell Pre-processor). ADDED_CFLAGS += -Wp,-lang-c ADDED_CFLAGS += -DVXWORKS=1 ADDED_CFLAGS += -Wall -Winline -O2 -fomit-frame-pointer -fno-strength-reduce ADDED_CFLAGS += -DNO_DEFLATE=1 The functions in gzio.o are nice. I don't think you have to modify any of the source. fwiw, Bill Pringlemeir - -- ..Are we having FUN yet...? vxWorks FAQ, "http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html" --------------------------- Newsgroups: comp.os.vxworks Subject: Re: about virtual memory exhausted error Date: Wed, 22 Aug 2001 10:57:31 -0400 From: "Kevin Livingston" Message-ID: References: <488e459a.0108212131.7bfebe80@posting.google.com> Sometimes windows gets flakey after running for a while, and "a while" depends on your system it could be days, or just since lunch. And rebooting will usually allow everything to straighten out. I would just see if rebooting would fix your problem before changing settings or buying memory. good luck. Kevin "John" wrote in message news:488e459a.0108212131.7bfebe80@posting.google.com... > Hello, > > Welcome to the world of Windoze... Basically, your swap space is being > exhausted by the compiler (compiling complex code can do this, > especially it seems C++ code). To solve the problem you have a choice: > > 1) Increase the size of the swap space (paging file) using the Windows > control panel. It is normally on the System tool's performance tab, > but since this is Windows, it could have been moved anywhere :-( > > 2) Make sure all other apps are shut down, including any of those > system tray ones that clutter up Windows. These all take memory, and > when doing big compilations you'll want every byte you can find to > keep it fast. > > 3) Upgrade to a real operating system ;-) > > I don't know how Win2K handles virtual memory, but it there is usually > an option to specify the size of the page file manually, set it to be > nice and large. > > You might also want to make sure that you have plenty of physical > memory in your machine too... For complex C++ compilations I would > suggest 256Mb or more (especially given the price of RAM these days). > It will make things happen much more smoothly... > > HTH, > > John... > > Yasuyuki Kino wrote in message news:... > > I'm using Tornado 2.0 on Windows 2000 > > > > When I try to compile c++ source code with "-O2" compiler option, > > cc1plus.exe sends an error message "virtual memory exhousted". > > > > The memory is still left. > > > > Why is it? --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Memory allocation Date: 22 Aug 2001 09:27:34 -0700 From: john_94501@yahoo.com (John) Organization: http://groups.google.com/ Message-ID: <488e459a.0108220827.5d3f1f6@posting.google.com> References: <1a86250f.0108220042.345c2668@posting.google.com> Hello, This one is simple... the first time you use stdout (or stdin, or stderr), the standard I/O system allocates the FILE structures for the three standard I/O channels for calling task. These structures are attached to the task's TCB. So, when you call it from the shell, it is the shell task (or the WDB agent if you are using the host shell) that gets these new FILE structures. When your test finishes the FILE structures still exist, so you see the memory leakage. Now, if you spawn the test, when it is finished the task exits, and because the FILE structures were associated with the TCB, they are freed. Hence no apparent memory leak. HTH, John... ike@1freeemail.com (Ike Robinson) wrote in message news:<1a86250f.0108220042.345c2668@posting.google.com>... > Hello, > > I use > VxWorks (for Motorola MCP750 - MPC 750) version 5.4. > Kernel: WIND version 2.5. > > Here's the small example/test: > void test() { > FILE *f=stdout; > } > > When I run memShow, execute test and run memShow again, then I see > that I have 80 bytes less free. Does anybody know why these 80 bytes > are allocated when I use stdout (even in declaration)? > Also when I use 'sp test' (spawn task) memShow shows the same values > before and after 'sp test'. Does anybody know why in such case there > is no free memory decrease? > > Maybe there are some docs/faqs which I can read to find an > explanation? > > > Thanks in advance, > Ike --------------------------- Newsgroups: comp.os.vxworks Subject: Re: memory allocation problems Date: Wed, 22 Aug 2001 12:39:14 -0700 From: Joe Durusau Organization: Lockheed Martin Corporation Message-ID: <3B840A62.B0F5F126@lmco.com> References: <3B83B0A1.891C8E46@yahoo.com> Well, according to WRS documentation, the only possible reason is that not enough memory is available. You might check the global variable errno and see what it contains. Usually, the error table is automatically included, but I don't know about T II. Speaking only for myself, Joe Durusau Aleks wrote: > > Hi all, > > I use PPC603 and in the beginning some partition must be created (C > code). > memPartCreate() returns NULL... What additional diagnostic could I get > in order to find why? > Is anything in addition to printErrno()? In manual for printErrno() is > said, that > error-status symbol table must be included in the system. How can I > check it? > > Thanks in advance, > Aleks --------------------------- Newsgroups: comp.os.vxworks Subject: Newbie: inhibit / trap VME bus error? Date: 22 Aug 2001 10:23:14 -0700 From: david.frye@powerconv.alstom.com (Dave) Organization: http://groups.google.com/ Message-ID: <48743300.0108220923.1f16c6b0@posting.google.com> Hello, I have a system where another card may cause random VME bus errors when my card (running VxWorks) attempts to read its shared memory. This produces a bus error on my card and the tasks associated with this bus error are suspended. Is it possible to have a bus error without these tasks being suspended but still be warned of this error? If so, how? Thanks in advance! Dave --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Exception handling in VxWorks Date: 22 Aug 2001 10:38:10 -0700 From: vivekp@hotmail.com (pvpartha1) Organization: http://groups.google.com/ Message-ID: <623ba2de.0108220938.75fe0a79@posting.google.com> References: <623ba2de.0108210727.75b5534f@posting.google.com> Lee DeRaud wrote in message news:... > On 21 Aug 2001 08:27:07 -0700, vivekp@hotmail.com (pvpartha1) wrote: > > >Is it possible to trap data exception errors or bus errors and suspend > >the thread that causes that to happen? How is this done? > >any help would be appreciated > > In what architecture? Most of them already have bus error and/or data > access exceptions (e.g. misaligned accesses) enabled by default, and > suspension of the offending task is (part of) the default handling. > (Note that, much to the surprise of everyone here, FPU exceptions > (e.g. divide-by-zero) are *not* enabled in M68K.) > > Lee The architecture iam running on is PPC604. What iam trying to do is the following Whenever a task exception occurs suspend the task and do a stack trace and task trace . Write contnts to a file and move on. Is it possible to overload the default exception handler? How is it done ? --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Problem in scaning user input ! Date: Wed, 22 Aug 2001 18:44:48 GMT From: mark_santelli@srtelecom.com (Mark) Organization: Bell Sympatico Message-ID: <3b83fd6e.189763295@news.bellglobal.com> References: How are you running your program? Are you spawning it? What do you do exactly?? On 31 Jul 2001 00:33:06 -0700, mridulgupta1@yahoo.com (Mridul Gupta) wrote: >hi all, > i am a new user to VxWorks and i wrote a interactive program >which takes inputs from user. the problem is that the scanf doesnt >works ... it seems like the shell is eating the input in between. I >tried suspending the shell task but it didnt work. i also tried >changing priorities but nothing worked ... please tell me how do i >take user input ?? > >Thanx > >Mridul --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Debugging when the debugger crashes Date: 22 Aug 2001 11:57:54 -0700 From: john_94501@yahoo.com (John) Organization: http://groups.google.com/ Message-ID: <488e459a.0108221057.525dcd88@posting.google.com> References: <2psil9.q9e.ln@127.0.0.1> <3B7E0220.9ACC02DE@covad.net> <3B822F37.6B80B863@agilent.com> <7cdtl9.lr4.ln@127.0.0.1> <488e459a.0108212043.29e13e3b@posting.google.com> > We're currently using excHookAdd(). Our strategy, which was developed by > trial and error on another MIPS platform, is currently pretty horrible and > uses a helper task; this helper task, in one place, waits for the main > task to be suspended, resumes it, waits for it to be suspended *again*, > and resumes it again. This did work, but only just. > > Our requirements are that when an exception arrives, we have to call an > exception handler in that task. The exception handler needs to be passed > the CPU state of the task when the exception occurred, and has to be able > to either longjmp out or change the CPU state and continue executing. > Ideally the solution should be portable across multiple VxWorks platforms. > > In another post you mention _func_excBaseHook. It looks like just what I > want --- assuming it works, of course. When I call it, do I need to chain > on to the old routine? No. If your handler decides that it is not an exception it can handle, all it needs to do is return FALSE and the standard handler will continue. If your handler does want to return after handling the exception (as opposed to calling longjmp() for example), then it simply returns TRUE and the primary exception handler will exit immediately, without suspending the task or even printing a message about the exception. HTH, John... --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Newbie: inhibit / trap VME bus error? Date: Wed, 22 Aug 2001 13:52:59 -0500 From: Andrew Johnson Organization: APS, Argonne National Laboratory Message-ID: <3B83FF8B.EE700172@aps.anl.gov> References: <48743300.0108220923.1f16c6b0@posting.google.com> Dave wrote: > > I have a system where another card may cause random VME bus errors > when my card (running VxWorks) attempts to read its shared memory. > This produces a bus error on my card and the tasks associated with > this bus error are suspended. Ouch - get that card fixed, it'll be much easier in the long run. > Is it possible to have a bus error without these tasks being suspended > but still be warned of this error? > If so, how? You don't want to go there - the Bus Error means that a read or write cycle couldn't be completed, and the data read or written has been lost. How can the CPU continue if it was supposed to read a value from some location, because it needs the value when executing some following instruction. What you might be meaning is to re-try the cycle that caused the bus error. You might be able to do that on some CPU families (68K probably, PPC no, others unknown) - you'll have to replace the vxWorks Bus Error exception handler with one of your own, and read the CPU's User Manual to find out how to go about rerunning the cycle that caused the exception. I strongly suggest that you keep track of what you're doing, because you need to fall through to the real Bus Error handler if there's really nothing at that address (ie if the retry also gets a bus error). Oh, and beware any write posting that your VME interface chip may do - you'll need to disable this to get it to work. As you can probably see, this won't be easy - better to get the hardware fixed if at all possible. - - Andrew - -- The world is such a cheerful place when viewed from upside-down It makes a rise of every fall, a smile of every frown --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Debugging when the debugger crashes Date: 22 Aug 2001 12:30:00 -0700 From: vi_jay2k@yahoo.com (Jay) Organization: http://groups.google.com/ Message-ID: <6c517362.0108221129.58e6015f@posting.google.com> References: <2psil9.q9e.ln@127.0.0.1> vxworksintent@tao-group.com (David Given) wrote in message news:<2psil9.q9e.ln@127.0.0.1>... > I've got a particularly annoying problem on my MIPS-based device. I have > Tornado attached to it via ethernet; unfortunately, my application crashes > so hard that it takes the VxWorks ethernet driver down with it. (I think > it's locking up with interrupts locked.) This means, of course, that the > debugger doesn't work any more. > > I suspect it's jumping into the wild blue yonder and stomping all over > memory. I can tell this because I see gibberish appear on the screen once > it's locked. But I can't tell any more about it, because I can't debug it. > > What I'd really like is ICE or a JTAG debugger, but unfortunately my > hardware doesn't support it; all I can do is use the ethernet debugger. > Does anyone have any suggestions as to how I might be able to tell what's > going on here? The main thing I'd like to know is the last thing it was > doing before it died... I dont know if someone has already covered this. Do you have a serial line? If you have (and if its not dead yet) the you can get into vxWorks shell and execute some of the commands to figure out whats the problem is. Here are some of the commands: i - displays the status of all tasks. Typically all tasks need to be in PEND mode For e.g. telnet deamon will be pending on the socket. tt - task trace ti - task info checkStack - displays the stack info for all tasks. You would know if some task has overshoot its stack. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: memory allocation problems Date: Wed, 22 Aug 2001 12:45:19 -0700 From: "Dan Gold" Message-ID: <3b840c55$0$77221$e2e8da3@nntp.cts.com> References: <3B83B0A1.891C8E46@yahoo.com> "Aleks" wrote in message news:3B83B0A1.891C8E46@yahoo.com... > Hi all, > > I use PPC603 and in the beginning some partition must be created (C > code). > memPartCreate() returns NULL... What additional diagnostic could I get > in order to find why? > Is anything in addition to printErrno()? In manual for printErrno() is > said, that > error-status symbol table must be included in the system. How can I > check it? Just how 'in the beginning' are you trying to create it ? You need to make sure that memPartLibInit() in usrRoot() has at least been called. And in your code, before doing the memPartCreate(), you might just add a memShow(0) to see what memory looks like before you make your call. hth, Dan Gold gold@ensemble.com --------------------------- Newsgroups: comp.os.vxworks Subject: TFFS in Tornado 2 for PPC Date: Wed, 22 Aug 2001 20:49:34 +0000 (UTC) From: "Johan Bostedt" Organization: Mailgate.ORG Server - http://www.Mailgate.ORG Message-ID: I am trying to implement a TFFS (flash file system) for Tornado on an DMV-179 board. Following the instructions in the TFFS manual I copied a sysTffs file to my BSP directory and changed the parameters to fit with the HW. When I compile this code I get an error message during linking about the symbols mountFTL and formatFTL. These symbols are not in the PPC604 library and not in any other library. What am I doing wrong. I know a similar message was posted back in 1999, but I haven't found any answer yet regards Johan Bostedt - ---------------------------------------------------------------------- Johan Bostedt mailto:johan@racomna.se Phone: +46(0)18- 12 40 70 Racomna AB http://www.racomna.se Direct: +46(0)18- 12 40 95 Salagatan 16 A Mobile: +46(0)70-657 64 70 753 30 Uppsala Fax: +46(0)18- 12 40 74 SWEDEN - -- Posted from [213.136.50.230] via Mailgate.ORG Server - http://www.Mailgate.ORG --------------------------- Newsgroups: comp.os.vxworks Subject: SV: TFFS in Tornado 2 for PPC Date: Wed, 22 Aug 2001 23:05:06 +0200 From: "Johan Bostedt" Organization: Mailgate.ORG Server - http://www.Mailgate.ORG Message-ID: <9m16q8$e7q$1@newsreader.mailgate.org> References: Sender: johan.bostedt@racomna.se It seems like a file "ftllite.o" is missing in target/lib/objPPC604gnuvx, when comparing how things are defined for mountNFTL and formatNFTL. But how am I going to get that file? I'm using a Dy4179+ BSP for Tornado 2.0 (rel. 1.2) /Johan Johan Bostedt skrev i diskussionsgruppsmeddelandet:A78E22C5F0C113438E5B6AA109F34E5D032E00@konsum.r acomna.local... I am trying to implement a TFFS (flash file system) for Tornado on an DMV-179 board. Following the instructions in the TFFS manual I copied a sysTffs file to my BSP directory and changed the parameters to fit with the HW. When I compile this code I get an error message during linking about the symbols mountFTL and formatFTL. These symbols are not in the PPC604 library and not in any other library. What am I doing wrong. I know a similar message was posted back in 1999, but I haven't found any answer yet regards Johan Bostedt - ---------------------------------------------------------------------- Johan Bostedt mailto:johan@racomna.se Phone: +46(0)18- 12 40 70 Racomna AB http://www.racomna.se Direct: +46(0)18- 12 40 95 Salagatan 16 A Mobile: +46(0)70-657 64 70 753 30 Uppsala Fax: +46(0)18- 12 40 74 SWEDEN - -- Posted from [213.136.50.230] via Mailgate.ORG Server - http://www.Mailgate.ORG --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Anyone get VxWorks to boot quickly Date: Wed, 22 Aug 2001 16:44:42 -0500 From: "Brad Crabtree" Organization: Raytheon Company Message-ID: <7LVg7.8$Ta2.107@dfw-service2.ext.raytheon.com> References: Thanks to all for the replies, very helpful. We are also working with local WindRiver for help. I promise to post any lessons learned for quick boot when we complete. Thanks again. > -------------------------------------------- > Bradley D. Crabtree > b-crabtree@raytheon.com > > > --------------------------- Newsgroups: comp.os.vxworks Subject: TFFS in Tornado 2 for PPC Date: Wed, 22 Aug 2001 22:06:05 GMT From: Andray Kaganovsky Organization: @Home Message-ID: References: <9m16q8$e7q$1@newsreader.mailgate.org> "Johan Bostedt" wrote in news:9m16q8$e7q$1 @newsreader.mailgate.org: > It seems like a file "ftllite.o" is missing in target/lib/objPPC604gnuvx, > when comparing how things are defined for mountNFTL and formatNFTL. > But how am I going to get that file? > I'm using a Dy4179+ BSP for Tornado 2.0 (rel. 1.2) > /Johan > > > Johan Bostedt skrev i > diskussionsgruppsmeddelandet:A78E22C5F0C113438E5B6AA109F34E5D032E00@konsum.r > acomna.local... > I am trying to implement a TFFS (flash file system) for Tornado on an > DMV-179 board. Following the instructions in the TFFS manual I copied a > sysTffs file to my BSP directory and changed the parameters to fit with > the HW. When I compile this code I get an error message during linking > about the symbols mountFTL and formatFTL. These symbols are not in the > PPC604 library and not in any other library. > What am I doing wrong. > I know a similar message was posted back in 1999, but I haven't found > any > answer yet > regards Johan Bostedt > ---------------------------------------------------------------------- > Johan Bostedt mailto:johan@racomna.se Phone: +46(0)18- 12 40 70 > Racomna AB http://www.racomna.se Direct: +46(0)18- 12 40 95 > Salagatan 16 A Mobile: +46(0)70-657 64 70 > 753 30 Uppsala Fax: +46(0)18- 12 40 74 > SWEDEN > > > > -- > Posted from [213.136.50.230] > via Mailgate.ORG Server - http://www.Mailgate.ORG > [AndrayK]. TrueFFS used to be an optional product that needed to be purchased separately. I don't know if it's still the case. Andray --------------------------- Newsgroups: comp.os.vxworks,comp.os.linux Subject: porting linux kernel code to vxworks? Date: 22 Aug 2001 15:14:07 -0700 From: dragonli@yahoo.com (June) Organization: http://groups.google.com/ Message-ID: Is there anybody tried porting linux kernel code (ipv4) to vxworks? Can you point me to some guideline, articles, website, whatever? Thanks. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Newby Info Date: 22 Aug 2001 15:44:43 -0700 From: deepa@eventhelix.com (Deepa Ahluwalia) Organization: http://groups.google.com/ Message-ID: <6a9a3562.0108221444.5a16de83@posting.google.com> References: You will find articles about embedded and real-time software programming at the URL: http://www.eventhelix.com/RealtimeMantra/ Examples of Real-time and Embedded Software development can be found at: http://www.eventhelix.com/ThoughtProjects/ Deepa http://www.EventHelix.com Real-time Software Tools and Techniques --------------------------- Newsgroups: comp.os.vxworks Subject: DHCP server on multi i/faces Date: 22 Aug 2001 15:46:47 -0700 From: chrismooremobile@yahoo.com (Chris Moore) Organization: http://groups.google.com/ Message-ID: Hi, I'm looking at using the DHCP server that comes with VxW (the supported one). But looks like there is a problem. If I have multiple interfaces (say two ethernets) and both of those subnets have DHCP clients on them, it means that the DHCP pool is common for both of these?!?! Thats useless. I need to have two different pools, ie: 192.168.2.x for eth1 and 192.168.50.x for eth2. But if I receive a DHCP request it does not differentiate which interface it came in from does it? Or does it go "Hmm, I rec'd a DHCP on eth1, eth1s ip is 192.168.2.x (mask 0xFFFFFF00) so I should only serve up addresses where the network component (192.168.2) matches..." Does anyone know if it defintely works like this? The documentation is great but doesn't explain anything useful. Rgds, Chris --------------------------- Newsgroups: comp.os.vxworks Subject: Telnet sessions Date: Wed, 22 Aug 2001 17:38:37 -0600 From: Les Hammer Organization: Agilent Technologies Message-ID: <3B84427C.57B71C13@agilent.com> Hi, Does anyone have experience in getting more than one telnet session to work with VxWorks? When I try to open a second telnet session I get a "Sorry, the system is engaged" error, even though I used telnetInit(2) in my usrAppInit.c, and I have the void telnetInit(int maxConnNum) code. Also, does anybody have experience in a telnet session that goes to a limited user interface, rather than the VxWorks shell? Thanks, Les Hammer Agilent Technologies --------------------------- Newsgroups: comp.os.vxworks Subject: Re: about virtual memory exhausted error Date: 23 Aug 2001 10:32:10 +0900 From: Yasuyuki Kino Organization: OMRON SOFTWARE Co., Ltd. Message-ID: References: <488e459a.0108212131.7bfebe80@posting.google.com> Hello, I have 256Mb memory, and swap space is set as 1000Mb, but compiler sends error message. I checked memory in compiling, and 128Mb or more memory remained at that time .... ... Thank you. kino. >>>>> In <488e459a.0108212131.7bfebe80@posting.google.com>=20 >>>>> =E2=80=9CJohn=E2=80=9D =3D john_94501@yahoo.com (John) wrote: > Hello, > Welcome to the world of Windoze... Basically, your swap space is being > exhausted by the compiler (compiling complex code can do this, > especially it seems C++ code). To solve the problem you have a choice: > 1) Increase the size of the swap space (paging file) using the Windows > control panel. It is normally on the System tool's performance tab, > but since this is Windows, it could have been moved anywhere =EF=BF=BD=EF= =BF=BD=EF=BF=BD=EF=BF=BD=EF=BF=BD=EF=BF=BD=EF=BF=BD=EF=BF=BD=EF=BF=BD=EF=BF= =BD=EF=BF=BD=EF=BF=BD=EF=BF=BD=EF=BF=BD=EF=BF=BD=EF=BF=BD=EF=BF=BD=EF=BF=BD= =EF=BF=BD=EF=BF=BD=EF=BF=BD=EF=BF=BD=EF=BF=BD=EF=BF=BD=EF=BF=BD=EF=BF=BD=EF= =BF=BD=EF=BF=BD=EF=BF=BD=EF=BF=BD=EF=BF=BD=EF=BF=BD=EF=BF=BD=EF=BF=BD=EF=BF= =BD=EF=BF=BD=EF=BF=BD=EF=BF=BD=EF=BF=BD=EF=BF=BD=EF=BF=BD=EF=BF=BD=EF=BF=BD= =EF=BF=BD=EF=BF=BD=EF=BF=BD=EF=BF=BD=EF=BF=BD=EF=BF=BD=EF=BF=BD=EF=BF=BD=EF= =BF=BD=EF=BF=BD=EF=BF=BD=EF=BF=BD=EF=BF=BD=EF=BF=BD=EF=BF=BD=EF=BF=BD=EF=BF= =BD=EF=BF=BD=EF=BF=BD=EF=BF=BD=EF=BF=BD=EF=BF=BD=EF=BF=BD=EF=BF=BD=EF=BF=BD= =EF=BF=BD=EF=BF=BD=EF=BF=BD=EF=BF=BD=EF=BF=BD=EF=BF=BD=EF=BF=BD=EF=BF=BD=EF= =BF=BD=EF=BF=BD=EF=BF=BD=EF=BF=BD=EF=BF=BD=EF=BF=BD=EF=BF=BD=EF=BF=BD=EF=BF= =BD=EF=BF=BD=EF=BF=BD=EF=BF=BD=EF=BF=BD=EF=BF=BD=EF=BF=BD=EF=BF=BD=EF=BF=BD= =EF=BF=BD=EF=BF=BD=EF=BF=BD=EF=BF=BD=EF=BF=BD=EF=BF=BD=EF=BF=BD=EF=BF=BD=EF= =BF=BD=EF=BF=BD=EF=BF=BD=EF=BF=BD=EF=BF=BD=EF=BF=BD=EF=BF=BD=EF=BF=BD=EF=BF= =BD=EF=BF=BD=EF=BF=BD=EF=BF=BD=EF=BF=BD=EF=BF=BD=EF=BF=BD=EF=BF=BD=EF=BF=BD= =EF=BF=BD=EF=BF=BD=EF=BF=BD=EF=BF=BD=EF=BF=BD=EF=BF=BD=EF=BF=BD=EF=BF=BD=EF= =BF=BD=EF=BF=BD > 2) Make sure all other apps are shut down, including any of those > system tray ones that clutter up Windows. These all take memory, and > when doing big compilations you'll want every byte you can find to > keep it fast. > 3) Upgrade to a real operating system =EF=BF=BD=EF=BF=BD=EF=BF=BD=EF=BF= =BD=EF=BF=BD=EF=BF=BD=EF=BF=BD=EF=BF=BD=EF=BF=BD=EF=BF=BD=EF=BF=BD=EF=BF=BD= =EF=BF=BD=EF=BF=BD=EF=BF=BD=EF=BF=BD=EF=BF=BD=EF=BF=BD=EF=BF=BD=EF=BF=BD=EF= =BF=BD=EF=BF=BD=EF=BF=BD=EF=BF=BD=EF=BF=BD=EF=BF=BD=EF=BF=BD=EF=BF=BD=EF=BF= =BD=EF=BF=BD=EF=BF=BD=EF=BF=BD=EF=BF=BD=EF=BF=BD=EF=BF=BD=EF=BF=BD=EF=BF=BD= =EF=BF=BD=EF=BF=BD=EF=BF=BD=EF=BF=BD=EF=BF=BD=EF=BF=BD=EF=BF=BD=EF=BF=BD=EF= =BF=BD=EF=BF=BD=EF=BF=BD=EF=BF=BD=EF=BF=BD=EF=BF=BD=EF=BF=BD=EF=BF=BD=EF=BF= =BD=EF=BF=BD=EF=BF=BD=EF=BF=BD=EF=BF=BD=EF=BF=BD=EF=BF=BD=EF=BF=BD=EF=BF=BD= =EF=BF=BD=EF=BF=BD=EF=BF=BD=EF=BF=BD=EF=BF=BD=EF=BF=BD=EF=BF=BD=EF=BF=BD=EF= =BF=BD=EF=BF=BD=EF=BF=BD=EF=BF=BD=EF=BF=BD=EF=BF=BD=EF=BF=BD=EF=BF=BD=EF=BF= =BD=EF=BF=BD=EF=BF=BD=EF=BF=BD=EF=BF=BD=EF=BF=BD=EF=BF=BD=EF=BF=BD=EF=BF=BD= =EF=BF=BD=EF=BF=BD=EF=BF=BD=EF=BF=BD=EF=BF=BD=EF=BF=BD=EF=BF=BD=EF=BF=BD=EF= =BF=BD=EF=BF=BD=EF=BF=BD=EF=BF=BD=EF=BF=BD > I don't know how Win2K handles virtual memory, but it there is usually > an option to specify the size of the page file manually, set it to be > nice and large. > You might also want to make sure that you have plenty of physical > memory in your machine too... For complex C++ compilations I would > suggest 256Mb or more (especially given the price of RAM these days). > It will make things happen much more smoothly... > HTH, > John... --------------------------- Newsgroups: comp.os.vxworks Subject: MIPS, breakpoints, watchpoints problem Date: Thu, 23 Aug 2001 04:44:14 +0000 (UTC) From: "ssaraf" Organization: Mailgate.ORG Server - http://www.Mailgate.ORG Message-ID: <65256AB1.001A3501.00@sandesh.hss.hns.com> Hi vxWorkers, I have two questions regarding breakpoints/watchpoints in vxWorks. We have a QED-RM 7000(MIPS) target and I was wondering how b and bh are implemented in vxworks. I thought b would just place a break instruction at the address and then do all the good stuff in the exception handler. When I do a b somefunction, and call that function, the breakpoint is hit, but 'l thatfunction' or d 'that function' doesn't show break or the opcode for break. How does it work then? The other problem is bh someaddress,3(which is supposed to be a hardware breakpoint on any read/write access) doesn't work. Does anyone actually have watchpoints working on a RM-7000 target? I wrote some assembler code basically mtc0, dmtc0's to $18,$19,$24(the watch and the mask registers, respectively) but I still don't get a Iwatch or a DWatch exception when I try to read/write the address mentioned in the watch registers. I did something like: x=malloc(10); on the shell. wrote some assembler to do : 1) put physical address(using the bsp VIRT_TO_PHY macro) of x in watch register1 using dmtc0 access type is store,load,instruction 2) disable the mask register. then on the shell did: 3) did a sp bzero, address_of_x,10 nothing happened, I should have a got an exception here? :) Any hints appreciated. Thanks! - -- Posted from [202.54.26.202] via Mailgate.ORG Server - http://www.Mailgate.ORG --------------------------- Newsgroups: comp.os.vxworks Subject: MIPS, breakpoints, watchpoints problem Date: 22 Aug 2001 21:55:45 -0700 From: suman@linux-delhi.org (suman saraf) Organization: http://groups.google.com/ Message-ID: Hi vxWorkers, I have two questions regarding breakpoints/watchpoints in vxWorks. We have a QED(now PMC)-RM 7000(MIPS) target and I was wondering how b and bh are implemented in vxworks. I thought b would just place a break instruction at the address and then do all the good stuff in the exception handler. When I do a b somefunction, and call that function, the breakpoint is hit, but 'l thatfunction' or d 'that function' doesn't show break or the opcode for break. How does it work then? The other problem is bh someaddress,3(which is supposed to be a hardware breakpoint on any read/write access) doesn't work. Does anyone actually have watchpoints working on a RM-7000 target? I wrote some assembler code basically mtc0, dmtc0's to $18,$19,$24(the watch and the mask registers, respectively) but I still don't get a Iwatch or a DWatch exception when I try to read/write the address mentioned in the watch registers. I did something like: x=malloc(10); on the shell. wrote some assembler to do : 1) put physical address(using the bsp VIRT_TO_PHY macro) of x in watch register1 using dmtc0 access type is store,load,instruction 2) disable the mask register. then on the shell did: 3) did a sp bzero, address_of_x,10 nothing happened, I should have a got an exception here? :) Any hints appreciated. Thanks! --------------------------- Newsgroups: comp.os.vxworks Subject: Re: about virtual memory exhausted error Date: 22 Aug 2001 22:49:40 -0700 From: john_94501@yahoo.com (John) Organization: http://groups.google.com/ Message-ID: <488e459a.0108222149.62f8998a@posting.google.com> References: <488e459a.0108212131.7bfebe80@posting.google.com> Hello, Well, if this happens reliably every time you compile one file it could be a compiler error... try sending the file to Wind River's support people to see if they can reproduce it. HTH, John... Yasuyuki Kino wrote in message news:... > Hello, > > I have 256Mb memory, and swap space is set as 1000Mb, > but compiler sends error message. > > I checked memory in compiling, and 128Mb or more memory > remained at that time .... > > ... Thank you. > > kino. > > > >>>>> In <488e459a.0108212131.7bfebe80@posting.google.com> > >>>>> ?John? = john 94501@yahoo.com (John) wrote: > > Hello, > > > Welcome to the world of Windoze... Basically, your swap space is being > > exhausted by the compiler (compiling complex code can do this, > > especially it seems C++ code). To solve the problem you have a choice: > > > 1) Increase the size of the swap space (paging file) using the Windows > > control panel. It is normally on the System tool's performance tab, > > but since this is Windows, it could have been moved anywhere ????????????????????? > ???????????????????????????? > ???????????????????????????? > ???????????????????????????? > ???????????????????????????? > ???????????? > > > 2) Make sure all other apps are shut down, including any of those > > system tray ones that clutter up Windows. These all take memory, and > > when doing big compilations you'll want every byte you can find to > > keep it fast. > > > 3) Upgrade to a real operating system ????????????? > ???????????????????????????? > ???????????????????????????? > ???????????????????????????? > ??????????????? > > > I don't know how Win2K handles virtual memory, but it there is usually > > an option to specify the size of the page file manually, set it to be > > nice and large. > > > You might also want to make sure that you have plenty of physical > > memory in your machine too... For complex C++ compilations I would > > suggest 256Mb or more (especially given the price of RAM these days). > > It will make things happen much more smoothly... > > > HTH, > > > John... --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Telnet sessions Date: 22 Aug 2001 23:01:35 -0700 From: john_94501@yahoo.com (John) Organization: http://groups.google.com/ Message-ID: <488e459a.0108222201.e663389@posting.google.com> References: <3B84427C.57B71C13@agilent.com> Hello, This is a common topic on here... writing a simple telnet session is in fact very, very easy. All you need to do is listen for TCP connections on a known port (by default for telnet this is port 23). Once the connection is established, you can set the standard I/O for your task to the socket's file descriptor and use normal C I/O functions. Here's a little code fragment for listening for connections on a specified port, no warranty of course, use at your own risk, etc, etc: #include "vxWorks.h" #include "errnoLib.h" #include "inetLib.h" #include "ioLib.h" #include "netinet/in.h" #include "netinet/tcp.h" #include "socket.h" #include "sockLib.h" #include "stdio.h" #include "string.h" #include "unistd.h" int openSocket ( int port ) { int optVal = 1; /* turn feature on */ int f; int s; struct sockaddr_in sin, from; int len; memset((char *)&sin, 0, sizeof(sin)); memset((char *)&from, 0, sizeof(from)); f = socket(AF_INET, SOCK_STREAM, 0); if (f < 0) { perror("socket"); return -1; } /* Make socket reusable */ if (setsockopt (f, SOL_SOCKET, SO_REUSEADDR, (char *) &optVal, sizeof (optVal)) == ERROR) { perror ("setsockopt"); return -1; } sin.sin_family = AF_INET; sin.sin_port = htons(port); if (bind(f,(struct sockaddr *) &sin, sizeof(sin)) < 0) { perror("bind"); return -1; } if (listen(f, 5) < 0) { perror("listen"); return -1; } len = sizeof(from); s = accept(f,(struct sockaddr *)&from, &len); if (s <= 0) { perror("accept"); return -1; } /* Set keepalive state on the connection socket */ if (setsockopt (s, SOL_SOCKET, SO_KEEPALIVE, (char *) &optVal, sizeof (optVal)) == ERROR) { perror ("setsockopt"); return -1; } close(f); return(s); } Assuming it doesn't return -1, you have a new file descriptor for your connection. Spawn a new task to handle that connection and pass it this file descriptor so it can communicate with the remote telnet client. If you want to allow more connections, simply get the main task to call openSocket() again, passing it the same port number. HTH, John... Les Hammer wrote in message news:<3B84427C.57B71C13@agilent.com>... > Hi, > > Does anyone have experience in getting more than one > telnet session to work with VxWorks? When I try to > open a second telnet session I get a "Sorry, the system > is engaged" error, even though I used telnetInit(2) in my > usrAppInit.c, and I have the > void telnetInit(int maxConnNum) > code. > > Also, does anybody have experience in a telnet session > that goes to a limited user interface, rather than the VxWorks > shell? > > Thanks, > > Les Hammer > Agilent Technologies --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Exception handling in VxWorks Date: Wed, 22 Aug 2001 23:13:48 -0700 From: "Albert H Chen" Message-ID: References: <623ba2de.0108210727.75b5534f@posting.google.com> <623ba2de.0108220938.75fe0a79@posting.google.com> Try using "sigvec()" or "sigaction()" see the following example: #define TASK_PRI 254 #define PIPE_NAME "/pipe/server" #define NUM_MSGS 10 /* setjmp buffer */ LOCAL jmp_buf env; extern void i (int); STATUS recoverExcDemo (); /************************************************************** * sigHandler -- Returns task to known state. * * Called when signal posted. * */ void sigHandler ( int sig, /* signal number */ int code, /* additional code */ SIGCONTEXT *sigContext /* context of task before siganl */ ) { switch (sig) { case SIGBUS: printErr ("\nSignal SIGBUS (bus error) received\n"); break; case SIGILL: printErr ("\nSignal SIGILL received\n"); break; case SIGSEGV: printErr ("\nSignal SIGSEGV received\n"); break; } longjmp (env, sig); } /************************************************************** * */ STATUS demo_raise () { printf ("causing bus error\n"); /* Deliberately raise bus error */ (* (FUNCPTR) NULL_ADDR)(); return OK; } /************************************************************** * */ STATUS demo_recover () { int val = 0; SIGVEC sig; int setjmpReturnValue; sig.sv_handler = (VOIDFUNCPTR) sigHandler; sig.sv_mask = 0; sig.sv_flags = 0; val = 50; printf ("demo_recover...\n"); setjmpReturnValue = setjmp (env); if (setjmpReturnValue == 0) { /* direct invocation of setjmp */ /* install signal handler for each of the following signals */ sigvec (SIGBUS, &sig, NULL); sigvec (SIGILL, &sig, NULL); sigvec (SIGSEGV, &sig, NULL); printf ("demo_recover...setup sigvec\n"); } else { /* from a call to longjmp */ printf ("recover from signal %d. val = %d \n", setjmpReturnValue, val); return OK; } /* raise the exception */ demo_raise (); return OK; } STATUS recoverDemo () { demo_recover (); return OK; } HTH, Albert "pvpartha1" wrote in message news:623ba2de.0108220938.75fe0a79@posting.google.com... > Lee DeRaud wrote in message news:... > > On 21 Aug 2001 08:27:07 -0700, vivekp@hotmail.com (pvpartha1) wrote: > > > > >Is it possible to trap data exception errors or bus errors and suspend > > >the thread that causes that to happen? How is this done? > > >any help would be appreciated > > > > In what architecture? Most of them already have bus error and/or data > > access exceptions (e.g. misaligned accesses) enabled by default, and > > suspension of the offending task is (part of) the default handling. > > (Note that, much to the surprise of everyone here, FPU exceptions > > (e.g. divide-by-zero) are *not* enabled in M68K.) > > > > Lee > > > The architecture iam running on is PPC604. What iam trying to do is > the following > > > Whenever a task exception occurs suspend the task and do a stack trace > and task trace . Write contnts to a file and move on. Is it possible > to overload the default exception handler? How is it done ? --------------------------- Newsgroups: comp.os.vxworks Subject: SV: TFFS in Tornado 2 for PPC Date: Thu, 23 Aug 2001 08:17:55 +0200 From: "Johan Bostedt" Organization: Mailgate.ORG Server - http://www.Mailgate.ORG Message-ID: <9m276r$g6e$2@newsreader.mailgate.org> References: <9m16q8$e7q$1@newsreader.mailgate.org> Sender: johan.bostedt@racomna.se Yes it's optional and purchased separately, but still... --------------------------- Newsgroups: comp.os.vxworks Subject: Re: about virtual memory exhausted error Date: 23 Aug 2001 16:23:40 +0900 From: Yasuyuki Kino Organization: OMRON SOFTWARE Co., Ltd. Message-ID: References: <488e459a.0108212131.7bfebe80@posting.google.com> <488e459a.0108222149.62f8998a@posting.google.com> Hello, As a matter of fact, I have already asked to windriver support. However, since a reply was not yet obtained, I asked this newsgroup. I try to look for how to avoid it. Thank you. - -- kino >>>>> In <488e459a.0108222149.62f8998a@posting.google.com> >>>>> "John" = john_94501@yahoo.com (John) wrote: > Hello, > Well, if this happens reliably every time you compile one file it > could be a compiler error... try sending the file to Wind River's > support people to see if they can reproduce it. > HTH, > John... --------------------------- Newsgroups: comp.os.vxworks Subject: Several Questions... Date: Thu, 23 Aug 2001 00:55:16 -0700 From: "Albert H Chen" Message-ID: 1. There is a certain portion of code in the bsp that belongs both in the bootrom and also in the working image. How do I do something along the lines of following: void func(){ if(boot_rom) printf("In bootrom\n"); else printf("Not in bootrom\n"); } 2. Does intLock prevent task preemption? 3. Why can't you do semGive() of mutal-exclusion semaphores? Thanks, Albert --------------------------- Newsgroups: comp.os.vxworks Subject: Strict timing Date: Thu, 23 Aug 2001 10:02:12 +0200 From: "Andrea della Porta" Organization: Centro Servizi Interbusiness Message-ID: <9m2cif$82j$1@fe2.cs.interbusiness.it> Does anyone know how to obtain precise task_delay() when several task is running? How do priority level among tasks affect this delay()? I've to simulate I2C bus waveforms but it seems that the timing passed to task_delay() is not respected. Thanks in advance, Andrea della Porta --------------------------- Newsgroups: comp.os.vxworks Subject: SLIP protocol under Windows 2000/NT Date: Thu, 23 Aug 2001 11:28:40 +0300 From: "Robert Berger" Organization: Intracom S.A., GREECE Message-ID: <9m2fga$s1n$1@newsserv.intranet.gr> I just got VxWorks/Tornado and I am trying to use connect to my PC. Unfortunately there is no ethernet interface on my BSP and the SLIP protocol does not seem to work. Can someone help on how to connect a BSP using SLIP to a PC running Windows 2000/NT ? regards, Robert --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Debugging when the debugger crashes Date: Thu, 23 Aug 2001 09:36:50 +0100 From: James Marshall Organization: Agilent Technologies UK Ltd Message-ID: <3B84C0A2.E2914F3D@agilent.com> References: <2psil9.q9e.ln@127.0.0.1> <3B7E0220.9ACC02DE@covad.net> <3B822F37.6B80B863@agilent.com> <7cdtl9.lr4.ln@127.0.0.1> <3B82809D.775FD159@agilent.com> <488e459a.0108212114.6a0a2485@posting.google.com> Okay - I understand where I went wrong: right at the beginning when I assumed that NMI was an "exception". As the Intel Architecture manual implies (yeah I didn't RTFM carefully) an "asynchronous exception" doesn't exist - - it's called an interrupt! Replacing the NMI vector looked like a good solution until WindRiver support pointed out that no VxWorks functions can be called since NMI can happen while the kernel is in a critical section with ints locked. This is even more strict than an ISR. excJobAdd() uses a message queue so I can't call it from within my vectored NMI handler. Hmm...this is non-trivial. I guess that's why it requires a task coming into context before VxWorks picks up the indication of an NMI exception. It looks like the _func_excBaseHook might be my better option. It gets round the problem of signal handlers superseding exception handling. Many thanks for your helpful suggestions, John. This has been a great learning experience for me. James Marshall. John wrote: > Hello, > > James Marshall wrote in message news:<3B82809D.775FD159@agilent.com>... > > > I use excHookAdd() to add my exception handler as I found out on the group archive. One problem I've had > > with it is the way VxWorks handles exceptions (on x86 anyway): the exception isn't reported/handled > > until a task runs. > > An "exception" cannot occur unless a task is running (unless it > happens in an ISR, and that is a panic situation that causes the OS to > reboot). It strikes me that the NMI should in fact be hooked in using > not the excConnect() function, but the intConnect() one. Since I > cannot see an NMI in the standard header file for the x86 family, it > will be treated as a generic exception vector (which vector number is > it on?). You can change this - see solution 2 below. > > > I tried to find a way of resuming an excepting task that has been suspended and came up with this: > > > > - in the exception handler call excJobAdd(), passing it a pointer to a function that spawns a task at > > priority 1, and passing the current task id. [excJobAdd() works like netJobAdd(), passing a function to > > be called from tExcTask, and is declared in excLib.h] > > > > - in the spawned task, wait until your excepting task is suspended (yes, I had to add a taskDelay(1)), > > then resume it and exit the spawned task. > > This is perhaps a little complicated. A couple of simpler solutions > for your case might be the following: > > 1) Use the older hook (_func_excBaseHook) instead of the one you're > using right now. > > This hook is called with four parameters: > > int vecNum, > ESF0 * pESF, > REG_SET * pRegs, > EXC_INFO * pExcInfo > > The return value should be FALSE if the exception was not handled, or > TRUE if it was. So, if you detect your special NMI exception, just > return TRUE and the exception handler will exit without suspending the > task. > > 2) Better still, simply replace the vector for the NMI. If you know > the vector number (and I can't see it in the header files for x86 so > I'm not sure where the NMI vector is), simply connect a routine to > that vector using intConnect() - yes, intConnect() is used for > exception handler connection on x86 platforms. > > Your handler routine can simply be a return, and the offending NMI > exceptions will be ignored. > > -- > > Something for others to be aware of... in the case of a normal class > of exception, simply resuming the task will not do any good. When a > CPU takes a synchronous exception, the program counter is normally set > to the address of that exception. The intention is that the handler > will fix the cause of the exception (think paging memory on big > operating systems; the exception happens and the OS gets the missing > page off the disk) and then retry the instruction. > > But all is not lost. There are a number of ways of recovering from > these types of exception, such as page faults. Take a look in the > reference manual at the following routines: taskRestart() or > setjmp()/longjmp(). These can be called from a hook routine, or from a > signal handler (all the conventional exceptions are hooked into > signals too if a signal handler has been registered by the task). > > The way to choose? Signals are task specific; each task must register > to receive them, but can easily do its own thing to handle an error. > The hook routine is the same for all exception types, and all tasks. > This might be what you need, but means that the hook routine has a lot > more work to do... > > HTH, > > John... --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Several Questions... Date: Thu, 23 Aug 2001 11:37:40 +0200 From: Urban Lindberg Organization: Ericsson Message-ID: <3B84CEE4.B407DB0B@ks.ericsson.se> References: 1. Comparing the address of sysInit and RAM_LOW_ADRS should tell you if its boot_rom or not. VxWorks is loaded at RAM_LOW_ADRS, while the boot rom code is copied to RAM_HIGH_ADRS. Another way is to insert a variable which is initialized to something (TRUE, FALSE) in a common file and in any of the bootrom specific files set it to another value. 2. Normally, yes. But exceptions will preempt any task. Don't use it unless you realy need to. 3. Huh, what do you mean? You have to do a semGive for all semTake if you want to release the resource (unless you use semMGiveForce, which should only be used as a debugging aid). Se docu for semMLib. HTH Urban Albert H Chen wrote: > > 1. There is a certain portion of code in the bsp that belongs both in the > bootrom and also in the working image. > How do I do something along the lines of following: > > void func(){ > if(boot_rom) printf("In bootrom\n"); > else printf("Not in bootrom\n"); > } > > 2. Does intLock prevent task preemption? > > 3. Why can't you do semGive() of mutal-exclusion semaphores? > > Thanks, > > Albert --------------------------- Newsgroups: comp.os.vxworks Subject: ioctl in END Driver. Date: 23 Aug 2001 03:37:01 -0700 From: bondi_ameya@yahoo.com (Ameya) Organization: http://groups.google.com/ Message-ID: <57c7939b.0108230237.6c09c64e@posting.google.com> hi vxworks experts, i have an xxxIoctl in my END driver. when i call ioctl from the usrappinit, i expect the xxxIoctl to be called but this does not happen. i tired the following... 