From vxwexplo-errs@csg.lbl.gov Tue Feb 1 19:13:46 2000 From: "Raymund Christopher A. Toribio" Date: Tue Feb 1 19:13:49 PST 2000 Subject: relocation overflow Hello, Does anyone know about relocation overflow? Im trying to make an object file and during compilation I got that error. I am using ccmips and my architecture is MIPS with a cputype of R4650. But when I tried using gcc or ccsimso I did not get that error. The ROM_SIZE is set to its maximum. Is there anything that I need to alter/configure in VxWorks? in the configuration files? Thank you very much. Best Regards, rc mailto:rtoribio@tspi.com.ph From vxwexplo-errs@csg.lbl.gov Tue Feb 1 23:49:45 2000 From: "Chuljin Ahn" Date: Tue Feb 1 23:49:48 PST 2000 Subject: reply to relocation overflow Subject: relocation overflow Submitted-by: "Raymund Christopher A. Toribio" Hello, Does anyone know about relocation overflow? Im trying to make an object file and during compilation I got that error. I am using ccmips and my architecture is MIPS with a cputype of R4650. But when I tried using gcc or ccsimso I did not get that error. The ROM_SIZE is set to its maximum. Is there anything that I need to alter/configure in VxWorks? in the configuration files? Thank you very much. Best Regards, rc mailto:rtoribio@tspi.com.ph ----------------------------------------- Dear Mr. Toribio, If ccmips generates an object image in COFF format, the "--split-by-reloc" option of $(LD) might be of help. In my case, ldsh, linker for SH processor, generated relocation overflow error because COFF couldn't have 65535 relocations in a single section. With the option, the error disappeared. Check the $(LD) manual about the option. I wish it might be useful to you. Chuljin Ahn ------------------------------------------------------------------ Ahn, Chuljin (ahnmace@lge.co.kr) Engineer IK Business Team, Digital Venture Business, LG Electronics, Inc. Jinwee Myun, Pyongtaek City, Kyungki Do, Rep. of Korea, 451-713 Phone:+82-333-610-5330, FAX:+82-333-610-5352, PCS:+82-19-223-6599 ------------------------------------------------------------------ From vxwexplo-errs@csg.lbl.gov Wed Feb 2 04:03:17 2000 From: Vxworks Exploder Date: Wed Feb 2 04:03:20 PST 2000 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Wed Feb 2 04:03:10 PST 2000 Subject: Re: cannot boot VxWorks on MCPN750 Subject: libwww on VxWorks Subject: tests Subject: serial line problem Subject: Est backend & T2.0 Subject: heap management Subject: Exciting Embedded Real-Time programmer Opportunity ! Subject: Re: vxWorks memory protection Subject: writing an END on vxWorks 5.4 Subject: loadModule of objects in memory Subject: taskVarInfo (SimPC) Subject: Re: CORBA on VxWorks Subject: PHP on WindWeb/VxWorks? Subject: Re: Maximum Number of Open File Descriptors Subject: Debugger Problem Subject: Re: vxWorks memory protection Subject: A "Tracy Riddle" is spamming c.o.vxworks posters Subject: Re: tests Subject: help Subject: suggestions: good books on real time programming? Subject: Good reference books? Subject: Re: A "Tracy Riddle" is spamming c.o.vxworks posters Subject: Efficient select() usage. Subject: Re: Efficient select() usage. Subject: serial port output Subject: Re: writing an END on vxWorks 5.4 Subject: Re: CORBA on VxWorks Subject: Re: A "Tracy Riddle" is spamming c.o.vxworks posters Subject: fopen on VxSim Subject: VxSim communicate to WinNT Subject: Ethernet on PPC860 using scc1-4 Subject: Audio driver for UCB1200/UCB1300 Subject: Is there another x86 Boot method? Subject: Real-Time CORBA Trade Study made public Subject: Re: A "Tracy Riddle" is spamming c.o.vxworks posters Subject: Re: fopen on VxSim Subject: Re: DCHP only for IP address Subject: Re: DCHP only for IP address Subject: Re: Question about memory confilict Subject: Task Trace in ExcHook routine? Subject: SNTP v. 4 ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: cannot boot VxWorks on MCPN750 Date: Mon, 31 Jan 2000 15:11:52 +0100 From: "Achim Zimmer" Organization: Colt Telecom GmbH Message-ID: <874562$n9n$1@crusher.de.colt.net> References: <86shh2$54q3@nntp.cig.mot.com> <87289k$s3h$1@news1.xs4all.nl> I am not sure if this can help you, but we had a similar problem We had an ADS860 board from Motorola. The image from T2 dosn't work, from T1 it did. The bootloader from T2 works fine. The problem was a file called "sysalib.s" in our vxWorks configuration BSP for the ADS board. It contains the first entrypoint for vxWorks. When we have compiled the image an warning was shown in the last line like this "could not find _sysinit, set it to default 0x10000" After we add the file to the vxWorks build, it works fine. Check this and let us know, if it was the error !! Achim schrieb in im Newsbeitrag: 87289k$s3h$1@news1.xs4all.nl... > Tim Berger wrote: > : Hello, > > : Booting the MCPN750 board with the Tornado II boot loader fails after > : programming the flash memory; there's no response from the board after we > : reboot. However, the Tornado I boot loader starts up successfully. Then, > : when we try to load the VxWorks image, the board freezes after the "Starting > : at 0x100000" line. We've tried using version 1 and 2 VxWorks images with > : the same result. Here's a log of the serial port output just before the > : board freezes: > > Are you building both the T1 and the T2 bootrom yourself? If so, what are > the differences in the sources? Are there any differences in the memory > layout (you can check these with the xxx-obj- tools)? > I don't have any experience with this board, so these are just some > suggenstions. > > Groeten, > Johan > > -- > o o o o o o o . . . ___________________________________ > o _____ || Johan Borkhuis | > .][__n_n_|DD[ ====_____ | johan@borksoft.xs4all.nl | > >(________|__|_[_________]_|________________________________| > _/oo OOOOO oo` ooo ooo 'o!o!o o!o!o` > === VxWorks page: http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html === --------------------------- Newsgroups: comp.os.vxworks Subject: libwww on VxWorks Date: Mon, 31 Jan 2000 19:37:06 +0530 From: Yusuf Motiwala Organization: Lucent Technologies, Columbus, Ohio Message-ID: <3895970A.B9E260B8@lucent.com> Hi, Is there any port of libwww available for vxworks. Regards, Yusuf --------------------------- Newsgroups: comp.os.vxworks,comp.realtime Subject: tests Date: 31 Jan 2000 20:43:38 +0000 From: Prateek Sadhukhan Organization: Hughes Network Systems Message-ID: Hi, I have been working on [realtime systems] VxWorks recently and have done some coding of network interface drivers etc. I want to know if there is some online tests to test my knowledge of Vxworks [apart from lurking onto this newsgroup :-)]. Need not be programming contests but questions which are basic to vxworks orr realtime OS s in general. thanks and regards prateek --------------------------- Newsgroups: comp.os.vxworks Subject: serial line problem Date: Mon, 31 Jan 2000 16:27:37 +0100 From: Herve Mathieu Organization: INRIA =?iso-8859-1?Q?Rh=F4ne=2DAlpes?= Message-ID: <3895A9E9.BAB2F855@inrialpes.fr> I'm using MVME167 board with VxWorks 5.3. A sensor is connected to a rs232 line, without XON/XOFF (hard or soft), at 9600 bauds. The problem is that some characters are loose. The quantity of data is less than the input buffer size (about 800 chars). So I'm very surprise to loose characters, 68040 should be fast enough to get all characters! Has anybody such a problem, or something better, a solution ? Hervé Mathieu - -- /~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~\ Hervé MATHIEU - INRIA-Rhône-Alpes - Moyens robotiques 655 avenue de l'europe - ZIRST - 38330 Montbonnot Tel 04 76 61 52 84 - Fax 04 76 61 52 12 Tel +33 4 76 61 52 84 - Fax +33 4 76 61 52 12 Email: herve.mathieu@inrialpes.fr