1) i know the device name "abcd" is correct by using the endFindByName ( "abcd",0); 1) An explicit call to the muxIoctl works fine. but how does the ioctl link with the muxIoctl? muxIoctl(pEnd, EIOCGFLAGS,&data) 2) i started a server and a client.they work fine but a call to ioctl using the sockets fd always gives me a fd=-1. lemme know. regards ameya --------------------------- Newsgroups: comp.os.vxworks Subject: any one would help me to solve the PNP problem? Date: 23 Aug 2001 04:01:03 -0700 From: virsala@yahoo.com (robert chen) Organization: http://groups.google.com/ Message-ID: hi all: Now we are doning a project which needs to add a pnp function to it. Because we do not use this technology before,any one would tell me how to implement this function? Also you can give me some useful links on the web about this subject. Any suggestions are appreciated. Robert lee --------------------------- End of New-News digest ********************** From vxwexplo-errs@csg.lbl.gov Thu Aug 23 20:41:35 2001 From: karunakaran nair Date: Thu Aug 23 20:41:37 PDT 2001 Subject: C , C++ and vxWorks tasks how to work with vxworks tasks in a C++ scenario ? say I want to have a task, spawned from a member function, and this task has to wait on a semaphore sent from C function of a different file, and then sequence thru to do some activities calling some member functions ? i see the task to be in READY state when I spawn it from the member function. here is my design. extern SEM_ID mysemaphore void myclass::memberfunction(void) { spawn mytask } void mytask(void) { while(1) { wait on semaphore ( sent from another c file ) call some myclass member functions to do something } } __________________________________________________ Do You Yahoo!? Make international calls for as low as $.04/minute with Yahoo! Messenger http://phonecard.yahoo.com/ From vxwexplo-errs@csg.lbl.gov Fri Aug 24 04:03:23 2001 From: Vxworks Exploder Date: Fri Aug 24 04:03:26 PDT 2001 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Fri Aug 24 04:03:10 PDT 2001 Subject: Problems with Fast Ethernet at Ads8260 Subject: Re: SV: TFFS in Tornado 2 for PPC (DY 4 DMV-179) Subject: Re: Anyone get VxWorks to boot quickly Subject: Re: multiple developers & env vars in Tornado (was dependency generation) Subject: SIU9 Errata on MPC860? Subject: logInit question Subject: Re: Strict timing Subject: Multicast Subject: SV: SV: TFFS in Tornado 2 for PPC (DY 4 DMV-179) Subject: CPX8216 Subject: Re: Newbie: inhibit / trap VME bus error? Subject: NTP 4.1 for vxworks 5.4 Subject: Re: Multicast Subject: WEHT protection domains? Subject: Re: CS8900 driver - multiple units Subject: WRS and their DHCP server. Subject: Re: Several Questions... Subject: Re: Newbie: inhibit / trap VME bus error? Subject: Re: Several Questions... Subject: PPC6750 Subject: What is the module format of vxworks on earth? ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Problems with Fast Ethernet at Ads8260 Date: 23 Aug 2001 04:20:00 -0700 From: ruchama@septier.com (Ruchama Goldshmidt) Organization: http://groups.google.com/ Message-ID: <5df340e0.0108230320.3cd21bb5@posting.google.com> Hello! I am working with the ads8260 board and trying to find the source for the Lxt970 fast Ethernet PHY, because I am working also at a new board with the same cpu but with a new PHY called Lxt971 and I want to see which changes should I make and without the source I cannot proceed! So if anyone have this code or something that could help me I will be gratefull!! Ruchama Goldshmidt Septier Communication Ltd. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: SV: TFFS in Tornado 2 for PPC (DY 4 DMV-179) Date: 23 Aug 2001 11:22:24 GMT From: ac151@FreeNet.Carleton.CA (David Clarke) Organization: The National Capital FreeNet Message-ID: <9m2p1g$anh$1@freenet9.carleton.ca> References: <9m16q8$e7q$1@newsreader.mailgate.org> Reply-To: ac151@FreeNet.Carleton.CA (David Clarke) "Johan Bostedt" (johan@racomna.se) writes: > It seems like a file "ftllite.o" is missing in target/lib/objPPC604gnuvx, > when comparing how things are defined for mountNFTL and formatNFTL. > But how am I going to get that file? > I'm using a Dy4179+ BSP for Tornado 2.0 (rel. 1.2) > /Johan Johan: As of Release 1.1 the DY4 179+ BSP should support TFFS. All that you need do is #define INCLUDE_TFFS in config.h and that should enable TFFS (from the perspective of the BSP). As was already noted, TFFS is an optional component offered by WindRiver. If you INCLUDE_TFFS in the BSP without having the WRS portion of TFFS installed you'll see linker errors similar to the ones you've reported. Once TFFS is installed, the linker errors should go away. Regards, David Clarke (The DY4 179+ T2 BSP is currently at release 1.4). - -- - ------------------------------- /''' -------------------------------- David Clarke c-00 ac151@Freenet.carleton.ca > - ------------------------------- - -------------------------------- --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Anyone get VxWorks to boot quickly Date: Thu, 23 Aug 2001 14:16:27 +0200 From: "Frank Wolf" Message-ID: References: <7LVg7.8$Ta2.107@dfw-service2.ext.raytheon.com> Hi there! We've had the same problem and found the following solution for us: In file 'bootInit.c' are two local functions that needed lots of time even when cache was enabled! LOCAL void copyLongs (source, destination, nlongs); LOCAL void fillLongs (buf, nlongs, val); We've replaced both of them to use the built-in DMA controller of our processor to do that job! This made our startup about 5x-10x faster. (I assume this depends on the processor/DMA controller used; we've used Hitachi SH3) Frank "Brad Crabtree" schrieb im Newsbeitrag news:7LVg7.8$Ta2.107@dfw-service2.ext.raytheon.com... > Thanks to all for the replies, very helpful. We are also working with local > WindRiver for help. I promise to post any lessons learned for quick boot > when we complete. > > Thanks again. > > -------------------------------------------- > > Bradley D. Crabtree > > b-crabtree@raytheon.com > > > > > > > > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: multiple developers & env vars in Tornado (was dependency generation) Date: Thu, 23 Aug 2001 09:50:18 -0400 From: "Joe Ott" Organization: Verio Message-ID: References: Kevin, We solved this problem by using the subst command. Check your windows help for more information on the subst command. It's been around since DOS days. We all have our own development directory but they are all mapped to the same drive letter. For example my development directory may be K:\thisproject\jott then I would use the subst command "subst o: K:\thisproject\jott" to map it to drive letter O:. Other team members would do the same. This allowed us to develop a workspace file and project files based on a single drive letter that was the same for everyone. We also version controlled the workspace file and project files so every one could keep up to date. These we set up so that they would be read/write when fetched from PVCS since Tornado expects these to be read/write. HTH, Joe Ott "Kevin Livingston" wrote in message news:w0ug7.19$oR1.339@client... > I believe some people have talked about having multiple developers working > on the same Tornado project. We would like to do that as well. Our current > "attempted" solution is to use environment variables to define where include > directories are. (e.g. using -I on the compile line). Tornado has the very > handy macro PRJ_DIR which can be used. however, if it is used, then the > project is saved and then reopened. (or just reopened by anyone with > tornado) PRJ_DIR will become expanded - and thus defeat the purpose, because > when the project is saved it will have absolute directories, and make it > difficult for any other developers. > > I thought I had a solution by using OS defined environment variables. As > tornado does not expand these variables when a project is opened. However, > I believe that tornado uses the information in the compile line: included > directories, to help it form its dependency lists. But, when doing this > tornado cannot understand foreign environment variables well, only its own > (e.g. PRJ_DIR) and frequently generates faulty dependencies - either > defeating the purpose of having dependencies, or corrupting the information > in the make file to a point where the project cannot be built at all. > > it is not a requirement to use environment variables, we would just like to > have multiple developers working on the same project, and currently the only > major thing holding us back, is setting the compile line for the project. > (as well as the other settings which need directories - e.g. project libs.) > > does anyone else experience this behavior? or have a workaround or > suggestions? > (I already use forward slashes; there are no spaces in my source between # > and include; I have a VPATH etc. -- the problem now appears with tornado - > and I guess technically it is not a problem, they do not have to support OS > environment variables - but it is the easiest (no cost) way to get multiple > developers seamlessly involved.) > If it helps we are using Tornado 2.0.2 with the Host running windows NT. > > thanks for your help > Kevin > > > --------------------------- Newsgroups: comp.os.vxworks Subject: SIU9 Errata on MPC860? Date: Thu, 23 Aug 2001 14:04:42 GMT From: lourens@mecalc.co.za (Lourens Geldenhuys) Organization: Mecalc (Pty) Ltd Message-ID: <3b850a78.9441496@modem.mecalc> Reply-To: lourens@mecalc.co.za We are running the ads860 BSP on a custom board and I am experiencing a phenomenon similar to errata SIU9, (CPU Receives Machine Check After Writing to the PLPRCR), but not entirely. Always on boot-up (cold boot), the 860 stops executing instructions after a write to the PLPRCR, but never on a warm boot (hard-reset). In other words, if I power-up my boards, the processor stops dead after writing to the PLPRCR, but then doing a hard-reset allows it to boot fine. Could it be that my XFC capacitor values are bad? I can confirm that the 860's SPLL has indeed locked after a cold boot, because I can see the CLKOUT on the scope. Has anybody else experienced this? If it is indeed the SIU9 errata, does WindRiver have a workaround patch available? Any advice? Lourens --------------------------- Newsgroups: comp.os.vxworks Subject: logInit question Date: Thu, 23 Aug 2001 15:58:19 +0300 From: "AE" Message-ID: <3b84f0c1@news.barak.net.il> Hi I would like to redirect logging messages (generated by logMsg() ) to a remote PC and read those messages using HyperTerm. HyperTerm can listen to a TCP/IP socket. How can I create a TCP/IP device and pass it to logInit(). Thanks. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Strict timing Date: Thu, 23 Aug 2001 11:24:30 -0400 From: Bhavesh Patel Organization: Mailgate.ORG Server - http://www.Mailgate.ORG Message-ID: <3B85202E.F4561436@fwion.com> References: <9m2cif$82j$1@fe2.cs.interbusiness.it> Sender: bpatel@fwion.com use high precision timestamping by enabling "INCLUDE_TIMESTAMP" in your config.h. However, for this your bsp should support timestamp driver. bhavesh Andrea della Porta wrote: > Does anyone know how to obtain precise task_delay() when several task is > running? How do priority level among tasks affect this delay()? I've to > simulate I2C bus waveforms but it seems that the timing passed to > task_delay() is not respected. > > Thanks in advance, > Andrea della Porta --------------------------- Newsgroups: comp.os.vxworks Subject: Multicast Date: Thu, 23 Aug 2001 10:33:01 -0500 From: Dinker Charak Organization: Fermi National Accelerator Laboratory Message-ID: <3B85222D.5ABF1466@fnal.gov> Hello, We need to support multicast. Should board selection be constrained by this in anyway? We are trying to choose between MVME230x and MVME5100 boards. dinker --------------------------- Newsgroups: comp.os.vxworks Subject: SV: SV: TFFS in Tornado 2 for PPC (DY 4 DMV-179) Date: Thu, 23 Aug 2001 17:53:16 +0200 From: "Johan Bostedt" Organization: Mailgate.ORG Server - http://www.Mailgate.ORG Message-ID: <9m38tl$iqq$2@newsreader.mailgate.org> References: <9m16q8$e7q$1@newsreader.mailgate.org> <9m2p1g$anh$1@freenet9.carleton.ca> Sender: johan.bostedt@racomna.se TFFS is installed and every configuration aspect in the book is implemented. But still it didn't compile. However, I found a solution on this eminent web site: http://www.xs4all.nl/~borkhuis/vxworks/mbx860.html#1.2 Luckily I had the libPPCEC603gnuvx.a archive with the missing ftllite.o in it. I extracted it and added the file to my libPPC604gnuvx.a and vxWorks compiled again! I have made several tests too and TFFS seems to behave as expected! regards, Johan Bostedt --------------------------- Newsgroups: comp.os.vxworks Subject: CPX8216 Date: 23 Aug 2001 10:32:41 -0700 From: jdas@alphion.com (JD) Organization: http://groups.google.com/ Message-ID: <27750684.0108230932.11eab69e@posting.google.com> Does anyone have experience using CPX8216 chassis with MCP750ha system slot card with MCPN7xxx in non system slots in running VxWorks 5.4? My specific question is what other software do I need for doing domain management as well as SM based back plane communication across the domain? If possible we do not want to migrate to AE. Any help will be apprecited. Thanks. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Newbie: inhibit / trap VME bus error? Date: Thu, 23 Aug 2001 11:37:44 -0600 From: "Mark A. Naivar" Organization: LANL Message-ID: <3B853F68.C8F3E2B8@alvie-mail.lanl.gov> References: <48743300.0108220923.1f16c6b0@posting.google.com> Reply-To: mnaivar@lanl.gov Dave wrote: > Hello, > > I have a system where another card may cause random VME bus errors > when my card (running VxWorks) attempts to read its shared memory. > This produces a bus error on my card and the tasks associated with > this bus error are suspended. > > Is it possible to have a bus error without these tasks being suspended > but still be warned of this error? > If so, how? > > Thanks in advance! > > Dave Dave, I was able to make my own version of VxMemProbe that's a lot faster and recovers from bus errors. Here's what worked for me: jmp_buf myEnv; void mySigHandler(int sigNum) { longjmp(myEnv, ERROR); } // Call this once before calling the Probe() routine below. void myByteProbeInit() { struct sigaction newSig, oldSig; newSig.sa_handler = mySigHandler; newSig.sa_mask = 0; newSig.sa_flags = 0; sigaction(SIGBUS, &newSig, &oldSig); sigaction(SIGSEGV, &newSig, &oldSig); } STATUS myByteProbe(char *addr, int mode, char *data) { if(setjmp(myEnv) == 0) { if(mode == READ) *data = *addr; else *addr = *data; } else return (ERROR); } Hope this helps. - -Mark --------------------------- Newsgroups: comp.os.vxworks Subject: NTP 4.1 for vxworks 5.4 Date: Thu, 23 Aug 2001 14:06:35 -0400 From: "Hilton" Organization: Nortel Message-ID: <9m3gna$r9e$1@bcarh8ab.ca.nortel.com> Hi, I've been able to compile ntp 4.1 for vxworks 5.4 for the i386 simulator and when loading ntpd, I encounter an unresolved symbol for adjtime(). This function isn't available in vxworks, but the vxworks clock_settime() function is good enough for our use. The porting notes written by Casey Crellin seem outdated for the distribution, but it indicates that we have the choice of using either adjtime() or clock_settime() by modifying the ntp_machine.h file. The comments in that file (which has been updated since his port) indicate that everything in that file has been ported to autoconf (except for SYS_WINNT). Although configure is specified to use clock_settime, (ie. "#define USE_CLOCK_SETTIME 1" in config.h), ntpd doesn't appear to do anything with it and still tries to use adjtime(). 1. Is this right? or is there something that needs to be done to force ntpd to use clock_settime() ? After looking through the source, there doesn't appear to be any code to account for this. 2. We're using a software clock. Is the usrTime library with the adjtime() function for vxworks only good for hardware clocks? It seems like overkill to adapt it for a s/w clock. 3. If the v4.1 code indeed doesn't properly account for this, has anyone made changes to make it work? Thanks, Hilton --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Multicast Date: Thu, 23 Aug 2001 14:42:23 -0500 From: John McClenny Organization: Posted via Supernews, http://www.supernews.com Message-ID: References: <3B85222D.5ABF1466@fnal.gov> In article <3B85222D.5ABF1466@fnal.gov>, dinker@fnal.gov says... > Hello, > > We need to support multicast. Should board selection be constrained by > this in anyway? We are trying to choose between MVME230x and MVME5100 > boards. > > dinker > Any modern Ethernet chip has multicast and multicast filtering support. I don't know what chips are on these boards, but you shouldn't have a problem with any current design. --------------------------- Newsgroups: comp.os.vxworks Subject: WEHT protection domains? Date: Fri, 24 Aug 2001 01:07:03 GMT From: rabbitispoor@mindspring.com (Timothy Miller) Organization: MindSpring Enterprises Message-ID: <3b85a870.88609450@news.mindspring.com> Reply-To: rabbitispoor@mindspring.com Has anyone used this much ballyhooed feature? --------------------------- Newsgroups: comp.os.vxworks Subject: Re: CS8900 driver - multiple units Date: 23 Aug 2001 19:28:07 -0700 From: shimon@san.rr.com (Shimon Gersten) Organization: http://groups.google.com/ Message-ID: References: "Shimon Gersten" wrote in message news:... > The CS8900 driver distributed with Tornado 2 (if_cs.c) for x86 CPU claims to > only support a single unit. > Is there a way to support more than one CS8900 device? VxWorks CS8900 drivers in WindRiver distribution and Cirrus web site only support one device. I modified the latest (8 May 01) BSD 4.4 style driver from Cirrus site to support multiple CS8900 devices. It is tested with 2 devices and in 16 bit memory mode. --------------------------- Newsgroups: comp.os.vxworks Subject: WRS and their DHCP server. Date: 23 Aug 2001 21:04:53 -0700 From: chrismooremobile@yahoo.com (Chris Moore) Organization: http://groups.google.com/ Message-ID: Hi, Heres a question. 1) Normally the DHCP server gets inited in usrNetwork.c. Now I've cut it out and got it running from a CLI. This works to a point. The thing binds correctly and reads its tables correctlt and the task spawns correctly but it does not respond to any DHCP queries. Now, if I shift it back into the usrNetwork.c it works fine. I then went through it bit by bit and there is no problem (hell, its only a few lines of code) but it just does not work. If I init it down in the usrNetwork.c and then SPAWN it from my CLI, it works fine. Any ideas at all?!?! I can't trace the damn stuff cause of no code. Also, there is no function to DELETE a lease entry, only add them... Chris --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Several Questions... Date: Fri, 24 Aug 2001 00:19:51 -0700 From: "Denis" Organization: University of California, Los Angeles Message-ID: <9m4v3e$117$1@persian.noc.ucla.edu> References: Albert, "Albert H Chen" wrote in message news:AB2h7.4016$64.2190486@newsrump.sjc.telocity.net... > 3. Why can't you do semGive() of mutal-exclusion semaphores? you can't, accept you have to do it from the same task that took it. binary semaphores can be given from any task. denis --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Newbie: inhibit / trap VME bus error? Date: Fri, 24 Aug 2001 00:30:08 -0700 From: "Denis" Organization: University of California, Los Angeles Message-ID: <9m4vmn$18k$1@persian.noc.ucla.edu> References: <48743300.0108220923.1f16c6b0@posting.google.com> <3B853F68.C8F3E2B8@alvie-mail.lanl.gov> Mark, "Mark A. Naivar" wrote in message news:3B853F68.C8F3E2B8@alvie-mail.lanl.gov... > Dave wrote: > [snip] > STATUS > myByteProbe(char *addr, int mode, char *data) > { > if(setjmp(myEnv) == 0) > { > if(mode == READ) *data = *addr; > else *addr = *data; > } > else > return (ERROR); > } so, why does your solution work? Is it supposed to always work, or is it only because it is a compiler/os combination ? perhaps it does not work in certain cases? here is why i am asking: if i remember correctly, longjmp is only guaranteed to work if the function which contains the setjmp() haven't terminated already. in your case, it looks like it has. thanks, denis --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Several Questions... Date: Fri, 24 Aug 2001 01:17:25 -0700 From: "Albert H Chen" Message-ID: References: <3B84CEE4.B407DB0B@ks.ericsson.se> Oops... I meant: 3. Why can't you do semGive() of mutal-exclusion semaphores in an ISR? Thanks for the answers... Albert "Urban Lindberg" wrote in message news:3B84CEE4.B407DB0B@ks.ericsson.se... > 1. Comparing the address of sysInit and RAM_LOW_ADRS should tell you if > its boot_rom or not. VxWorks is loaded at RAM_LOW_ADRS, while the boot > rom code is copied to RAM_HIGH_ADRS. Another way is to insert a variable > which is initialized to something (TRUE, FALSE) in a common file and in > any of the bootrom specific files set it to another value. > > 2. Normally, yes. But exceptions will preempt any task. Don't use it > unless you realy need to. > > 3. Huh, what do you mean? You have to do a semGive for all semTake if > you want to release the resource (unless you use semMGiveForce, which > should only be used as a debugging aid). Se docu for semMLib. > > HTH > > Urban > > Albert H Chen wrote: > > > > 1. There is a certain portion of code in the bsp that belongs both in the > > bootrom and also in the working image. > > How do I do something along the lines of following: > > > > void func(){ > > if(boot_rom) printf("In bootrom\n"); > > else printf("Not in bootrom\n"); > > } > > > > 2. Does intLock prevent task preemption? > > > > 3. Why can't you do semGive() of mutal-exclusion semaphores? > > > > Thanks, > > > > Albert --------------------------- Newsgroups: comp.os.vxworks Subject: PPC6750 Date: Fri, 24 Aug 2001 14:15:49 +0530 From: Deepa Organization: Lucent Technologies, Columbus, Ohio Message-ID: <3B86143D.2D529DA6@lucent.com> Hi, When we build our code with the BSP for the PPC 6750(rel 2) board using VxWorks Ver 5.4, Tornado Ver2, we get the following error. The error message that is displayed by the Wind River project tool is: 'list must have an even number of elements' This error is traced back to the prj_vxWorks.tcl file. Could any one please tell us whether you have a patch for the same? Thanks Deepa. --------------------------- Newsgroups: comp.os.vxworks Subject: What is the module format of vxworks on earth? Date: Fri, 24 Aug 2001 17:48:55 +0800 From: "survey liu" Organization: Bentium Ltd. (CN99) Message-ID: <9m57r9$mr6$1@mail.cn99.com> Hi,all I have many problems on this OS: 1. While reading the manual on function ld(), you would see the following descriptions: - --This command loads an object module from a file or from standard input. The object module - --must be in UNIX a.out format. External references in the module are resolved during loading. - --The syms parameter determines how symbols are loaded; possible values are: - -- 0 - Add global symbols to the system symbol table. - -- 1 - Add global and local symbols to the system symbol table. - -- -1 - Add no symbols to the system symbol table. " But, how could vxWorks realize the shared lib management while using format a.out? It seems vxWorks support ELF, too. 2. In additional, there a lot of utilities in /host/***/bin like aoutToBinDec,elfHex,elfBsd ... So, do I really need these utilities? If yes, how could I select the format of my output files using TonadoII? Any differnce? 3. Also, I want to know how the vxWorks manages its symbol table. Where is the array standTbl located? I can find out it by "lkup". 4. Since the compiler and debugger of vxWorks are all base on GNU toolchain, could the programs running under vxWorks cause a "core dump"? If yes, while would the core file be? THT - -- Best regards, Survey Liu 4/F Wearnes Technology Bldg. No.10 Kefa Rd.,Science & Industry Park Nanshan Shenzhen 518057,PRC Tel: +86-755-6635333-4300 Fax: +86-755-6635550 Email: survey.liu@utstar.com Web: http://www.utstar.com --------------------------- End of New-News digest ********************** From vxwexplo-errs@csg.lbl.gov Fri Aug 24 07:07:33 2001 From: "Vamsi Krishna K" Date: Fri Aug 24 07:07:35 PDT 2001 Subject: Netowrk download - Atlas board hi, I am trying to build bootrom.hex for Atlas board using build - bootrom option in Tornado2 IDE. I want to download the VxWorks image using NE2000 add-on card. I am not able to build bootrom.hex with INCLUDE_ENE, INCLUDE_BSD #defined in config.h. I have undefined INCLUDE_END. What exactly must be done to build the image for atlas board. Please give any suggestions. thanks Vamsi. From vxwexplo-errs@csg.lbl.gov Fri Aug 24 07:42:11 2001 From: John Moore Date: Fri Aug 24 07:42:13 PDT 2001 Subject: Re: C , C++ and vxWorks tasks Not a lot to go on... Did the semaphore get created ? ( semBCreate, semCCreate ...) Check the return value of the semTake in your while loop. It is probably an error and the task is just spinning (hence the READY state). Also, I think you'll have to pass the task a pointer to your myclass object to be able call myclass methods. jmoore VxWorks > how to work with vxworks tasks in a C++ scenario ? say > I want to have a task, spawned from a member function, > and this task has to wait on a semaphore sent from C > function of a different file, and then sequence thru > to do some activities calling some member functions ? > i see the task to be in READY state when I spawn it > from the member function. here is my design. > > extern SEM_ID mysemaphore > void myclass::memberfunction(void) > { > > spawn mytask > > } > > void mytask(void) > { > > while(1) { > wait on semaphore ( sent from another c file ) > > call some myclass member functions to do something > > } > > } From vxwexplo-errs@csg.lbl.gov Sat Aug 25 04:03:40 2001 From: Vxworks Exploder Date: Sat Aug 25 04:03:42 PDT 2001 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Sat Aug 25 04:03:10 PDT 2001 Subject: Re: Several Questions... Subject: Re: Debugging multiple tasks Subject: royalties Subject: Re: What is the module format of vxworks on earth? Subject: -frepo vs. -fno-implicit-templates for code bloat in VxWorks5.4 ?? Subject: Re: MPC857T : UT bit set in PDPAR register Subject: Re: GDB problems with Tornado 2 for Mips (libR4000gnuvx.a) Subject: help Subject: Re: memory allocation problems Subject: Re: PPC6750 Subject: Re: GDB problems with Tornado 2 for Mips (libR4000gnuvx.a) Subject: Re: Several Questions... Subject: Re: What is the module format of vxworks on earth? Subject: How to format the disk Subject: Re: royalties Subject: How to format the disk Subject: Build demo gizmo.out hang on Windows Subject: Re: Telnet sessions Subject: Device Drivers for MPC8260 UART Subject: Re: -frepo vs. -fno-implicit-templates for code bloat in VxWorks5.4 ?? Subject: Re: Device Drivers for MPC8260 UART Subject: Re: logInit question ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: Several Questions... Date: Fri, 24 Aug 2001 14:01:50 +0200 From: Urban Lindberg Organization: Ericsson Message-ID: <3B86422E.BB8A04B0@ks.ericsson.se> References: <3B84CEE4.B407DB0B@ks.ericsson.se> That's because mutex semaphores are assosiated with the task that has taken the semaphore. ISR's are not guaranteed to execute in the task that has taken the semaphore. On some platforms they even have their own stack. And of course if it was allowed you should not use it, the task that took the semaphore does it to protect a critical region of code, if an ISR could give that semaphore, that region would be unprotected and might cause undesired behaviour. Urban Albert H Chen wrote: > > Oops... I meant: > > 3. Why can't you do semGive() of mutal-exclusion semaphores in an ISR? > > Thanks for the answers... > > Albert > [Snip] --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Debugging multiple tasks Date: Fri, 24 Aug 2001 14:33:19 +0200 From: "Werner Schiendl" Message-ID: <3b864a47$0$25440$6e365a64@newsreader02.highway.telekom.at> References: <3b81713e$0$21924$6e365a64@newsreader02.highway.telekom.at> Hi, I think the difference is basically that the method I mentioned relies on the debugger to set the breakpoint. The b command on the other side works without the debugger (WDB). In case the debugger is downloaded to the target, you can in both situations use the debugger to do source level debugging. regards werner --------------------------- Newsgroups: comp.os.vxworks Subject: royalties Date: Fri, 24 Aug 2001 09:38:55 -0400 From: david lindauer Message-ID: <3B8658EE.53C1A14C@notifier-is.net> hi, when we bought VXWorks, we had an agreement with WindRiver that the royalties would be around $10 per product sold. This seemed reasonable. Unfortunately the person who made that agreement has moved on; now wind river wants to charge us $500 per product sold. Needless to say we cannot afford this... the hardware for our product only costs about that and we'd have to do tremendous markup and price ourselves out of the market. Is this really the going price for royalties? Do people really pay that much? David --------------------------- Newsgroups: comp.os.vxworks Subject: Re: What is the module format of vxworks on earth? Date: Fri, 24 Aug 2001 15:25:15 +0200 From: "Werner Schiendl" Message-ID: <3b865669$0$18118$6e365a64@newsreader02.highway.telekom.at> References: <9m57r9$mr6$1@mail.cn99.com> Hi, "survey liu" wrote in message news:9m57r9$mr6$1@mail.cn99.com... > Hi,all > > I have many problems on this OS: > > 1. While reading the manual on function ld(), you would see the following > descriptions: > --This command loads an object module from a file or from standard input. > The object module > > --must be in UNIX a.out format. External references in the module are > resolved during loading. > > --The syms parameter determines how symbols are loaded; possible values are: > > -- 0 - Add global symbols to the system symbol table. > -- 1 - Add global and local symbols to the system symbol table. > -- -1 - Add no symbols to the system symbol table. " > > But, how could vxWorks realize the shared lib management while using format > a.out? It seems vxWorks support ELF, too. > The ld function loads an object file of a file format supported by the target and links it with the kernel. No funny shared libs management involved there, it is just linked 'statically' while loading. This works perfect with a.out, too. > 2. In additional, there a lot of utilities in /host/***/bin like > aoutToBinDec,elfHex,elfBsd ... So, do I really need these utilities? If yes, > how could I select the format of my output files using TonadoII? Any > differnce? > The output format is predefined for all VxWorks architectures. Depending on your CPU architecture, you are bound to use that format. (Unless you are willing to go through a lot of haszles). The output format used for your particular output format should be documented. (VxWorks User's Guide - Appendices) > 3. Also, I want to know how the vxWorks manages its symbol table. Where is > the array standTbl located? I can find out it by "lkup". > If you build a image with symbol table built in, a symbol table is built into the image and loaded with the image. The symbol table is set up, when VxWorks itself has booted. Don't know about the mechanice, though. However, you do not necessarily need an symbol table, because the Tornado development environment is able to do the linking based on the symbol information located on your host machine. > 4. Since the compiler and debugger of vxWorks are all base on GNU toolchain, > could the programs running under vxWorks cause a "core dump"? If yes, while > would the core file be? > Isn't core dumping a linux (maybe unix) idiom? VxWorks handles exceptions in it's own way. See excLib in the manuals for more details. You can trap CPU exceptions caused by your programm and do what is to your desire. hth werner --------------------------- Newsgroups: comp.os.vxworks Subject: -frepo vs. -fno-implicit-templates for code bloat in VxWorks5.4 ?? Date: Fri, 24 Aug 2001 14:00:00 GMT From: shorton3@yahoo.com (Stephen & Dona Horton) Organization: Excite@Home - The Leader in Broadband http://home.com/faster Message-ID: <3b875869.5674259@news.plano1.tx.home.com> Hello Folks, I'm currently working on a project with very size/space sensitive requirements. I'm trying to minimize the amount of code bloat caused by using heavy STL templates (I'm using GCC 2.7.2 in VxWorks 5.4). I've read several threads in the VxExploder archives regarding this issue, but the results I've had in attacking this problem have not been good. My software consists of the VxWorks image file (without templates) and several dozen application module .out files (with templates). After applying both the -frepo and the -fno-implicit-templates compiler options, I noticed that the resulting .out file sizes are the same. Is this expected? Note that I've applied several compiler-related patches from WindRiver's web site (including the patch regarding -frepo). Several questions: 1. Does compiling .out files using -frepo mean that the .rpo files are linked into the .out file? 2. Assuming that the .rpo files/template instances are linked into the .out file, why should the 2 .out files above be the same size? Note that in the case of the no-implicit-templates file, I have included my file containing all 'template class xxxx' instantiations in a separate .out file that I'm planning on loading before any other .out files. It seems that the .out file compiled and linked with .frepo should be larger to contain the template instantiations. 3. When I load the .out files, they have unresolved symbols for all templates. Should I assume that the .rpo files must be loaded separately then? Or is something broken in my attempt at this? Switching directions somewhat... I was hoping to get the -frepo option working, because I attempted the no-implicit-templates solution first, but I was having trouble getting the syntax correct for all the template instantiations (especially for the STL templates...multimap, basic_string, etc. - I guess because I'm using templates as parameters to templates sometimes - yuck!). I've read the section in Mumit's STL newbies guide, and I plan on looking at the ACE source code today (I noticed that they appear to be using this technique). But: 4. Does anyone have any concrete, more complicated examples of how the STL template instantiations should be declared? 5. Is it okay to put all template instantiations into one .out file and load it first? Is there a more advisable approach to getting this technique to work? Any information on this topic will be greatly appreciated, Stephen --------------------------- Newsgroups: comp.os.vxworks Subject: Re: MPC857T : UT bit set in PDPAR register Date: 24 Aug 2001 08:20:52 -0700 From: google@gkec.freeserve.co.uk (Graham) Organization: http://groups.google.com/ Message-ID: <6c0ce808.0108240720.5065837d@posting.google.com> References: <9kngm3$fvk$1@news2.kornet.net> Set up the ATM/Utopia first setting up PDPAR as appropriate then set up the FEC. "Ogre" wrote in message news:<9kngm3$fvk$1@news2.kornet.net>... > UT bit in PDPAR register determines whether the parameter RAM¡¯s page 4 > (SCC4) operates in serial or UTOPIA mode. > I want to use MPC857T in simultaneous MII and UTOPIA operation (UTOPIA muxed > mode). > > My source is following as .. > > -------------------------------------- > sysFecEnetEnable() > { > 01 : /* mask IRQ7 off, as it is shared with MII_TX_CLK */ > 02 : *SIMASK &= ~SIMASK_IRM7 ; > > 03 : /* also clear any pending interrupt */ > 04 : *SIPEND |= SIPEND_IRQ7 ; > > 05 : *SDCR |= SDCR_FAID_BR6 ; > > 06 : *PDPAR = 0xC000 ; or 06 : *PDPAR = 0x4000 ; > 07 : *PDPAR = 0x0080 ; > 08 : *PDDIR = 0x0080 ; > 09 : *UTMODE &= 0xFFFFFF7F ; > } > > motFecStart() > { > ..... > > 55 : sysFecEnetEnable() ; > 56 : MOT_FEC_PINMUX_ENABLE_REVD ; > ... > 70 : MOT_FEC_ETH_ENABLE_REVD ; > } > ----------------------------------------- > > 01 ~ 05 is default. > 06 set UT bit and ATM bit. or 06 set UT bit. > 07 set MDC for ethernet management clock. > 09 set muxed mode. > 56 set FEC_PINMUX bit > 70 set FEC_PINMUX and ETHER_EN bit. > > but, at 06, system crashed. > To use simultaneous MII and UTOPIA, MPC857T has UT = 1, FEC_PINMUX = 1, > SPLIT = 0(MUX) configuration. > but, crashed.. -_-; > > I hope this problem will be solved. > thanks any response. > > Ogre --------------------------- Newsgroups: comp.os.vxworks Subject: Re: GDB problems with Tornado 2 for Mips (libR4000gnuvx.a) Date: Fri, 24 Aug 2001 15:07:33 GMT From: "Nihad Alibegovic" Organization: Road Runner Message-ID: References: <__Af7.2724$Qh2.545742@typhoon.san.rr.com> Thank you for the piece of code Girish. It worked well for what I wanted to do. The tWdbTask exception is now gone. However, I noticed couple more things in Tornado 2 GDB that are causing me a lot of headaches: 1. When I load symbols in GDB and then try to attach to the task the break occurs in the task at the usual location. However, the code pane is showing the last piece of code compiled in that is known. In other words, instead of showing assembly for semQPut() which is where the break occured it puts me into the last .c file I linked into the image. So if I then try to single step the GDB crashes again. In Tornado 1 the GDB did the right thing by jumping to assembly of semQPut() being that it does not have C/C++ code for semQPut(). I found a workaround for this by switching the code pane into assembly view from straight C/C++ view. This then puts me into proper spot and I can step or setup my breakpoint and continue. 2. OK, so GDB goes of and gets register contents but it never goes and gets floating point registers on the processor. I have INCLUDE_FLOATING_POINT and INCLUDE_HW_FP enabled in the BSP and on the target when I do taskRegsShow() I get contents of floating point registers but I cannot get them through GDB. So far Tornado 2 has been less than stellar to say the least. It is actually impacting my development time instead of speeding things up. Any thoughts by anyobody on the above two? In further examination of header files I noticed that in $(WIND_BASE)/target/h/arch/mips there is a dbgXXX.h file which defines DBG_NO_SINGLE_STEP to 1???????? What is the deal with this. Is it possible that Wind River distributed the libR4000gnuvx.a with this kind of definition and that this library does not support single stepping on purpose????? "Girish V. Gulawani" wrote in message news:Q0Dg7.3740$j04.316204@nnrp.gol.com... > > there are infact 2 changes in the same file - > /target/config/comps/src/usrWdbTask.c > .1 > in function vxTaskRegsGet : comment out the > /* DO NOT REMOVE THIS CALL TO taskRegsGet (SPR #26387) */ > taskRegsGet (pContext->contextId,&dummy); > -- this one you comment out. > .2 > check for valid delete hook function. i dont know who changes it & makes it > invalid entry, (perhaps cache?). > _ _wdbTaskDeleteHook() > { > ..... > void(*hook()); > void(* myhook()); > hook= TCB->WdbExitHook; > myhook=wdbCtxExitNotifyHook; > if (hook!= myhook) > hook=myhook; > ..... > } > i hope this should work for you too. keep us posted here. > > > "Nihad Alibegovic" wrote in message > news:V5vg7.6338$Qh2.1774693@typhoon.san.rr.com... > > This could be the very same problem. Please get me some code to try. I > > will try what you put in below and then anything else you give me. We > have > > had terrible time with this so far and like yours it has now been 2 weeks > > without reply. I will also check out your SPR. Thank you very much for > > your help. > > > > "Girish V. Gulawani" wrote in message > > news:gS6g7.3713$j04.314817@nnrp.gol.com... > > > perhaps i have come across similar problem earlier. that was related to > > SPR > > > 26387 (unpublished). > > > the problem is in getting mips registers while attaching task through > gdb. > > & > > > have already communicated to wrs. but ofcourse no reply. > > > to this spr 26387 details are - > > > in the WRS distributed source file > > > /target/config/comps/src/usrWdbTask.c the function vxTaskRegsGet() > fails. > > > vxTaskRegGet () > > > { > > > TCB *tcb; > > > ... > > > tcb = taskTcb() > > > switch(regSetType) { > > > case WDB_REG_SET_IU: > > > { > > > REG_SET dummy; > > > taskRegsGet(context, &dummy); **** problem *** > > > *ppRegSet = (char*)tcb->regs; > > > break; > > > } ... > > > } > > > if you comment of the problem statement it works. then there is one more > > > place you have to change that is delete hook in the same file. however i > > > dont have the code right now. > > > if you feel it is the same problem do post here. i'll respond asap. > > > my debugger is working properly now. but still wrs support has not > replied > > > and its over 3/4 months. > > > > > > > > > "Nihad Alibegovic" wrote in message > > > news:__Af7.2724$Qh2.545742@typhoon.san.rr.com... > > > > Hello everyone, > > > > > > > > we ported our BSP from vxWorks 5.3 to vxWorks5.4 on Tornado 2. When > it > > > was > > > > all said and done everything works except GDB. Target attaches fine > we > > > can > > > > put break points but as soon as we try to single step we get tWdb > > > exception. > > > > It is worth noting that if I put a break point in from the host shell > > but > > > > single step in target resident shell things work well. So, I guess > this > > > is > > > > not just a GDB problem. Additionally, I have same problem when trying > > to > > > > attach to the task. It is as if the host and target memories are > > > completely > > > > out of sync and host is accessing memory it is not supposed to access > on > > > the > > > > target. We use Network to debug which means we are in task mode > > debugging > > > > only. > > > > > > > > The platform we run on is Mips 4310 CPU so we use libR4000gnuvx.a. > We > > > have > > > > all the WDB includes in our BSP that we need as certified by Wind > River. > > > > Wind River has helped very little in this issue in the last week. > > While, > > > at > > > > this does anyone get quick support from these guys. In 2 years > working > > > with > > > > vxWorks they have the slowest customer support of any company I ever > > > worked > > > > with. > > > > > > > > Nihad Alibegovic > > > > MTS3 > > > > nalibegovic@hns.com > > > > > > > > > > > > > > > > > > > > > > > > > > > > --------------------------- Newsgroups: comp.os.vxworks Subject: help Date: Fri, 24 Aug 2001 09:50:54 -0400 From: "Jing Ma" Message-ID: <3b865bb1.0@alognews.analogic.com> Hi, I have a problem with Wind Web server. I used some source code which from (Tornado\target\unsupported\http\tutorials) to create a Vxworks image with Wind web Server 1.0. it works fine. After we installed Wind Web sever 2.0 , it doesn't work. Any ideas would be appreciated. Thanks! jing --------------------------- Newsgroups: comp.os.vxworks Subject: Re: memory allocation problems Date: Fri, 24 Aug 2001 17:53:02 +0000 From: alekso Organization: ericsson Message-ID: <3B86947E.5030505@yahoo.com> References: <3B83B0A1.891C8E46@yahoo.com> <3b840c55$0$77221$e2e8da3@nntp.cts.com> >>I use PPC603 and in the beginning some partition must be created (C >>code). >>memPartCreate() returns NULL... What additional diagnostic could I get >>in order to find why? >>Is anything in addition to printErrno()? In manual for printErrno() is >>said, that >>error-status symbol table must be included in the system. How can I >>check it? >> > > Just how 'in the beginning' are you trying to create it ? You need to make > sure that memPartLibInit() in usrRoot() has at least been called. And in > your code, before doing the memPartCreate(), you might just add a memShow(0) > to see what memory looks like before you make your call. > > hth, > Dan Gold gold@ensemble.com 1. allocating memory pool from system partition by memPartAlloc 2. Using returned to memory block we got from previous step do memPartCreate. memShow shows us that there is enough memory for the system. Now we found out, that system not allows us create partition with the size about 6 kb. Is any limitation in VxWorks for patition size specified in memPartCreate function call? - - I cannot find any reference in the manual.. BR Aleks --------------------------- Newsgroups: comp.os.vxworks Subject: Re: PPC6750 Date: 24 Aug 2001 09:11:39 -0700 From: john_94501@yahoo.com (John) Organization: http://groups.google.com/ Message-ID: <488e459a.0108240811.38188e0@posting.google.com> References: <3B86143D.2D529DA6@lucent.com> Hello, Are you compiling with any unusual flags on your make command line? If I remember correctly, --always-print-commands will cause this problem to appear (because the TCL script makes use make too, and it is not expecting this additional output). HTH, John... Deepa wrote in message news:<3B86143D.2D529DA6@lucent.com>... > Hi, > When we build our code with the BSP for the PPC 6750(rel 2) board using > VxWorks Ver 5.4, Tornado Ver2, we get the following error. > The error message that is displayed by the Wind River project > tool is: > > 'list must have an even number of elements' > > This error is traced back to the prj_vxWorks.tcl file. > > Could any one please tell us whether you have a patch for the same? > > Thanks > Deepa. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: GDB problems with Tornado 2 for Mips (libR4000gnuvx.a) Date: Fri, 24 Aug 2001 16:13:47 GMT From: "Girish V. Gulawani" Organization: Global Online Japan // Exodus Communications KK. Message-ID: <%2vh7.3992$j04.319360@nnrp.gol.com> References: <__Af7.2724$Qh2.545742@typhoon.san.rr.com> > 1. When I load symbols in GDB and then try to attach to the task the break > occurs in the task at the usual location. However, the code pane is showing > the last piece of code compiled in that is known. In other words, instead > of showing assembly for semQPut() which is where the break occured it puts > me into the last .c file I linked into the image. So if I then try to > single step the GDB crashes again. In Tornado 1 the GDB did the right thing > by jumping to assembly of semQPut() being that it does not have C/C++ code > for semQPut(). I found a workaround for this by switching the code pane > into assembly view from straight C/C++ view. This then puts me into proper > spot and I can step or setup my breakpoint and continue. i am not sure though but i faced similar problem. it was due to mimatch in the core file. you specify this core file in the target server configuration. should there be mismatch the debug symbols and source will be differ. > 2. OK, so GDB goes of and gets register contents but it never goes and gets > floating point registers on the processor. I have INCLUDE_FLOATING_POINT > and INCLUDE_HW_FP enabled in the BSP and on the target when I do > taskRegsShow() I get contents of floating point registers but I cannot get > them through GDB. assuming your flavour of r4k has h/w floating point then you included options are okay. then check if you are compiling with -soft-float option?? also are you spawning the task with FP_TASK option set?? > In further examination of header files I noticed that in > $(WIND_BASE)/target/h/arch/mips there is a dbgXXX.h file which defines > DBG_NO_SINGLE_STEP to 1???????? > > What is the deal with this. Is it possible that Wind River distributed the > libR4000gnuvx.a with this kind of definition and that this library does not > support single stepping on purpose????? dont remember seeing such option. also i have been using the gdb with mips r4k target without any problem. so i dont think wrs would stop us from doing single step...i'd guess. anyway could you post here the those 2-3 magic lines of DBG_NO...??? --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Several Questions... Date: 24 Aug 2001 09:16:05 -0700 From: john_94501@yahoo.com (John) Organization: http://groups.google.com/ Message-ID: <488e459a.0108240816.5a82884f@posting.google.com> References: <3B84CEE4.B407DB0B@ks.ericsson.se> Hello, > 3. Why can't you do semGive() of mutal-exclusion semaphores in an ISR? This should be obvious if you've understood the key features of a mutex semaphore (read the programmer's guide section again). Let me explain briefly... A mutex semaphore is used to protect resources. As such, it is always taken first, and then given back (think of it as token that allows the holder to do some work). The implementation ensures that only the current holder can give it back. Now, since an ISR can never be the holder (semTake() is not allowed in an ISR), there is no way that it would make sense to give back the token. HTH, John... > Thanks for the answers... > > Albert > > "Urban Lindberg" wrote in message > news:3B84CEE4.B407DB0B@ks.ericsson.se... > > 1. Comparing the address of sysInit and RAM_LOW_ADRS should tell you if > > its boot_rom or not. VxWorks is loaded at RAM_LOW_ADRS, while the boot > > rom code is copied to RAM_HIGH_ADRS. Another way is to insert a variable > > which is initialized to something (TRUE, FALSE) in a common file and in > > any of the bootrom specific files set it to another value. > > > > 2. Normally, yes. But exceptions will preempt any task. Don't use it > > unless you realy need to. > > > > 3. Huh, what do you mean? You have to do a semGive for all semTake if > > you want to release the resource (unless you use semMGiveForce, which > > should only be used as a debugging aid). Se docu for semMLib. > > > > HTH > > > > Urban > > > > Albert H Chen wrote: > > > > > > 1. There is a certain portion of code in the bsp that belongs both in > the > > > bootrom and also in the working image. > > > How do I do something along the lines of following: > > > > > > void func(){ > > > if(boot_rom) printf("In bootrom\n"); > > > else printf("Not in bootrom\n"); > > > } > > > > > > 2. Does intLock prevent task preemption? > > > > > > 3. Why can't you do semGive() of mutal-exclusion semaphores? > > > > > > Thanks, > > > > > > Albert --------------------------- Newsgroups: comp.os.vxworks Subject: Re: What is the module format of vxworks on earth? Date: 24 Aug 2001 09:26:55 -0700 From: john_94501@yahoo.com (John) Organization: http://groups.google.com/ Message-ID: <488e459a.0108240826.cdf5995@posting.google.com> References: <9m57r9$mr6$1@mail.cn99.com> Hello, 1) Historically, VxWorks used a.out. Today things are a little more complex since some architectures defined other formats as part of their ABIs. So, there's a list that looks something like this (from memory, so it might be wrong): 68K, x86: a.out ARM & i960: COFF MIPS & PPC: ELF The newer VxWorks AE uses ELF exclusively. As for "shared library management," VxWorks does not support shared libraries in the Unix/Windows sense. Instead, .o files may be loaded and unloaded at any time, and their code shared. For a 5.x system, all the code lives in a single address space, so it is inherently shared. For VxWorks AE, there are shared library domains to allow sharing of code between isolated application spaces. 2) You cannot select the output format easily; each architecture supports one format. The tools are there in general to convert from the OMF to raw binary or hex files so that PROM programmers can read the information. 3) You don't need to know anything about the internals of the symbol table to make use of it. In fact, making use of the data structures directly could give you some serious problems if you upgrade to newer versions of the OS in the future. Instead, you should use the symLib API to query and manipulate the symbol table. 4) No, VxWorks will not generate core dump files. Core dump files are nothing to do with the GNU compiler; they are a feature of Unix, and very much oriented towards a process model operating system. HTH, John... "survey liu" wrote in message news:<9m57r9$mr6$1@mail.cn99.com>... > Hi,all > > I have many problems on this OS: > > 1. While reading the manual on function ld(), you would see the following > descriptions: > --This command loads an object module from a file or from standard input. > The object module > > --must be in UNIX a.out format. External references in the module are > resolved during loading. > > --The syms parameter determines how symbols are loaded; possible values are: > > -- 0 - Add global symbols to the system symbol table. > -- 1 - Add global and local symbols to the system symbol table. > -- -1 - Add no symbols to the system symbol table. " > > But, how could vxWorks realize the shared lib management while using format > a.out? It seems vxWorks support ELF, too. > > 2. In additional, there a lot of utilities in /host/***/bin like > aoutToBinDec,elfHex,elfBsd ... So, do I really need these utilities? If yes, > how could I select the format of my output files using TonadoII? Any > differnce? > > 3. Also, I want to know how the vxWorks manages its symbol table. Where is > the array standTbl located? I can find out it by "lkup". > > 4. Since the compiler and debugger of vxWorks are all base on GNU toolchain, > could the programs running under vxWorks cause a "core dump"? If yes, while > would the core file be? > > THT > > -- > > Best regards, > Survey Liu > > 4/F Wearnes Technology Bldg. No.10 > Kefa Rd.,Science & Industry Park > Nanshan Shenzhen 518057,PRC > Tel: +86-755-6635333-4300 > Fax: +86-755-6635550 > Email: survey.liu@utstar.com > Web: http://www.utstar.com --------------------------- Newsgroups: comp.os.vxworks Subject: How to format the disk Date: Fri, 24 Aug 2001 11:36:34 -0500 From: "Shesha Bhushan" Organization: Arizona State University Message-ID: <9m6704$q2b$1@news.asu.edu> Hi Gurus, I have connected a scsi hardisk to a Single board Computer that runs vxWorks OS. The hard disk was previously formated as NTFS. I want to format this hard disk with dosFs. How to format this previously formated disk Thanks Shesha Bhushan --------------------------- Newsgroups: comp.os.vxworks Subject: Re: royalties Date: Fri, 24 Aug 2001 13:40:49 -0500 From: Andrew Johnson Organization: APS, Argonne National Laboratory Message-ID: <3B869FB1.8717B0A0@aps.anl.gov> References: <3B8658EE.53C1A14C@notifier-is.net> david lindauer wrote: > > when we bought VXWorks, we had an agreement with WindRiver that the > royalties would be around $10 per product sold. This seemed > reasonable. Unfortunately the person who made that agreement has moved > on; now wind river wants to charge us $500 per product sold. Needless > to say we cannot afford this... the hardware for our product only costs > about that and we'd have to do tremendous markup and price ourselves out > of the market. Is this really the going price for royalties? Depends on your volume. You're actually buying a target license for each product you sell. In small quantities, they can and do charge that kind of price. If you buy licenses in bulk, the price goes down as the quantity goes up. $10 is on the extremely cheap end though, you'd need some pretty high volumes to get that price I guess (but I've never bought any in significant quantities so don't believe what I write). A rather unfortunate and expensive misunderstanding if you're any significant way into the development process... - - Andrew - -- The world is such a cheerful place when viewed from upside-down It makes a rise of every fall, a smile of every frown --------------------------- Newsgroups: comp.os.vxworks Subject: How to format the disk Date: Fri, 24 Aug 2001 11:39:34 -0500 From: "Shesha Bhushan" Organization: Arizona State University Message-ID: <9m675o$q2o$1@news.asu.edu> Hi Gurus, I have connected a scsi hardisk to a Single board Computer that runs vxWorks OS. The hard disk was previously formated as NTFS. I want to format this hard disk with dosFs. How to format this previously formated disk. Tx a Bunch Shesha Bhushan --------------------------- Newsgroups: comp.os.vxworks Subject: Build demo gizmo.out hang on Windows Date: Fri, 24 Aug 2001 18:44:00 GMT From: Hong Hsu Message-ID: <3B86A116.1EEB961B@bellatlantic.net> Hi, I wonder this is right place to get a help for Tornado Prototyper 2.0 for Windows. If not, would you please point out reference for that? Here is my situation. The installation of Tornado Prototyper on my PC was successful. After generating the cobble.o, Building demo gizmo.out hang in Build Output window with following message: 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_M ULTI_THREAD -D_REENTRANT -I. -ID:\TORNADO\target\h -DCPU=SIMNT -c D:\Tornado\target\proj\gizmo\cobble.c vxrm D:\Tornado\target\proj\gizmo\prjObjs.lst In Output Window there is no nmsimpc statement. (I use standalone PC with Intel Pentium 4 processor and try to run the gizmo.out on simulator). I also notice that Winoldap.exe is running at this point and never goes away even the Wtxregd.exe is killed. What is Winoldap.exe doing and how to quit it? Did I miss something here in order to compile, link and run demo gizmo? Many thanks, 8/24 --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Telnet sessions Date: Fri, 24 Aug 2001 13:43:45 -0600 From: Les Hammer Organization: Agilent Technologies Message-ID: <3B86AE71.4F0A9AC0@agilent.com> References: <3B84427C.57B71C13@agilent.com> <488e459a.0108222201.e663389@posting.google.com> John wrote: > Here's a little code fragment for listening for connections on a > specified port, no warranty of course, use at your own risk, etc, etc: Thanks John. I've added some send() and recv() code in the middle. It appears to be working. Les --------------------------- Newsgroups: comp.os.vxworks Subject: Device Drivers for MPC8260 UART Date: Fri, 24 Aug 2001 15:57:42 -0400 From: "j" Organization: UUNET Canada News Reader Service Message-ID: <2myh7.15838$Z2.192063@nnrp1.uunet.ca> I am gonna write device drivers for MPC8260 using VxWorks to program SCC as a UART. Can anybody help me by providing some inforamtion or websites where I can find information and/or examples. john81s@yahoo.com Thanks. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: -frepo vs. -fno-implicit-templates for code bloat in VxWorks5.4 ?? Date: 24 Aug 2001 13:21:49 -0700 From: shorton3@yahoo.com (Stephen Horton) Organization: http://groups.google.com/ Message-ID: <63de7ccb.0108241221.1ceb3672@posting.google.com> References: <3b875869.5674259@news.plano1.tx.home.com> shorton3@yahoo.com (Stephen & Dona Horton) wrote in message news:<3b875869.5674259@news.plano1.tx.home.com>... > Hello Folks, > > I'm currently working on a project with very size/space sensitive > requirements. I'm trying to minimize the amount of code bloat caused > by using heavy STL templates (I'm using GCC 2.7.2 in VxWorks 5.4). > I've read several threads in the VxExploder archives regarding this > issue, but the results I've had in attacking this problem have not > been good. > > My software consists of the VxWorks image file (without templates) and > several dozen application module .out files (with templates). After > applying both the -frepo and the -fno-implicit-templates compiler > options, I noticed that the resulting .out file sizes are the same. Is > this expected? Note that I've applied several compiler-related patches > from WindRiver's web site (including the patch regarding -frepo). > Several questions: > > 1. Does compiling .out files using -frepo mean that the .rpo files are > linked into the .out file? > 2. Assuming that the .rpo files/template instances are linked into the > .out file, why should the 2 .out files above be the same size? Note > that in the case of the no-implicit-templates file, I have included my > file containing all 'template class xxxx' instantiations in a separate > .out file that I'm planning on loading before any other .out files. It > seems that the .out file compiled and linked with .frepo should be > larger to contain the template instantiations. > 3. When I load the .out files, they have unresolved symbols for all > templates. Should I assume that the .rpo files must be loaded > separately then? Or is something broken in my attempt at this? > > Switching directions somewhat... I was hoping to get the -frepo option > working, because I attempted the no-implicit-templates solution first, > but I was having trouble getting the syntax correct for all the > template instantiations (especially for the STL templates...multimap, > basic_string, etc. - I guess because I'm using templates as parameters > to templates sometimes - yuck!). I've read the section in Mumit's STL > newbies guide, and I plan on looking at the ACE source code today (I > noticed that they appear to be using this technique). But: > > 4. Does anyone have any concrete, more complicated examples of how the > STL template instantiations should be declared? > 5. Is it okay to put all template instantiations into one .out file > and load it first? Is there a more advisable approach to getting this > technique to work? > > Any information on this topic will be greatly appreciated, > Stephen Upon further investigation of the VxWorks compile process (and re-reading Mumit's STL newbies guide (http://www.xraylith.wisc.edu/~khan/software/stl/STL.newbie.html), I noticed that the 'collect' phase of the linker is not 'noticably' running. Could this be my problem? Is the collect functionality built-into the linker? (I noticed that there is no collect2.exe file included with VxWorks...) thanks, Stephen --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Device Drivers for MPC8260 UART Date: Fri, 24 Aug 2001 14:20:15 -0700 From: "Dan Gold" Message-ID: <3b86c5a8$0$23793$e2e8da3@nntp.cts.com> References: <2myh7.15838$Z2.192063@nnrp1.uunet.ca> "j" wrote in message news:2myh7.15838$Z2.192063@nnrp1.uunet.ca... > I am gonna write device drivers for MPC8260 using VxWorks to program SCC as > a UART. > Can anybody help me by providing some inforamtion or websites where I can > find information and/or examples. Why re-invent the wheel ? You get the SCC UART driver from WRS when you buy the 8260 BSP. If you don't get the source, just ask your local WRS person. hth, Dan Gold gold@ensemble.com --------------------------- Newsgroups: comp.os.vxworks Subject: Re: logInit question Date: Fri, 24 Aug 2001 20:00:40 -0700 From: "Denis" Message-ID: <9m748j$k2hk$1@ID-44494.news.dfncis.de> References: <3b84f0c1@news.barak.net.il> hello, "AE" wrote in message news:3b84f0c1@news.barak.net.il... [...snip...] | How can I create a TCP/IP device and pass it to logInit(). while we dont actually use logMsg() here, i just looked at the logInit(), and the function allows you to specify a file descriptor to which the messages will go. so, it is only logical that you'd be able to create a socket, and perform logInit, and give that descriptor as a 'fd' argument. so, i think 'creating a tcp/ip device' in your question merely amounts to opening a socket, making sure it is connected (you could do both udp and tcp, however udp is unreliable, and probably is not very well suited for logging information, but it depends on the circumstances). if you do not know how to make a socket, then go google for it (plenty of tutorials around), or just get one of those Stevens Unix network programming book. hth, denis --------------------------- End of New-News digest ********************** From vxwexplo-errs@csg.lbl.gov Sat Aug 25 04:04:03 2001 From: "vanitha ramaswami" Date: Sat Aug 25 04:04:06 PDT 2001 Subject: Command Line Options to Create Project in Tornado Hello, Is there any command line option available to create a project in Tornado ?. It's very tedious to create a project thro' the GUI when we have so many directories. - Vanitha Get 250 color business cards for FREE! http://businesscards.lycos.com/vp/fastpath/ From vxwexplo-errs@csg.lbl.gov Sun Aug 26 04:03:16 2001 From: Vxworks Exploder Date: Sun Aug 26 04:03:18 PDT 2001 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Sun Aug 26 04:03:04 PDT 2001 Subject: Re: Build demo gizmo.out hang on Windows Subject: dec 21143 networking Subject: mpc8260 smc uart driver problem Subject: VisionClick Convert Elf error message Subject: Re: Multicast ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: Build demo gizmo.out hang on Windows Date: Sat, 25 Aug 2001 18:03:09 GMT From: Hong Hsu Message-ID: <3B87E904.39F9405E@bellatlantic.net> References: <3B86A116.1EEB961B@bellatlantic.net> <002001c12cd5$6c26c1a0$c80ba8c0@jopasana.co.in> David, Thanks for the response. I build gizmo.out either using Build -> Rebuild All or right click on gizmo Files, then choose Build 'gizmo.out', the Build Output window will pop up. at this point there are two new programs appeared in Task Manager window. One is Winoldap and another is Make. By using TaskInfo2000 tool, I saw two new processes runcmd.exe and make.exe are running. The gizmo.out will never be generated. No error message. If I quit the Tornado Prototyper at this point, the make.exe in D:\tornado\host\x86-win32\bin stay forever and Winoldap program stay forever. You even can not kill them. Thank you again, honghsu@bellatlantic.net 8/25 David Joseph wrote: > Hi > I couldnt send this to the newsgroup because I experienced some problems. > > Let me see if I understand your problems > 1) On building gizmo.out, your program hangs and > 2) You noticed that Winoldap.exe is running at this point > > well all I can say for > 1) is that try to 'Build All' ..... (menu) Build->Rebuild All or just click > on the icon that looks like an 'A' hovering above a tub of water > > For Problem 2) I think the appearance of Winoldap.exe has nothing to do with > Tornado since I tried to monitor the processes list from just before I > started Tornado till I exited it, going through building ,downloading and > running the application. I didnot get any occurance of Winoldap.exe > anywhere. Just as information it is Windows NT Non-Win16 application shell. > Try doing the same and see. > > Hope this solves your problem > If it dosen't, send a mail telling me about it. > Regards > David > > ----- Original Message ----- > From: Hong Hsu > Newsgroups: comp.os.vxworks > Sent: Saturday, August 25, 2001 12:14 AM > Subject: Build demo gizmo.out hang on Windows > > > Hi, I wonder this is right place to get a help for Tornado > > Prototyper 2.0 for Windows. > > If not, would you please point out reference for that? > > > > Here is my situation. The installation of Tornado Prototyper on my PC > > was successful. After > > generating the cobble.o, Building demo gizmo.out hang in Build Output > > window with following message: > > > > 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_M > > ULTI_THREAD -D_REENTRANT -I. -ID:\TORNADO\target\h -DCPU=SIMNT -c > > D:\Tornado\target\proj\gizmo\cobble.c > > vxrm D:\Tornado\target\proj\gizmo\prjObjs.lst > > > > In Output Window there is no nmsimpc statement. (I use standalone PC > > with Intel Pentium 4 processor and try to run the gizmo.out on > > simulator). I also notice that Winoldap.exe is running at this point > > and never goes away even the Wtxregd.exe is killed. What is > > Winoldap.exe doing and how to quit it? > > > > Did I miss something here in order to compile, link and run demo gizmo? > > > > Many thanks, > > 8/24 > > > > > > > > --------------------------- Newsgroups: comp.os.vxworks Subject: dec 21143 networking Date: Sat, 25 Aug 2001 18:49:53 GMT From: "Kelly Hornsby" Organization: Excite@Home - The Leader in Broadband http://home.com/faster Message-ID: I have a PowerPC design based on the Sandpoint 755 system. It uses a DEC/Intel 21143 as the NIC, but I'm having problems with it. the board enters both HwInit 1 and 2, and I get no MuxLoad error messages from it. However, it does not seem to be performing correctly. I can see the device lights for it (link light) for traffic, but I can get no pings from it or to it. Is there any easy way to attack this... whether the information is not correct inside of it, or whether the IP stack is incorrect? I'm open for ideas. for example, is there a way to force an Ethernet transfer? --------------------------- Newsgroups: comp.os.vxworks Subject: mpc8260 smc uart driver problem Date: 25 Aug 2001 19:04:51 -0700 From: kevinatnj@yahoo.com.cn (kewin) Organization: http://groups.google.com/ Message-ID: <221caa76.0108251804.27c2531@posting.google.com> i am writing a UART driver for M8260 SMC based on VxWorks.The problem met me when I initialize the SMC.i use one rx BD and one tx BD.The rx BD's offset from the base of dual port RAM is 0x0800,and the tx BD's offset is 0x0808. rx BD and tx BD look like the following: rxBD txBD status length buffer pointer status length buffer pointer 0xB0 0x01 0x******** 0xB0 0x01 0x******** After write the valuse,i use CPCR to issue the command, INIT RX AND TX COMMAND. Now the problem happens.The txBD status changes to 0x00,and length to 0x00 too. I have try to exchange the position of rxBD and txBD.And this time is that the rxBD status change to 0x00.Till now i have only initialized the RBASE and TBASE in dual port RAM,and i didn't enable the SMCMR's enable tx and rx bits. what are the maybe reasons? Any advice will be welcome. --------------------------- Newsgroups: comp.os.vxworks Subject: VisionClick Convert Elf error message Date: 26 Aug 2001 02:50:38 -0700 From: leonardmatherson@hotmail.com (leonard matherson) Organization: http://groups.google.com/ Message-ID: Hello everybody, I have built .out and .o files and I am now trying to use the download mechanism in visionclick 7.0. When I try to convert these files I get this error message:- convert v7.4C Copyright (c) 1996-2000 Wind River HSI convert ELF file D:\Tornado\target\proj\testfec\ppc860diab\testfec.out to CLICK file ffffffff Extracting symbols from 'D:\Tornado\target\proj\testfec\ppc860diab\testfec.out' error #81: ELF format error, non-executable file Any pointers? Thanks in advance, Leonard. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Multicast Date: Sun, 26 Aug 2001 12:51:42 +0200 From: "Leonid Rosenboim" Organization: Verio Message-ID: References: <3B85222D.5ABF1466@fnal.gov> I think you should nto worry about the CPU board, the driver will support Multicast one way or another. But I strongly suggest that if your Multicast is going to be heavy (say more the 500kbis/sec) you should be EXTREMELY WEARY about your network infrastructire. Namely, beware that most cheaper switches and hubs will "flood" multicasts. The problem with many Ethernet drivers is that they go into promiscous or "all-multicast" mode when even a single group is joined, letting the IP layer filter out the wanted groups. This is not an issue if the overall MC traffic is low or if the Ethernet switch it is connected to, is aware of IGMP, and send MC to those ports which have asked for them. Leonid "Dinker Charak" wrote in message news:3B85222D.5ABF1466@fnal.gov... > Hello, > > We need to support multicast. Should board selection be constrained by > this in anyway? We are trying to choose between MVME230x and MVME5100 > boards. > > dinker --------------------------- End of New-News digest ********************** From vxwexplo-errs@csg.lbl.gov Sun Aug 26 13:21:07 2001 From: Jim Way Date: Sun Aug 26 13:21:09 PDT 2001 Subject: Re: royalties --------------------------- Newsgroups: comp.os.vxworks Subject: Re: royalties Date: Fri, 24 Aug 2001 13:40:49 -0500 From: Andrew Johnson Organization: APS, Argonne National Laboratory Message-ID: <3B869FB1.8717B0A0@aps.anl.gov> References: <3B8658EE.53C1A14C@notifier-is.net> david lindauer wrote: > Is this really the going price for royalties? David, I'm not the Keeper of Beans at my company, but from what I understand, the royalties for my product are about $80-90 per unit sold. I think our sales forecasts were in the "thousands" range for the first few years. Royalties are a function of volume [as Andrew Johnson pointed out] PLUS certain add-in modules that will increase your payments. One suggestion would be to carefully examine what features you have added from the vxWorks capabilities that cost you extra money and see if you can build them yourself or build them another way that avoids using the cost-extra code. Another suggestion would be to get your Marketing team to come up with a three or five year sales forecast, then go back to WindRiver and see if they'll reevaluate their royalty calculations. Another more heretical suggestion would be to see how difficult another OS would be to incorporate into your product. Yes, any change is painful, but sometimes you have to be willing to walk away from a deal to strengthen your negotiating position. Best of luck! 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 Sun Aug 26 15:33:19 2001 From: karunakaran nair Date: Sun Aug 26 15:33:21 PDT 2001 Subject: Re: C , C++ and vxWorks tasks jmoore, what are the restrictions applicable when using semaphores ? i mean the guidelines. thanks, KN --- the vxWorks Users Group Exploder wrote: > Submitted-by vxwexplo-errs@csg.lbl.gov Fri Aug 24 > 07:42:11 2001 > Submitted-by: John Moore > > Not a lot to go on... > > Did the semaphore get created ? ( semBCreate, > semCCreate ...) > > Check the return value of the semTake in your while > loop. It is probably an > error and the task is just spinning (hence the READY > state). > > Also, I think you'll have to pass the task a pointer > to your myclass object to > be able call myclass methods. > > jmoore > > > > > VxWorks > > > how to work with vxworks tasks in a C++ scenario > ? say > > I want to have a task, spawned from a member > function, > > and this task has to wait on a semaphore sent from > C > > function of a different file, and then sequence > thru > > to do some activities calling some member > functions ? > > i see the task to be in READY state when I spawn > it > > from the member function. here is my design. > > > > extern SEM_ID mysemaphore > > void myclass::memberfunction(void) > > { > > > > spawn mytask > > > > } > > > > void mytask(void) > > { > > > > while(1) { > > wait on semaphore ( sent from another c file ) > > > > call some myclass member functions to do > something > > > > } > > > > } > > > > > ********** > > This is a user group mailing list for vxWorks > related topics > see http://www-vxw.lbl.gov/vxworks for posting > and subscription info __________________________________________________ Do You Yahoo!? Make international calls for as low as $.04/minute with Yahoo! Messenger http://phonecard.yahoo.com/ From vxwexplo-errs@csg.lbl.gov Sun Aug 26 15:41:08 2001 From: karunakaran nair Date: Sun Aug 26 15:41:11 PDT 2001 Subject: on timers vxworkers, what is aux clock ? i understand that it is some extra timer that sits on the hardware. is there any way to know what is the minimum resolution i can get from the board from the vxworks shell. guess this question is confusing. when to use system clock for timing, and when to go for aux clock. from where does the watchdog timer get its timing ? system clock ?? i need to be more confident on the timers because i want to have some high precision time stamps in my code. will help me a lot. thanks, KN __________________________________________________ Do You Yahoo!? Make international calls for as low as $.04/minute with Yahoo! Messenger http://phonecard.yahoo.com/ From vxwexplo-errs@csg.lbl.gov Sun Aug 26 15:46:29 2001 From: karunakaran nair Date: Sun Aug 26 15:46:31 PDT 2001 Subject: launches i still launch vxworks from my unix shell prompt. and then i have to spawn a windshell, and then i have to go to my correct directory if that is not from where i launched, and then i have to run my script. looks like i am going to the days of vacuum tubes. way back with another tool, i used to compile, build, download, and run from a single GUI. is this possible in the UNIX environment and vxworks Tornado ? and then, occasionally, wind shell hangs, and i have to go for a hard reset. has some body better ways of working to share with me. or, are these common problems ? thanks,KN __________________________________________________ Do You Yahoo!? Make international calls for as low as $.04/minute with Yahoo! Messenger http://phonecard.yahoo.com/ From vxwexplo-errs@csg.lbl.gov Sun Aug 26 19:56:29 2001 From: "Mike Anderson" Date: Sun Aug 26 19:56:31 PDT 2001 Subject: RE: on timers VxWorks Greetings! > Submitted-by: karunakaran nair > > vxworkers, > what is aux clock ? i understand that it is some > extra timer that sits on the hardware. is there any > way to know what is the minimum resolution i can get > from the board from the vxworks shell. guess this > question is confusing. when to use system clock for > timing, and when to go for aux clock. from where does > the watchdog timer get its timing ? system clock ?? > i need to be more confident on the timers because i > want to have some high precision time stamps in my > code. will help me a lot. > thanks, > KN > The auxilliary clock is an optional support item in your BSP. Typically, it is implemented if you have more than one hardware timer on your board. There is typically no way to know that the min/max resolution is from the shell other than to start setting values and see which ones give you an error return status. However, if you look at the BSP, you should find the code that is used to set the aux clock up. That code will typically indicate what the min and max resolution of the aux clock is. The watchdog timers use the system clock for their timer. The "spy" facility uses the aux clock. WindView will typically use a third timer called the timestamp. This timer is typically put into a free-running mode that generates an interrupt each time the counter wraps to zero. This would be a good option for high-res timestamping. Look at the WindView user's guide to see the details. HTH, Mike Anderson -- ======================================================= T h e P T R G r o u p, I n c. ======================================================= Embedded, Real-Time Solutions and Training Mike Anderson mailto:mike@ThePTRGroup.com Chief Scientist http://www.ThePTRGroup.com V: (703) 585-9384 F: (703) 430-3748 ======================================================= "Software development is like making a baby... You can't make a baby in one month by impregnating nine women. Some things just take time." From vxwexplo-errs@csg.lbl.gov Mon Aug 27 00:44:30 2001 From: Kirti Mahajan Date: Mon Aug 27 00:44:32 PDT 2001 Subject: Problem in accessing PMC module Hello All, Our target board is PPC (MVE3604). We are facing the problem in accessing PMC-RF (VMIC5579) module having 64MB memory. Where in the PCI Memory space and I/O space device wil map...? While pciFindDevice() and pciFindClass routings work on giving device id and vender id. pciConfigIn... and pciConfgOut...() routines are also working. vxWorks Image has been configured to acess VME32 adress space, VME24 & VME16 address space range. Part of BSP Configuration file is as - /**********************************************************************/ #ifndef EXTENDED_VME /* Psuedo PREP memory map as seen from CPU */ #define CPU_PCI_ISA_IO_ADRS 0x80000000 /* base of ISA I/O space */ #define CPU_PCI_ISA_MEM_ADRS 0xc0000000 /* base of ISA mem space */ #define CPU_PCI_IO_ADRS 0x81000000 /* base of PCI I/O space */ #define CPU_PCI_MEM_ADRS 0xc1000000 /* base of PCI mem space */ #else /* Extended VME memory map as seen from the CPU */ #define CPU_PCI_ISA_IO_ADRS 0xfe000000 /* base of ISA I/O space */ #define CPU_PCI_ISA_MEM_ADRS 0xfd000000 /* base of ISA mem space */ #define CPU_PCI_IO_ADRS CPU_PCI_ISA_IO_ADRS /* base of PCI I/O space */#define CPU_PCI_MEM_ADRS CPU_PCI_ISA_MEM_ADRS /* base of PCI mem space */#endif #define CPU_PCI_ISA_IO_SIZE 0x00010000 /* 64 kbytes */ #define CPU_PCI_IO_SIZE 0x00800000 /* 8 meg */ #define CPU_PCI_IO_UPPER_ADRS (CPU_PCI_ISA_IO_ADRS>>16) #define CPUCRA_HI (CPU_PCI_ISA_IO_ADRS>>16) #define CPUCRA_LO 0x0800 #define CPU_PCI_ISA_MEM_SIZE 0x00010000 /* 64 kbytes */ #define CPU_PCI_MEM_SIZE 0x01000000 /* 16 meg */ #define CPU_PCI_MEM_UPPER_ADRS (CPU_PCI_ISA_MEM_ADRS>>16) /* /* * PCI MASTER MEMORY WINDOW LIMITS * * These values are strictly defined by the base memory addresses and window * sizes of the spaces defined above. These values must be correct for the * sysBusProbe() memory range checks for the PCI bus to work properly. */ #ifndef EXTENDED_VME # define PCI_MSTR_LO_ADRS (CPU_PCI_ISA_IO_ADRS) # define PCI_MSTR_HI_ADRS (CPU_PCI_MEM_ADRS + CPU_PCI_MEM_SIZE) #else # define PCI_MSTR_LO_ADRS (CPU_PCI_MEM_ADRS) # define PCI_MSTR_HI_ADRS (CPU_PCI_IO_ADRS + CPU_PCI_IO_SIZE) #endif /* EXTENDED_VME */ #ifndef INCLUDE_MPIC /* * All starting addresses are correct except for CPU_PCI_IACK_ADRS. * This address was selected to allow a minimum 8k memory range * for the MMU table entry. */ #define CPU_PCI_IACK_ADRS 0xbfffe000 /* covers PCI IACK space */ /* base = 0xbffffff0; */ #define CPU_PCI_IACK_SIZE 0x00002000 /* 8 kbytes */ #define MPIC_ADDR(reg) (MPIC_BASE_ADRS + reg) #define MPIC_GLOBAL_CONFIG_REG 0x01020 #define RESET_CNTRLR 0x80000000 #endif /* INCLUDE_MPIC */ /* Base address of HW devices as seen from CPU */ #define FALCON_BASE_ADRS 0xfef80000 #define FALCON_REG_SIZE 0x00010000 #define FALCON_BASE_UPPER_ADRS (FALCON_BASE_ADRS>>16) #define FALCON_DRAM_ATTR 0xfef80010 #define RAVEN_BASE_ADRS 0xfeff0000 #define RAVEN_REG_SIZE 0x00010000 #define FLASH_BASE_ADRS 0xFF000000 #define FLASH_MEM_SIZE 0x01000000 /* MPIC configuration defines */ #define MPIC_BASE_ADRS 0xfc000000 #define MPIC_REG_SIZE 0x00040000 #ifndef EXTENDED_VME # define MPIC_PCI_BASE_ADRS ( MPIC_BASE_ADRS - CPU_PCI_ISA_MEM_ADRS ) #else # define MPIC_PCI_BASE_ADRS MPIC_BASE_ADRS #endif /* Extended VME config */ /* memory map as seen on the PCI bus */ #define PCI_CNFG_ADRS 0x00800000 /* base of PCI config space */ #ifndef EXTENDED_VME # define PCI_IO_ADRS 0x01000000 /* base of PCI I/O address */ # define PCI_MEM_ADRS 0x01000000 /* base of PCI MEM address */ # define PCI2DRAM_BASE_ADRS 0x80000000 /* memory seen from PCI bus */ #else # define PCI_IO_ADRS 0x00000000 /* base of PCI I/O address */ # define PCI_MEM_ADRS CPU_PCI_MEM_ADRS /* base of PCI MEM address */ # define PCI2DRAM_BASE_ADRS 0x00000000 /* memory seen from PCI bus */ #endif /* Extended VME config */ * * Primary PCI bus configuration space address and data register addresses * as seen by the CPU on the local bus. */ #ifndef EXTENDED_VME # define PCI_PRIMARY_CAR 0x80000CF8 /* PCI config address register */ # define PCI_PRIMARY_CDR 0x80000CFC /* PCI config data register */ #else # define PCI_PRIMARY_CAR 0xFE000CF8 /* PCI config address register */ # define PCI_PRIMARY_CDR 0xFE000CFC /* PCI config data register */ #endif /* Extended VME config */ /* * PCI Config Space device addresses based on their device number * * Bit 32 is set to enable CONFIG_DATA accesses to PCI Cycles */ #define CNFG_START_SEARCH 0x5800 /* PCI Space starting offset */ #define CNFG_RAVEN_ADRS CPU_PCI_ISA_IO_ADRS /* Raven PCI and MPIC ASIC*/#define CNFG_IBC_ADRS 0x80005800 /* IBC */ #define CNFG_SCSI_ADRS 0x80006000 /* SCSI */ #define CNFG_UNIVERSE_ADRS 0x80006800 /* VMEbus Bridge */ #define CNFG_LN_ADRS 0x80007000 /* Ethernet Device */ #define CNFG_VGA_ADRS 0x80007800 /* Graphics Device */ #define CNFG_PMC1_ADRS 0x80008000 /* PMC Slot1 */ #define CNFG_SCSI2_ADRS 0x80008800 /* Secondary SCSI */ #define CNFG_LN2_ADRS 0x80009000 /* Secondary Ethernet */ #define CNFG_PMC2_ADRS 0x80009800 /* PMC Slot 2 or PCIX */ /*****************************************************************/ Any help ior suggestion are welcom Thanks and Regards From vxwexplo-errs@csg.lbl.gov Mon Aug 27 04:04:12 2001 From: Vxworks Exploder Date: Mon Aug 27 04:04:15 PDT 2001 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Mon Aug 27 04:03:07 PDT 2001 Subject: Re: VisionClick Convert Elf error message Subject: Re: dec 21143 networking Subject: vxTicks Being Cleared to Zero? Subject: Re: vxTicks Being Cleared to Zero? Subject: Re: vxTicks Being Cleared to Zero? Subject: Re: HELP - vx ... FTP-Server is innocent, it's a dosFS2 problem Subject: Re: How to format the disk Subject: Re: Strict timing Subject: Re: dec 21143 networking Subject: PPP problem Subject: Re: VisionClick Convert Elf error message Subject: Re: VisionClick Convert Elf error message ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: VisionClick Convert Elf error message Date: Sun, 26 Aug 2001 12:23:00 -0400 From: George Varndell Organization: Newsfeeds.com http://www.newsfeeds.com 90,000+ UNCENSORED Newsgroups. Message-ID: <3B892264.B33A0774@hotmail.com> References: leonard matherson wrote: > > Hello everybody, > > I have built .out and .o files and I am now trying to > use the download mechanism in visionclick 7.0. > > When I try to convert these files I get this error message:- > > convert v7.4C Copyright (c) 1996-2000 Wind River HSI > convert ELF file D:\Tornado\target\proj\testfec\ppc860diab\testfec.out to CLICK > file ffffffff > Extracting symbols from 'D:\Tornado\target\proj\testfec\ppc860diab\testfec.out' > error #81: ELF format error, non-executable file You don't convert .o or .out files. When you build your project, don't you get a file named vxWorks? That's what you need to convert. Regards, George Varndell - -- Embedded Systems Expertise Varndell Engineering, LLC. http://www.varndellengineering.com 81 Baltimore St. Suite 206 Cumberland MD. 21502 - -----= Posted via Newsfeeds.Com, Uncensored Usenet News =----- http://www.newsfeeds.com - The #1 Newsgroup Service in the World! - -----== Over 80,000 Newsgroups - 16 Different Servers! =----- --------------------------- Newsgroups: comp.os.vxworks Subject: Re: dec 21143 networking Date: 26 Aug 2001 09:33:58 -0700 From: john_94501@yahoo.com (John) Organization: http://groups.google.com/ Message-ID: <488e459a.0108260833.2d00181c@posting.google.com> References: Hello, I have a couple of suggestions... Speak to your local Wind FAE and make sure that you have the latest version of the driver (this driver is a particularly complex one since there are so many variations in the hardware). While about it, ask him/her if you can get the source code (it contains a lot of debug, but needs to be enabled at compile time). If you can't get the source, try for a version of the .o built with debug enabled. HTH, John... "Kelly Hornsby" wrote in message news:... > I have a PowerPC design based on the Sandpoint 755 system. It uses a > DEC/Intel 21143 as the NIC, but I'm having problems with it. the board > enters both HwInit 1 and 2, and I get no MuxLoad error messages from it. > However, it does not seem to be performing correctly. I can see the device > lights for it (link light) for traffic, but I can get no pings from it or to > it. > > Is there any easy way to attack this... whether the information is not > correct inside of it, or whether the IP stack is incorrect? I'm open for > ideas. for example, is there a way to force an Ethernet transfer? --------------------------- Newsgroups: comp.os.vxworks Subject: vxTicks Being Cleared to Zero? Date: Sun, 26 Aug 2001 22:30:20 GMT From: "Pat DiRienzo" Message-ID: <0Mei7.168$td.80985@typhoon1.gnilink.net> We are using the system clock as a time of day clock on our system (single board computer and other VME cards) by using clock_gettime calls. The clock works fine (time of day correctly displayed on our system screen) until it reaches a certain point in our code, after monitoring vxTicks I found that it was being set to zero and causing the time of day clock on our system to display an incorrect time. I understand that vxTicks (the vxWorks global) is updated by tickAnnounce from our system timer ISR and that vxTicks is used by vxWork's to provide a time of day thru clock_gettime calls. vxAbsTicks remained OK thru all of this. So why is vxTicks being set to zero, It does not appear to be a wrapping problem. pdirienzo@sierradatasys.com pat.dirienzo@verizon.net --------------------------- Newsgroups: comp.os.vxworks Subject: Re: vxTicks Being Cleared to Zero? Date: Mon, 27 Aug 2001 01:43:29 GMT From: Harvey Taylor Organization: MTS Internet Message-ID: <3B89C10A.70BE@despam.autobahn.mb.ca> References: <0Mei7.168$td.80985@typhoon1.gnilink.net> In article <0Mei7.168$td.80985@typhoon1.gnilink.net>, Pat DiRienzo wrote: > > We are using the system clock as a time of day clock on our system (single > board computer and other VME cards) by using clock_gettime calls. The clock > works fine (time of day correctly displayed on our system screen) until it > reaches a certain point in our code, after monitoring vxTicks I found that > it was being set to zero and causing the time of day clock on our system to > display an incorrect time. I understand that vxTicks (the vxWorks global) > is updated by tickAnnounce from our system timer ISR and that vxTicks is > used by vxWork's to provide a time of day thru clock_gettime calls. > vxAbsTicks remained OK thru all of this. So why is vxTicks being set to > zero, It does not appear to be a wrapping problem. > Most probable reason is a coding error - some module writing improperly through a pointer. <$0.02> - -het - -- Everyone has the right to recognition everywhere as a person before the law. UDHR:Article 6: http://www.autobahn.mb.ca/~het/udhr.html H.E. Taylor http://www.autobahn.mb.ca/~het/ --------------------------- Newsgroups: comp.os.vxworks Subject: Re: vxTicks Being Cleared to Zero? Date: 26 Aug 2001 21:48:17 -0700 From: john_94501@yahoo.com (John) Organization: http://groups.google.com/ Message-ID: <488e459a.0108262048.27208f31@posting.google.com> References: <0Mei7.168$td.80985@typhoon1.gnilink.net> Hello, Check for taskDelay() calls with negative delays... from what I remember of previous posts to this group, such calls have the side effect that the tick counter is zeroed again... HTH, John... "Pat DiRienzo" wrote in message news:<0Mei7.168$td.80985@typhoon1.gnilink.net>... > We are using the system clock as a time of day clock on our system (single > board computer and other VME cards) by using clock_gettime calls. The clock > works fine (time of day correctly displayed on our system screen) until it > reaches a certain point in our code, after monitoring vxTicks I found that > it was being set to zero and causing the time of day clock on our system to > display an incorrect time. I understand that vxTicks (the vxWorks global) > is updated by tickAnnounce from our system timer ISR and that vxTicks is > used by vxWork's to provide a time of day thru clock_gettime calls. > vxAbsTicks remained OK thru all of this. So why is vxTicks being set to > zero, It does not appear to be a wrapping problem. > > pdirienzo@sierradatasys.com > pat.dirienzo@verizon.net --------------------------- Newsgroups: comp.os.vxworks Subject: Re: HELP - vx ... FTP-Server is innocent, it's a dosFS2 problem Date: Mon, 27 Aug 2001 10:47:46 +0200 From: "Leonid Rosenboim" Organization: Verio Message-ID: References: <3B77EB38.DBD20CFA@stoye.de> <3B83499C.BC841720@stoye.de> This is a multi-part message in MIME format. - ------=_NextPart_000_003C_01C12EE5.B5B7E1C0 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable This bug was discovered sometime in early 2000, and WRS supposedly have = a patch for this. There is also a work around - if the chkdsk is run = frequently enough (i.e. after some files have been deleted) it will = reoranize the directory and "pack" its entires, so there wont be a whole = cluster full of deleted entries. "Rainer Stelzer" wrote in message = news:3B83499C.BC841720@stoye.de... Hi vxworkers,=20 the FTP Server (neither the delivered nor the unsupported one) wasn't = the problem.=20 The problem we encountered, is a known one: SPR.No. 31480.=20 =20 SPR# 31480=20 Submitted by: rodw=20 Date: May 24 2000=20 Title: DosFS2.0 : Files being lost=20 Problem_Description=20 When a cluster chain for a directory contains a cluster where every=20 directory entry has been deleted; if a new file is then created in = that=20 directory,=20 the remaining entries in the cluster are changed from deleted = entries to=20 unused entries.=20 This has the effect of preventing software from finding files beyond = this=20 point.=20 If the check disk utility=20 is then run the files are actually deleted and the lost fat chains = are=20 recovered.=20 =20 I think every one using T2 and dosFS2 should know that, cause this = behavior=20 results in complex, long term error situations that are hard to find.=20 And that's the reason why in my opinion this kind of errors should be = reported by wrs=20 in an urgent e-mail, rather than polling windserve by the users.=20 (Which is a problem, because of what to search for ?)=20 We spent three day's to localize and reproduce an error that was found = and corrected=20 in 8/2000 !=20 rgds=20 rainer=20 vxworks, tornado=20 = =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=20 Gebr=FCder Stoye GmbH=20 department: traffic control systems=20 name: Rainer Stelzer=20 adr: Longericher Strasse 177=20 city: Cologne, Germany=20 zip: 50739=20 mobil: 0174/8194860=20 phone: ++49/221/1707019=20 fax: ++49/221/1707010=20 mailto: r.stelzer@stoye.de=20 www: http://www.stoye.de=20 = =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=20 =20 - ------=_NextPart_000_003C_01C12EE5.B5B7E1C0 Content-Type: text/html; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable

This bug was discovered sometime in = early 2000,=20 and WRS supposedly have a patch for this. There is also a work around - = if the=20 chkdsk is run frequently enough (i.e. after some files have been = deleted) it=20 will reoranize the directory and "pack" its entires, so there wont be a = whole=20 cluster full of deleted entries.
 
"Rainer Stelzer" <r.stelzer@stoye.de> wrote in = message=20 news:3B83499C.BC841720@stoye.de<= /A>...
Hi=20 vxworkers,=20

the FTP Server (neither the delivered nor the unsupported one) = wasn't the=20 problem.
The problem we encountered, is a known one: SPR.No. = 31480.=20
 =20

SPR# 31480
Submitted by: rodw
Date: = May 24=20 2000
Title: DosFS2.0 : Files being lost=20 =

           =             &= nbsp;           &n= bsp;     =20 Problem_Description=20

 When a cluster chain for a directory contains a cluster = where=20 every
directory entry has been deleted; if a new file is = then=20 created in that
directory,
  the = remaining=20 entries in the cluster are changed from deleted entries to =
unused=20 entries.
  This has the effect of preventing = software from=20 finding files beyond this
point.
  If = the check=20 disk utility
  is then run the files are actually = deleted=20 and the lost fat chains are
recovered.
 =20

I think every one using T2 and dosFS2 should know that, cause this = behavior=20
results in complex, long term error situations that are hard to = find.=20

And that's the reason why in my opinion this kind of errors should = be=20 reported by wrs
in an urgent e-mail, rather than polling windserve = by the=20 users.
(Which is a problem, because of what to search for ?) =
We spent=20 three day's to localize and reproduce an error that was found and = corrected=20
in 8/2000 !=20

rgds=20

rainer=20

vxworks, tornado =
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=3D=3D=20
            = Gebr=FCder Stoye GmbH
department: traffic control systems=20
      name: Rainer Stelzer=20
       adr: Longericher Strasse 177=20
      city: Cologne, Germany=20
       zip: 50739=20
     mobil: 0174/8194860 =
    =20 phone: ++49/221/1707019
       fax:=20 ++49/221/1707010
    mailto: r.stelzer@stoye.de=20
       www: http://www.stoye.de=20 =
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=3D=3D
 =20

- ------=_NextPart_000_003C_01C12EE5.B5B7E1C0-- --------------------------- Newsgroups: comp.os.vxworks Subject: Re: How to format the disk Date: Mon, 27 Aug 2001 09:47:07 +0200 From: =?iso-8859-1?Q?Torbj=F6rn?= Frostemark Organization: FOI Message-ID: <3B89FAFA.87A3E305@foi.se> References: <9m675o$q2o$1@news.asu.edu> This is a multi-part message in MIME format. - --------------16A6D528ECA7D49AF095B26A Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Shesha Bhushan wrote: > Hi Gurus, > I have connected a scsi hardisk to a Single board Computer that runs > vxWorks OS. The hard disk was previously formated as NTFS. I want to format > this hard disk with dosFs. How to format this previously formated disk. Do not change anything in the original sysScsi.c (do not try to mount the disk) and set SCSI ID to 2 (so this example will work). From a windshell, run following lines: pSpd20 = scsiPhysDevCreate(pSysScsiCtrl, 2, 0, 0, 0, 0, 0, 0) pSbd0 = scsiBlkDevCreate(pSpd20, 0, 0) dosFsMkfsOptions(6) dosFsMkfs("/", pSbd0) Now you should be able to mount the disk. Torbjorn - --------------16A6D528ECA7D49AF095B26A Content-Type: text/x-vcard; charset=us-ascii; name="torfr.vcf" Content-Transfer-Encoding: 7bit Content-Description: Card for Torbjörn Frostemark Content-Disposition: attachment; filename="torfr.vcf" begin:vcard n:Frostemark;Torbjörn tel;cell:+46-70-249 76 27 tel;fax:+46-(13-37 80 49 tel;work:+46-13-37 80 14 x-mozilla-html:TRUE org:FOI - Swedish Defence Research Agency;Electronic Warfare Systems adr:;;Box 1165;SE-581 11 Linköping;;;SWEDEN version:2.1 email;internet:torfr@foi.se title:Research Engineer fn:Torbjörn Frostemark end:vcard - --------------16A6D528ECA7D49AF095B26A-- --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Strict timing Date: Mon, 27 Aug 2001 10:55:06 +0200 From: "Leonid Rosenboim" Organization: Verio Message-ID: References: <9m2cif$82j$1@fe2.cs.interbusiness.it> The taskDelay() call timing precision depends on the task priority, you are correct to suspect that higher-priority task will add delay after the task awakens until it actually runs. There is a model called "Rate Monolythic Scheduling" or RMS, which clearly describes the relation beween priority, and response time. But regarding I2C specifically, I have written such code a while ago, as part of an Ethernet driver, and as I recall, the I2C bus is quite asynchronous, so these delays should not be too much of a worry. Of course, in certain critical spots in the I2C code, you might to lock out interrupts to avoid someone preempting the code, but these sections should be just a couple of lines of code. "Andrea della Porta" wrote in message news:9m2cif$82j$1@fe2.cs.interbusiness.it... > Does anyone know how to obtain precise task_delay() when several task is > running? How do priority level among tasks affect this delay()? I've to > simulate I2C bus waveforms but it seems that the timing passed to > task_delay() is not respected. > > Thanks in advance, > Andrea della Porta > > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: dec 21143 networking Date: Mon, 27 Aug 2001 11:04:35 +0200 From: "Leonid Rosenboim" Organization: Verio Message-ID: References: Beside getting source code and the lines, you can first try to diagnose the problem with ifShow(), just by looking at the counters, and by using a packet sniffer on the network to check if the target is transmitting at all. From your description, my best guess would be that you have not configured the device interrupt correctly. This is the most typical issue. "Kelly Hornsby" wrote in message news:lrSh7.112792$w5.11673633@news1.rdc1.ga.home.com... > I have a PowerPC design based on the Sandpoint 755 system. It uses a > DEC/Intel 21143 as the NIC, but I'm having problems with it. the board > enters both HwInit 1 and 2, and I get no MuxLoad error messages from it. > However, it does not seem to be performing correctly. I can see the device > lights for it (link light) for traffic, but I can get no pings from it or to > it. > > Is there any easy way to attack this... whether the information is not > correct inside of it, or whether the IP stack is incorrect? I'm open for > ideas. for example, is there a way to force an Ethernet transfer? > > > > --------------------------- Newsgroups: comp.os.vxworks Subject: PPP problem Date: Mon, 27 Aug 2001 10:40:37 +0200 From: Emmanuel Herbreteau Organization: SEPRO Robotique (BEP) Message-ID: <3B8A0784.255C0A51@sepro-robotique.com> Hi "vx-Workers", I'm using PPP to connect a PC to my target (PowerPC with VxWorks 5.4) throught a RS232 running at 38400 bauds (without hardware flow control). The following code is the code used to initialize the PPP interface : pppOpt.flags=OPT_SILENT_MODE |OPT_PASSIVE_MODE |OPT_DEFAULTROUTE |OPT_NO_MRU |OPT_NO_IP |OPT_NO_PC |OPT_NO_PAP |OPT_NO_CHAP; pppOpt.netmask = "255.255.255.0"; pppInit(0,"/tyCo/0","192.168.254.254","192.168.254.1",38400,&pppOpt,NULL); ... It works but, when establishing connection, Windows NT always reports eleven (11) CRC errors. Then the connection works flawlessy ! (after a delay of 30 seconds, because of those errors) Any ideas about those "CRC errors" ? - -- Regards Emmanuel --------------------------- Newsgroups: comp.os.vxworks Subject: Re: VisionClick Convert Elf error message Date: 27 Aug 2001 01:57:34 -0700 From: leonardmatherson@hotmail.com (leonard matherson) Organization: http://groups.google.com/ Message-ID: References: <3B892264.B33A0774@hotmail.com> I am trying to download with visionclick a very small program that sets registers etc. Thanks, Leonard. George Varndell wrote in message news:<3B892264.B33A0774@hotmail.com>... > leonard matherson wrote: > > > > Hello everybody, > > > > I have built .out and .o files and I am now trying to > > use the download mechanism in visionclick 7.0. > > > > When I try to convert these files I get this error message:- > > > > convert v7.4C Copyright (c) 1996-2000 Wind River HSI > > convert ELF file D:\Tornado\target\proj\testfec\ppc860diab\testfec.out to CLICK > > file ffffffff > > Extracting symbols from 'D:\Tornado\target\proj\testfec\ppc860diab\testfec.out' > > error #81: ELF format error, non-executable file > > You don't convert .o or .out files. > When you build your project, don't you get a file named vxWorks? > That's what you need to convert. > > Regards, > George Varndell > -- > Embedded Systems Expertise Varndell Engineering, LLC. > http://www.varndellengineering.com 81 Baltimore St. Suite 206 > Cumberland MD. 21502 > > > -----= Posted via Newsfeeds.Com, Uncensored Usenet News =----- > http://www.newsfeeds.com - The #1 Newsgroup Service in the World! > -----== Over 80,000 Newsgroups - 16 Different Servers! =----- --------------------------- Newsgroups: comp.os.vxworks Subject: Re: VisionClick Convert Elf error message Date: Mon, 27 Aug 2001 06:11:17 -0400 From: George Varndell Organization: Newsfeeds.com http://www.newsfeeds.com 90,000+ UNCENSORED Newsgroups. Message-ID: <3B8A1CC5.689A6DA@hotmail.com> References: <3B892264.B33A0774@hotmail.com> leonard matherson wrote: > > I am trying to download with visionclick a very small program > that sets registers etc. That's fine, a good idea even, but you must still build and convert an elf file. Click is telling you that the file you asked it to convert is not a fully linked elf file. I'm guessing your .out file is a relocatable, partial link of a few object modules. Are you linking with -r? Regards, George Varndell - -- Embedded Systems Expertise Varndell Engineering, LLC. http://www.varndellengineering.com 81 Baltimore St. Suite 206 Cumberland MD. 21502 - -----= Posted via Newsfeeds.Com, Uncensored Usenet News =----- http://www.newsfeeds.com - The #1 Newsgroup Service in the World! - -----== Over 80,000 Newsgroups - 16 Different Servers! =----- --------------------------- End of New-News digest ********************** From vxwexplo-errs@csg.lbl.gov Mon Aug 27 09:18:57 2001 From: =?iso-8859-1?q?ram=20mohan?= Date: Mon Aug 27 09:19:00 PDT 2001 Subject: Re: Problem in accessing PMC module hi, we r also trying to use one of pmc sites to extend internal(primary) pci bus on to secondary pci bus(on vme backplane) thru p0 connector but we have no clue how to get this concept to work it seems u've done some work on this can u help us thanks in advance Mohan --- the vxWorks Users Group Exploder wrote: > Submitted-by vxwexplo-errs@csg.lbl.gov Mon Aug 27 > 00:44:30 2001 > Submitted-by: Kirti Mahajan > > > > Hello All, > > Our target board is PPC (MVE3604). We are facing > the problem in accessing > PMC-RF (VMIC5579) module having 64MB memory. > Where in the PCI Memory space and I/O space device > wil map...? > While pciFindDevice() and pciFindClass routings > work on giving device id > and vender id. > pciConfigIn... and pciConfgOut...() routines are > also working. > > vxWorks Image has been configured to acess VME32 > adress space, VME24 & > VME16 address space range. > > > Part of BSP Configuration file is as - > /**********************************************************************/ > > #ifndef EXTENDED_VME > > /* Psuedo PREP memory map as seen from CPU */ > > #define CPU_PCI_ISA_IO_ADRS 0x80000000 /* > base of ISA I/O space > */ > #define CPU_PCI_ISA_MEM_ADRS 0xc0000000 /* > base of ISA mem space > */ > #define CPU_PCI_IO_ADRS 0x81000000 /* > base of PCI I/O space > */ > #define CPU_PCI_MEM_ADRS 0xc1000000 /* > base of PCI mem space > */ > > #else > > /* Extended VME memory map as seen from the CPU */ > > #define CPU_PCI_ISA_IO_ADRS 0xfe000000 /* > base of ISA I/O space > */ > #define CPU_PCI_ISA_MEM_ADRS 0xfd000000 /* > base of ISA mem space > */ > #define CPU_PCI_IO_ADRS CPU_PCI_ISA_IO_ADRS > /* base of PCI I/O > space */#define CPU_PCI_MEM_ADRS > CPU_PCI_ISA_MEM_ADRS /* base of > PCI mem space */#endif > > #define CPU_PCI_ISA_IO_SIZE 0x00010000 /* > 64 kbytes */ > #define CPU_PCI_IO_SIZE 0x00800000 /* 8 > meg */ > #define CPU_PCI_IO_UPPER_ADRS > (CPU_PCI_ISA_IO_ADRS>>16) > #define CPUCRA_HI > (CPU_PCI_ISA_IO_ADRS>>16) > #define CPUCRA_LO 0x0800 > > #define CPU_PCI_ISA_MEM_SIZE 0x00010000 /* > 64 kbytes */ > #define CPU_PCI_MEM_SIZE 0x01000000 /* > 16 meg */ > #define CPU_PCI_MEM_UPPER_ADRS > (CPU_PCI_ISA_MEM_ADRS>>16) > > /* > /* > * PCI MASTER MEMORY WINDOW LIMITS > * > * These values are strictly defined by the base > memory addresses and > window > * sizes of the spaces defined above. These values > must be correct for > the > * sysBusProbe() memory range checks for the PCI bus > to work properly. > */ > > #ifndef EXTENDED_VME > # define PCI_MSTR_LO_ADRS > (CPU_PCI_ISA_IO_ADRS) > # define PCI_MSTR_HI_ADRS (CPU_PCI_MEM_ADRS + > CPU_PCI_MEM_SIZE) > #else > # define PCI_MSTR_LO_ADRS (CPU_PCI_MEM_ADRS) > # define PCI_MSTR_HI_ADRS (CPU_PCI_IO_ADRS + > CPU_PCI_IO_SIZE) > #endif /* EXTENDED_VME */ > > #ifndef INCLUDE_MPIC > > /* > * All starting addresses are correct except for > CPU_PCI_IACK_ADRS. > * This address was selected to allow a minimum 8k > memory range > * for the MMU table entry. > */ > > #define CPU_PCI_IACK_ADRS 0xbfffe000 /* > covers PCI IACK space > */ > /* > base = 0xbffffff0; */ > #define CPU_PCI_IACK_SIZE 0x00002000 /* 8 > kbytes */ > > #define MPIC_ADDR(reg) (MPIC_BASE_ADRS + > reg) > #define MPIC_GLOBAL_CONFIG_REG 0x01020 > #define RESET_CNTRLR 0x80000000 > > #endif /* INCLUDE_MPIC */ > /* Base address of HW devices as seen from CPU */ > > #define FALCON_BASE_ADRS 0xfef80000 > #define FALCON_REG_SIZE 0x00010000 > #define FALCON_BASE_UPPER_ADRS > (FALCON_BASE_ADRS>>16) > #define FALCON_DRAM_ATTR 0xfef80010 > #define RAVEN_BASE_ADRS 0xfeff0000 > #define RAVEN_REG_SIZE 0x00010000 > #define FLASH_BASE_ADRS 0xFF000000 > #define FLASH_MEM_SIZE 0x01000000 > > /* MPIC configuration defines */ > > #define MPIC_BASE_ADRS 0xfc000000 > #define MPIC_REG_SIZE 0x00040000 > #ifndef EXTENDED_VME > # define MPIC_PCI_BASE_ADRS ( MPIC_BASE_ADRS - > CPU_PCI_ISA_MEM_ADRS ) > #else > # define MPIC_PCI_BASE_ADRS MPIC_BASE_ADRS > #endif /* Extended VME config */ > > /* memory map as seen on the PCI bus */ > > #define PCI_CNFG_ADRS 0x00800000 /* > base of PCI config > space */ > #ifndef EXTENDED_VME > # define PCI_IO_ADRS 0x01000000 /* > base of PCI I/O address > */ > # define PCI_MEM_ADRS 0x01000000 /* > base of PCI MEM address > */ > # define PCI2DRAM_BASE_ADRS 0x80000000 /* > memory seen from PCI > bus */ > #else > # define PCI_IO_ADRS 0x00000000 /* > base of PCI I/O address > */ > # define PCI_MEM_ADRS CPU_PCI_MEM_ADRS /* > base of PCI MEM > address */ > # define PCI2DRAM_BASE_ADRS 0x00000000 /* > memory seen from PCI > bus */ > #endif /* Extended VME config */ > > * > * Primary PCI bus configuration space address and > data register addresses > * as seen by the CPU on the local bus. > */ > > #ifndef EXTENDED_VME > # define PCI_PRIMARY_CAR 0x80000CF8 /* PCI > config address register > */ > # define PCI_PRIMARY_CDR 0x80000CFC /* PCI > config data register > */ > #else > # define PCI_PRIMARY_CAR 0xFE000CF8 /* PCI > config address register > */ > # define PCI_PRIMARY_CDR 0xFE000CFC /* PCI > config data register > */ > #endif /* Extended VME config */ > > > /* > * PCI Config Space device addresses based on their > device === message truncated === From vxwexplo-errs@csg.lbl.gov Mon Aug 27 13:22:22 2001 From: "David Laight" Date: Mon Aug 27 13:22:24 PDT 2001 Subject: Re: vxTicks being cleared to zero > Check for taskDelay() calls with negative delays... from what I > remember of previous posts to this group, such calls have the side > effect that the tick counter is zeroed again... The guys who wrote the vxWorks timer code couldn't decide how to use modulo 2^32-1 arithmetic for the timers (1). Instead if ANY timeout requests an expiry time such that vxTicks + timeout >= 2^32 then vxTicks is set to zero and the expiry time in all the pending timouts 'fixed'. This is entirely broken! As well as confusing any code using vxTicks for other purposes (ok it's an unpublished variable....) it also breaks all the ansi time routines. If the system clock interrupts at 100Hz (a moderate sensible value) vxTicks wraps after about 16 months. Run at 1kHz (not unreasonable with current processor speeds) and you have under 50 days. Now you want your router/set top box/xxx to stay up 100%, but vxWorks will corrupt the system date/time in under 2 months! (run at just over 2kHz and select() will break badly for other reasons.) (1) save the absolute expiry time in the timeout slot and check: if ((int)(expiry_time - now) <= 0) expired() this is fine provided you limit timeouts to 2^31 ticks. David PS: I may be unemployed but I don't want to fix these bugs :-) From vxwexplo-errs@csg.lbl.gov Mon Aug 27 13:55:51 2001 From: Carl Lionberger Date: Mon Aug 27 13:55:53 PDT 2001 Subject: rebootHook list order problem I'd like to set up a rebootHook that does some network activity, but the VxWorks documentation states that this cannot be done because the reboot Hook list is performed in the order of addition, so the network is shut down before any of the user routines get to run. One might have hoped they would run the list in reverse order of addition! Does anyone have any suggestions? Thanks, Carl -- // // Carl Lionberger CALionberger@lbl.gov // Computer Systems Engineer Voice (510) 486-7503 // Engineering Division Fax (510) 486-4544 // Berkeley National Laboratory (LBNL) // From vxwexplo-errs@csg.lbl.gov Mon Aug 27 22:32:26 2001 From: Anuradha Pani Date: Mon Aug 27 22:32:28 PDT 2001 Subject: example of a tcp server hi, could anyone give any pointers to any links or sites which have a tcp server example for vxworks? thanks anuradha __________________________________________________ Do You Yahoo!? Make international calls for as low as $.04/minute with Yahoo! Messenger http://phonecard.yahoo.com/ From vxwexplo-errs@csg.lbl.gov Tue Aug 28 03:05:34 2001 From: "Sheetal Khandkar" Date: Tue Aug 28 03:05:37 PDT 2001 Subject: Problem in UGL - mouse clicks not recognized Hi, I have downloaded the graphics prototyper from the windriver site and trying to use the windML functions. My problem is none of the event trapping APIs like "winEventGet" are trapping mouse clicks. The keyboard button pressed events are recognised, but no mouse events. Is any configuration required to enable mouse clicks? Another thing, in the documentation they mention something about the windML configuration tool which should be available in the Tornado>Tools>windML menu. My copy does not include this. Also the uglInit.h file which is supposed to be there in the tornado\target\src\ugl\config\ directory, is not existing and neither is ugl_winLib.o. Thanks and Regards, -Sheetal. From vxwexplo-errs@csg.lbl.gov Tue Aug 28 03:39:19 2001 From: "Kuruvilla, Sujit V (Sujit)" Date: Tue Aug 28 03:39:22 PDT 2001 Subject: VxWorks timers Hi, I have a query regarding the timer routines timer_create(), timer_connect() etc. Does vxworks internally implement a timer by associating a watch dog timer with it. Meaning if I create 3 timers using timer_create() does that involve using 3 watch dog timers by vxworks to implement it. If it uses one watch dog timer per timer_create timer, can't I implement my own timer library without using so many watch dog timers? Thanks in Advance, Regards, Sujit From vxwexplo-errs@csg.lbl.gov Tue Aug 28 04:03:32 2001 From: Vxworks Exploder Date: Tue Aug 28 04:03:34 PDT 2001 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Tue Aug 28 04:03:12 PDT 2001 Subject: S_iosLib_INVALID_FILE_DESCRIPTOR error Subject: S_iosLib_INVALID_FILE_DESCRIPTOR Subject: Re: vxTicks Being Cleared to Zero? Subject: Tornado AE Subject: Re: SV: TFFS in Tornado 2 for PPC Subject: Re: Build demo gizmo.out hang on Windows Subject: BSP for MPC860 from Thales Computers. Subject: creation of vxWorks image without Tornado GUI *.wpj file... Subject: Re: UDP Viewer Subject: Re: Build demo gizmo.out hang on Windows Subject: FORMAT Hard Disk Subject: wqeqweqwe Subject: Could not open H:/.wind/shellHistory.dat to store shell history. Subject: Re: UDP Viewer Subject: Where's WindRiver's Embedded Software Book Subject: Re: creation of vxWorks image without Tornado GUI *.wpj file... Subject: Re: Could not open H:/.wind/shellHistory.dat to store shell history. Subject: Re: creation of vxWorks image without Tornado GUI *.wpj file... Subject: Re: Telnet sessions Subject: Re: Strict timing Subject: Re: SLIP protocol under Windows 2000/NT Subject: Re: Tornado AE Subject: Re: creation of vxWorks image without Tornado GUI *.wpj file... Subject: Re: creation of vxWorks image without Tornado GUI *.wpj file... Subject: Re: creation of vxWorks image without Tornado GUI *.wpj file... Subject: Re: creation of vxWorks image without Tornado GUI *.wpj file... Subject: How to stop all tasks when exception or breakpoint ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: S_iosLib_INVALID_FILE_DESCRIPTOR error Date: Mon, 27 Aug 2001 17:14:13 +0530 From: Deepa Organization: Lucent Technologies, Columbus, Ohio Message-ID: <3B8A328D.B504D31D@lucent.com> Hi, I am trying to load the VxWorks image over the network. At the point where it tries to load the symbol table, it gives the S_iosLib_INVALID_FILE_DESCRIPTOR error. What could this mean? Thanks Deepa. --------------------------- Newsgroups: comp.os.vxworks Subject: S_iosLib_INVALID_FILE_DESCRIPTOR Date: Mon, 27 Aug 2001 17:20:54 +0530 From: Deepa Organization: Lucent Technologies, Columbus, Ohio Message-ID: <3B8A341E.10EA536F@lucent.com> Hi, I am trying to load the VxWorks image over the network. At the point where it tries to load the symbol table, it gives the S_iosLib_INVALID_FILE_DESCRIPTOR error. What could this mean? Thanks Deepa. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: vxTicks Being Cleared to Zero? Date: 27 Aug 2001 06:25:44 -0700 From: mvi76@yahoo.com (Martin Ivanov) Organization: http://groups.google.com/ Message-ID: <5f86e92.0108270525.6b754c21@posting.google.com> References: <0Mei7.168$td.80985@typhoon1.gnilink.net> "Pat DiRienzo" wrote in message news:<0Mei7.168$td.80985@typhoon1.gnilink.net>... > We are using the system clock as a time of day clock on our system (single > board computer and other VME cards) by using clock_gettime calls. The clock > works fine (time of day correctly displayed on our system screen) until it > reaches a certain point in our code, after monitoring vxTicks I found that > it was being set to zero and causing the time of day clock on our system to > display an incorrect time. I understand that vxTicks (the vxWorks global) > is updated by tickAnnounce from our system timer ISR and that vxTicks is > used by vxWork's to provide a time of day thru clock_gettime calls. > vxAbsTicks remained OK thru all of this. So why is vxTicks being set to > zero, It does not appear to be a wrapping problem. > > pdirienzo@sierradatasys.com > pat.dirienzo@verizon.net Do you use posix-timers? I had some painfull experience with posix timers - if you load such timer with some bad value, VxTicks may get reset. Unproper usage of taskDelay() can also be the origin. Regards Martin --------------------------- Newsgroups: comp.os.vxworks Subject: Tornado AE Date: Mon, 27 Aug 2001 15:38:41 +0100 From: Graham Marshall Organization: Bazaar Technologies Message-ID: <3B8A5B71.F9C6A93A@hotmail.com> Hi Guys, My company is interested in obtaining Tornado AE for x86. Does anyone have any information (other than the marketing) about this product? I am particularly interested in: * C++ Support (including which version of gcc does it use?) * Does Protected Domain support require hooks to be added by the developer or does it come for 'free'? * Is it an extension of VxWorks 5.4 or is it a rewrite? * Are the tools robust and useful? I'd be really grateful for any feedback - can you post it to the group in the interests of sharing? Thanks in advance, Bruce --------------------------- Newsgroups: comp.os.vxworks Subject: Re: SV: TFFS in Tornado 2 for PPC Date: 27 Aug 2001 14:39:24 GMT From: Johan Borkhuis Organization: Agere Systems Message-ID: References: <9m16q8$e7q$1@newsreader.mailgate.org> "Johan Bostedt" wrote: > It seems like a file "ftllite.o" is missing in > target/lib/objPPC604gnuvx, when comparing how things are defined for > mountNFTL and formatNFTL. But how am I going to get that file? > I'm using a Dy4179+ BSP for Tornado 2.0 (rel. 1.2) Take a look at this: http://www.xs4all.nl/~borkhuis/vxworks/mbx860.html#2-A As also noted in another message: TFFS is a product that has to be purchased seperate. This can be used to check, but it should not be used to bypass purchase of the product! 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: Build demo gizmo.out hang on Windows Date: Mon, 27 Aug 2001 09:26:24 -0500 From: "tazboy78" Organization: http://extra.newsguy.com Message-ID: <9mdl9o02fa7@enews4.newsguy.com> References: <3B86A116.1EEB961B@bellatlantic.net> Try turning off any virus scanners. I have seen virus scanners screw up build processes before. "Hong Hsu" wrote in message news:3B86A116.1EEB961B@bellatlantic.net... > Hi, I wonder this is right place to get a help for Tornado > Prototyper 2.0 for Windows. > If not, would you please point out reference for that? > > Here is my situation. The installation of Tornado Prototyper on my PC > was successful. After > generating the cobble.o, Building demo gizmo.out hang in Build Output > window with following message: > > 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_M > ULTI_THREAD -D_REENTRANT -I. -ID:\TORNADO\target\h -DCPU=SIMNT -c > D:\Tornado\target\proj\gizmo\cobble.c > vxrm D:\Tornado\target\proj\gizmo\prjObjs.lst > > In Output Window there is no nmsimpc statement. (I use standalone PC > with Intel Pentium 4 processor and try to run the gizmo.out on > simulator). I also notice that Winoldap.exe is running at this point > and never goes away even the Wtxregd.exe is killed. What is > Winoldap.exe doing and how to quit it? > > Did I miss something here in order to compile, link and run demo gizmo? > > Many thanks, > 8/24 > > > > --------------------------- Newsgroups: comp.os.vxworks Subject: BSP for MPC860 from Thales Computers. Date: Mon, 27 Aug 2001 15:06:43 GMT From: christa@ist.co.za (Christa de Villiers) Message-ID: <3b8a602a.20632598@news.is.co.za> Hi there, We intend using the MPC860 PowerQUICC board from Thales Computers, but need to port their BSP for VxWorks 5.3.1, Tornado 1.0.1 to Tornado 2.0.2 /VxWorks5.4. I would like to get into contact with anyone who has recently done this, to get an idea of how big the task is. We do not have any BSP development experience. Thanks in advance, Christa --------------------------- Newsgroups: comp.os.vxworks Subject: creation of vxWorks image without Tornado GUI *.wpj file... Date: Mon, 27 Aug 2001 21:15:33 +0530 From: rajendra Organization: Lucent Technologies Message-ID: <3B8A6B1D.BA36F098@lucent.com> Hi, Could somebody give me some guide lines as to how we can create vxworks image for a specifed BSP without using the Tornado GUI ? when we use the Tornado GUI to create the project we get this *.wpj file in the project directory. What is the *.wpj file is all about ? What kind of dependencies does it enforce on the make file ? Make file seems to depend on this *.wph file to build the vxWorks image... Any help would be greately appreciated. Thanks in advance.... rgds, rajendra. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: UDP Viewer Date: Mon, 27 Aug 2001 06:54:35 -0700 From: Joe Durusau Organization: Lockheed Martin Corporation Message-ID: <3B8A511B.9D3B77B7@lmco.com> References: <3b88d945$1@news.barak.net.il> If you want to do something with the data, get out your local windows programmer. If you just want to see the packets, look for sniffer utilities on the net. Speaking only for myself, Joe Durusau AE wrote: > > Hi > > I would like to send messages from my system (running VxWorks) to a remote > PC using a UDP socket. > > Unfortunately HyperTerm can only listen to a TCP/IP socket and not to UDP/IP > socket. > > Does anyone know of a window application that can display messages sent > through a UDP socket? > > Thanks. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Build demo gizmo.out hang on Windows Date: Mon, 27 Aug 2001 17:04:01 GMT From: Hong Hsu Message-ID: <3B8A7E2B.4000300@bellatlantic.net> References: <3B86A116.1EEB961B@bellatlantic.net> <9mdl9o02fa7@enews4.newsguy.com> Thanks for the writting. There is no virus scan program which is running. Thanks again, -Hong tazboy78 wrote: > Try turning off any virus scanners. > > I have seen virus scanners screw up build processes before. > > > "Hong Hsu" wrote in message > news:3B86A116.1EEB961B@bellatlantic.net... > >> Hi, I wonder this is right place to get a help for Tornado >> Prototyper 2.0 for Windows. >> If not, would you please point out reference for that? >> >> Here is my situation. The installation of Tornado Prototyper on my PC >> was successful. After >> generating the cobble.o, Building demo gizmo.out hang in Build Output >> window with following message: >> >> 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_M >> ULTI_THREAD -D_REENTRANT -I. -ID:\TORNADO\target\h -DCPU=SIMNT -c >> D:\Tornado\target\proj\gizmo\cobble.c >> vxrm D:\Tornado\target\proj\gizmo\prjObjs.lst >> >> In Output Window there is no nmsimpc statement. (I use standalone PC >> with Intel Pentium 4 processor and try to run the gizmo.out on >> simulator). I also notice that Winoldap.exe is running at this point >> and never goes away even the Wtxregd.exe is killed. What is >> Winoldap.exe doing and how to quit it? >> >> Did I miss something here in order to compile, link and run demo gizmo? >> >> Many thanks, >> 8/24 >> >> >> >> --------------------------- Newsgroups: comp.os.vxworks Subject: FORMAT Hard Disk Date: Mon, 27 Aug 2001 11:51:38 -0500 From: "Shesha Bhushan" Organization: Arizona State University Message-ID: <9me50g$gkb$1@news.asu.edu> Hi Gurus, I have connected a scsi hardisk to a Single board Computer that runs vxWorks OS. The hard disk was previously formated as NTFS. I want to format this hard disk with dosFs. How to format this previously formated disk. Tx a Bunch Shesha Bhushan shesha@asu.edu --------------------------- Newsgroups: comp.os.vxworks Subject: wqeqweqwe Date: Mon, 27 Aug 2001 11:52:08 -0500 From: "Shesha Bhushan" Organization: Arizona State University Message-ID: <9me51h$gkd$1@news.asu.edu> --------------------------- Newsgroups: comp.os.vxworks Subject: Could not open H:/.wind/shellHistory.dat to store shell history. Date: Mon, 27 Aug 2001 14:42:35 -0700 From: "Denis" Message-ID: <9meeo3$1te32$1@ID-44494.news.dfncis.de> hello, when closing the shell, i get a following message "Could not open H:/.wind/shellHistory.dat to store shell history." i am on win2k, tornado 1.01. i do have a c:\tornado\.wind\ directory. where does the "H:/" come from? do i not have some variable set? i'd really like that shellHistory feature thing to work. also, i do have WIND_BASE=C:/Tornado thanks, denis --------------------------- Newsgroups: comp.os.vxworks Subject: Re: UDP Viewer Date: Mon, 27 Aug 2001 14:59:37 -0700 From: "Denis" Message-ID: <9mefo4$1opae$1@ID-44494.news.dfncis.de> References: <3b88d945$1@news.barak.net.il> <3B8A511B.9D3B77B7@lmco.com> "Joe Durusau" wrote in message news:3B8A511B.9D3B77B7@lmco.com... | If you want to do something with the data, get out your local | windows programmer. [snip] or get out your local unix programmer, and go download cygwin :) denis --------------------------- Newsgroups: comp.os.vxworks Subject: Where's WindRiver's Embedded Software Book Date: 27 Aug 2001 15:58:38 -0700 From: ryeung@earthlink.net (Raymond Yeung) Organization: http://groups.google.com/ Message-ID: They announced the publication of such book early this year. Anyone know when it would be available in book store? --------------------------- Newsgroups: comp.os.vxworks Subject: Re: creation of vxWorks image without Tornado GUI *.wpj file... Date: Mon, 27 Aug 2001 20:09:04 -0400 From: joe durusau Organization: BELLSOUTH.net Message-ID: <3B8AE120.7E9426DA@bellsouth.net> References: <3B8A6B1D.BA36F098@lucent.com> I haven't done it for vxWorks, but I've done it for a downloadable project. You really need to start by finding all the files that get included (and indirectly included) by the Makefile. Then you can strip out all the junk, and get what you want. The .wpj file itself appears to be a distracter. The biggest problem for me was getting rid of all the extra directories that wind seems to be in love with. Sorry, I don't have a cookbook type answer. Speaking only for myself, Joe Durusau rajendra wrote: > Hi, > > Could somebody give me some guide lines as to how we can create vxworks image > for a specifed BSP without using the Tornado GUI ? > > when we use the Tornado GUI to create the project we get this *.wpj file in the > project directory. > > What is the *.wpj file is all about ? What kind of dependencies does it enforce > on the make file ? Make file seems to depend on this *.wph file to build the > vxWorks image... > > Any help would be greately appreciated. > > Thanks in advance.... > > rgds, > rajendra. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Could not open H:/.wind/shellHistory.dat to store shell history. Date: Mon, 27 Aug 2001 18:59:45 -0700 From: "Denis" Message-ID: <9metqb$1tajs$1@ID-44494.news.dfncis.de> References: <9meeo3$1te32$1@ID-44494.news.dfncis.de> "Denis" wrote in message news:9meeo3$1te32$1@ID-44494.news.dfncis.de... | hello, | | when closing the shell, i get a following message "Could | not open H:/.wind/shellHistory.dat to store shell history." | | i am on win2k, tornado 1.01. i do have a c:\tornado\.wind\ | directory. where does the "H:/" come from? do i not have | some variable set? i'd really like that shellHistory | feature thing to work. | | also, i do have WIND_BASE=C:/Tornado hate to reply to my own posts. just in case anyone cares, it turned out that i had a "HOMEDRIVE" variable set to H:, and tornado being um..not very smart, i think :) looked at the "HOME" part of it, and thought it was my $HOME (which it in no way was, and i am on win2k anyway) so...that's what my problem was... denis --------------------------- Newsgroups: comp.os.vxworks Subject: Re: creation of vxWorks image without Tornado GUI *.wpj file... Date: Mon, 27 Aug 2001 22:03:02 -0700 From: DrDiags Organization: laserlink.net Message-ID: <3B8B2606.5FE1F697@covad.net> References: <3B8A6B1D.BA36F098@lucent.com> Reply-To: drdiags@covad.net Rajendra, On the Windows version of Tornado, there is a section under the Tornado Help that states: cmpScriptLib NAME cmpScriptLib - TCL scripting commands for project manipulation TCL PROCEDURES cmpProjOpen - open a project cmpProjClose - close the currently open project cmpProjCreate - create a project and open it cmpProjCopy - create a copy of an existing project and open it cmpProjDelete - delete the currently open project cmpProjHandleGet - get the handle to the currently open project cmpBuild - build the project cmpBuildRuleSet - set the build rule for the project cmpBuildSpecSet - set the active buildSpec cmpBuildSpecCreate - create a new buildSpec cmpBuildMacroSet - set a build macro cmpBuildMacroGet - get a build macro cmpBuildFlagsAdd - add flags to the build cmpAdd - add components to vxWorks cmpRemove - remove components from vxWorks cmpListGet - get a list of components in vxWorks cmpParamSet - set a parameter in the current project cmpParamGet - get a parameter cmpFileListGet - get a list of files in the project cmpFileAdd - add a file to the project cmpFileRemove - remove a file from the project cmpLoad - dynamically load components autoscale - automatically scale vxWorks cmpUnavailListGet - list unavailable components cmpValidate - validate the component configuration cmpTest - test the component descriptor files for errors cmpInfo - get info about a set of components DESCRIPTION Tornado comes with a project facility that allows one to graphically organize source files into projects, change the build rules, and configure the operating system. See the Tornado Users Guide for details on what facilities the project facility offers. This library provides a way to perform these tasks in a script. To use this library, one must first source this library from wtxtcl. As an example, suppose you wanted to do a clean build of your project every night. One could write a script as follows: example.tcl: source [wtxPath host resource app-config Project cmpScriptLib.tcl] cmpProjOpen myProjectFile.wpj cmpBuild clean cmpBuild One can execute the script from a UNIX or Windows command shell by typing "wtxtcl example.tcl." Alternatively, one can type the scripting commands directly into the wtxtcl shell. SEE ALSO cmpScriptLib, dataDocLib Of course, you can build vxWorks images using the old Tornado 1.0.1 build method. Make a copy of the BSP directory and modify the config.h file to enable or disable the components you want for your vxWorks image. Then you can just use the BSP makefile and make your vxWorks kernel image or bootrom. rajendra wrote: > > Hi, > > Could somebody give me some guide lines as to how we can create vxworks image > for a specifed BSP without using the Tornado GUI ? > > when we use the Tornado GUI to create the project we get this *.wpj file in the > project directory. > > What is the *.wpj file is all about ? What kind of dependencies does it enforce > on the make file ? Make file seems to depend on this *.wph file to build the > vxWorks image... > > Any help would be greately appreciated. > > Thanks in advance.... > > rgds, > rajendra. - -- http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html (vxWorks FAQ home page) --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Telnet sessions Date: 28 Aug 2001 07:26:17 GMT From: Johan Borkhuis Organization: Agere Systems Message-ID: References: <3B84427C.57B71C13@agilent.com> Les Hammer wrote: > Hi, > > Does anyone have experience in getting more than one > telnet session to work with VxWorks? When I try to > open a second telnet session I get a "Sorry, the system > is engaged" error, even though I used telnetInit(2) in my > usrAppInit.c, and I have the > void telnetInit(int maxConnNum) > code. There is an unsupported telnet server available from WindRiver, just ask your FAE for this code. You can also add protocols to other ports, and add your functions to this. > Also, does anybody have experience in a telnet session > that goes to a limited user interface, rather than the VxWorks > shell? Using this telnet server it is possible to add your own module to it. 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: Strict timing Date: 28 Aug 2001 07:37:51 GMT From: Johan Borkhuis Organization: Agere Systems Message-ID: References: <9m2cif$82j$1@fe2.cs.interbusiness.it> "Andrea della Porta" wrote: > Does anyone know how to obtain precise task_delay() when several task is > running? How do priority level among tasks affect this delay()? I've to > simulate I2C bus waveforms but it seems that the timing passed to > task_delay() is not respected. The taskDelay function delay the processing for a number of ticks, unless another (higher priority) task is running. If this is not the case you can calculate the delay using the following: Minimal delay = (ticks - 1) * tick-time Maximal delay = ticks * tick-time So for a delay of 5 ticks and a tick-time of 10 msec the minimal delay is 40 msec and the maximum is 50 msec (if no higher priority task is running). If you specify a delay of 0 you allow other task of higher or equal priority to run. In your case it would be best to specify the highest priority for your timing task, and also specify a higher system clock frequency, for example 500 instead of 60 (the default). This however will cost you some processing because you do a lot more timer interrupts. Groeten, Johan - -- o o o o o o o . . . _____________________________ o _____ || Johan Borkhuis | .][__n_n_|DD[ ====_____ | borkhuis@agere.com | >(________|__|_[_________]_|__________________________| _/oo OOOOO oo` ooo ooo 'o!o!o o!o!o` === VxWorks FAQ: http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html === --------------------------- Newsgroups: comp.os.vxworks Subject: Re: SLIP protocol under Windows 2000/NT Date: 28 Aug 2001 07:38:58 GMT From: Johan Borkhuis Organization: Agere Systems Message-ID: References: <9m2fga$s1n$1@newsserv.intranet.gr> "Robert Berger" wrote: > I just got VxWorks/Tornado and I am trying to use connect to my PC. > Unfortunately there is no ethernet interface on my BSP and the SLIP > protocol does not seem to work. > > Can someone help on how to connect a BSP using SLIP to a PC running > Windows 2000/NT ? You can use the normal serial interface, use WDB on the serial interface or use PPP. 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: Tornado AE Date: 28 Aug 2001 00:57:49 -0700 From: eeyore@dial.pipex.com (Eeyore) Organization: http://groups.google.com/ Message-ID: <2c7641d6.0108272357.57940f6a@posting.google.com> References: <3B8A5B71.F9C6A93A@hotmail.com> Hello, Been looking at it too; try to get an eval copy to play with it if you can. Graham Marshall wrote in message news:<3B8A5B71.F9C6A93A@hotmail.com>... > Hi Guys, > > My company is interested in obtaining Tornado AE for x86. > > Does anyone have any information (other than the marketing) about this > product? > > I am particularly interested in: > * C++ Support (including which version of gcc does it use?) A custom version created by Wind River called gcc-2.96. It is similar to FSF versions of the 2.9x ilk. Claims to have better C++ support than T2; haven't really tested that out much. > * Does Protected Domain support require hooks to be added by the > developer or does it come for 'free'? No hooks needed; this is not like the old VxVMI product. Create them either dynamically using the API, or even at build time and have them boot up with the OS and start running automatically. > * Is it an extension of VxWorks 5.4 or is it a rewrite? Basically an extension. All the things you know and love in 5.4 are there, plus some new goodies like the protection domains. Also, resource reclamation and object naming. > * Are the tools robust and useful? Unlike T2, there's a single model for both command line and GUI. This is cool if you hate GUIs in general (like me), but find them convenient for some things (I admit it, some things are easier from a GUI). Some of the hierarchical project model seems interesting too, though I'm still exploring exactly what that can do. > I'd be really grateful for any feedback - can you post it to the group > in the interests of sharing? I'd be interested in hearing other people's comments too. Eeyore. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: creation of vxWorks image without Tornado GUI *.wpj file... Date: 28 Aug 2001 07:57:54 GMT From: Johan Borkhuis Organization: Agere Systems Message-ID: References: <3B8A6B1D.BA36F098@lucent.com> rajendra wrote: > Hi, > > Could somebody give me some guide lines as to how we can create vxworks > image for a specifed BSP without using the Tornado GUI ? > > when we use the Tornado GUI to create the project we get this *.wpj > file in the project directory. > > What is the *.wpj file is all about ? What kind of dependencies does it > enforce on the make file ? Make file seems to depend on this *.wph file > to build the vxWorks image... You can do a "make vxWorks" from your BSP directory. If you want to modify any included funtion you can change it in config.h. 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: creation of vxWorks image without Tornado GUI *.wpj file... Date: Tue, 28 Aug 2001 12:40:59 +0400 From: "Arthur V. Grigorian" Organization: Auriga, Inc. Message-ID: <3B8B591B.C70D7AAE@auriga.ru> References: <3B8A6B1D.BA36F098@lucent.com> Hello, Another quick HOWTO on the subj (in case you hate reading manual ;) - In the Tornado GUI create a project and configure it according to your requirements; then build it and make sure it works. - Exit Tornado GUI and get the prjParams.h file from your project directory. - Copy BSP directory anywhere you like (not to spoil the original) and put your prjParams.h file into it; then enter this directory and execute something like the following: (example for Windows based Tornado) %WIND_BASE%\host\x86-win32\bin\torvars.bat make vxWorks.st ADDED_MODULES="myModule.o" \ ADDED_CFLAGS=" -DPRJ_BUILD -DUSER_APPL_INIT={myAppInit();}" where "myModule.o" is your object file(s) and "myAppInit" is the main initialization function for your software. Note that in this example I presume standalone vxWorks configuration (vxWorks.st). - Note that Tornado GUI doesn't use project configuration to build bootrom, so you should not use -DPRJBUILD either if you need to build your software specific bootrom. To do that you can replace some BSP files/configuration (like config.h) by your ones in the BSP build directory and run something like that: %WIND_BASE%\host\x86-win32\bin\torvars.bat make bootrom ADDED_MODULES="myBootModule.o" Hope that helps, __ R2 >rajendra wrote: > > Hi, > > Could somebody give me some guide lines as to how we can create vxworks image > for a specifed BSP without using the Tornado GUI ? > > when we use the Tornado GUI to create the project we get this *.wpj file in the > project directory. > > What is the *.wpj file is all about ? What kind of dependencies does it enforce > on the make file ? Make file seems to depend on this *.wph file to build the > vxWorks image... > > Any help would be greately appreciated. > > Thanks in advance.... > > rgds, > rajendra. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: creation of vxWorks image without Tornado GUI *.wpj file... Date: Tue, 28 Aug 2001 14:25:17 +0530 From: rajendra Organization: Lucent Technologies Message-ID: <3B8B5C75.16E67930@lucent.com> References: <3B8A6B1D.BA36F098@lucent.com> Hi Johan, Thanks for the reply. But I was referring to a bootable vxWorks image with the project specific code in elf format. (i.e this image should contain the plain vxWorks and the project specific code as well.) I suppose doing a simple Make in the BSP directory gives only the plain vxWorks image. Can I edit the Makefile in the BSP directory and include all the project specific *.obj files ? How do I include all the required components ? thanks and best regards, rajendra. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: creation of vxWorks image without Tornado GUI *.wpj file... Date: 28 Aug 2001 10:34:41 GMT From: Johan Borkhuis Organization: Agere Systems Message-ID: References: <3B8A6B1D.BA36F098@lucent.com> <3B8B5C75.16E67930@lucent.com> rajendra wrote: > Hi Johan, > > Thanks for the reply. > But I was referring to a bootable vxWorks image with the project > specific code in elf format. (i.e this image should contain the plain > vxWorks and the project specific code as well.) > > I suppose doing a simple Make in the BSP directory gives only the plain > vxWorks image. Can I edit the Makefile in the BSP directory and include > all the project specific *.obj files ? How do I include all the > required components ? Yes, you can do that by adding your object files to the MACH_EXTRA macro. Also if you place a C-file in the BSP directory it will get included in the build. I do not prefer this method, as you are adding specific stuff to something that should be generic. What you also could do is base your project on an example project (under target\proj) and do a make vxWorks there. You still have to modify the Makefile for this.. Groeten, Johan - -- o o o o o o o . . . _____________________________ o _____ || Johan Borkhuis | .][__n_n_|DD[ ====_____ | borkhuis@agere.com | >(________|__|_[_________]_|__________________________| _/oo OOOOO oo` ooo ooo 'o!o!o o!o!o` === VxWorks FAQ: http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html === --------------------------- Newsgroups: comp.os.vxworks Subject: How to stop all tasks when exception or breakpoint Date: Tue, 28 Aug 2001 18:56:28 +0800 From: "Jo" Organization: SEEDNet News Service Message-ID: <9mftbe$lgj$1@news.seed.net.tw> Hi, please give me advises, thanks: The Environment: Tornado 2.0 MIPS R4650 library, connect host and target via ethernet, I don't have any H/W ICE tools. I found that my BSP might not support system mode, so when breakpoint or exception happens, only 1 tasks attach and halt. but what I want is to halt all tasks to know other tasks status. I tried to add some #define in BSP config.h, but failed. Anyone can tell me how to stop all tasks ? should I change BSP config.h to support system mode ? what component should I add ? any reference manual or web page ? Thanks a lot, jo - ----------------------------------------------------- --------------------------- End of New-News digest ********************** From vxwexplo-errs@csg.lbl.gov Tue Aug 28 05:43:24 2001 From: "David Laight" Date: Tue Aug 28 05:43:27 PDT 2001 Subject: Re: rebootHook list order problem > I'd like to set up a rebootHook that does some network activity, but the > VxWorks documentation states that this cannot be done because the reboot > Hook list is performed in the order of addition, so the network is shut > down before any of the user routines get to run. One might have hoped > they would run the list in reverse order of addition! Does anyone have > any suggestions? Yep - 'tis broken! Unfortunately it works 'as documented' so I guess you won't get WRS to do anything about it. Like many othervxWorks issues..... Suggestions: 1) reverse the list prior to calling reboot - I recall that some BSP code is called during the reboot, maybe before the hooks? 2) write some code to add an item to the other end of the list! Ok - both need you to reverse-engineer rebootHookAdd(), but I don't recall that it is one of the worlds most complex routines! David From vxwexplo-errs@csg.lbl.gov Tue Aug 21 10:06:06 2001 From: "David Laight" Date: Tue Aug 28 06:22:42 PDT 2001 Subject: Re: dosfs file system corruptions > But in fact, I know one such bug in DosFs 2.0, which is related to > directories and causes lost cluster chains. Suppose you have a directory > "MyDir1", and a file "MyFile1", you can legally do "ename > ("MyFile1","MyDir1/MyFile1"), that is fine. But if you do > rename("MyDir1","MyDir1/MyDir2") you will loose all subdirectories and files do you mean the 'recursive one'? > under "MyDir1", which will be diagnozed as lost chains by CHKDSK later. DosFs 1 will lose the old contents of mydir if you do: rename( "myfile", "mydir" ) this is VERY easy to do with NFS... mv myfile mydir So is dosfs 2 any better here? dosFs 1 also gets confused by empty files. I've forgotten the exact sequence, but open, close, reboot (i think) leaves a FAT block allocated to the file and free list. I wrote a file system checker in order to try and isolate some corruptions - they stopped happening if we checked the fs..... but we never could reproduce the error, just suddenly got VERY corrupt FS. I read somewhere that the vxworks CHKDSK code uses the second copy of the FAT as temporary storage. What does it do if there is only one copy - probably sensible on compact flash etc. David From vxwexplo-errs@csg.lbl.gov Tue Aug 28 06:22:39 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 GAA24478 for ; Tue, 28 Aug 2001 06:22:39 -0700 (PDT) Received: from SpamWall.lbl.gov (localhost [127.0.0.1]) by postal1.lbl.gov (8.11.2/8.11.2) with ESMTP id f7SDMcx15572 for ; Tue, 28 Aug 2001 06:22:38 -0700 (PDT) Received: from wiproecmx1.wipro.com (wiproecmx1.wipro.com [164.164.31.5]) by SpamWall.lbl.gov (8.11.2/8.11.2) with ESMTP id f7SDMZ515562 for ; Tue, 28 Aug 2001 06:22:36 -0700 (PDT) Received: from ecvwall11.wipro.com (ecvwall1.wipro.com [164.164.23.6]) by wiproecmx1.wipro.com (8.11.3/8.11.3) with SMTP id f7SNeR116952 for ; Tue, 28 Aug 2001 18:40:28 -0500 (GMT) Received: from jerry ([192.168.85.139]) by ecmail.mail.wipro.com (Netscape Messaging Server 4.15) with ESMTP id GIS51L00.QMV for ; Tue, 28 Aug 2001 18:50:09 +0530 Message-ID: <009901c12fc4$6cb08dd0$8b55a8c0@wipro.com> Reply-To: "Jerry P Nicholas" From: "Jerry P N" To: "VxWorks" Subject: VxWorks Telnet task Date: Tue, 28 Aug 2001 18:52:02 +0530 Organization: Wipro Technologies MIME-Version: 1.0 Content-Type: multipart/mixed; boundary="----=_NextPart_000_0095_01C12FF2.863E1050" X-Priority: 1 X-MSMail-Priority: High X-Mailer: Microsoft Outlook Express 5.50.4807.1700 X-MimeOLE: Produced By Microsoft MimeOLE V5.50.4807.1700 This is a multi-part message in MIME format. ------=_NextPart_000_0095_01C12FF2.863E1050 Content-Type: multipart/alternative; boundary="----=_NextPart_001_0096_01C12FF2.863E1050" ------=_NextPart_001_0096_01C12FF2.863E1050 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable Hi I am trying to initialise the Telnet Server using "telnetInit( ) " = function call ( by including Configuration macro INCLUDE_TELNET) but the = task gets created as shown below NAME ENTRY TID PRI STATUS PC SP = ERRNO DELAY tTelnetd _telnetd efde50 2 DELAY 13651e efddd4 = 30065 37 the task is created with status as "Delay" and priority of "2" it shows = the error as "S_taskLib_NAME_NOT_FOUND" I have tried 1) Resuming the task 2) Verifying the task 3) Verifying the = ID of the task 4)Activating the Task. the result of the same is in the Text file attached with this mail=20 The results observed were that=20 -> taskId exists, -> the priority of the task can be changed but this does not change the = Delay status. -> if the telnettask is deleted and spawned again with a different = priority we get the message that "Telnet already initialised" How can this problem be rectified=20 regards Jerry ------=_NextPart_001_0096_01C12FF2.863E1050 Content-Type: text/html; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable
Hi
   I am trying to initialise = the Telnet=20 Server using "telnetInit( ) " function call ( by including Configuration = macro INCLUDE_TELNET) but the task gets created as shown = below
 