http://www.inrialpes.fr/iramr/people/hmathieu/ ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ --------------------------- Newsgroups: comp.os.vxworks Subject: Est backend & T2.0 Date: Mon, 31 Jan 2000 17:20:18 +0100 From: "PG" Organization: Ascom Switzerland, Bern, Switzerland Message-ID: <874co4$368$1@ascomax.hasler.ascom.ch> Hi all, I would like to know how to launch the estppc backend in Tornado 2. I work with the EST evaluation board for PPC8260. I've installed the BSP and the ppcbackend. I've done a "make ESTBackEnd.vxWorks", and I've downloaded the ESTBackEnd.ab & ESTBackEnd.bdx with Vision Click. In tornado2, I select the estppc backend (in target server) and launch the target Server.I obtain the following message Target Server cannot load the backend Target Server will exit Does somebody know the solution to launch correctly the EST target server in Tornado 2? Thanks in advance P.Grangier --------------------------- Newsgroups: comp.os.vxworks Subject: heap management Date: Mon, 31 Jan 2000 16:34:25 GMT From: dixon@crd.ge.com Organization: Deja.com - Before you buy. Message-ID: <874dih$ogg$1@nnrp1.deja.com> I am working with a colleague to resolve a performance problem that we believe to be caused by heap fragmentation. There have been several posts discussing this problem. We've tried to replace the standard heap management code with Doug Lea's implementation that we found at www.pso.com. There's a comment in this code that it is not compatible with vxworks partition management. We initially ignored this warning, and replaced the standard routines with Doug's. When the system boots we are getting errors that appear to be related to the comment. When I look at the replacement code, it really doesn't create partitions. The comment goes on to suggest renaming the replacement routines and using the new names in our code. This solution doesn't seem to be an option, because we don't have source for what we believe to be the biggest culprit. Is the solution as simple as really creating partitions? I've found one reference to a replacment memPartLib (www.seaweed.com), but it's not available for x86 at least according to their web page. Does anyone have any suggestions? TIA Standard disclaimers apply Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Exciting Embedded Real-Time programmer Opportunity ! Date: Mon, 31 Jan 2000 12:16:42 -0200 From: "Bandwidth" Message-ID: <874fs1$1ae$1@bob.news.rcn.net> We are looking for a Embedded Real-time programmer using C language for a Contract assignment. This is at one of the high growth Companies in Virginia. We thought this may be of interest to some of you. Here are a few more details Position Description: 6-10 years of software development experience in C in an embedded environment plus solid knowledge of ATM, IP, and Frame Relay networking and protocols. Experience with Motorola family such as MPC 860 processor is a plus. Candidates should also have a working knowledge of real time operating systems such as nucleus, pSOS or vxWorks It will be a pleasure to talk to you about this. Please send your resume and contact information urgently. Thank you, Yog Bandwidth, Inc. E-mail Address(es): bandwidthinc@erols.com Business Information: Company: Bandwidth, Inc. Phone: (301) 519-0240 Toll Free 800-954-6190 Fax: (301) 330-0205 --------------------------- Newsgroups: comp.os.vxworks Subject: Re: vxWorks memory protection Date: Mon, 31 Jan 2000 13:10:10 -0500 From: usirsiwal@nexabit.com Message-ID: <3895D001.BA62B37F@nexabit.com> References: <3891F5F8.176AEAB9@nexabit.com> <86tphl$aoo$1@nnrp1.deja.com> Thanx John! I am using motorolla 603. I disable mmu and cache before updating pte. That should make sure that TLBs are always in sync. Do we know if vxWorks ever writes to a task's stack space? If it does than that will explain the issue I am seeing. - -Umesh john_94501@my-deja.com wrote: > > In taskSwitchHook, I check if this is a safe task. If so, I write > > protect the old task stack space in the page table, > > and if unprotect the stack space for new task. But, sometimes I get > data > > access exception for resons unknown > > to me. > > Two things come to mind: > > 1) Have you taken care of invalidating any TLB entries that may exist? > You don't say which PPC you are using, but even the lowliest of 8xx > members have some TLB entries. If you are changing the access > permissions of pages you need to flush the TLB of any affected entries > to make sure that the new task doesn't see those instead of reading the > translation tables in memory again. > > 2) If you have your cache in copyback mode, you want to ensure that you > are flushing the lines containing the updated page tables back to memory > since the cache is probably bypassed when a TLB miss occurs. > > Hope that helps, > > John... > > Sent via Deja.com http://www.deja.com/ > Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: writing an END on vxWorks 5.4 Date: Mon, 31 Jan 2000 20:07:22 +0100 From: "Ludovic LEMOAN" Organization: UUNet Message-ID: <874mca$9e7$1@jaydee.iway.fr> I can't attach a second driver on ip stack because i can't set the IP Mask and the IP Address for the specific driver i'm writen. If you've an idea to explain the reason why, it will be really great!!! Sincerely, Ludovic LE MOAN --------------------------- Newsgroups: comp.os.vxworks Subject: loadModule of objects in memory Date: 31 Jan 2000 19:27:42 GMT From: toni@devsoft.devsoft.com (Anton Hartl) Organization: Customer of SpaceNet GmbH Message-ID: Hi, [PPC target with 5.3.1] I try to loadModule() an object that I already have in memory. The memDrv is used to accomplish this: a file "/mem/" is opened to obtain a file descriptor for the object which is passed to loadModule(). However the only result I get is: Incorrect ELF header size: 31771 load module failed I've already verified that I indeed get to the object through the "/mem/" file. See below for the code sample. Does someone have another idea on how to accomplish this? The ramdisk would be an option; but I'd really like to avoid that. - -Toni - -------------------------------------------------------------------------- #include #include #include memld() { char *p; int mfd; char mfile[128]; char databuf[2 * 1024]; memDrv(); memDevCreate("/mem/", 0, sysMemTop()); /* allocate buffer to hold a string */ p = (char *)malloc(2 * 1024); memset(p, 0, 2 * 1024); strcpy(p, "Hello World\n"); /* that is the corresponding memory file, read back the data */ sprintf(mfile, "/mem/%d", (int)p); mfd = open(mfile, O_RDONLY, 0644); read(mfd, databuf, sizeof(databuf)); /* this works */ printf("%12s\n", databuf); /* now the same for the object */ memset(p, 0, 2 * 1024); mfd = open("memld.o", 0xffff, 0644); read(mfd, p, 2 * 1024); /* p points to the memory area that hold the the object file; try loading this through the memory device */ sprintf(mfile, "/mem/%d", (int)p); mfd = open(mfile, O_RDONLY, 0644); if (loadModule(mfd, LOAD_GLOBAL_SYMBOLS) == 0) printf("load module failed\n"); } compile and run: -> ld < memld.o -> memld Hello World Incorrect ELF header size: 31771 load module failed value = 19 = 0x13 - -------------------------------------------------------------------------- - -- /dev Software Gmbh | Anton Hartl | Dumbo: You're an unusual cop. Gabelsbergerstr. 