NAME    =20  ENTRY       =20 TID      PRI    =20 STATUS     PC     =    =20 SP         = ERRNO    =20 DELAY
tTelnetd    _telnetd=20     efde50   2    =    =20 DELAY      13651e   efddd4=20     30065         = 37
 
the  task is created with status = as "Delay"=20 and priority of "2" it shows the error as =20 "S_taskLib_NAME_NOT_FOUND"
 
I have tried 1) Resuming the task 2) = Verifying the=20 task 3) Verifying the ID of the task 4)Activating=20 the Task.
 
the result of the same is in the Text = file attached with this mail
 
The results observed were = that 
-> taskId exists,
-> the priority of the task can = be changed=20 but this does not change the Delay status.
-> if the telnettask is deleted and = spawned=20 again with a different priority we get the message that "Telnet already=20 initialised"
 
How can this problem be rectified =
 
regards
Jerry
------=_NextPart_001_0096_01C12FF2.863E1050-- ------=_NextPart_000_0095_01C12FF2.863E1050 Content-Type: text/plain; name="telnet.txt" Content-Transfer-Encoding: quoted-printable Content-Disposition: attachment; filename="telnet.txt" -> i NAME ENTRY TID PRI STATUS PC SP ERRNO = DELAY ---------- ------------ -------- --- ---------- -------- -------- = ------- ----- tExcTask _excTask feb204 0 PEND 187eba feb174 = 0 0 tLogTask _logTask fe8900 0 PEND 187eba fe886c = 0 0 tRlogind _rlogind effd00 2 DELAY 13651e effa20 = 30065 37 tTelnetd _telnetd efde50 2 DELAY 13651e efddd4 = 30065 37 =20 -> taskIdVerify 0xefde50 value =3D 0 =3D 0x0 -> taskIsSuspended 0xefde50 value =3D 0 =3D 0x0 -> taskResume 0xefde50 value =3D 0 =3D 0x0 -> i NAME ENTRY TID PRI STATUS PC SP ERRNO = DELAY ---------- ------------ -------- --- ---------- -------- -------- = ------- ----- tExcTask _excTask feb204 0 PEND 187eba feb174 = 0 0 tLogTask _logTask fe8900 0 PEND 187eba fe886c = 0 0 tRlogind _rlogind effd00 2 DELAY 13651e effa20 = 30065 15 tTelnetd _telnetd efde50 2 DELAY 13651e efddd4 = 30065 14 -> taskActivate 0xefde50 value =3D 0 =3D 0x0 -> i NAME ENTRY TID PRI STATUS PC SP ERRNO = DELAY ---------- ------------ -------- --- ---------- -------- -------- = ------- ----- tExcTask _excTask feb204 0 PEND 187eba feb174 = 0 0 tLogTask _logTask fe8900 0 PEND 187eba fe886c = 0 0 tRlogind _rlogind effd00 2 DELAY 13651e effa20 = 30065 5 tTelnetd _telnetd efde50 2 DELAY 13651e efddd4 = 30065 4 -> ------=_NextPart_000_0095_01C12FF2.863E1050 Content-Type: text/plain; name="Wipro_Disclaimer.txt" Content-Transfer-Encoding: 7bit Content-Disposition: attachment; filename="Wipro_Disclaimer.txt" ----------------------------------------------------------------------------------------------------------------------- Information transmitted by this E-MAIL is proprietary to Wipro Limited and is intended for use only by the individual or entity to which it is addressed, and may contain information that is privileged, confidential or exempt from disclosure under applicable law. If you are not the intended recipient or it appears that this mail has been forwarded to you without proper authority, you are notified that any use or dissemination of this information in any manner is strictly prohibited. In such cases, please notify us immediately at mailto:mailadmin@wipro.com and delete this mail from your records. ------------------------------------------------------------------------------------------------------------------------ ------=_NextPart_000_0095_01C12FF2.863E1050-- From vxwexplo-errs@csg.lbl.gov Tue Aug 28 08:18:49 2001 From: =?iso-8859-1?q?ram=20mohan?= Date: Tue Aug 28 08:18:51 PDT 2001 Subject: prob in executing c++ progs in target shell hi all, we r using tornado/ppc combination(dev on windows) there is no prob in downloading and executing c progs but when we change the same c prog to c++ prog we are getting error sayng Undefined symbol xxxx where xxxx is our entry point. we've included all cpp modules in boot image. can anyone help us out from here? thanks in advance Mohan From vxwexplo-errs@csg.lbl.gov Tue Aug 28 08:41:34 2001 From: =?iso-8859-1?q?ram=20mohan?= Date: Tue Aug 28 08:41:36 PDT 2001 Subject: Re: Command Line Options to Create Project in Tornado hi, there is no command to create proj proj is not something like compiling to have some command like cc anyway where r u facing prob when u have multiple dirs i mean can u describe wht's ur prob is? regards Mohan --- the vxWorks Users Group Exploder wrote: > Submitted-by vxwexplo-errs@csg.lbl.gov Sat Aug 25 > 04:04:03 2001 > Submitted-by: "vanitha ramaswami" > > > Hello, > Is there any command line option available to > create a project in Tornado ?. It's very tedious to > create a project thro' the GUI when we have so many > directories. > > > - Vanitha > > > Get 250 color business cards for FREE! > http://businesscards.lycos.com/vp/fastpath/ > > ********** > > This is a user group mailing list for vxWorks > related topics > see http://www-vxw.lbl.gov/vxworks for posting > and subscription info From vxwexplo-errs@csg.lbl.gov Tue Aug 28 09:17:44 2001 From: "Mike Anderson" Date: Tue Aug 28 09:17:46 PDT 2001 Subject: RE: example of a tcp server VxWorks Greetings! > Submitted-by: Anuradha Pani > > hi, > > could anyone give any pointers to any links or sites > which have a tcp server example for vxworks? > The examples from the Stevens book will work pretty much out of the box on VxWorks. Otherwise hit the WindSurf website because I believe that there are a couple of examples there as well. HTH, Mike -- ======================================================= T h e P T R G r o u p, I n c. ======================================================= Embedded, Real-Time Solutions and Training Mike Anderson mailto:mike@ThePTRGroup.com Chief Scientist http://www.ThePTRGroup.com V: (703) 585-9384 F: (703) 430-3748 ======================================================= "Software development is like making a baby... You can't make a baby in one month by impregnating nine women. Some things just take time." From vxwexplo-errs@csg.lbl.gov Tue Aug 28 11:55:34 2001 From: Rohit_Kapur@raytheon.com Date: Tue Aug 28 11:55:36 PDT 2001 Received: by ressud-as01.res.ray.com(Lotus SMTP MTA v4.6.6 (890.1 7-16-1999)) id 85256AB6.0067E8A5 ; Tue, 28 Aug 2001 14:54:57 -0400 X-Lotus-FromDomain: RES From: Rohit_Kapur@raytheon.com To: vxwexplo@lbl.gov (the vxWorks Users Group Exploder) Message-ID: <85256AB6.0067E72F.00@ressud-as01.res.ray.com> Date: Tue, 28 Aug 2001 14:45:18 -0400 Subject: unsubscribe From vxwexplo-errs@csg.lbl.gov Tue Aug 28 14:22:46 2001 From: Anuradha Pani Date: Tue Aug 28 14:22:48 PDT 2001 Subject: question on tcp server again hi everybody, firstof all, thanks to all those who replied to my mail. i do have access to vxWorks training manual and i have gone through the example, so let me make my question more specific. i have include below my server code. this is what i do: 1. i start the server either directly from the command line, or by spawing from another task. 2. everything works fine, i mean the client sends messages and the server responds correctly. the problem occurs when i delete the server task by doing a taskDelete(). the server socket never gets closed. so the next time i bring up my server, i get a bind error, and the error being 'address in use'. 3. i can actually see the server socket, port and IP address still in use by doing a inetstatShow(), even after i have done a taskDelete(). I also verify that the task is deleted by typing 'i' on windSh. so because of this problem i have to recompile using a different port number for the server port, which is definitely not what i want. i just don't know what to do at this point, any suggestion would be truly welcome. thanks again anuradha /****************************************************/ /* File: tcpServer2.c - Server application 2 */ #include #include #include #include #include #include #include #include #include #include #include #include #include #include "signal.h" #include "tcpHeader.h" void tcpServerWorkTask (int, char*, u_short); int tcpServer2(void) { struct sockaddr_in serverAddr; struct sockaddr_in clientAddr; int sockAddrSize; int sockfd; int newfd; int taskId; int taskCount = 0; char workName[16]; int status; int localhostlen; char localhost[100]; int optval; int errno; /* Get local host name */ status = gethostname(localhost, localhostlen); /* Debug Statement */ printf("\n\nServer started...\n"); printf("......................\n"); /* set up local address */ sockAddrSize = sizeof (struct sockaddr_in); bzero ((char*) &serverAddr, sockAddrSize); serverAddr.sin_family = AF_INET; serverAddr.sin_port = htons (SERVER_PORT); serverAddr.sin_addr.s_addr = htonl (INADDR_ANY); /* Debug Statement */ printf("Local address initialization done correctly\n"); /* Create a TCP socket */ if ( (sockfd = socket(AF_INET,SOCK_STREAM,0) ) < 0) { printf("Error in creating server socket, exiting...\n"); close(sockfd); return -1; } /* Debug Statement */ printf("Server socket with #%d created successfully on \"%s\" at port %d\n", sockfd,localhost, SERVER_PORT); /* Set socket options */ /* For some reason, the SO_REUSEADDR option doesn't work */ optval = 100; status = setsockopt (sockfd, SOL_SOCKET, SO_REUSEADDR, (char*) &optval, sizeof (optval)); if(status== 0) printf("Setting socket options successful\n"); /* Bind the socket to local address */ if (bind(sockfd, (struct sockaddr*)&serverAddr,sockAddrSize) < 0) { printf("Error in binding socket, exiting...\n"); errno = errnoGet(); printf("Error = %d\n", errno); close(sockfd); return -1; } /* Debug Statement */ printf("Binding successful for server\n"); /* listen to incoming connections */ if (listen (sockfd, 5) < 0) { printf("Error listening, exiting...\n"); close(sockfd); return -1; } /* Debug Statement */ printf("Server listening successfully\n"); printf("Ready to accept calls now\n"); /* accept new connections */ for(; ;) { if ((newfd = accept(sockfd, (struct sockaddr*) &clientAddr, &sockAddrSize)) <0) { printf("Accept error, exiting...\n"); close(sockfd); return -1; } sprintf(workName, "tTcpWork%d", taskCount++); /* Debug statement */ printf("Spawning serverChildTask #%d\n", taskCount); if ((taskId = taskSpawn(workName, 10,0, 4096, (FUNCPTR) tcpServerWorkTask, newfd, (int) inet_ntoa(clientAddr.sin_addr), ntohs (clientAddr.sin_port),0,0,0,0,0,0,0)) < 0) { printf("Error in spawning task, exiting....\n"); close (newfd); } printf("Server spawned task with task id = %d\n", taskId); taskDelay(1); } /* end of infinite for loop */ close(sockfd); printf("CLOSED SERVER SOCKET\n"); } /* end of tcpServer */ __________________________________________________ Do You Yahoo!? Make international calls for as low as $.04/minute with Yahoo! Messenger http://phonecard.yahoo.com/ From vxwexplo-errs@csg.lbl.gov Tue Aug 28 15:04:00 2001 From: "Mike Anderson" Date: Tue Aug 28 15:04:03 PDT 2001 Subject: RE: question on tcp server again VxWorks Greetings! > Submitted-by: Anuradha Pani > [snip] > > the problem occurs when i delete the server task by > doing a taskDelete(). the server socket never gets > closed. so the next time i bring up my server, i get a > bind error, and the error being 'address in use'. > After you get the fd from the socket call, you will need to call setsockopt with the SO_REUSEADDR option. From the VxWorks documentation: SO_REUSEADDR -- Reusing a Socket Address Specify the SO_REUSEADDR option to bind a stream socket to a local port that may be still bound to another stream socket: int optval = 1; setsockopt (sock, SOL_SOCKET, SO_REUSEADDR, &optval, sizeof (optval)); The value at optval is an integer (type int), either 1 (on) or 0 (off). When the SO_REUSEADDR option is turned on, applications may bind a stream socket to a local port even if it is still bound to another stream socket, if that other socket is associated with a "zombie" protocol control block context not yet freed from previous sessions. The uniqueness of port number combinations for each connection is still preserved through sanity checks performed at actual connection setup time. If this option is not turned on and an application attempts to bind to a port which is being used by a zombie protocol control block, the bind( ) call fails. HTH, Mike -- ======================================================= T h e P T R G r o u p, I n c. ======================================================= Embedded, Real-Time Solutions and Training Mike Anderson mailto:mike@ThePTRGroup.com Chief Scientist http://www.ThePTRGroup.com V: (703) 585-9384 F: (703) 430-3748 ======================================================= "Software development is like making a baby... You can't make a baby in one month by impregnating nine women. Some things just take time." From vxwexplo-errs@csg.lbl.gov Tue Aug 28 17:30:20 2001 From: Sean McNeil Date: Tue Aug 28 17:30:22 PDT 2001 Subject: Consulting available Located in the San Francisco area, I have been very busy supporting several VxWorks-based products including a web tablet, set-top box, and an internet appliance. My current commitments are beginning to wind down to the point of having time to offer my services. If you are looking for someone with advanced working knowledge of VxWorks and embedded programming, I'm looking forward to hearing from you. Thanks, Sean Sean McNeil Consulting sean@mcneil.com From vxwexplo-errs@csg.lbl.gov Tue Aug 28 18:53:54 2001 From: "Tan, Tat Kin" Date: Tue Aug 28 18:53:56 PDT 2001 Subject: ifShow counter Hello fellow vxWorks gurus, The command at windsh prompt: -> ifShow "cs" cs (unit number 0): Flags: (0x63) UP BROADCAST ARP RUNNING Type: ETHERNET_CSMACD Internet address: 10.254.17.100 Broadcast address: 10.254.17.255 Netmask 0xff000000 Subnetmask 0xffffff00 Ethernet address is 00:02:b3:3b:fc:38 Metric is 0 Maximum Transfer Unit size is 1500 512 packets received; 0 packets sent 512 multicast packets received 2162 multicast packets sent 0 input errors; 0 output errors 0 collisions; 0 dropped value = -16 = 0xfffffff0 = _func_taskRegsShowRtn + 0xffd33990 -> will generally print out various counters of network packets, ie: how many packets the device has tranmitted and received. One of the counters, indicating packet sent is always stayed at "0" (zero). This does not happened in other devices. I wonder if there is anyone been facing the similar problem? Or is it only to that particular network device (a bug???) causing zero value. The driver itself has a similar function name csShow() in which i have cheched and found that the "packet sent" indicator is also showing value "0". Can anyone explain this? By the way the device is CS8900 ethernet chip. thanks in advance. With Warm Regards, Tat Kin Tan From vxwexplo-errs@csg.lbl.gov Wed Aug 29 04:03:38 2001 From: Vxworks Exploder Date: Wed Aug 29 04:03:41 PDT 2001 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Wed Aug 29 04:03:11 PDT 2001 Subject: ifGetAddr problem Subject: Re: vxTicks Being Cleared to Zero? Subject: Re: Anyone get VxWorks to boot quickly Subject: question on porting a protocol stack Subject: modified "copy()" api problem Subject: [Q] ping problem between vxsim0 and vxsim1 simulators for solaris 7 using ppp Subject: Re: modified "copy()" api problem Subject: could not boot VxWorks bootImage on a MDP860T Pro Subject: Re: UDP Viewer Subject: reboot blues Subject: Re: ifGetAddr problem Subject: Re: ifGetAddr problem Subject: Re: DHCP server on multi i/faces Subject: Re: Tornado AE Subject: Does anybody know how to do this ??? Subject: Re: creation of vxWorks image without Tornado GUI *.wpj file... Subject: Why not bring up the network interface! Subject: Re: question on porting a protocol stack ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: ifGetAddr problem Date: Tue, 28 Aug 2001 11:46:08 GMT From: Gururaj patkar Organization: Optimum Online Message-ID: <3B8B83FF.96CFF27@optonline.net> Hi, Can somebody detail me on the internal working of ifGetAddr? I guess the vxworks software is trying to access a linked list internally to extract "ip address" in string form? Also, is ifAddrGet buggy? In the beginning using ifAddrGet in my code works fine, after the system is running for a short while, intermittently and after a while consistently ifAddrGet fails and returns ERROR. Both the arguments to ifAddrGet are valid in the runtime code. when it fails in the runtime code at that moment if I try to execute ifAddrGet from TARGET SHELL it fails too!, again both the parameters- the ethernet interace string and buffer supplied are valid. thanks for any help. gururaj --------------------------- Newsgroups: comp.os.vxworks Subject: Re: vxTicks Being Cleared to Zero? Date: Tue, 28 Aug 2001 10:27:13 +0100 From: "Dave Korn" Organization: Lumber Cartel (tinlc) Members #2234-2237 (owing to browser refresh) Message-ID: References: <0Mei7.168$td.80985@typhoon1.gnilink.net> Pat DiRienzo wrote in message <0Mei7.168$td.80985@typhoon1.gnilink.net>... >We are using the system clock as a time of day clock on our system (single >board computer and other VME cards) by using clock_gettime calls. The clock >works fine (time of day correctly displayed on our system screen) until it >reaches a certain point in our code, after monitoring vxTicks I found that >it was being set to zero and causing the time of day clock on our system to >display an incorrect time. I understand that vxTicks (the vxWorks global) >is updated by tickAnnounce from our system timer ISR and that vxTicks is >used by vxWork's to provide a time of day thru clock_gettime calls. >vxAbsTicks remained OK thru all of this. So why is vxTicks being set to >zero, It does not appear to be a wrapping problem. If you're using SCSI and PPc processors, it's a known bug that I discovered happening when doing large (> 1 Mb) reads or writes, although I can't say if it might not happen under other conditions as well. To test if this is what's happening, taskSuspend the tScsiTask from the command shell at some point before the vxTicks variable gets reset to zero in the normal flow of your program code; your application will suspend at some point waiting for a scsi operation to complete. At that point, taskSuspend your application and taskResume the scsi task; if the vxTicks variable then resets to zero, you know it's this bug and not anything that your app is doing. The solution is to get onto WRS support and shout at them until they give you the source code to clockLib, and replace vxTicks with vxAbsTicks throughout. Here's the analysis of the bug I reported to WRS: - -------snip!------- The problem we have been experiencing with the SCSI driver is not a transient failure of a single call to clock_gettime but a permanent mis-setting of the clock. Judging from the disassembly of clock_gettime, after checking that the clockLib has been init'ed, clock_gettime accesses a structure called _clockRealtime which looks like: struct posix_clock { int ticks_per_sec; // I made these names up but they int unknown; // int tick_offset; // seem to describe the data correctly int moreunknowns[?]; // } _clockRealtime; It subtracts the value in tick_offset from the vxTicks variable, probably in order to establish a time zero at the point when clockLibInit is called, and then divides the remainder by ticks_per_sec, before building the tv_secs and tv_nano values to return to the calling application. - -> lkup "_clockRealtime" _clockRealtime 0x001fc270 bss (suilven:vxWorks.sym) value = 0 = 0x0 - -> vxTicks vxTicks = 0x1f5870: value = 3883 = 0xf2b - -> d 0x1fc270, 4, 4 001fc270: 00000000 00000000 00000000 00000000 *................* value = 21 = 0x15 - -> d 0x1fc270, 4, 4 001fc270: 00000000 00000000 00000000 00000000 *................* value = 21 = 0x15 - -> vxTicks vxTicks = 0x1f5870: value = 5488 = 0x1570 - -> clockLibInit 0 value = 0 = 0x0 - -> vxTicks vxTicks = 0x1f5870: value = 5690 = 0x163a - -> d 0x1fc270, 4, 4 001fc270: 0000003c 00fe502a 000015eb 00000000 *.....P*........* value = 21 = 0x15 The problem seems to be that after reading 1Mb or more from the SCSI drive in a single operation, vxTicks has been reset to zero. However the member tick_offset in _clockRealtime remains with its old value. When clock_gettime is called thereafter the subtraction marked at (A) below produces a small negative result, which is eventually passed back to the application as a huge positive integer. This situation persists until the value of vxTicks finally increments past and exceeds the value of tick_offset again; at that point the value returned by clock_gettime wraps around to zero again. This problem only occurs after a single read from the SCSI hard drive of 1Mb or more; smaller reads don't cause the problem even if they cumulatively amount to over a megabyte. The zeroing of vxTicks appears to be performed by the scsi driver itself (tScsiTask) since that can be suspended from the vxshell, while the application issues the read command. The app. then pends waiting for tScsiTask to complete; while it pends, it can also be suspended from the shell. At that point vxTicks has not yet been zeroed. If the tScsiTask is then resumed from the shell, the read goes ahead, and vxTicks becomes zeroed; this happens while the app. is still suspended and pended, and so most probably is done by tScsiTask. DaveK - ------------- Commented disassembly of clock_gettime follows ------------- (gdb) disassemble clock_gettime Dump of assembler code for function clock_gettime: ; function prologue 0x159b1c : stwu r1,-24(r1) 0x159b20 : mflr r0 0x159b24 : stw r29,12(r1) 0x159b28 : stw r30,16(r1) 0x159b2c : stw r31,20(r1) 0x159b30 : stw r0,28(r1) ; save entry params: posix clockid in r3 => r29 0x159b34 : mr r29,r3 ; and struct timespec * in r4 => r31 0x159b38 : mr r31,r4 ; init the clock lib (if not already done) 0x159b3c : bl 0x159974 ; error code returned ? 0x159b40 : cmpwi cr1,r29,0 ; if not, skip around 0x159b44 : beq cr1,0x159b54 ; else get errno pointer to r3 0x159b48 : bl 0x1621b4 <__errno> ; choose error code 22 (EINVAL) because probably clockid was wrong 0x159b4c : li r0,22 ; and exit, setting errno 0x159b50 : b 0x159b64 ; no error, so continue: check valid struct timespec * supplied. 0x159b54 : cmpwi cr1,r31,0 0x159b58 : bne cr1,0x159b70 ; no, so get errno pointer 0x159b5c : bl 0x1621b4 <__errno> ; choose error code 14 (EFAULT) because it meas Bad address 0x159b60 : li r0,14 ; and set errno (join here from previous error path) 0x159b64 : stw r0,0(r3) ; set error return STATUS 0x159b68 : li r3,-1 ; and exit from function. 0x159b6c : b 0x159c2c ; point r9 at data area at 0x001f0000 0x159b70 : lis r9,31 ; and r10 at 0x00200000 0x159b74 : lis r10,32 ; r8 points to 0x001FC270 = _clockRealtime 0x159b78 : addi r8,r10,-15760 ; load r9 from 0x001f5870 = vxTicks 0x159b7c : lwz r9,22640(r9) ; load r0 from word at (_clockRealTime+8) = tick_offset 0x159b80 : lwz r0,8(r8) ; load r11 from word at (_clockRealTime+0) = ticks_per_sec 0x159b84 : lwz r11,-15760(r10) (A) ; r0 = vxTicks - _clockRealtime.tick_offset 0x159b88 : subf r0,r0,r9 ; r0 = r0 / _clockRealtime.ticks_per_second 0x159b8c : divwu r0,r0,r11 ; store tv_sec member of timespec 0x159b90 : stw r0,0(r31) ; now repeat the first code section above, but calculate nanos 0x159b94 : lwz r0,8(r8) 0x159b98 : lwz r11,-15760(r10) ; divide by ticks_per 0x159b9c : subf r9,r0,r9 ; modulo operation.. 0x159ba0 : divwu r0,r9,r11 0x159ba4 : mullw r0,r0,r11 0x159ba8 : lwz r11,4(r8) 0x159bac : subf r9,r0,r9 0x159bb0 : mullw r9,r9,r11 0x159bb4 : lwz r0,0(r31) 0x159bb8 : stw r9,4(r31) 0x159bbc : lwz r9,12(r8) 0x159bc0 : lis r11,15258 0x159bc4 : add r0,r0,r9 0x159bc8 : stw r0,0(r31) ; store tv_nanos member of timespec 0x159bcc : lwz r9,4(r31) 0x159bd0 : lwz r0,16(r8) 0x159bd4 : ori r11,r11,51711 0x159bd8 : add r9,r9,r0 0x159bdc : cmpw cr1,r9,r11 0x159be0 : stw r9,4(r31) 0x159be4 : ble cr1,0x159c00 0x159be8 : lis r11,-15259 0x159bec : ori r11,r11,13824 0x159bf0 : lwz r9,0(r31) 0x159bf4 : lwz r0,4(r31) 0x159bf8 : addi r9,r9,1 0x159bfc : b 0x159c1c 0x159c00 : cmpwi cr1,r9,0 0x159c04 : bge cr1,0x159c28 0x159c08 : lis r11,15258 0x159c0c : ori r11,r11,51712 0x159c10 : lwz r9,0(r31) 0x159c14 : lwz r0,4(r31) 0x159c18 : addi r9,r9,-1 0x159c1c : stw r9,0(r31) 0x159c20 : add r0,r0,r11 0x159c24 : stw r0,4(r31) 0x159c28 : li r3,0 ; exit from function here.... 0x159c2c : lwz r0,28(r1) 0x159c30 : mtlr r0 ; function epilogue..... 0x159c34 : lwz r29,12(r1) 0x159c38 : lwz r30,16(r1) 0x159c3c : lwz r31,20(r1) 0x159c40 : addi r1,r1,24 0x159c44 : blr End of assembler dump. - -------snip!------- DaveK - -- " So far, science's crowning achievement, the Atomic Bomb, has not even been able to alleviate the misery and torture of endless centuries of world poverty." - George Hammond, self-proclaimed discoverer of the scientific proof of god. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Anyone get VxWorks to boot quickly Date: Tue, 28 Aug 2001 10:38:57 +0100 From: "Dave Korn" Organization: Lumber Cartel (tinlc) Members #2234-2237 (owing to browser refresh) Message-ID: References: Brad Crabtree wrote in message ... >I have a need to get a Tornado 2.0 VxWorks ROM image (about 4MB in size) >loaded and booted on a Power PC VME board in 3-4 seconds. It is currently >taking about 6. Does anyone have any lessons learned on getting a VxWorks >image to boot quickly? > Use less PTEs and more BATs. Init'ing the PTEs takes loads of time at bootup, so if you can map your memory through a BAT you'll save loads. If that's not practical because you have too much or too complicated memory structure, maybe you could only map a chunk of ram to get started with through one of the bats, and then when the system is up and running map the rest through PTEs and add it to the system memory partition. DaveK - -- " So far, science's crowning achievement, the Atomic Bomb, has not even been able to alleviate the misery and torture of endless centuries of world poverty." - George Hammond, self-proclaimed discoverer of the scientific proof of god. --------------------------- Newsgroups: comp.os.vxworks Subject: question on porting a protocol stack Date: Tue, 28 Aug 2001 08:55:57 -0400 From: "Yong Zhang" Organization: Marconi Message-ID: <9mg49u$6t0$1@newsfeed.pit.comms.marconi.com> Hi, I want to know if it is very common to spawn a new task for a protocol stack you bought from an outside vender? The protocol stack seems usually use a function like while(1){.....}. How should I set the parameters for the task? Is there some difference between a stack which need a lots computation power and a stack which need a little computation power? Let's say they are equal important. Will the while(1) waste a lot cpu time? Thanks! Yong --------------------------- Newsgroups: comp.os.vxworks Subject: modified "copy()" api problem Date: Tue, 28 Aug 2001 09:36:07 -0400 From: Bhavesh Patel Organization: Mailgate.ORG Server - http://www.Mailgate.ORG Message-ID: <3B8B9E47.76C048B9@fwion.com> Sender: bpatel@fwion.com Hi all, I am using vxworks5.4/Tornado2.0/BSP based on PPC8260. Also we are using TFFS for the 16MB intel strataflash (28F128J3W) I am using modified version of copy() api from usrLib.c wherein 1) I have increased the buffer size from 1K to 16K (care has been taken to make sure the task stack size is appropriate, in my case it is 32k) 2) In case write() succeeds after the read(), immediately flush by invoking ioctl with FIOFLUSH option. 3)replaced the creat() call in the beginning of copy api to open() with the flags (CREATE | WRITE). With the above modified copy api the writes from ramdisk to intel strataflash tffs system occurs very fast (3 to 4 sec) for 2MB file, the modified copy() returns successfully without any error, HOWEVER after the copy if I check the size of the file it is still the PREVIOUS one. It seems the file was not overwritten (care had been taken to unlock the lock bits in the intel strataflash, unlock Tffs BEFORE doing the copy). The modified copy fails ONLY in scenario where I am overwriting an EXISTING file. To solve the problem, ITEM#3 was surprisingly the problem, reverting back the call to creat() made the modified copy work. This is puzzling to me why open() fails and creat() succeeds in making the copy work. Even though the copy works it is very slow (even with 16k buffer used from local stack) can somebody shed light on this problem and explain what is wrong with the call to open() instead of creat(). Thanks bhavesh --------------------------- Newsgroups: comp.os.vxworks Subject: [Q] ping problem between vxsim0 and vxsim1 simulators for solaris 7 using ppp Date: Tue, 28 Aug 2001 14:00:16 GMT From: "ÃÖ°­¿À" Organization: Dacom Message-ID: Hi all. I wonder here is right place for posting my problem. I started two simulator in tornado prototyper plus on solaris 7 using ppp vxsim0 and vxsim1. vxsim0's IP is "127.0.1.0" and vxsim1's IP is "127.0.1.1". host's IPs are "150.150.53.30" and "127.0.1.254"(ppp setting). The problem is: host can ping "127.0.1.0","127.0.1.1" and "127.0.1.254"; vxsim0 can ping "127.0.1.0","127.0.1.254", but not "127.0.1.1" ; vxsim1 can ping "127.0.1.1","127.0.1.254", but not "127.0.1.0" ; Can somebody explain it? Thanks a lot. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: modified "copy()" api problem Date: Tue, 28 Aug 2001 10:14:04 -0400 From: Bhavesh Patel Organization: Mailgate.ORG Server - http://www.Mailgate.ORG Message-ID: <3B8BA72C.2506974A@fwion.com> References: <3B8B9E47.76C048B9@fwion.com> Sender: bpatel@fwion.com a correction, the open() was done with "WRITE" flag only. bhavesh > --------------------------- Newsgroups: comp.os.vxworks Subject: could not boot VxWorks bootImage on a MDP860T Pro Date: Tue, 28 Aug 2001 18:34:41 +0200 From: Christophe Couturier Organization: Guest of France Telecom Oleane's newsreading service Message-ID: <3B8BC820.8131FCBB@st2e.com> I've got a problem using vxWorks on a MDPPro board with a MPC860T. When booting, the vxWorks bootimage is correctly downloaded but fails to execute. On the hyperterminal screen, I receive the following message: "muxDevStart failed for device entry 0!" The EEPROM is then reseted to it's factory default settings. The CPU is also recognized as a 855T. What does that mean? What's the solution? Thanks Christophe --------------------------- Newsgroups: comp.os.vxworks Subject: Re: UDP Viewer Date: 28 Aug 2001 13:18:11 -0400 From: vanbaren@falcon.si.com (vanbaren_gerald) Organization: Smiths Industries Message-ID: <3b8bd253$1@news.si.com> References: <3b88d945$1@news.barak.net.il> <3B8A511B.9D3B77B7@lmco.com> We use ethereal running on linux. It works very well. It is advertised as running on NT, but we didn't have much luck with that. Note that you need a hub, not a switch, to sniff packets unless you are only sniffing your own packets. Switches go port-to-port so a sniffer isn't going to see any packets not addressed to it unless you get into SMON, and I don't want to go there. gvb Joe Durusau writes: > If you want to do something with the data, get out your local >windows programmer. If you just want to see the packets, look for >sniffer >utilities on the net. >Speaking only for myself, >Joe Durusau >AE wrote: >> >> Hi >> >> I would like to send messages from my system (running VxWorks) to a remote >> PC using a UDP socket. >> >> Unfortunately HyperTerm can only listen to a TCP/IP socket and not to UDP/IP >> socket. >> >> Does anyone know of a window application that can display messages sent >> through a UDP socket? >> >> Thanks. - -- +-----------------------------------------------------------------------+ | Jerry Van Baren / vanbaren_gerald@si.com / Grand Rapids, MI | | My employer is a company. Companies are artifacts of a legal system. | |_____________Artifacts are incapable of having opinions._______________| --------------------------- Newsgroups: comp.os.vxworks Subject: reboot blues Date: Tue, 28 Aug 2001 13:52:05 -0400 From: "William Grigg" Message-ID: <9mglo5$689$1@news.draper.com> I have just finished reading the thread "debugging when the debugger crashes". It sounds similar to my current situation: 1. I am running VxWorks 5.4 on an Ampro 4Ge board (x86 architecture) 2. There is a Fastcom serial board that is configured to run at 1.5Mhz, and is being used to handle about 10K bytes/sec input on one port, and about 250 bytes/sec on the other. 3. The COM2 port is used for a very low rate serial input also. 4. The system will run just fine for between 5 minutes and 45 minutes grabbing the data, crunching it, and producing the correct result. 5. The system randomly (between 5 and 45 minutes) crashes by rebooting. 6. It does not leave me a clue as to why the "crash reboot". 7. I have tons of internal monitors that are looking for serial input/output buffer overflows, and the like. But nothing gets triggered to give an indication of what might be going on. Any suggestions? I noticed that one of the thread responses said: "....unless it happens in an ISR, and that is a panic situation that causes the OS to reboot...." I did in fact write the serial driver for the Fastcom board (well, modified the stc16652 driver anyway), and have wondered if that could be the problem. Does it sound like a good guess that there is some sort of "end condition" in the serial driver that I should be looking for? TIA --------------------------- Newsgroups: comp.os.vxworks Subject: Re: ifGetAddr problem Date: Tue, 28 Aug 2001 18:00:03 GMT From: Tony Dal Santo Message-ID: References: <3B8B83FF.96CFF27@optonline.net> Gururaj patkar wrote: > Also, is ifAddrGet buggy? > In the beginning using ifAddrGet in my code works fine, after the system > is running for a short while, intermittently and after a while > consistently ifAddrGet fails and returns ERROR. Make sure the buffer for the IP address (second argument) is at least of size INET_ADDR_LEN (18). If it is smaller than that, your memory (stack/global) is getting stomped. Odds are, you are only allocating 16 bytes (the REAL minimum) for this string. Tony --------------------------- Newsgroups: comp.os.vxworks Subject: Re: ifGetAddr problem Date: Tue, 28 Aug 2001 14:22:03 -0400 From: Bhavesh Patel Organization: Mailgate.ORG Server - http://www.Mailgate.ORG Message-ID: <3B8BE14B.7A6811ED@fwion.com> References: <3B8B83FF.96CFF27@optonline.net> Sender: bpatel@fwion.com No I am allocating more than 16 bytes, in fact it is a 48byte buffer. When I said valid I meant it was adhering to the requirements in the vxworks reference manual for this api. bhavesh --------------------------- Newsgroups: comp.os.vxworks Subject: Re: DHCP server on multi i/faces Date: 28 Aug 2001 11:35:02 -0700 From: chrismooremobile@yahoo.com (Chris Moore) Organization: http://groups.google.com/ Message-ID: References: Okay, well looks like I'll answer my own question. Yes! It does only hand out from the appropriate pool depending on what interface your request came in. Yes! You can also init it on more than one iface (you pass it a list of device ptrs from the ifunit() func) and lastly Yes! It sucks big time as you cannot stop the server (unless you want to kill the task) but looking at some code I have it is filled with shutdown routines that are just not made availble (or prob tested). Yes! Also when it calls your hooks to dump the current leases it dumps ALL THE CURRENT leases, not the ones that have changed, ALL OF THEM! EVERY 10 MINUTES! Hmm, great design (not). Nor can you remove any address pools from an active system. Fabulous. Chris chrismooremobile@yahoo.com (Chris Moore) wrote in message news:... > Hi, > > I'm looking at using the DHCP server that comes with VxW (the > supported one). But looks like there is a problem. > > If I have multiple interfaces (say two ethernets) and both of those > subnets have DHCP clients on them, it means that the DHCP pool is > common for both of these?!?! Thats useless. > > I need to have two different pools, ie: 192.168.2.x for eth1 and > 192.168.50.x for eth2. But if I receive a DHCP request it does not > differentiate which interface it came in from does it? Or does it go > "Hmm, I rec'd a DHCP on eth1, eth1s ip is 192.168.2.x (mask > 0xFFFFFF00) so I should only serve up addresses where the network > component (192.168.2) matches..." > > Does anyone know if it defintely works like this? The documentation is > great but doesn't explain anything useful. > > Rgds, > > Chris --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Tornado AE Date: Tue, 28 Aug 2001 14:19:00 -0500 From: Andrew Johnson Organization: APS, Argonne National Laboratory Message-ID: <3B8BEEA4.264EE3A4@aps.anl.gov> References: <3B8A5B71.F9C6A93A@hotmail.com> <2c7641d6.0108272357.57940f6a@posting.google.com> Eeyore wrote: > > Graham Marshall wrote: > > > > * C++ Support (including which version of gcc does it use?) > > A custom version created by Wind River called gcc-2.96. It is similar > to FSF versions of the 2.9x ilk. Claims to have better C++ support > than T2; haven't really tested that out much. WRS subcontract their GCC work to cygnus which is now part of Red Hat. The gcc-2.96 version that comes with AE is supposedly compiled from the same source as the gcc that is shipped with RH Linux 7.0. - - Andrew - -- The world is such a cheerful place when viewed from upside-down It makes a rise of every fall, a smile of every frown --------------------------- Newsgroups: comp.os.vxworks Subject: Does anybody know how to do this ??? Date: Tue, 28 Aug 2001 12:35:03 -0500 From: "Shesha Bhushan" Organization: Arizona State University Message-ID: <9mgrtu$n9m$1@news.asu.edu> Hi Gurus, I have connected a scsi hardisk to a Single board Computer that runs vxWorks OS. The hard disk was previously formated as NTFS. I want to format this hard disk with dosFs. How to format this previously formated disk. Tx a Bunch Shesha Bhushan shesha@asu.edu --------------------------- Newsgroups: comp.os.vxworks Subject: Re: creation of vxWorks image without Tornado GUI *.wpj file... Date: Tue, 28 Aug 2001 17:01:09 -0400 From: Bhavesh Patel Organization: Mailgate.ORG Server - http://www.Mailgate.ORG Message-ID: <3B8C0695.D299088C@fwion.com> References: <3B8A6B1D.BA36F098@lucent.com> Sender: bpatel@fwion.com a better way would be to load all the object modules via script files, (obviously the object modules are built separately in different Tornado projects). The script file is given as part of boot parameters. this way vxworks.st is kept separate from application software. bhavesh --------------------------- Newsgroups: comp.os.vxworks Subject: Why not bring up the network interface! Date: 28 Aug 2001 22:57:33 -0700 From: wys205@sina.com (John Wang) Organization: http://groups.google.com/ Message-ID: Hi, I create a bootable vxworks in a floppy disk with vxworks.st ,but vxworks.st is not initilized the network interface.I try usrNetInit() manually ,this only bring up loopback driver(lo0),and not bring up the ethernet netcard driver. Why? my network is realtek8139(compatible with NE2000) Thanks in Advance! wys205 --------------------------- Newsgroups: comp.os.vxworks Subject: Re: question on porting a protocol stack Date: Wed, 29 Aug 2001 06:24:18 GMT From: Ephraim Gadsby Organization: - minimal Message-ID: References: <9mg49u$6t0$1@newsfeed.pit.comms.marconi.com> On Tue, 28 Aug 2001 08:55:57 -0400, "Yong Zhang" wrote: >Hi, > >I want to know if it is very common to spawn a new task for a protocol stack >you bought from an outside vender? The protocol stack seems usually use a >function like while(1){.....}. > >How should I set the parameters for the task? Is there some difference >between a stack which need a lots computation power and a stack which need a >little computation power? Let's say they are equal important. Will the >while(1) waste a lot cpu time? Presumably somewhere inside the while loop is at least one blocking call, so probably not, the use of while(1) implies nothing. It may use a lot of cpu time, but it need not waste it, unless it is badly written. I hope you don't mind me saying, but this is a very naive sort of question for someone undertaking this kind of work. I think you would be well advised to seek help from from your collleages, and try to get some VxWorks training, or do some background reading on RTOSs. --------------------------- End of New-News digest ********************** From vxwexplo-errs@csg.lbl.gov Wed Aug 29 06:17:50 2001 From: Rohit_Kapur@raytheon.com Date: Wed Aug 29 06:17:52 PDT 2001 Subject: Re: comp.os.vxworks newsdigest unsubscribe Rohit_Kapur@raytheon.com From vxwexplo-errs@csg.lbl.gov Wed Aug 29 08:18:40 2001 From: "Mike Anderson" Date: Wed Aug 29 08:18:42 PDT 2001 Subject: RFC2233.mib? VxWorks Greetings! Does anyone have a copy of the rfc2233.mib file for the VxWorks SNMP support? When I try to rebuild the agent, the make file looks for this mib and it's not in the supplied WRS mib tree. TIA, Mike -- ======================================================= T h e P T R G r o u p, I n c. ======================================================= Embedded, Real-Time Solutions and Training Mike Anderson mailto:mike@ThePTRGroup.com Chief Scientist http://www.ThePTRGroup.com V: (703) 585-9384 F: (703) 430-3748 ======================================================= "Software development is like making a baby... You can't make a baby in one month by impregnating nine women. Some things just take time." From vxwexplo-errs@csg.lbl.gov Wed Aug 29 08:37:00 2001 From: "Sinn, Larry" Date: Wed Aug 29 08:37:02 PDT 2001 Subject: RE: royalties When we negotiated a price for the license (more than 5 years ago), the formula WRS used also included the cost of the product. I would push that point. HTH 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 -----Original Message----- From: vxwexplo@lbl.gov [mailto:vxwexplo@lbl.gov] Sent: Sunday, August 26, 2001 1:32 PM To: vxworks_users@csg.lbl.gov Subject: Re: royalties Submitted-by vxwexplo-errs@csg.lbl.gov Sun Aug 26 13:21:07 2001 Submitted-by: Jim Way --------------------------- Newsgroups: comp.os.vxworks Subject: Re: royalties Date: Fri, 24 Aug 2001 13:40:49 -0500 From: Andrew Johnson Organization: APS, Argonne National Laboratory Message-ID: <3B869FB1.8717B0A0@aps.anl.gov> References: <3B8658EE.53C1A14C@notifier-is.net> david lindauer wrote: > Is this really the going price for royalties? David, I'm not the Keeper of Beans at my company, but from what I understand, the royalties for my product are about $80-90 per unit sold. I think our sales forecasts were in the "thousands" range for the first few years. Royalties are a function of volume [as Andrew Johnson pointed out] PLUS certain add-in modules that will increase your payments. One suggestion would be to carefully examine what features you have added from the vxWorks capabilities that cost you extra money and see if you can build them yourself or build them another way that avoids using the cost-extra code. Another suggestion would be to get your Marketing team to come up with a three or five year sales forecast, then go back to WindRiver and see if they'll reevaluate their royalty calculations. Another more heretical suggestion would be to see how difficult another OS would be to incorporate into your product. Yes, any change is painful, but sometimes you have to be willing to walk away from a deal to strengthen your negotiating position. Best of luck! Jim From vxwexplo-errs@csg.lbl.gov Wed Aug 29 11:29:15 2001 From: Bruce Wilson Date: Wed Aug 29 11:29:17 PDT 2001 Subject: Force 3750 cPCI Init We're using VxWorks 5.4, Tornado2, and trying to use a Force 3750 cPCI board. Has anybody been using the Force 3750 board with devices in other slots? Apparently, those devices are on bus 1, but when we tried calling frcPciConfigReadDword( bus=1, dev=25, func=0, regOffset=0, &temp) we got a "machine check" error. We had no trouble initializing the devices in other slots when using the Force 3603 Compact PCI board. Everything was on bus 0. What do we need now, to get the devices initialized on bus 1 for the 3750? Bruce A. Wilson U.C. Lawrence Livermore National Lab P.O. Box 808 (L-278) Livermore, CA 94550 voice: (925) 422-4145 fax: (925) 423-1243 From vxwexplo-errs@csg.lbl.gov Wed Aug 29 12:30:57 2001 From: "Mike Anderson" Date: Wed Aug 29 12:30:59 PDT 2001 Subject: Does OSPF work at all in VxWorks? VxWorks Greetings! [Flame on] OK. I've installed SNMP and OSPF from the CDs. Upgraded to T2CP4, installed the latest driver patches, etc. I include both OSPF and SNMP into the build and remake the O/S. SNMP is there. OSPF is there. But, they don't seem to talk to each other. I started the OSPF and initialized an interface. I can dump the OSPF mib to memory using m2OspfGeneralGroupGet and I see my router ID and area number, etc. So that seems to be working. But, the OSPF mib is not visible in the SNMP tree when you walk the MIB. TSR 248225 outstanding for over 1 month with no answer as to how you can get the OSPF info gettable/settable from SNMP. Anyone got an answer/fix so we can access the OSPF mib frmo SNMP? Wind doesn't seem to know... So, I went back to the snmpv1 tree and found the commented out lines in snmp2Mib.mib about OSPF support and uncommented them. I then tried to rebuld. No good. The SNMP agent code is not compatible with T2CP4. Sigh... I go back to the drawing board and reinstall and bring it to to T2CP3. But, the OSPF code (I'm working for someone who has both the SNMP and OSPF sources) won't rebuild under T2CP3 because it needs the NPT support that came out before T2CP4 and that CD wasn't supplied with the OSPF source, nor is it available on WindSurf. Heavy sigh... So, let me recap: 1) distributed binaries of SNMP and OSPF won't talk to each other in spite of the OSPF code having a mib (ospf_mib). 