51 | Phone: +49 (0)89-28659843 | Mickey Mouse: You're an unusal 80333 Muenchen | Fax: +49 (0)89-28659845 | killer, too. Germany | Email: toni@devsoft.com | John Woo, The Killer --------------------------- Newsgroups: comp.os.vxworks Subject: taskVarInfo (SimPC) Date: Tue, 1 Feb 2000 00:19:13 +0100 From: "Mario Semo" Organization: Telenor magnet GmbH Message-ID: <87558r$ok7$1@newsmaster01.magnet.at> I want to create "Thread(Task)LocalStorage". this is a concept from NT/CE/OS2 world. the idea is simple: i create a TLS-Handle. then i can store in each thread with this Handle a void *, and i can query the void * of a given thread(task). e.g.: handle=TlsAlloc // once for all my threads. // each thread. TlsSet(handle,"data") ... and somewhere else ptr=TlsGet(handle,threadid) // returns the data stored in a thread to this handle. ok, my idea was to use TaskVars. but this does not work at all, or at least i do not understand this concept at all. i have a taskVarInfo sample. the following samples calls a functions which lists all taskVars. (none is defined. i never called taskVarAdd). the structure is: call function to list (1) call another function which just calls currentTaskId call function to list again (2) and the fist call says 0 taskVars (as expected) and the 2nd call says 1 taskVar. what is the reason? compile call: gen thr EMU (results in: ccsimpc -mpentium -U__WINNT -UWIN32 -U__WINNT__ -UWINNT -U__MINGW32__ -U_WIN 32 -U__WIN32 -U__WIN32__ -U__WIN32 -ansi -g -fno-builtin -fno-defer-pop -Wal l -DRW_MULTI_THREAD -D_REENTRANT -I. -IF:/Tornado/target/h -I. -DCPU=SIMNT - -c thr.cpp -o thr.o) and loading and running vxMain results in: - -> vxMain with id() currentThreadId=79150928 taskVarnAnz=0 with currentid() currentThreadId=79150928 taskVarnAnz=1 0 ... 0x467684 value = 0 = 0x0 - -- regards, Mario Semo. http://www.kirchnersoft.com begin 666 thr.cpp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rganization: @Home Network Message-ID: <38963E7B.5E8A68E7@home.com> References: <85u217$f63$1@nnrp1.deja.com> Check out TAO - The Ace Orb. It's at http://www.cs.wustl.edu/~schmidt/TAO.html Free, fastest ORB, ported to VxWorks. // Wally st1734 wrote: > > I'm looking for CORBA vendors on VXWORKS. > > Any pointers ? > > Sent via Deja.com http://www.deja.com/ > Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: PHP on WindWeb/VxWorks? Date: Tue, 01 Feb 2000 02:10:50 GMT From: robert_light@my-deja.com Organization: Deja.com - Before you buy. Message-ID: <875fb7$io9$1@nnrp1.deja.com> Has anyone ported PHP to work with WindWeb on VxWorks? Can anyone point me to WindWeb documentation that shows how cgi works with WindWeb? Please reply to light@buypages.com via email. Thanks - Bob Light Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Maximum Number of Open File Descriptors Date: Mon, 31 Jan 2000 20:31:31 -0600 From: Andrew Johnson Organization: APS, Argonne National Laboratory Message-ID: <38964583.996F5558@aps.anl.gov> References: <874u13$n833@overload.lbl.gov> "Blake, Don" wrote: > > What is the maximum number of open file descriptors in VxWorks? Is it > configurable? Yes, see config/all/configAll.h: #define NUM_FILES 50 /* max 50 files open simultaneously */ - - Andrew - -- Complexity comes for free, Simplicity you have to work for. --------------------------- Newsgroups: comp.os.vxworks Subject: Debugger Problem Date: Tue, 01 Feb 2000 03:46:26 GMT From: "Jeff Byrd" Organization: AT&T Worldnet Message-ID: I am working with Tornado 1.0.1 and have a problem with the debugger. When I place my source files in different directories, the debugger cannot seem to find them. When all the source files are in the same directory, no problem. Any ideas? Thanks --------------------------- Newsgroups: comp.os.vxworks Subject: Re: vxWorks memory protection Date: Tue, 01 Feb 2000 04:14:23 GMT From: john_94501@my-deja.com Organization: Deja.com - Before you buy. Message-ID: <875mis$nm4$1@nnrp1.deja.com> References: <3891F5F8.176AEAB9@nexabit.com> <86tphl$aoo$1@nnrp1.deja.com> <3895D001.BA62B37F@nexabit.com> In article <3895D001.BA62B37F@nexabit.com>, usirsiwal@nexabit.com wrote: > Thanx John! > I am using motorolla 603. > I disable mmu and cache before updating pte. That should make > sure that TLBs are always in sync. Disabling the MMU does not cause the TLB to be flushed - any pre-existing translations will still be there when you switch the MMU back on. You must explicitly flush it if you change PTEs which may be in there. The normal symptom of this is that you can still access memory that should be inaccessible though (since it will be the outgoing task's accesses that are in the TLB). It is quite possible that you have TLB entries still after your change, but I don't think that is your problem... > Do we know if vxWorks ever writes to a task's stack space? If it does than > that will explain the issue I am seeing. During a context switch the outgoing task's stack is used to run the scheduling code (including your hook function). You cannot make that task's stack inaccessible in a hook routine. To achieve what you want to achieve is actually quite difficult because of this; in fact I'm not sure it is possible without re-writing a lot of the rescheduling code! Sorry, John... Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: A "Tracy Riddle" is spamming c.o.vxworks posters Date: Tue, 01 Feb 2000 05:43:34 GMT From: Grant Taylor Organization: Picante Consulting Message-ID: Sender: gtaylor@habanero.picante.com She appears to be using addresses posted to this group in early December. I've gotten two identical mails from her thus far; they consist of the usual "I heard good things about you. I'm a headhunter with an exciting job in your specialty...". So steer clear of this person: > Tracy Riddle > TR&Associates > 8895 Towne Centre Drive > Suite 105, #250 > San Diego, Ca 92122 > 858-453-5020 P > 858-453-2925 F - -- Grant Taylor - gtaylor@picantecom - http://www.picante.com/~gtaylor/ Linux Printing HOWTO: http://www.picante.com/~gtaylor/pht/ --------------------------- Newsgroups: comp.os.vxworks,comp.realtime Subject: Re: tests Date: Tue, 01 Feb 2000 09:44:05 +0200 From: Leonid Rosenboim Organization: BitBand Technologies Ltd. http://www.bitband.com Message-ID: <38968EC5.E89355E4@bitband.com> References: Reply-To: leonid@bitband.com I think Wind River has some sort of tests for Authorized Consultants... never seen them but you can get more details in the "Partners" section of www.wrs.com. Prateek Sadhukhan wrote: > Hi, > > I have been working on [realtime systems] VxWorks recently and have > done some coding of network interface drivers etc. I want to know if > there is some online tests to test my knowledge of Vxworks [apart > from lurking onto this newsgroup :-)]. Need not be programming > contests but questions which are basic to vxworks orr realtime OS s in > general. > > thanks and regards > prateek --------------------------- Newsgroups: comp.os.vxworks Subject: help Date: Tue, 1 Feb 2000 16:18:15 +0800 From: "bob" Organization: Motorola CIG Message-ID: <8764hb$h041@nntp.cig.mot.com> Hi, all I find there are some micro define in the VxWorks about ehternet, such as, #define IFF_UP 0x1 /* interface is up */ #define IFF_BROADCAST 0x2 /* broadcast address valid */ ...., in the head file if.h. I remember I found them in the certain RFCxxxx. But I forgot which RFc it is in. Who can help me and remind me? thank you very much. --------------------------- Newsgroups: comp.os.vxworks Subject: suggestions: good books on real time programming? Date: Tue, 18 Jan 2000 19:27:28 -0800 From: "test" Organization: Ye 'Ol Disorganized NNTPCache groupie Message-ID: <948252560.860495@sj-nntpcache-3> Any suggestion on good reference type books to read up before attempting to dive into vxWorks app programming ? (other than the vxWorks manual) thanks waseem --------------------------- Newsgroups: comp.os.vxworks Subject: Good reference books? Date: Tue, 18 Jan 2000 19:33:07 -0800 From: "Waseem Siddiqi" Organization: Ye 'Ol Disorganized NNTPCache groupie Message-ID: <948252829.834506@sj-nntpcache-3> Any suggestions on good reference type books to get for