2) T2CP4 is incompatible with the SNMP from the CD *and* the envoy 9.1 code. It is also incompatible with OSPF (lots of defines missing for the OSPF mib in m2lib.h) You cannot rebuild any of it out of the box with T2CP4. 3) T2CP3 is missing the NPT files necessary to try and rebuild the SNMP or OSPF code. Curses... foiled again. 4) So, the customer has spent megabucks on the sources that won't build. TSRs are outstanding for over a month with no hope of resolution. 5) The building instructions for both products are hopeless. The online docs don't match. They seem to be written for completely different products. 6) In fairness, at least the local WRS FAEs are responsive and they truly want to help, but the problem is out of their domain of expertise. However, they did give me a pointer to the missing NPT stuff for T2CP3. So, I'll be back to that tack as soon as I can get it downloaded. But then, what will happen when I put T2CP4 back in place? Now, I've been using VxWorks for almost 14 years. I've tought WRS's Tornado, BSP and device driver classes for Wind. I got a 97% on the Tornado certified engineer test for pete's sake. I have reason to believe that I'm a fairly competent engineer. However, after 25 years in this business, maybe I'm just getting old. But, my personal opinion is that this whole section of Wind's offering is a piece of sh*t and not worth the powder it would take to blow it to hell. Please, someone convince me I'm wrong. Point out where I've gone wrong and I'll happily print a retraction and a mea culpa. But, from what I've seen in the past few months, I think I hear penguins calling. Tux, please save a seat for me... [Flame off] Regards, Mike -- ======================================================= T h e P T R G r o u p, I n c. ======================================================= Embedded, Real-Time Solutions and Training Mike Anderson mailto:mike@ThePTRGroup.com Chief Scientist http://www.ThePTRGroup.com V: (703) 585-9384 F: (703) 430-3748 ======================================================= "Software development is like making a baby... You can't make a baby in one month by impregnating nine women. Some things just take time." From vxwexplo-errs@csg.lbl.gov Thu Aug 30 04:04:55 2001 From: Vxworks Exploder Date: Thu Aug 30 04:04:58 PDT 2001 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Thu Aug 30 04:04:32 PDT 2001 Subject: Can not open C:\Tornado\host..... Subject: Re: question on porting a protocol stack Subject: Re: Does anybody know how to do this ??? Subject: Re: Several Questions... Subject: How do breakpoints work??? Subject: Re: Why not bring up the network interface! Subject: Re: Can not open C:\Tornado\host..... Subject: Re: How do breakpoints work??? Subject: CPX8216 cPCI Chasis Subject: ftpLib Problem with ASCII file transfer Subject: Re: creation of vxWorks image without Tornado GUI *.wpj file... Subject: Tornado II / config.h Subject: Re: How do breakpoints work??? Subject: memShow Subject: tornado run cannot find symbol Subject: Re: memShow Subject: How do you build vxWorks.st in Tornado 2.0 on a PC? Subject: Re: How do breakpoints work??? Subject: Halt When A Variable Changes? Subject: errno in VxWorks Subject: Re: How do breakpoints work??? Subject: Re: How do breakpoints work??? Subject: Re: Tornado II / config.h Subject: Re: errno in VxWorks Subject: Re: How do you build vxWorks.st in Tornado 2.0 on a PC? Subject: vxIMRClear() and vxIMRSet() on i960Hx and Jx processor ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Can not open C:\Tornado\host..... Date: Wed, 29 Aug 2001 14:59:36 +0300 From: "Kokkalidou Giota" Organization: Intracom S.A., GREECE Message-ID: <9mima0$25t$1@newsserv.intranet.gr> Hello everybody! Well, I am trying to make build at a downloadable project and while processing the build operation I have the follow message: make: *** No rule to make target `C:Tornadohost?-win32libgcc-libi386-pc-mingw32egcs-2.90.2 9..\..\..\..\i386-pc-mingw32\sys-include\memory.h', needed by `octet.o'. Stop. As I can see the path it is trying to open is wrong, but why??? Thanks in advance, Giota... --------------------------- Newsgroups: comp.os.vxworks Subject: Re: question on porting a protocol stack Date: Wed, 29 Aug 2001 08:43:56 -0400 From: "Yong Zhang" Organization: Marconi Message-ID: <9minvd$s3m$1@newsfeed.pit.comms.marconi.com> References: <9mg49u$6t0$1@newsfeed.pit.comms.marconi.com> Thanks for the answer. Can you recommend some books or materials for me to read on this topic? Thanks! Yong Ephraim Gadsby wrote in message news:bpvootgtens8npqhapaheg4q5pr0f7o2qi@4ax.com... > On Tue, 28 Aug 2001 08:55:57 -0400, "Yong Zhang" > wrote: > > >Hi, > > > >I want to know if it is very common to spawn a new task for a protocol stack > >you bought from an outside vender? The protocol stack seems usually use a > >function like while(1){.....}. > > > >How should I set the parameters for the task? Is there some difference > >between a stack which need a lots computation power and a stack which need a > >little computation power? Let's say they are equal important. Will the > >while(1) waste a lot cpu time? > > Presumably somewhere inside the while loop is at least one blocking > call, so probably not, the use of while(1) implies nothing. It may use > a lot of cpu time, but it need not waste it, unless it is badly > written. > > I hope you don't mind me saying, but this is a very naive sort of > question for someone undertaking this kind of work. I think you would > be well advised to seek help from from your collleages, and try to get > some VxWorks training, or do some background reading on RTOSs. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Does anybody know how to do this ??? Date: 29 Aug 2001 06:47:29 -0600 From: taashlo@sandia.gov Organization: Sandia National Laboratories, Albuquerque, NM USA Message-ID: References: <9mgrtu$n9m$1@news.asu.edu> "Shesha Bhushan" writes: > Hi Gurus, > I have connected a scsi hardisk to a Single board Computer that runs > vxWorks OS. The hard disk was previously formated as NTFS. I want to format > this hard disk with dosFs. How to format this previously formated disk. > > Tx a Bunch > Shesha Bhushan > shesha@asu.edu Which version of VxWorks are you running? I'm using version 5.4 (that came with Tornado 2.0) and I found the answer to your question in the VxWorks Programmer's Guide. Chapter 3 - I/O System, has a "SCSI Drivers" section that covers configuring SCSI. And then Chapter 4 - Local File Systems starts off by discussing dosFs. Example 3-4 "Configuring SCSI Drivers" on pages 134-135 should get you started. Tad Ashlock --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Several Questions... Date: Wed, 29 Aug 2001 15:22:46 +0200 From: "Jonathan ILIAS" Organization: University of Angers, France. Message-ID: <9miq7q$30e1$1@news.univ-angers.fr> References: > 2. Does intLock prevent task preemption? No, I think there is another function to do this. See in "VxWorks Reference Manual : Libraries" at intArchLib|intLock -> Warnings. It is for forbidden to call any system routine into an intLock/intUnlock block. As they say : "Locking out interrupts does not prevent rescheduling". Jonathan ILIAS --------------------------- Newsgroups: comp.os.vxworks Subject: How do breakpoints work??? Date: Wed, 29 Aug 2001 09:10:31 -0400 From: "Granum, Allen [CAR:CF81:EXCH]" Organization: Nortel Networks Message-ID: <3B8CE9C7.5005E14D@americasm01.nt.com> Wondering how the whole breakpoint mechanism works on a PPC/vxWorks system? I have always thought that when a breakpoint was set, that some code was inserted at the location of the breakpoint which would halt the execution. However, there is skeptasism since further investigation of the breakpoint location using the l and d commands does not show any modification to the code where the breakpoint was set. Can anyone enlighten me? - -- Allen Granum Nortel Networks Wireless Internet BSP Development Team Email: agranum@nortelnetworks.com Phone: (613)763-4324 --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Why not bring up the network interface! Date: Wed, 29 Aug 2001 07:49:52 -0700 From: DrDiags Organization: laserlink.net Message-ID: <3B8D0110.4C5044DE@covad.net> References: Reply-To: drdiags@covad.net John, Try the FAQ listed below. This very topic was just recently discussed on this newsgroup, so maybe a Google search will provide you with that thread. I think a couple of tips are to make sure your interface card is listed in the (o)ther field in the boot parameters and there are a couple of INCLUDE_STANDALONE type defines that have to be set in your config.h file. (Check for exact spelling in the configAll.h file). HTH. John Wang wrote: > > Hi, > I create a bootable vxworks in a floppy disk with vxworks.st ,but > vxworks.st > is not initilized the network interface.I try usrNetInit() manually > ,this only bring up loopback driver(lo0),and not bring up the ethernet > netcard driver. > Why? my network is realtek8139(compatible with NE2000) > > Thanks in Advance! > > wys205 - -- http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html (vxWorks FAQ home page) --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Can not open C:\Tornado\host..... Date: Wed, 29 Aug 2001 07:57:09 -0700 From: DrDiags Organization: laserlink.net Message-ID: <3B8D02C5.41C83883@covad.net> References: <9mima0$25t$1@newsserv.intranet.gr> Reply-To: drdiags@covad.net Kokkalidou, Looks like your WIND_BASE variable has / slashes instead of \ slashes (running from command line). Or if you are running from the IDE, the \ should be /. Either way, the interpreter for the environment is clobbering the characters "x86" with the "?" and the "Tornado\host" or "Tornado/host" becomes "Tornadohost". What does the torVars.bat file set the WIND_BASE to? Did you add paths to the project, make sure they are correct for your build env. HTH. Kokkalidou Giota wrote: > > Hello everybody! > > Well, I am trying to make build at a downloadable project and while > processing the build operation I have the follow message: > make: *** No rule to make target > `C:Tornadohost?-win32libgcc-libi386-pc-mingw32egcs-2.90.2 > 9..\..\..\..\i386-pc-mingw32\sys-include\memory.h', needed by `octet.o'. > Stop. > As I can see the path it is trying to open is wrong, but why??? > > Thanks in advance, > > Giota... - -- http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html (vxWorks FAQ home page) --------------------------- Newsgroups: comp.os.vxworks Subject: Re: How do breakpoints work??? Date: 29 Aug 2001 15:13:09 GMT From: Johan Borkhuis Organization: Agere Systems Message-ID: References: <3B8CE9C7.5005E14D@americasm01.nt.com> "Granum, Allen [CAR:CF81:EXCH]" wrote: > I have always thought that when a breakpoint was set, that some code > was inserted at the location of the breakpoint which would halt the > execution. However, there is skeptasism since further investigation of > the breakpoint location using the l and d commands does not show any > modification to the code where the breakpoint was set. As far as I know these routines know that you inserted a breakpoint there, and will show the original code. You can try your own code to read these memory locations and see what is written 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: CPX8216 cPCI Chasis Date: 29 Aug 2001 08:51:03 -0700 From: jdas@alphion.com (JD) Organization: http://groups.google.com/ Message-ID: <27750684.0108290751.5b36423b@posting.google.com> Hi, Anyone have any experience running VxWorks 5.4 on MCP750HA board in CPX8216 platform? Thanks for any insight. --------------------------- Newsgroups: comp.os.vxworks Subject: ftpLib Problem with ASCII file transfer Date: Wed, 29 Aug 2001 15:57:51 GMT From: ranjana@surfree.com (Ranjana Shah) Organization: WWW.US.INTER.NET Message-ID: <3b8d0cc0.8178597@news.surfree.com> Reply-To: ranjana@surfree.com Has anyone performed file transfers in ASCII mode using ftpLib functions under VxWorks? So far I have been unable to do so. I am using a no frills FTP server on the PC communicating with the ftp client on my VxWorks target which uses the ftpLib function calls. I know the high level function ftpXfer() defaults to binary mode (Type I) transfers. So, I am using the low level calls (as shown below) so that I can issue the "Type A" command to set ASCII mode. Note that the FTP commands ASCII and BINARY returned with errors. FtpRetrCtrlSock = ftpHookup(FTP_HOST_NAME); ftpLogin(FtpRetrCtrlSock, "username", "password", ""); ftpCommand (FtpRetrCtrlSock,"TYPE A", 0, 0, 0, 0, 0, 0); ftpCommand (FtpRetrCtrlSock, "CWD %s", (int) Dir_Path, 0, 0, 0, 0, 0); FtpRetrDataSock = ftpDataConnInit(FtpRetrCtrlSock); ftpCommand (FtpRetrCtrlSock, "RETR %s", (int) FileToDwnld, 0, 0, 0, 0, 0); FtpRetrDataSock = ftpDataConnGet (FtpRetrDataSock); Then I do a sequence of reads using: nBytes = read(FtpRetrDataSock, Buf, DATA_SIZE); But no matter what I do, I get a binary mode transfer, because the retrieved file still has CRLF characters, which should have been converted to LF only if ASCII transfer occurred. If anyone sees anything wrong with the above sequence of calls, or if anyone has done file transfers using ftpLib in ASCII mode, please tell me how. Thanks. Ranjana Shah Telephonics Corp. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: creation of vxWorks image without Tornado GUI *.wpj file... Date: Wed, 29 Aug 2001 11:38:34 -0500 From: Dinker Charak Organization: Fermi National Accelerator Laboratory Message-ID: <3B8D1A8A.81F5DC1B@fnal.gov> References: <3B8A6B1D.BA36F098@lucent.com> <3B8B2606.5FE1F697@covad.net> > > Hi, > > > > Could somebody give me some guide lines as to how we can create vxworks image > > for a specifed BSP without using the Tornado GUI ? > > what i had done was write a simple batch file which set the evn vars like WIND_* ones and the in the bsp dir do a make vxWorks, etc ... i never touched the gui ... the idea behind the wpj is that lot many projects can be built on one bsp with out messing the bsp ... so each custom change to configuartion was stored in prjParam.c and so on ... you could copy the bsp you want to use in another dir in target/config and play with it ... it was low level fiddling with bsp and i liked it that way ... regards, dinker --------------------------- Newsgroups: comp.os.vxworks Subject: Tornado II / config.h Date: Wed, 29 Aug 2001 11:45:57 -0500 From: Dinker Charak Organization: Fermi National Accelerator Laboratory Message-ID: <3B8D1C45.69601F61@fnal.gov> hello, in my config.h i have #define EXTENDED_VME i want to undef it using tornado ... the VxWorks tab allows me only to configure the parameters ... when i see the value of a parameter, i see it's value as set in #ifdef EXTENDED_VME ... i want it to be from #ifndef EXTENDED_VME part ... if i manually change in config.h i still see same values ... somehow rebuilding dependencies is not cleaning it ... anyone done fiddled with #define-s thru Tornado II? please advice, regards, dinker --------------------------- Newsgroups: comp.os.vxworks Subject: Re: How do breakpoints work??? Date: Wed, 29 Aug 2001 17:00:45 GMT From: Harvey Taylor Organization: MTS Internet Message-ID: <3B8D3B06.1725@despam.autobahn.mb.ca> References: <3B8CE9C7.5005E14D@americasm01.nt.com> In article <3B8CE9C7.5005E14D@americasm01.nt.com>, Allen Granum wrote: > > Wondering how the whole breakpoint mechanism works on a PPC/vxWorks system? > I don't know about the PPC, but AFAIK there are three possible mechanisms. Insert an interrupt or illegal op code. Run the processor in Trace mode and watch for the address. Use external hardware to monitor the address bus & trigger an int. > I have always thought that when a breakpoint was set, that some code was inserted > at the location of the breakpoint which would halt the execution. However, the > breakpoint location using the l and d commands does not show any modification > to the code where the breakpoint was set. > > Can anyone enlighten me? > As someone else pointed out, does the tool know about your breakpoint? WALP it. [Write A Little Program] - -het - -- Everyone has the right to recognition everywhere as a person before the law. UDHR:Article 6: http://www.autobahn.mb.ca/~het/udhr.html H.E. Taylor http://www.autobahn.mb.ca/~het/ --------------------------- Newsgroups: comp.os.vxworks Subject: memShow Date: Wed, 29 Aug 2001 15:58:11 -0400 From: "Kevin Livingston" Message-ID: memShow shows a lot of very good statistics about memory allocation and usage. is there anyway to programatically get the raw numbers that memShow is outputting, without haveing to do something silly like parse the text that memShow returns. All I want to do is to look at memory allocation before my program runns and then again after, and perform some comparisons / math on the values. I would be interested in available memory or allocated memory, perferably both, but with one you can easily get the other. is there an easy way to do this that I am just missing. thanks for your help Kevin --------------------------- Newsgroups: comp.os.vxworks Subject: tornado run cannot find symbol Date: Wed, 29 Aug 2001 20:33:12 GMT From: rjdejoie@kmrmail.kmr.ll.mit.edu (Bob DeJoie) Message-ID: <3b8d4d4c.160796543@news> Hi, I have am having problems with the Tornado run command being able to find my downloaded applications entry point: Configuration: I am running T2 on MPC750 boards. I have two boards in one chassis that operate as a pair using VxMP to sharing state information and to communicate between each other. When development is complete the application will run as a stand-alone application, during development I am using a serial port for the console and for the connection to Tornado. To debug the processor pair I am running two instances of Tornado on one NT box. I have the OS burned into FLASH and download my application during development. Problem: When I download the application to both targets and then click the run icon to run them I often get a "symbol not found errror" on one or the other of the targets. If I try again it may or may not find the symbol. I have found that if I do a "lkup symbol" in the tornado shell the shell always find the symbol and then the run icon will usually find the symbol. Can anyone shed any light on this behavior? Bob ================================================ Bob DeJoie Kwajalein Missile Range Republic of the Marshall Islands email: rjdejoie@kmrmail.kmr.ll.mit.edu     phone: 805-355-5855 fax: 805-355-3833 Time Zone: GMT +12 hours                      ================================================ --------------------------- Newsgroups: comp.os.vxworks Subject: Re: memShow Date: Wed, 29 Aug 2001 20:41:27 GMT From: ranjana@surfree.com (Ranjana Shah) Organization: WWW.US.INTER.NET Message-ID: <3b8d5147.25724289@news.surfree.com> References: Reply-To: ranjana@surfree.com You may get what you need using memPartInfoGet() function of memLib, e.g.: struct MEM_PART_STATS MemStats; memPartInfoGet(memSysPartId, &MemStats); Then MemStats.numBytesAlloc and MemStats.numBytesFree will provide you with some of the information you want. memPartFindMax( ) gives the size of the largest available free block Also sysPhysMemTop() provides you top of physical memory. Ranjana Shah "Kevin Livingston" wrote: >memShow shows a lot of very good statistics about memory allocation and >usage. > >is there anyway to programatically get the raw numbers that memShow is >outputting, without haveing to do something silly like parse the text that >memShow returns. All I want to do is to look at memory allocation before my >program runns and then again after, and perform some comparisons / math on >the values. >I would be interested in available memory or allocated memory, perferably >both, but with one you can easily get the other. > >is there an easy way to do this that I am just missing. > >thanks for your help >Kevin > > --------------------------- Newsgroups: comp.os.vxworks Subject: How do you build vxWorks.st in Tornado 2.0 on a PC? Date: 29 Aug 2001 16:02:19 -0500 From: "john cobarruvias" Message-ID: <3b8d585b$1_1@newsa.ev1.net> Reply-To: "john cobarruvias" Sender: "john cobarruvias" How do you build vxWorks.st in Tornado 2.0 on a PC? There is no option in Build VxWorks for VxWorks.st. Anyone know how to do this? John Cobarruvias --------------------------- Newsgroups: comp.os.vxworks Subject: Re: How do breakpoints work??? Date: Wed, 29 Aug 2001 17:36:20 -0400 From: "Mike Kordik" Message-ID: References: <3B8CE9C7.5005E14D@americasm01.nt.com> If you set a breakpoint and the tool replaces the instruction with some sort of trap it could list the original instruction when you do a listing. I have seen this behaviour on several different chips/platforms. This makes more sense in a listing thant a memory dump but I guess there is no reason why it couldn't do it on a dump. Just a thought. - -Mike "Granum, Allen [CAR:CF81:EXCH]" wrote in message news:3B8CE9C7.5005E14D@americasm01.nt.com... > Wondering how the whole breakpoint mechanism works on a PPC/vxWorks system? > > I have always thought that when a breakpoint was set, that some code was inserted at the location of the breakpoint which would halt the execution. However, there is skeptasism since further investigation of the > breakpoint location using the l and d commands does not show any modification to the code where the breakpoint was set. > > Can anyone enlighten me? > > -- > Allen Granum > Nortel Networks > Wireless Internet > BSP Development Team > Email: agranum@nortelnetworks.com > Phone: (613)763-4324 > > > --------------------------- Newsgroups: comp.os.vxworks Subject: Halt When A Variable Changes? Date: Thu, 30 Aug 2001 01:01:44 GMT From: "Pat DiRienzo" Message-ID: Is it possible using the Tornado debugger to halt all tasks when a variable is changed to a specific value. I would like to halt the program when a memory location (for this variable) is set erroneously by a coding mistake or memory overwrite. I could then examine the code at the halted location to determine who corrupted this memory location. This is a debugging capability which could have saved my company hours of debug time and myself a lot of lost late nites and weekends. Maybe it is not possible with only a source level debugger such as Tornado because of the way they work? --------------------------- Newsgroups: comp.os.vxworks Subject: errno in VxWorks Date: 29 Aug 2001 19:21:30 -0700 From: npisolkar@yahoo.com (Nilesh) Organization: http://groups.google.com/ Message-ID: Hello! I am trying to do moniter existence of a file in a directory specified by the application. when I do stat() on that file, stat fails as exepected and returns errno as 0x388002. This seems strange to me. The last 2 bytes of the errno should tell me that file does not exist when not present in the file system. Does 0x8002 ( the last 2 bytes of the errno) mean something? I think I should get the last 2 bytes as 0x02 only. Could somebody help me out with this please? Am I wrong in interpreting the result? Thanks, Nilesh --------------------------- Newsgroups: comp.os.vxworks Subject: Re: How do breakpoints work??? Date: Wed, 29 Aug 2001 20:14:24 -0700 From: Lori Fraleigh Organization: Real-Time Innovations Message-ID: <3B8DAF90.DD2A0A88@rti.com> References: <3B8CE9C7.5005E14D@americasm01.nt.com> The tgtsvr generally caches the text segments for VxWorks. If you do an 'l' from WindShell, you won't see the changed instructions. If the target shell is available, perform an 'l' from it - you should see the instruction used to set the breakpoint. If your kernel has the proper components built in, you can also do an '@l' from WindShell. Specifying the '@' sign before the 'l' tells WindShell and the tgtsvr to execute the command on the target and return the response. Regards, Lori "Granum, Allen [CAR:CF81:EXCH]" wrote: > > Wondering how the whole breakpoint mechanism works on a PPC/vxWorks system? > > I have always thought that when a breakpoint was set, that some code was inserted at the location of the breakpoint which would halt the execution. However, there is skeptasism since further investigation of the > breakpoint location using the l and d commands does not show any modification to the code where the breakpoint was set. > > Can anyone enlighten me? > > -- > Allen Granum > Nortel Networks > Wireless Internet > BSP Development Team > Email: agranum@nortelnetworks.com > Phone: (613)763-4324 - -- Lori Fraleigh Director of Engineering Real-Time Innovations, Inc. http://www.rti.com Phone: 408-734-4200 x121 Fax: 408-734-5009 --------------------------- Newsgroups: comp.os.vxworks Subject: Re: How do breakpoints work??? Date: 29 Aug 2001 20:43:41 -0700 From: john_94501@yahoo.com (John) Organization: http://groups.google.com/ Message-ID: <488e459a.0108291943.4324cc16@posting.google.com> References: <3B8CE9C7.5005E14D@americasm01.nt.com> Hello, "Mike Kordik" wrote in message news:... > If you set a breakpoint and the tool replaces the instruction with some sort > of trap it could list the original instruction when you do a listing. I have > seen this behaviour on several different chips/platforms. This makes more > sense in a listing thant a memory dump but I guess there is no reason why it > couldn't do it on a dump. Actually there is a much less sinister (clever?) reason why the commands do not see the breakpoint... the shell is an unbreakable task, so all the time it is running the breakpoints are not installed. When a context switch causes a breakable task to run, the breakpoints will be resinstated. So, to see the breakpoint in memory simply spawn the d() command or l() command. That will then run in a breakable task, and you should see the magic code inserted to cause an exception. HTH, John... > Just a thought. > > -Mike > > "Granum, Allen [CAR:CF81:EXCH]" wrote in > message news:3B8CE9C7.5005E14D@americasm01.nt.com... > > Wondering how the whole breakpoint mechanism works on a PPC/vxWorks > system? > > > > I have always thought that when a breakpoint was set, that some code was > inserted at the location of the breakpoint which would halt the execution. > However, there is skeptasism since further investigation of the > > breakpoint location using the l and d commands does not show any > modification to the code where the breakpoint was set. > > > > Can anyone enlighten me? > > > > -- > > Allen Granum > > Nortel Networks > > Wireless Internet > > BSP Development Team > > Email: agranum@nortelnetworks.com > > Phone: (613)763-4324 > > > > > > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Tornado II / config.h Date: Wed, 29 Aug 2001 20:48:44 -0700 From: DrDiags Organization: laserlink.net Message-ID: <3B8DB79C.7277B774@covad.net> References: <3B8D1C45.69601F61@fnal.gov> Reply-To: drdiags@covad.net Dinker, There is an option under the Build tab to extend the dependency check. This will force the the dependency to revisit header files used during the initial project creation. This might help you pick up the changes you made to undefine the EXTENDED_VME. See the Read-only description for dependencies when you click the "Advanced" option from the Builds tab->right click the project and select this feature. This will cause the BSP files to be checked for changes. HTH. Dinker Charak wrote: > > hello, > > in my config.h i have > > #define EXTENDED_VME > > i want to undef it using tornado ... the VxWorks tab allows me only to > configure the parameters ... > > when i see the value of a parameter, i see it's value as set in #ifdef > EXTENDED_VME ... i want it to be from #ifndef EXTENDED_VME part ... > > if i manually change in config.h i still see same values ... somehow > rebuilding dependencies is not cleaning it ... > > anyone done fiddled with #define-s thru Tornado II? > > please advice, > > regards, > dinker - -- http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html (vxWorks FAQ home page) --------------------------- Newsgroups: comp.os.vxworks Subject: Re: errno in VxWorks Date: Wed, 29 Aug 2001 20:52:58 -0700 From: DrDiags Organization: laserlink.net Message-ID: <3B8DB89A.B7420C81@covad.net> References: Reply-To: drdiags@covad.net Nilesh, Look into using the routine for creating the statTbl, mkStatTbl is the name I think. It goes through all the headers and creates a table of errno's from each main component's error information. The upper word is the library and the lower is the error code. HTH. (If you have Windows, use the online help). Nilesh wrote: > > Hello! > I am trying to do moniter existence of a file in a directory > specified by the application. when I do stat() on that file, stat > fails as exepected and returns errno as 0x388002. This seems strange > to me. The last 2 bytes of the errno should tell me that file does not > exist when not present in the file system. Does 0x8002 ( the last 2 > bytes of the errno) mean something? I think I should get the last 2 > bytes as 0x02 only. > Could somebody help me out with this please? Am I wrong in > interpreting the result? > > Thanks, > > Nilesh - -- http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html (vxWorks FAQ home page) --------------------------- Newsgroups: comp.os.vxworks Subject: Re: How do you build vxWorks.st in Tornado 2.0 on a PC? Date: 30 Aug 2001 07:52:29 GMT From: Johan Borkhuis Organization: Agere Systems Message-ID: References: <3b8d585b$1_1@newsa.ev1.net> "john cobarruvias" wrote: > How do you build vxWorks.st in Tornado 2.0 on a PC? > > There is no option in Build VxWorks for VxWorks.st. vxWorks.st is the standard vxWorks image with a buildin symbol table. Just include the buildin symbol table, build your vxWorks image, and if you want you can copy it to vxWorks.st (but the last step is not really necessary:-). 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: vxIMRClear() and vxIMRSet() on i960Hx and Jx processor Date: Thu, 30 Aug 2001 08:41:15 GMT From: brede.thurmann@kongsberg-simrad.com (BT) Message-ID: <3b8dfc02.759480615@news.eunet.no> vxIMRClear() and vxIMRSet() on i960Hx and Jx processor is not an atomic function. I have checked my code and BSP where I am using these functions. After startup I am only using these function from one task so that they are not atomic should be OK. I have experienced some problems that might indicate that I have not been able to mask put an interrupt properly, I have therefore also checked my ram file for anny refernces to these function and I have not found any. The Intel 80960HA/HD/HT Specification Update states that there are several problems with clearing the interrupt mask. # Does anybodyknow why VxWorks has implemeted vxIMRClear() and vxIMRSet() in a no atomic way. (On the i960Cx it is implemented atomic by refering to sf1 instead of a ld/st the memory mapped addresses for IMSK as on i960Hx)? # Could it be that any VxWorks code mask's out interrupts without using these function and then fail my use of these functions since they are not atomic? # Has anybody experienced any problems masking out interrupts on the i960Hx processor? Brede --------------------------- End of New-News digest ********************** From vxwexplo-errs@csg.lbl.gov Thu Aug 30 05:37:52 2001 From: Rohit_Kapur@raytheon.com Date: Thu Aug 30 05:37:54 PDT 2001 Subject: Re: comp.os.vxworks newsdigest unsubscribe Rohit_Kapur@raytheon.com From vxwexplo-errs@csg.lbl.gov Thu Aug 30 08:33:19 2001 From: Ilia Bosis Date: Thu Aug 30 08:33:21 PDT 2001 Subject: RE:Halt When A Variable Changes? This is less of the debugger issue, but of the hardware on Your dispose. Trapping memory change by the software means only requires breakpoint after EVERY instruction and so virtually impossible. However, that may be easily done with a little dedicated hardware and if You have that hardware on disposal and the debugger that can make advantage of it, You can do it. Most of the modern 32-bit CPU provide the needed hardware on the chip, making it available either via the BDM or privileged, supervisor-mode instructions. Of course, the details depends greatly on what hardware You have: the specific CPU/microcontroller chip, ICE, BDM box, etc. From my experience, CrossWind can't do it - AFAIK it's the cross-platform software-only debugger and as such must rely on the lowest common denominator that is no debugging hardware at all. On another hand, VisionIce connected to the PowerQUICC CPU via the BDM box, do that with elegance and effortlessly. Check Your hardware documentation for details. Hope, this helps. Ilia -------------------------------------------- Ilia Bosis Software Engineer PacketLight Networks E-mail: Ilia_Bosis@packetlight.com Web: www.packetlight.com > -----Original Message----- > From: vxwexplo@lbl.gov [mailto:vxwexplo@lbl.gov] > Sent: Thursday, August 30, 2001 1:12 PM > To: vxworks_users@csg.lbl.gov > Subject: comp.os.vxworks newsdigest > > Newsgroups: comp.os.vxworks > Subject: Halt When A Variable Changes? > Date: Thu, 30 Aug 2001 01:01:44 GMT > From: "Pat DiRienzo" > Message-ID: > > Is it possible using the Tornado debugger to halt all tasks > when a variable > is changed to a specific value. I would like to halt the > program when a > memory location (for this variable) is set erroneously by a > coding mistake > or memory overwrite. I could then examine the code at the > halted location to > determine who corrupted this memory location. This is a debugging > capability which could have saved my company hours of debug > time and myself > a lot of lost late nites and weekends. Maybe it is not > possible with only a > source level debugger such as Tornado because of the way they work? > From vxwexplo-errs@csg.lbl.gov Fri Aug 31 04:03:19 2001 From: Vxworks Exploder Date: Fri Aug 31 04:03:23 PDT 2001 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Fri Aug 31 04:03:13 PDT 2001 Subject: fec problems,help needed Subject: Re: How do breakpoints work??? Subject: Re: Newbie: inhibit / trap VME bus error? Subject: Wind View and Windows 2000 Subject: Re: vxIMRClear() and vxIMRSet() on i960Hx and Jx processor Subject: EPFNOSUPPORT, (errno 46) Subject: vxWorks handling "string literals" as opposed to global data. Subject: Trouble with a socket call Subject: VxWorks package/library to use with MPC8260 Subject: Re: Halt When A Variable Changes? Subject: Re: VxWorks package/library to use with MPC8260 Subject: Re: fec problems,help needed Subject: Question to VxWorks Source customers - setting breakponts and source debugging for VxWorks "unbreakable" kernel tasks ???? Subject: memShow - value routines Subject: Re: VxWorks package/library to use with MPC8260 Subject: Re: vxWorks handling "string literals" as opposed to global data. Subject: Re: Trouble with a socket call Subject: flash bootrom and elfToBin Subject: Re: Wind View and Windows 2000 Subject: Re: VxWorks package/library to use with MPC8260 Subject: Re: Why not bring up the network interface! Subject: Re: fec problems,help needed Subject: simple standalone development environment Subject: Re: CPX8216 cPCI Chasis Subject: recommended G.703 interface ? Subject: Tornado prototyper(Solaris Version) - vxsim simulator ping problem. Subject: Re: Halt When A Variable Changes? Subject: Who can tell me how can I accelerate the speed of memory access Subject: Re: Halt When A Variable Changes? Subject: Re: How do breakpoints work??? Subject: WindView doesn't starts with Tornado Subject: Re: WindView doesn't starts with Tornado Subject: Ethernet cable disconnect Subject: Re: Ethernet cable disconnect ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: fec problems,help needed Date: 30 Aug 2001 06:59:36 -0700 From: kevinatnj@yahoo.com.cn (kewin) Organization: http://groups.google.com/ Message-ID: <221caa76.0108300559.5a9f9c53@posting.google.com> hi all i am debuging mpc860T bsp.i want to write and read PHY's register,but i can't make it.When i make this action ,there is nothing effect on the MII_DATA register of FEC which is mapped in dual port ram.And i try too to write value to MII_DATA directly in dual port RAM.There is the same results.The MII_DATA register's value is always 0xffffffff. Is there any problems on hareware or software.If software's problem,Will anybody tell me what i should do specially to initialize FEC. And advice will be welcome.Thanks a lot! --------------------------- Newsgroups: comp.os.vxworks Subject: Re: How do breakpoints work??? Date: Thu, 30 Aug 2001 10:45:25 -0400 From: "Macleod, Randy [CAR:CW31:EXCH]" Organization: Nortel Networks Message-ID: <3B8E5185.991FD6B0@americasm01.nt.com> References: <3B8CE9C7.5005E14D@americasm01.nt.com> <3B8DAF90.DD2A0A88@rti.com> Hi, Well, here is what I see using a telnet session to a VME PPC target. Find some function to break: - -> lkup "ifShow" ifShow 0x001e6884 text (cesnetvx.pserv.o.sym) value = 0 = 0x0 - -> b No breakpoints value = 0 = 0x0 - -> pData=0x001e6884 pData = 0x1ffebd0: value = 1992836 = 0x1e6884 = ifShow - -> sp printf,"At address %x, value = %x.\n",pData,*pData task spawned: id = 0x160fde8, name = t5 value = 23133672 = 0x160fde8 - -> At address 1e6884, value = 9421ffc8. # Add a breakpoint: - -> b 0x1e6884 value = 0 = 0x0 - -> b 0x001e6884: ifShow Task: all Count: 0 value = 0 = 0x0 - -> sp printf,"At address %x, value = %x.\n",pData,*pData task spawned: id = 0x160fde8, name = t6 value = 23133672 = 0x160fde8 - -> At address 1e6884, value = 9421ffc8. # Someone suggested that the behavior could be different # if there is more than one breakpoint... - -> b 0x1e6888 value = 0 = 0x0 - -> b 0x001e6888: ifShow +0x4 Task: all Count: 0 0x001e6884: ifShow Task: all Count: 0 value = 0 = 0x0 - -> sp printf,"At address %x, value = %x.\n",pData,*pData task spawned: id = 0x160fde8, name = t7 value = 23133672 = 0x160fde8 - -> At address 1e6884, value = 9421ffc8. So it sure looks like the instruction at the breakpoint address was not overwritten. Is VxWorks checking a breakpoint list every time the PC is incremented? // Randy "The Inquisitive Skeptic" MacLeod Lori Fraleigh wrote: > > The tgtsvr generally caches the text segments for VxWorks. If you > do an 'l' from WindShell, you won't see the changed instructions. > > If the target shell is available, perform an 'l' from it - you should > see the instruction used to set the breakpoint. If your kernel > has the proper components built in, you can also do an '@l' from > WindShell. Specifying the '@' sign before the 'l' tells WindShell > and the tgtsvr to execute the command on the target and return the > response. > > Regards, > Lori > > "Granum, Allen [CAR:CF81:EXCH]" wrote: > > > > Wondering how the whole breakpoint mechanism works on a PPC/vxWorks system? > > > > I have always thought that when a breakpoint was set, that some code was inserted at the location of the breakpoint which would halt the execution. However, there is skeptasism since further investigation of the > > breakpoint location using the l and d commands does not show any modification to the code where the breakpoint was set. > > > > Can anyone enlighten me? > > > > -- > > Allen Granum > > Nortel Networks > > Wireless Internet > > BSP Development Team > > Email: agranum@nortelnetworks.com > > Phone: (613)763-4324 > > -- > Lori Fraleigh Director of Engineering > Real-Time Innovations, Inc. http://www.rti.com > Phone: 408-734-4200 x121 Fax: 408-734-5009 - -- - - - - - The axis of the universe is irony. - - - - - --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Newbie: inhibit / trap VME bus error? Date: Thu, 30 Aug 2001 09:14:48 -0600 From: "Mark A. Naivar" Organization: LANL Message-ID: <3B8E5868.2060500@alvie-mail.lanl.gov> References: <48743300.0108220923.1f16c6b0@posting.google.com> <3B853F68.C8F3E2B8@alvie-mail.lanl.gov> <9m4vmn$18k$1@persian.noc.ucla.edu> Reply-To: mnaivar@lanl.gov - --------------070809030703080700010505 Content-Type: text/plain; charset=us-ascii; format=flowed Content-Transfer-Encoding: 7bit Denis wrote: >Mark, >"Mark A. Naivar" wrote in message >news:3B853F68.C8F3E2B8@alvie-mail.lanl.gov... > >>Dave wrote: >> > > [snip] > >>STATUS >>myByteProbe(char *addr, int mode, char *data) >>{ >> if(setjmp(myEnv) == 0) >> { >> if(mode == READ) *data = *addr; >> else *addr = *data; >> } >> else >> return (ERROR); >>} >> > > >so, why does your solution work? Is it supposed to always >work, or is it only because it is a compiler/os combination ? >perhaps it does not work in certain cases? > >here is why i am asking: > >if i remember correctly, longjmp is only guaranteed to work >if the function which contains the setjmp() haven't terminated >already. in your case, it looks like it has. > I have gotten this to work on 2-3 different CPUs, all of them were running VxWorks. I think I've used 2 different compilers with this code. Not exhaustive testing by any means, but I've never had any problems with it. If speed is not an issue, I'd recommend the vxMemProbe() routine instead. The memory access will cause the longjmp to be called if there is a bus error. So in all cases, the function hasn't terminated when the longjmp is called (it should be "hanging" at the memory access lines). Now that I'm looking at it, I left out a return(OK); line for the first part of the if() statement.... - -Mark - --------------070809030703080700010505 Content-Type: text/html; charset=us-ascii Content-Transfer-Encoding: 7bit

Denis wrote:
Mark,
"Mark A. Naivar" <mnaivar@alvie-mail.lanl.gov> wrote in message
news:3B853F68.C8F3E2B8@alvie-mail.lanl.gov...
Dave wrote:


   [snip]

STATUS
myByteProbe(char *addr, int mode, char *data)
{
        if(setjmp(myEnv) == 0)
        {
                if(mode == READ) *data = *addr;
                else *addr = *data;
        }
        else
                return (ERROR);
}


so, why does your solution work? Is it supposed to always
work, or is it only because it is a compiler/os combination ?
perhaps it does not work in certain cases?

here is why i am asking:

if i remember correctly, longjmp is only guaranteed to work
if the function which contains the setjmp() haven't terminated
already. in your case, it looks like it has.
I have gotten this to work on 2-3 different CPUs, all of them were running VxWorks.  I think I've used 2 different compilers with this code.  Not exhaustive testing by any means, but I've never had any problems with it.  If speed is not an issue, I'd recommend the vxMemProbe() routine instead.

The memory access will cause the longjmp to be called if there is a bus error.  So in all cases, the function hasn't terminated when the longjmp is called (it should be "hanging" at the memory access lines).

Now that I'm looking at it, I left out a return(OK); line for the first part of the if() statement....

- -Mark

- --------------070809030703080700010505-- --------------------------- Newsgroups: comp.os.vxworks Subject: Wind View and Windows 2000 Date: Thu, 30 Aug 2001 11:25:28 -0400 From: "Chris Lewis" Organization: NASA Goddard Space Flight Center (skates.gsfc.nasa.gov) Message-ID: <9mlls4$u$1@skates.gsfc.nasa.gov> When I start Tornado I get the error: couldn't read file "TRY=xxx.xxx.xxx.xxx\host\resource\tcl\WindView.win32.tcl": no such file or directory where xxx.xxx.xxx.xxx is the IP address of my host. The Wind View tool doesn't work. I tried changing the value of local host in the windows registry but it didn't help. The file is in the directory and WIND_BASE is set to C:\Tornado. Before switching to Windows 2000 Wind View worked fine. HELP! --------------------------- Newsgroups: comp.os.vxworks Subject: Re: vxIMRClear() and vxIMRSet() on i960Hx and Jx processor Date: Thu, 30 Aug 2001 19:59:28 +0200 From: "Leonid Rosenboim" Organization: Verio Message-ID: References: <3b8dfc02.759480615@news.eunet.no> It has been a while since I did i960 BSPs. but here is what I can remember: The VxWorks kernel never uses IMR, it inly locks interrupt globally, so the only place where these functions are used is the BSP and drivers, some of which need to mask their own interrupt. But even this use in a driver is quite rare. So the main purpise is to enable the interrupts of connected perihperals at startup. For this, the function does not need to be too smart. On the other hand, if you feel the need, you can rewrite them in sysALib.s anyway it pleases you. I think I have rewritten these functions in many of the BSPs I did, and I did some for Jx and Hx. HTH, Leonid "BT" wrote in message news:3b8dfc02.759480615@news.eunet.no... > vxIMRClear() and vxIMRSet() on i960Hx and Jx processor > is not an atomic function. > I have checked my code and BSP where I am using these > functions. After startup I am only using these function > from one task so that they are not atomic should be OK. > > I have experienced some problems that might indicate that > I have not been able to mask put an interrupt properly, I > have therefore also checked my ram file for anny refernces > to these function and I have not found any. > > The Intel 80960HA/HD/HT Specification Update states that > there are several problems with clearing the interrupt mask. > > # Does anybodyknow why VxWorks has implemeted vxIMRClear() > and vxIMRSet() in a no atomic way. (On the i960Cx it is > implemented atomic by refering to sf1 instead of a ld/st > the memory mapped addresses for IMSK as on i960Hx)? > > # Could it be that any VxWorks code mask's out interrupts > without using these function and then fail my use of these > functions since they are not atomic? > > # Has anybody experienced any problems masking out interrupts > on the i960Hx processor? > > Brede > > --------------------------- Newsgroups: comp.os.vxworks Subject: EPFNOSUPPORT, (errno 46) Date: Tue, 28 Aug 2001 20:21:07 +0100 From: "Paul D'Arcy" Organization: Nortel Networks Message-ID: <9mgqri$gr7$1@qnsgh006.europe.nortel.com> I have Wind River web server 2.0 running on a pentium platform. The webserver initialises with no problem, but on general use a EPFNOSUPPORT, (errno 46) appears. Example, I have the httpRpmFs configured to present static webpages(Simple webpages, plain html, no windweb func's etc.) On selecting this webpage, the webserver may present the page as it should or on some occassions(and eventually with repeated requesting of the page) present HTTP 404 - File not found Internet Explorer Doing a task printout from the vxWorks Shell informs that the tHTTPd has errno 46. One thing to note is that this same code works 100% on a StrongARM platform. Any help would be greatly appreciated. Regards Paul --------------------------- Newsgroups: comp.os.vxworks Subject: vxWorks handling "string literals" as opposed to global data. Date: Wed, 29 Aug 2001 11:31:10 +0100 From: "Paul D'Arcy" Organization: Nortel Networks Message-ID: <9mig5u$9pl$1@qnsgh006.europe.nortel.com> Hi all, I am using the following function as part of a WindRiver webserver implementation. /* httpSsiFnConfAdd( ) - add an entry to the SSI routine list Description: This routine adds the specified function pointer to the SSI subroutine configuration list. Note: The string is not copied. Only a pointer to its location is stored. Returns: ID of newly added SSI routine entry, or NULL. */ HTTP_GEN_LIST_ENTRY_ID httpSsiFnConfAdd ( char * szName, /* Template-tag name */ HTTP_SSI_FUNC pFunc /* Corresponding function pointer */ ) The following is an example of its use: httpSsiFnConfAdd("SsiButtonSubmit", buiSsiButtonSubmit); My question is due to the fact that the string is not copied and only a pointer to its location is, will this pointer always correctly point to the strings address or could the string get overwritten. Do I need to declare a global and pass this is as a parameter as follows for example /* global */ char strSsiButtonSubmit[] = "SsiButtonSubmit"; httpSsiFnConfAdd(strSsiButtonSubmit, buiSsiButtonSubmit); Any help would be greatly appreciated as how does vxWorks handle this kind of scenario??? Regards Paul --------------------------- Newsgroups: comp.os.vxworks Subject: Trouble with a socket call Date: Thu, 30 Aug 2001 18:33:08 +0000 (UTC) From: "Robert Gustavsson" Organization: Mailgate.ORG Server - http://www.Mailgate.ORG Message-ID: <032311034a4df8d221215b04c6b1b100.21187@mygate.mailgate.org> Hi! I am trying to run the tcp-example in the vxworks-documentation, but it doesn't work. I can compile the code without any problem but the socket-call in the server-part of the code fails. I'm booting vxworks over a network connection. Does anyone have some general hints about where to start looking for the possible cause of the trouble? Best regards, Robert. - -- Posted from du24-25.ppp.algonet.se [195.100.25.24] via Mailgate.ORG Server - http://www.Mailgate.ORG --------------------------- Newsgroups: comp.sys.powerpc.tech,comp.os.vxworks Subject: VxWorks package/library to use with MPC8260 Date: Thu, 30 Aug 2001 14:40:00 -0400 From: Pierre Olivier Organization: Bell Sympatico Message-ID: <3B8E8880.806B45BC@srtelecom.com> Hi there, We've been developping on the MPC860 with VxWorks 5.4 for a few years now, and are moving to the MPC8260 for the first time. We've been told by our WindRiver FAE that we already have the MPC8260 support library on the MPC860 ditstribution binaries, "hidden" under the guise of the 603 support library. Can anyone confirm or debunk this claim ? To me, it seems that even though the MPC8260 has a 603E core, the support library for the 8260 would be different, just like it is the case for the 403 and 860.... Please reply directly to pierre_olivier@srtelecom.com Thanks ! --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Halt When A Variable Changes? Date: Thu, 30 Aug 2001 08:42:22 -0400 From: david lindauer Message-ID: <3B8E34AE.FDF75A63@notifier-is.net> References: the problem is detecting the change. Some processors such as the x86 series have hardware debug registers that can detect a change to a memory address, but without that you pretty much can't pin it down short of single-stepping the processor and checking the address after each instruction. I doubt if tornado will support such processor-specific things, although if you had an x86 or some other processor with hardware debug registers you might be able to code a short routine into your program that initializes them and sets up an exception routine for them... don't know enough about the WRS operating system to know how close you can get to the instruction that caused the problem when you use this method to shut down tasking, depends on whether the exception handler can be multitasked I guess. Also there is an issue that you would need to hook exception vectors that are also needed by the tornado debugger... David Pat DiRienzo wrote: > Is it possible using the Tornado debugger to halt all tasks when a variable > is changed to a specific value. I would like to halt the program when a > memory location (for this variable) is set erroneously by a coding mistake > or memory overwrite. I could then examine the code at the halted location to > determine who corrupted this memory location. This is a debugging > capability which could have saved my company hours of debug time and myself > a lot of lost late nites and weekends. Maybe it is not possible with only a > source level debugger such as Tornado because of the way they work? --------------------------- Newsgroups: comp.sys.powerpc.tech,comp.os.vxworks Subject: Re: VxWorks package/library to use with MPC8260 Date: Thu, 30 Aug 2001 15:04:16 -0400 From: George Varndell Organization: Newsfeeds.com http://www.newsfeeds.com 90,000+ UNCENSORED Newsgroups. Message-ID: <3B8E8E30.EE2B7660@hotmail.com> References: <3B8E8880.806B45BC@srtelecom.com> Pierre Olivier wrote: > > Hi there, > > We've been developping on the MPC860 with VxWorks 5.4 for a few years > now, and are moving to the MPC8260 for the first time. We've been told > by our WindRiver FAE that we already have the MPC8260 support library on > the MPC860 ditstribution binaries, "hidden" under the guise of the 603 > support library. > > Can anyone confirm or debunk this claim ? To me, it seems that even > though the MPC8260 has a 603E core, the support library for the 8260 > would be different, just like it is the case for the 403 and 860.... > > Please reply directly to pierre_olivier@srtelecom.com Your trusty WindRiver FAE has spoken the truth. No doubt though, you'll need a patch or 2 before you get your new hardware up and running. Were you hoping to have to buy a new development seat? ;) Regards, George Varndell - -- Embedded Systems Expertise Varndell Engineering, LLC. http://www.varndellengineering.com 81 Baltimore St. Suite 206 Cumberland MD. 21502 - -----= Posted via Newsfeeds.Com, Uncensored Usenet News =----- http://www.newsfeeds.com - The #1 Newsgroup Service in the World! - -----== Over 80,000 Newsgroups - 16 Different Servers! =----- --------------------------- Newsgroups: comp.os.vxworks Subject: Re: fec problems,help needed Date: Thu, 30 Aug 2001 21:36:34 +0200 From: Patrick Deiber Organization: PatDB Message-ID: <3B8E95C1.CB26F7A@wanadoo.fr> References: <221caa76.0108300559.5a9f9c53@posting.google.com> Reply-To: patrick.deiber@wanadoo.fr Kewin, Do you have set the value of MII_SPEED to a value different to 0 ? Set MII_SPEED to 0x20 for example. kewin wrote: > hi all > i am debuging mpc860T bsp.i want to write and read PHY's > register,but i can't make it.When i make this action ,there is nothing > effect on the MII_DATA register of FEC which is mapped in dual port > ram.And i try too to write value to MII_DATA directly in dual port > RAM.There is the same results.The MII_DATA register's value is always > 0xffffffff. > Is there any problems on hareware or software.If software's > problem,Will anybody tell me what i should do specially to initialize > FEC. > And advice will be welcome.Thanks a lot! --------------------------- Newsgroups: comp.os.vxworks Subject: Question to VxWorks Source customers - setting breakponts and source debugging for VxWorks "unbreakable" kernel tasks ???? Date: Thu, 30 Aug 2001 16:54:15 -0400 From: "Alexander Povolotsky" Organization: Lucent Technologies, Columbus, Ohio Message-ID: <9mm968$qo1@nntpb.cb.lucent.com> I would like to ask those lucky guys who was purchased VxWorks source from WRS - did any one of you try getting read of "unbreakable" and actually setting breakpoints and doing source debugging for VxWorks kernel tasks (obviously after compiling with "-g" ) - dit it worked well ? Regards --------------------------- Newsgroups: comp.os.vxworks Subject: memShow - value routines Date: Thu, 30 Aug 2001 18:07:05 -0400 From: "Kevin Livingston" Message-ID: I have been looking for a way to get the information provided by memShow - except in a more useful format - e.g. numbers in variables instead of strings from printf's. I received some help from Ranjana Shah on this mailing list in the recent thread subject: memShow. Figuring there must be an API to do this I contacted WRS. To my complete suprise i found there is not, and that there is an outstanding SPR filed to get this functionality: Spr#28527---Is there an API or method to tell how much system memory is available in vxWorks? Problem_Description Customer has a workaround for his problem, but he wants an offical API that would do the following for him: The memShow routine dumps to STDOUT a bunch of memory stats, it doesn't return just the specific stat I want in a programmatically useful fashion, that is into a variable. This Spr is still open for an official fix. memShow obviously has all the information available, but they channel it directly into printf's instead of to variables where we could grab it first if we didn't want it printed. Anyway to make a long story short... WRS gave me their memShow code. So here is the original code (as I received it from WRS) and a modification of that code to return values and to remove all the printing - I also removed the shared memory code since that just called other routines anyway. It appears to work well enough. This code is of course provided "as is" and you accept all responsibility etc.etc. you know the speach. I will make no claim that it is pretty either - I just hacked on their code to get something good enough. (if you have any improvements or comments - please post them to the list so others may benifit) please give any credit to my employer BETA LaserMike 3 files attached: memShow.c (WRS original) memShowValue.h (BLM modification) memShowValue.c (BLM modification) Kevin An agent of WRS issued written permission for me to post this source to the comp.os.vxworks newgroup. Neither my employer nor myself makes any gaurentee either explicit or implicit about the behavior of this code nor will we accept any liability or responsibility for your use of this code [I know you are all smart enough to know all this already, and do not just blindly trust code, I in no way intend to sound patronizing, it does however, makes some people (like myself) feel better to have things disclosed in writing] Kevin Livingston Software Engineer BETA LaserMike begin 666 memShow.c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eferences: <3B8E8880.806B45BC@srtelecom.com> Hello, No, your FAE is telling you the right thing. The 603 in the name of the library refers to the core and not to the CPU part. The 750 and 7400 parts all use the '604' library for similar reasons. HTH, John... Pierre Olivier wrote in message news:<3B8E8880.806B45BC@srtelecom.com>... > Hi there, > > We've been developping on the MPC860 with VxWorks 5.4 for a few years > now, and are moving to the MPC8260 for the first time. We've been told > by our WindRiver FAE that we already have the MPC8260 support library on > the MPC860 ditstribution binaries, "hidden" under the guise of the 603 > support library. > > Can anyone confirm or debunk this claim ? To me, it seems that even > though the MPC8260 has a 603E core, the support library for the 8260 > would be different, just like it is the case for the 403 and 860.... > > Please reply directly to pierre_olivier@srtelecom.com > > Thanks ! --------------------------- Newsgroups: comp.os.vxworks Subject: Re: vxWorks handling "string literals" as opposed to global data. Date: 30 Aug 2001 17:16:41 -0700 From: john_94501@yahoo.com (John) Organization: http://groups.google.com/ Message-ID: <488e459a.0108301616.701732cf@posting.google.com> References: <9mig5u$9pl$1@qnsgh006.europe.nortel.com> Hello, Well, it isn't actually VxWorks that you need to know about. It is the compiler; the compiler will convert your string literal into a pointer to the appropriate bytes. Those bytes will be stored in the image along with your code and any other data you declare, and will not be changed unless your code writes to the space being pointed to (which, btw, you shouldn't really do in the case of literal strings). The more recent VxWorks AE will actually have these strings added to a section called 'rodata' (read-only data), and they will be forcibly write protected by the OS when loaded. That will prevent you from changing the string being pointed to (you'd get an exception if you tried). The only other case that could cause you a problem is if the code module containing the string literal was unloaded. In this case, the memory for it would be freed. HTH, John... "Paul D'Arcy" wrote in message news:<9mig5u$9pl$1@qnsgh006.europe.nortel.com>... > Hi all, > > I am using the following function as part of a WindRiver webserver > implementation. > > /* > httpSsiFnConfAdd( ) - add an entry to the SSI routine list > > Description: This routine adds the specified function pointer to the SSI > subroutine configuration list. > Note: The string is not copied. Only a pointer to its location is stored. > Returns: ID of newly added SSI routine entry, or NULL. > */ > > HTTP_GEN_LIST_ENTRY_ID httpSsiFnConfAdd > ( > char * szName, /* Template-tag name */ > HTTP_SSI_FUNC pFunc /* Corresponding function pointer */ > ) > > The following is an example of its use: > httpSsiFnConfAdd("SsiButtonSubmit", buiSsiButtonSubmit); > > My question is due to the fact that the string is not copied and only a > pointer to its location is, will this pointer always correctly point to the > strings address or could the string get overwritten. Do I need to declare a > global and pass this is as a parameter as follows for example > /* global */ > char strSsiButtonSubmit[] = "SsiButtonSubmit"; > > httpSsiFnConfAdd(strSsiButtonSubmit, buiSsiButtonSubmit); > > Any help would be greatly appreciated as how does vxWorks handle this kind > of scenario??? > Regards > Paul --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Trouble with a socket call Date: 30 Aug 2001 17:26:11 -0700 From: john_94501@yahoo.com (John) Organization: http://groups.google.com/ Message-ID: <488e459a.0108301626.656753d9@posting.google.com> References: <032311034a4df8d221215b04c6b1b100.21187@mygate.mailgate.org> Hello Robert, This has been said many times in here, but I'll say it again... Please do not post vague statements like "it doesn't work" - be more precise about what it is that does not work. You say it compiles OK; a good start. What happens next? Are you downloading the .o file, or linking it with the kernel image? If the former, does it download without errors? How do you run it? If you want a really general hint, then try using the debugger to debug it. Unfortunately, unless you give us some clue as to what's going wrong, that's probably about as good as it gets. I'm not trying to be nasty... people on this group will help you, myself included, but please try to give us as much information about what you're doing as possible. We don't get paid to post here, and we're probably as busy as you are, so having all the information in one message will give us a better chance of seeing the problem and helping you out. Also, as a general rule, you should include the version of Tornado and/or VxWorks you are using, the target architecture and in cases where the tools are concerned, your host OS too. Please post more details and we'll see what we can do to help, John... "Robert Gustavsson" wrote in message news:<032311034a4df8d221215b04c6b1b100.21187@mygate.mailgate.org>... > Hi! > > I am trying to run the tcp-example in the vxworks-documentation, but it > doesn't work. > I can compile the code without any problem but the socket-call in the > server-part > of the code fails. I'm booting vxworks over a network connection. > Does anyone have some general hints about where to start looking for the > possible > cause of the trouble? > > Best regards, Robert. --------------------------- Newsgroups: comp.os.vxworks Subject: flash bootrom and elfToBin Date: 30 Aug 2001 17:54:36 -0700 From: ryeung@earthlink.net (Raymond Yeung) Organization: http://groups.google.com/ Message-ID: This posting concerns programming boot code into flash boot section. I read some of the older threads here and found out how I can convert ELF format of bootrom_uncmp to binary that is immediately executable. I'm however seeing "garbages" at the beginning of the .bin file, which is the Copyright notice of Wind River. The following are some definitions in Makefile: ROM_BASE_ADRS = fff00000 ROM_TEXT_ADRS = fff00100 Basically, the PPC would jump to offset 0x100 to execute. Question: Should I skip the first 0x100 bytes of boot.bin file and start copying at offset 0x100 of the file to offset 0x100 on boot flash? Thanks Raymond --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Wind View and Windows 2000 Date: Thu, 30 Aug 2001 18:54:41 -0700 From: DrDiags Organization: laserlink.net Message-ID: <3B8EEE61.C753C629@covad.net> References: <9mlls4$u$1@skates.gsfc.nasa.gov> Reply-To: drdiags@covad.net Chris, I don't know the reason why you are getting this error, but I thought I remembered reading about this before in this newsgroup. I believe there was a patch for this, you could try going on WindSurf or maybe a google search on this newsgroup would give you an answer. Chris Lewis wrote: > > When I start Tornado I get the error: > > couldn't read file > "TRY=xxx.xxx.xxx.xxx\host\resource\tcl\WindView.win32.tcl": no such file or > directory > > where xxx.xxx.xxx.xxx is the IP address of my host. The Wind View tool > doesn't work. I tried changing the value of local host in the windows > registry but it didn't help. The file is in the directory and WIND_BASE is > set to C:\Tornado. > > Before switching to Windows 2000 Wind View worked fine. > > HELP! - -- http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html (vxWorks FAQ home page) --------------------------- Newsgroups: comp.sys.powerpc.tech,comp.os.vxworks Subject: Re: VxWorks package/library to use with MPC8260 Date: 30 Aug 2001 19:46:51 -0700 From: skipper@no-spam.calweb.com (Skipper Smith) Organization: Helpful Knowledge Consulting Message-ID: References: <3B8E8880.806B45BC@srtelecom.com> Reply-To: skipper@no-spam.calweb.com Pierre Olivier wrote: >Hi there, > >We've been developping on the MPC860 with VxWorks 5.4 for a few years >now, and are moving to the MPC8260 for the first time. We've been told >by our WindRiver FAE that we already have the MPC8260 support library on >the MPC860 ditstribution binaries, "hidden" under the guise of the 603 >support library. > >Can anyone confirm or debunk this claim ? To me, it seems that even >though the MPC8260 has a 603E core, the support library for the 8260 >would be different, just like it is the case for the 403 and 860.... From the CPU point of view, they are exactly the same- having been involved with the design of both of these chips, I can assure you it is true. I can also assure you that the 403 and 860 are extremely unrelated, CPU-wise, having been designed by completely separate design teams- one at IBM and one Motorola. - -- Skipper Smith Helpful Knowledge Consulting Worldwide Microprocessor Architecture Training PowerPC, ColdFire, 68K, CPU32 Hardware and Software /* Remove no-spam. from the reply address to send mail directly */ --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Why not bring up the network interface! Date: 30 Aug 2001 20:04:46 -0700 From: wys205@sina.com (John Wang) Organization: http://groups.google.com/ Message-ID: References: <3B8D0110.4C5044DE@covad.net> DrDiags : Thank for your help .I searched the comp.os.vxworks ,and as the posts change config.h parameters. As follow: ....... #define DEFAULT_BOOT_LINE \ "fd=0,0(0,0)host:/fd0/vxWorks.st h=192.168.0.68 e=192.168.0.69 u=target o=ene tn=target" ...... #define IO_ADRS_ENE 0x200 #define INT_LVL_ENE 0x0b #define INT_VEC_ENE (INT_NUM_IRQ0 + INT_LVL_ENE) /* Hardware jumper is used to set RJ45(Twisted Pair) AUI(Thick) BNC(Thin) */ ...... #ifdef INCLUDE_ENE #define INCLUDE_PCI /*include PCI bus library */ #define NETIF_USR_DECL \ IMPORT int eneattach(); #define NETIF_USR_ENTRIES \ {"ene", eneattach, (char *)NONE, 0, 0, 0}, #endif /*INCLUDE_ENE */ ....... All are done.But nothing to happen. :( I use eneattach() manually,ths system looks like in dead loop until I type ^c. What should i do? Regards John --------------------------- Newsgroups: comp.os.vxworks Subject: Re: fec problems,help needed Date: 30 Aug 2001 20:56:13 -0700 From: kevinatnj@yahoo.com.cn (kewin) Organization: http://groups.google.com/ Message-ID: <221caa76.0108301956.50b3a242@posting.google.com> References: <221caa76.0108300559.5a9f9c53@posting.google.com> <3B8E95C1.CB26F7A@wanadoo.fr> Patrick Deiber wrote in message news:<3B8E95C1.CB26F7A@wanadoo.fr>... > Kewin, > > Do you have set the value of MII_SPEED to a value different to 0 ? > Set MII_SPEED to 0x20 for example. > Thank you for your ringing the bell.I have set this value.And i have already initilize the other registers.I don't know why that is. --------------------------- Newsgroups: comp.os.vxworks Subject: simple standalone development environment Date: Fri, 31 Aug 2001 07:40:31 +0200 From: Rainer Stelzer Organization: Stoye GmbH, =?iso-8859-1?Q?K=F6ln?= Message-ID: <3B8F234F.56878750@stoye.de> Hi, we have the need to provide our customers a method to write own simple C applications to extend the software written by us. They do not need a powerfull Tornado environment. They do not need to use vxworks library calls. They do not need to use a debugger. They only need to compile single C Files into powerPC object format, and link them in a way, so that the linked file can be loaded by the target's object loader. - - Has somebody out there an idea of how to configure a simple standalone development enviroment with GNU tools on a winNT system , without using Tornado/vxworks components ? Thanx in advance. R.Stelzer ===================================================== Gebrüder Stoye GmbH department: traffic control systems name: Rainer Stelzer adr: Longericher Strasse 177 city: Cologne, Germany zip: 50739 mobil: 0173/7266090 phone: ++49/221/1707019 fax: ++49/221/1707010 mailto: r.stelzer@stoye.de www: http://www.stoye.de ===================================================== --------------------------- Newsgroups: comp.os.vxworks Subject: Re: CPX8216 cPCI Chasis Date: Fri, 31 Aug 2001 01:42:04 -0400 From: "Arturo Figueroa" Message-ID: <_qFj7.8728$W52.425555@weber.videotron.net> References: <27750684.0108290751.5b36423b@posting.google.com> Hi JD, Yes, Why? "JD" a écrit dans le message news: 27750684.0108290751.5b36423b@posting.google.com... > Hi, > > Anyone have any experience running VxWorks 5.4 on MCP750HA board in > CPX8216 platform? Thanks for any insight. --------------------------- Newsgroups: comp.os.vxworks Subject: recommended G.703 interface ? Date: Fri, 31 Aug 2001 08:25:13 +0200 From: Rainer Stelzer Organization: Stoye GmbH, =?iso-8859-1?Q?K=F6ln?= Message-ID: <3B8F2DC9.55F06D3@stoye.de> Hi, We must provide a realtime data communication link between two computer systems over a long distance. We choose a xDSL modem with a E1/G.703 interface. Now I'm looking for a PMC-Modul/VME Board with a vxworks driver. Because it is a special control application and not a common telecomunication task, we do not need a protocol like SS7, ATM or a hardware audio stream interface like SCA. We need a board with a chipset and a driver that makes it possible to access each single 64KBit by software. Can somebody recommend a board with this feature ? Thanx in advance. R.Stelzer vxworks tornado ===================================================== Gebrüder Stoye GmbH department: traffic control systems name: Rainer Stelzer adr: Longericher Strasse 177 city: Cologne, Germany zip: 50739 mobil: 0173/7266090 phone: ++49/221/1707019 fax: ++49/221/1707010 mailto: r.stelzer@stoye.de www: http://www.stoye.de ===================================================== --------------------------- Newsgroups: comp.os.vxworks Subject: Tornado prototyper(Solaris Version) - vxsim simulator ping problem. Date: Fri, 31 Aug 2001 06:47:18 GMT From: "ÃÖ°­¿À" Organization: Dacom Message-ID: Hi all. I got prototyper solaris version from WRS and installed to my server. Also installed license file provided by email. My machine is sun enterprise 3500, with solaris 7 11/99. I have one bug report and one critical problem with ping using PPP. 1. Bug Reporting: After extracting "setup.tar", the shell script file "SETUP" does not work properly unless "LANG" environment is set to "C". The following error message occured. $ env | grep LANG LANG=ko $ ./SETUP Tornado 2.0 Setup Copyright (C) 1999, Wind River Systems, Inc. All Rights Reserved expr: ±¸¹®¿À·ù <-- display korean words. ./SETUP: test: Àμö°¡ ¿¹»óµÊ <--- same korean words. let's change locale. $ export LANG=C (ksh) $./SETUP Tornado 2.0 Setup Copyright (C) 1999, Wind River Systems, Inc. All Rights Reserved (it works!!!) Actually, I'm using "LANG=ko" - 2bytes LANG, but it didn't work with "/bin/expr" in "tmpfreespace()". so I changed to "C", then it works well!! 2. ping problem reports using ppp for Solaris 2.7 I installed ppp packages to my server and started ppp daemon using tornado provided configuration file. ($WIND_BASE/target/config/solaris/asppp.cf) look at my screen capture. $ isainfo -b 64 $ pkginfo | egrep 'ppp|bnu' system SUNWapppr PPP/IP Asynchronous PPP daemon configuration files system SUNWapppu PPP/IP Asynchronous PPP daemon and PPP login service system SUNWbnur Networking UUCP Utilities, (Root) system SUNWbnuu Networking UUCP Utilities, (Usr) system SUNWpppk PPP/IP and IPdialup Device Drivers system SUNWpppkx PPP/IP and IPdialup Device Drivers (64-bit) $ cp $WIND_BASE/target/config/solaris/asppp.cf /etc $ /etc/init.d/asppp start $ ifconfig -a lo0: flags=849 mtu 8232 inet 127.0.0.1 netmask ff000000 hme0: flags=863 mtu 1500 inet 150.150.53.30 netmask ffffff80 broadcast 150.150.53.127 ipd0: flags=80c1 mtu 1500 inet 127.0.1.254 netmask ffffff00 It seems to be good state yet. Then I created my own bootable vxworks image by editing "$WIND_BASE/target/config/solaris/config.h" not using project. I added or modified following lines and compiled vxworks without error. #define INCLUDE_NETWORK #define INCLUDE_NET_INIT #undef INCLUDE_ULIP #define INCLUDE_PING #define INCLUDE_PPP #define INCLUDE_NET_SHOW #undef WDB_COMM_TYPE #define WDB_COMM_TYPE WDB_COMM_NETWORK.) I pressed "lauch -> VxSim" button and start vxsim0 target server. Look at the target server vxsim0 screen capture. Attaching network interface ppp0... ppp0: ppp 2.1.2 started by ppp0: Connect: ppp0 <--> /tyCo/1 ppp0: local IP address 127.0.1.0 ppp0: remote IP address 127.0.1.254 done. Attaching network interface lo0... done. NFS client support not included. VxWorks Copyright 1984-1998 Wind River Systems, Inc. CPU: SunOS 5.7 [sun4u] VxWorks: 5.4.2 BSP version: 1.2/0 Creation date: Aug 25 2001 WDB: Ready. In seesion logger for vxsim0 (verbose mode) tgtsvr.ex (vxsim0@songgang): Mon Aug 27 10:32:36 2001 Checking License ...OK Connecting to target agent... succeeded. Attaching C++ interface... succeeded. Attaching elf OMF reader for SIMSPARCSOLARIS CPU family... succeeded. Then started vxsim1 using same method (change to processor number 2 and ip address to127.0.1.1) Let's look at the my ping problem(also look routeShow result!) In vxsim0: - -> hostShow hostname inet address aliases - -------- ------------ ------- localhost 127.0.0.1 songgang 150.150.53.30 value = 0 = 0x0 - -> routeShow ROUTE NET TABLE destination gateway flags Refcnt Use Interface - ---------------------------------------------------------------------------- 0.0.0.0 127.0.1.254 3 1 0 ppp0 150.150.0.0 127.0.1.254 3 0 0 ppp0 - --------------------------------------------------------------------------- ROUTE HOST TABLE destination gateway flags Refcnt Use Interface --------------------------------------------------------------------------- - - 127.0.0.1 127.0.0.1 5 1 3 lo0 127.0.1.0 127.0.0.1 7 0 0 ppp0 127.0.1.254 127.0.1.0 5 3 67 ppp0 - --------------------------------------------------------------------------- value = 0 = 0x0 In vxsim1: - -> hostShow hostname inet address aliases - -------- ------------ ------- localhost 127.0.0.1 songgang 150.150.53.30 value = 0 = 0x0 - -> routeShow ROUTE NET TABLE destination gateway flags Refcnt Use Interface - ---------------------------------------------------------------------------- 0.0.0.0 127.0.1.254 3 1 0 ppp0 150.150.0.0 127.0.1.254 3 0 0 ppp0 --------------------------------------------------------------------------- - - ROUTE HOST TABLE destination gateway flags Refcnt Use Interface --------------------------------------------------------------------------- - - 127.0.0.1 127.0.0.1 5 1 1 lo0 127.0.1.1 127.0.0.1 7 0 0 ppp0 127.0.1.254 127.0.1.1 5 3 61 ppp0 --------------------------------------------------------------------------- - - value = 0 = 0x0 In vxsim0: - -> ping "127.0.1.0" PING 127.0.1.0: 56 data bytes no answer from 127.0.1.0 value = -1 = 0xffffffff (can't ping vxsim0 itself) - -> ping "127.0.1.1" PING 127.0.1.1: 56 data bytes no answer from 127.0.1.1 value = -1 = 0xffffffff (can't ping vxsim1) In vxsim1: (it was similar to vxsim0) - -> ping "127.0.1.1" PING 127.0.1.1: 56 data bytes no answer from 127.0.1.1 value = -1 = 0xffffffff - -> ping "127.0.1.0" PING 127.0.1.1: 56 data bytes no answer from 127.0.1.0 value = -1 = 0xffffffff So, How can I ping two vxsim simulator? I spend time almost two weeks using "routeAdd/routeDelete" commands. please tell me solution... Thanks in advance. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Halt When A Variable Changes? Date: Fri, 31 Aug 2001 07:01:31 GMT From: "Koen Van Hulle" Organization: Xeikon Message-ID: References: <3B8E34AE.FDF75A63@notifier-is.net> Pat, The only reliable method I know of that allows you to detect a variable being changed is using an ICE (In-Circuit Emulator). Software methods such as single stepping may hide the problem and its cause because they affect the timing of the system. I haven't used an ICE in a VxWorks environment yet however. Can someone who has tell us if Wind River's VisionICE offers such feature? ICE's don't come cheap, but thay can save you hours (and nights and weekends) of time! Koen "david lindauer" wrote in message news:3B8E34AE.FDF75A63@notifier-is.net... > the problem is detecting the change. Some processors such as the x86 series > have hardware debug registers that can detect a change to a memory address, but > without that you pretty much can't pin it down short of single-stepping the > processor and checking the address after each instruction. I doubt if tornado > will support such processor-specific things, although if you had an x86 or some > other processor with hardware debug registers you might be able to code a short > routine into your program that initializes them and sets up an exception routine > for them... don't know enough about the WRS operating system to know how close > you can get to the instruction that caused the problem when you use this method > to shut down tasking, depends on whether the exception handler can be > multitasked I guess. Also there is an issue that you would need to hook > exception vectors that are also needed by the tornado debugger... > > David > > Pat DiRienzo wrote: > > > Is it possible using the Tornado debugger to halt all tasks when a variable > > is changed to a specific value. I would like to halt the program when a > > memory location (for this variable) is set erroneously by a coding mistake > > or memory overwrite. I could then examine the code at the halted location to > > determine who corrupted this memory location. This is a debugging > > capability which could have saved my company hours of debug time and myself > > a lot of lost late nites and weekends. Maybe it is not possible with only a > > source level debugger such as Tornado because of the way they work? > --------------------------- Newsgroups: comp.os.vxworks Subject: Who can tell me how can I accelerate the speed of memory access Date: 31 Aug 2001 00:48:50 -0700 From: gong.chaomei@mail.zhongxing.com (gongchaomei) Organization: http://groups.google.com/ Message-ID: I am using Tornado II to make bootrom_uncmp. In my romInit.s ,I enable data cache to accelerate the speed of memory access described as follow: CPU: XPC8240RZU250C Memory clock frequency: 100M Hz CPU core frequency:250M Hz Memory type: SDRAM ---Kingmax PC133 (KSV884T4A1A-07) Memory size: 64M I enable data cache using BAT method value of BAT Register is: #define PROM_BASE 0xffF00000 #define PRAM_BASE 0x00000000 #define VROM_BASE PROM_BASE #define VRAM_BASE PRAM_BASE #define PORTX_BASE 0XFF000000 #define IBAT0L_VAL (PROM_BASE | BAT_READ_ONLY |BAT_WRITE_THROUGH) #define IBAT0U_VAL (VROM_BASE|BAT_VALID_SUPERVISOR|BAT_VALID_USER|BAT_BL_512K) #define DBAT0L_VAL (PROM_BASE | BAT_READ_ONLY |BAT_WRITE_THROUGH |BAT_CACHE_INHIBITED) #define DBAT0U_VAL IBAT0U_VAL #define IBAT1L_VAL (PRAM_BASE | BAT_READ_WRITE ) #define IBAT1U_VAL (VRAM_BASE|BAT_BL_64M|BAT_VALID_SUPERVISOR| BAT_VALID_USER) #define DBAT1L_VAL IBAT1L_VAL #define DBAT1U_VAL IBAT1U_VAL #define IBAT2L_VAL (PORTX_BASE | BAT_READ_WRITE |BAT_WRITE_THROUGH) #define IBAT2U_VAL (PORTX_BASE|BAT_BL_8M|BAT_VALID_SUPERVISOR| BAT_VALID_USER) #define DBAT2L_VAL IBAT2L_VAL #define DBAT2U_VAL IBAT2U_VAL #define IBAT3L_VAL (BAT_NO_ACCESS) #define IBAT3U_VAL (BAT_INVALID) #define DBAT3L_VAL (BAT_NO_ACCESS) #define DBAT3U_VAL (BAT_INVALID) #define BAT_BL_128K 0x00000000 #define BAT_BL_256K 0x00000004 #define BAT_BL_512K 0x0000000C #define BAT_BL_1M 0x0000001C #define BAT_BL_2M 0x0000003C #define BAT_BL_4M 0x0000007C #define BAT_BL_8M 0x000000FC #define BAT_BL_16M 0x000001FC #define BAT_BL_32M 0x000003FC #define BAT_BL_64M 0x000007FC #define BAT_BL_128M 0x00000FFC #define BAT_BL_256M 0x00001FFC /* supervisor/user valid mode definitions - Upper BAT*/ #define BAT_VALID_SUPERVISOR 0x00000002 #define BAT_VALID_USER 0x00000001 #define BAT_INVALID 0x00000000 /* WIMG bit settings - Lower BAT */ #define BAT_WRITE_THROUGH 0x00000040 #define BAT_CACHE_INHIBITED 0x00000020 #define BAT_COHERENT 0x00000010 #define BAT_GUARDED 0x00000008 /* Protection bits - Lower BAT */ #define BAT_NO_ACCESS 0x00000000 #define BAT_READ_ONLY 0x00000001 #define BAT_READ_WRITE 0x00000002 But when I test the memory speed I found out the speed of memory access with Instruction Cache and Data Cache is only the three times of the speed of memory access with Instruction. I estimate that the speed of memory access with I&D cache is about 64M Bytes/s --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Halt When A Variable Changes? Date: 31 Aug 2001 01:17:48 -0700 From: vincent_hue@yahoo.com (Vincent) Organization: http://groups.google.com/ Message-ID: <67bf868f.0108310017.3fa2de6b@posting.google.com> References: Pat, Look at the Shell user manual, and search for 'bh()' command. It's probably what you're looking for. It's available on x86, i960, some MIPS and PPC. HTH. Vincent "Pat DiRienzo" wrote in message news:... > Is it possible using the Tornado debugger to halt all tasks when a variable > is changed to a specific value. I would like to halt the program when a > memory location (for this variable) is set erroneously by a coding mistake > or memory overwrite. I could then examine the code at the halted location to > determine who corrupted this memory location. This is a debugging > capability which could have saved my company hours of debug time and myself > a lot of lost late nites and weekends. Maybe it is not possible with only a > source level debugger such as Tornado because of the way they work? --------------------------- Newsgroups: comp.os.vxworks Subject: Re: How do breakpoints work??? Date: 31 Aug 2001 01:23:01 -0700 From: john_94501@yahoo.com (John) Organization: http://groups.google.com/ Message-ID: <488e459a.0108310023.6022f974@posting.google.com> References: <3B8CE9C7.5005E14D@americasm01.nt.com> <3B8DAF90.DD2A0A88@rti.com> <3B8E5185.991FD6B0@americasm01.nt.com> Hello, An interesting approach, but flawed by one little detail. The command you are using to display the data is: > -> sp printf,"At address %x, value = %x.\n",pData,*pData You have used sp to try to avoid the unbreakable nature of the shell, but unfortunately printf does not evaluate '*pData' - the shell evaluates it, and then passes the result to printf. That means that it is still the shell that is reading the memory. To make this work, try the following: - -> sp d, pData, 1, 4 That should spawn the 'd' function which I don't believe checks the bp list before working (if you want to be sure, you have the source for it in target/src/usr/usrLib.c). Having a PPC604 target handy shows the following (this is AE, but should be the same for 5.4 - except for the shell prompt): [vxKernel] -> b 0x00178304 value = 0 = 0x0 [vxKernel] -> sp d, 0x00178304,1, 4 task spawned: tid = 0x301d38, name = pdt4 value = 3153208 = 0x301d38 [vxKernel] -> Dump memory for vxKernel (PD ID 0x203ef4) 0x00178300: 7fe00008 * ........* bdall value = 0 = 0x0 [vxKernel] -> sp d, 0x00178304,1, 4 task spawned: tid = 0x301d38, name = pdt5 value = 3153208 = 0x301d38 [vxKernel] -> Dump memory for vxKernel (PD ID 0x203ef4) 0x00178300: 9421fff0 * ! ........* A disassembly (without the bp installed) shows this: [vxKernel] -> l 0x00178304 Disassembly for vxKernel (PD ID 0x203ef4) pdShow: 0x00178304 9421fff0 stwu r1,-16(r1) 0x00178308 7c0802a6 mfspr r0,LR HTH, John... "Macleod, Randy [CAR:CW31:EXCH]" wrote in message news:<3B8E5185.991FD6B0@americasm01.nt.com>... > Hi, > > Well, here is what I see using a telnet session to a VME PPC target. > > Find some function to break: > -> lkup "ifShow" > ifShow 0x001e6884 text (cesnetvx.pserv.o.sym) > value = 0 = 0x0 > -> b > No breakpoints > value = 0 = 0x0 > -> pData=0x001e6884 > pData = 0x1ffebd0: value = 1992836 = 0x1e6884 = ifShow > -> sp printf,"At address %x, value = %x.\n",pData,*pData > task spawned: id = 0x160fde8, name = t5 > value = 23133672 = 0x160fde8 > -> At address 1e6884, value = 9421ffc8. > > # Add a breakpoint: > -> b 0x1e6884 > value = 0 = 0x0 > -> b > 0x001e6884: ifShow Task: all Count: 0 > value = 0 = 0x0 > -> sp printf,"At address %x, value = %x.\n",pData,*pData > task spawned: id = 0x160fde8, name = t6 > value = 23133672 = 0x160fde8 > -> At address 1e6884, value = 9421ffc8. > # Someone suggested that the behavior could be different > # if there is more than one breakpoint... > -> b 0x1e6888 > value = 0 = 0x0 > -> b > 0x001e6888: ifShow +0x4 Task: all Count: 0 > 0x001e6884: ifShow Task: all Count: 0 > value = 0 = 0x0 > -> sp printf,"At address %x, value = %x.\n",pData,*pData > task spawned: id = 0x160fde8, name = t7 > value = 23133672 = 0x160fde8 > -> At address 1e6884, value = 9421ffc8. > > So it sure looks like the instruction at the breakpoint > address was not overwritten. Is VxWorks checking a breakpoint > list every time the PC is incremented? > > // Randy "The Inquisitive Skeptic" MacLeod > > Lori Fraleigh wrote: > > > > The tgtsvr generally caches the text segments for VxWorks. If you > > do an 'l' from WindShell, you won't see the changed instructions. > > > > If the target shell is available, perform an 'l' from it - you should > > see the instruction used to set the breakpoint. If your kernel > > has the proper components built in, you can also do an '@l' from > > WindShell. Specifying the '@' sign before the 'l' tells WindShell > > and the tgtsvr to execute the command on the target and return the > > response. > > > > Regards, > > Lori > > > > "Granum, Allen [CAR:CF81:EXCH]" wrote: > > > > > > Wondering how the whole breakpoint mechanism works on a PPC/vxWorks system? > > > > > > I have always thought that when a breakpoint was set, that some code was inserted at the location of the breakpoint which would halt the execution. However, there is skeptasism since further investigation of the > > > breakpoint location using the l and d commands does not show any modification to the code where the breakpoint was set. > > > > > > Can anyone enlighten me? > > > > > > -- > > > Allen Granum > > > Nortel Networks > > > Wireless Internet > > > BSP Development Team > > > Email: agranum@nortelnetworks.com > > > Phone: (613)763-4324 > > > > -- > > 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: WindView doesn't starts with Tornado Date: Fri, 31 Aug 2001 10:38:32 +0200 From: "Jonathan ILIAS" Organization: University of Angers, France. Message-ID: <9mnias$27td$1@news.univ-angers.fr> Hi, I've got a little problem, Tornado's manuals may contain the solution. However, I ask you for it because . We are moving our PC's from Windows NT to Windows 2000. Then I needed to installed Tornado again. When I starts it, I get an error message : "Couldn't read file "TRY=localhost\host\ressource\tcl\Windview.win32.tcl" : no such file or directory". However, this file exists !!! Any idea about this ? PS : Sorry for my poor english, I'm not a native speaker (I'm french). - -- Jonathan ILIAS --------------------------- Newsgroups: comp.os.vxworks Subject: Re: WindView doesn't starts with Tornado Date: Fri, 31 Aug 2001 10:59:19 +0200 From: "Jonathan ILIAS" Organization: University of Angers, France. Message-ID: <9mnjhr$28kn$1@news.univ-angers.fr> References: <9mnias$27td$1@news.univ-angers.fr> All apologizes, I'm in a hurry and I didn't see the previous thread about this... (I searched for "Windview" and not for "Wind View")... --------------------------- Newsgroups: comp.os.vxworks Subject: Ethernet cable disconnect Date: Fri, 31 Aug 2001 08:58:34 GMT From: "Leo Hynninen" Organization: Sonera corp Internet services Message-ID: <_kIj7.69$Ob6.4465@read2.inet.fi> Hi. Is there any way to get information about ethernet cable? Is it connect or disconnect. Ours hardware dont support that. I mean I dont get information about ethernet carrier. Leo Hynninen --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Ethernet cable disconnect Date: Fri, 31 Aug 2001 06:44:45 -0400 From: George Varndell Organization: Newsfeeds.com http://www.newsfeeds.com 90,000+ UNCENSORED Newsgroups. Message-ID: <3B8F6A9C.6848187C@hotmail.com> References: <_kIj7.69$Ob6.4465@read2.inet.fi> Leo Hynninen wrote: > > Hi. > Is there any way to get information about ethernet cable? You mean something like BOOL isTheEthernetCableConnected()? > Is it connect or disconnect. > Ours hardware dont support that. I'll wager that it does. There are as many ways as there are combinations of MAC controller and PHY. > I mean I dont get information about ethernet carrier. You're not looking in the right place, and posting a vague and general question such as this to the wrong news group isn't getting you any closer to the answer. Please give some specifics. Regards, George Varndell - -- Varndell Engineering, LLC. Embedded Systems Expertise 81 Baltimore St. Suite 206 http://www.varndellengineering.com Cumberland MD. 21502 - -----= Posted via Newsfeeds.Com, Uncensored Usenet News =----- http://www.newsfeeds.com - The #1 Newsgroup Service in the World! - -----== Over 80,000 Newsgroups - 16 Different Servers! =----- --------------------------- End of New-News digest ********************** From vxwexplo-errs@csg.lbl.gov Fri Aug 31 07:15:21 2001 From: "David Laight" Date: Fri Aug 31 07:15:24 PDT 2001 Subject: Re: Ethernet cable disconnect > Is there any way to get information about ethernet cable? > Is it connect or disconnect. > Ours hardware dont support that. > I mean I dont get information about ethernet carrier. Look around the back of the machine... :-) It all depends on the ethernet controller and the type of cable. Some old ethernet drivers used these sort of 'tricks' to auto-detect the correct interface (AUI, thin, TP) at boot time. Unfortunately they are not 100% reliable - especially if you are bringing a system up after a power cut and the network hub (etc) isn't powered yet. For 10M ethernet, controllers with a built-in TP interface will have a register that indicates whether they are receiving 'link test' pulses from the far end. For 10M and an AUI connected transceiver (even if on board - most thin ethernet boards) the controller will report 'missing carrier' if the transceiver is absent, and probably if the cable is absent. (Actually this error is the lack of 'receive enable' during the transmit.) For 100M you need to read the transciever registers using the MII interface. Most vxWorks drivers make it hard to do any of the above! Actually, the easy way is to assume there are broadcast packets on the network and see if the count of packets is increasing! David From vxwexplo-errs@csg.lbl.gov Fri Aug 31 07:18:41 2001 From: "David Laight" Date: Fri Aug 31 07:18:44 PDT 2001 Subject: Re: Halt when a variable changes I managed to get this to trap on the strongArm using the CPUs hardware debug registers and intercepting the data fault vector. I had to disable the data trap in the ISR - before falling into the standard vxWorks fault handler. Because I hadn't linked this into the vxWorks debug code, it looked like a 'fatal' data fault - so I couldn't continue the task. However having the task suspended at the fault is usually good enough to let you work out what has happened. David From vxwexplo-errs@csg.lbl.gov Fri Aug 31 12:52:13 2001 From: jmk Date: Fri Aug 31 12:52:15 PDT 2001 Subject: vxWorks PCI Driver for PLX-9054 Hi all: I'm working on a vxWorks driver for a PC-104 Plus(PCI) board which uses the PLX-9054 chip. My system uses a WinSystems EBC-TXPlus with a Pentium processor and I am using the pcPentium BSP with modifications as my starting point. I'm doing preliminary testing with the PLX-9054 HDK, which includes a PCI board to evaluate the chip. I am able to read the PCI configuration space without problems and I can see the BARS. I am able to read the PLX registers through I/O space, but I am unable to perform the memory accesses to the memory mapped registers or the static ram on the evaluation board. My driver's HW init function is copied from the AIC7880 SCSI controller driver that comes with the BSP, with modifications for the PLX board. This is reading the Base Address Registers, calling sysMmuMapAdd, OR'ing the IO and Memory bits into the PCI Configuration Command Register. When I boot, the system locks up somewhere between sysHwInit and sysHwInit2. If I leave out the call to sysMmuMapAdd, the system boots, but the memory on the board is inaccessible the system bus errors when I try to read or write the contents of the BAR's. I then tried a version in which I manually put the mapping entries into sysPhysMemDesc in syslib.c instead of using sysMmuMapAdd. This version booted, but I only see 0xffff's when I look at the memory and I am unable to change the memory. The board is known good, as it was tried in a PC using the PLX monitor. I've gone through the archives of the newsgroup and have found some relevent stuff, but nothing that has solved my problem. I'm sure I'm leaving something out, but I'm not sure what it is. Any advice or suggestions would be very much appreciated. --joel From vxwexplo-errs@csg.lbl.gov Fri Aug 31 14:58:16 2001 From: Bruce Wilson Date: Fri Aug 31 14:58:18 PDT 2001 Subject: Re: Force 3750 cPCI Init Thanks to Karin Willers, and Chirag Patel of Force Computers for your help. Both of them recommended changing a setting in config.h: FROM: #define CPCI_BRIDGE_CONTROL PCI_21150_BCR_MAM TO: #define CPCI_BRIDGE_CONTROL 0 That solved the "machine check" problem. Additionally, I learned that peripheral boards when using the Force 3750 cpu board have device numbers between 9 - 15 now (and they're on bus 1). Whereas before, with the Force 3603, everything was on bus 0 and the peripheral board device numbers ranged from 25 - 31. --Bruce Wilson >Submitted-by vxwexplo-errs@csg.lbl.gov Wed Aug 29 11:29:15 2001 >Submitted-by: Bruce Wilson > >We're using VxWorks 5.4, Tornado2, and trying to use a Force 3750 cPCI board. >Has anybody been using the Force 3750 board with devices in other >slots? Apparently, those devices are on bus 1, but when we tried >calling frcPciConfigReadDword( bus=1, dev=25, func=0, regOffset=0, >&temp) we got a "machine check" error. > >We had no trouble initializing the devices in other slots when using >the Force 3603 Compact PCI board. Everything was on bus 0. > >What do we need now, to get the devices initialized on bus 1 for the 3750? > > >Bruce A. Wilson >U.C. Lawrence Livermore National Lab >P.O. Box 808 (L-278) >Livermore, CA 94550 >voice: (925) 422-4145 >fax: (925) 423-1243