embedded/vxWorks programming ? thanks waseem --------------------------- Newsgroups: comp.os.vxworks Subject: Re: A "Tracy Riddle" is spamming c.o.vxworks posters Date: Tue, 01 Feb 2000 10:29:02 -0500 From: Douglas Fraser Organization: Lucent Technologies Message-ID: <3896FBBE.1498BAA6@lucent.com> References: yup, me too... Doug Fraser PS: Grant, thanks for the HOWTO page. Unfortunately, Linux doesn't drive an HP IID in duplex mode very well. One of these days I need to look at the ghostscript driver in my spare time. (Did I say spare time? Ha!) --------------------------- Newsgroups: comp.os.vxworks Subject: Efficient select() usage. Date: Tue, 01 Feb 2000 21:44:07 +0530 From: Vijay Kumar Organization: MIEL Message-ID: <3897064F.228B3620@miel.mot.com> Hi, My requirement is as follows: - - I create sockets dynamically in my application on a need basis. - - I want to use select call to get indication on those sockets for read/write events. - - I want to use blocking select call. - - I should be able to update the fd's dynamically for the select call. - - I should be able to update the event for an fd dynamically. Has anybody tried this ? Please suggest me if U have a efficient solution for this problem. bye, +-------------------------------+ | VIJAY KUMAR | | | |Software Engineer | |MOTOROLA INDIA ELECTRONICS LTD.| |BANGALORE - 560042 | | | |Ph:91-80-5598615 extn. 3040 | |email: kvijay@miel.mot.com | +-------------------------------+ --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Efficient select() usage. Date: 1 Feb 2000 16:50:39 GMT From: Troy Cauble Organization: Bell Labs, Holmdel, NJ, USA Message-ID: <8772sv$m4r@nntpb.cb.lucent.com> References: <3897064F.228B3620@miel.mot.com> Vijay Kumar wrote: : Hi, : My requirement is as follows: : - I create sockets dynamically in my application on a need basis. : - I want to use select call to get indication on those sockets for : read/write events. : - I want to use blocking select call. : - I should be able to update the fd's dynamically for the select call. : - I should be able to update the event for an fd dynamically. : Has anybody tried this ? : Please suggest me if U have a efficient solution for this problem. A global fd_set, all accesses protected by a mutex semaphore. A routine to update the fd_set (minding the mutex). The task that blocks on select() will copy the fd_set (minding the mutex) to a local fd_set just before the select(). Use a pipe (as a member in the fd_set) to notify the task that the fd_set has changed. That is, have the update routine send a 1-byte message on the pipe to knock the task out of select(). - -troy --------------------------- Newsgroups: comp.os.vxworks Subject: serial port output Date: Tue, 01 Feb 2000 17:18:54 +0000 From: john Rose Organization: Cadence Design Systems, Inc. Message-ID: <3897157E.21F36862@cadence.com> Hi, again. I am adapting a board support package for and Arm Pid7t board. The board I am adapting it for is very similar, with only slightly different memory maps. The problem is that once I can start the bootrom I am getting nothing out of the serial port. At what stage of the bootup process do you normally get the usual blurb out of the serial port (press any key to abort bootup.... etc)? I know the serial port is working since I can add code to write to it and this is successful. I can follow the bootup process on a logic analyser and can watch while it copies memory from the remapped bootrom to the ram. The only problem is that I can't see far enough to see where the code crashes. The eventual crash is due to a memory abort. If I can establish a step that the program doesn't get to then this would be a great help in debugging this problem. Can anyone help? Thanks, John --------------------------- Newsgroups: comp.os.vxworks Subject: Re: writing an END on vxWorks 5.4 Date: Tue, 01 Feb 2000 17:50:44 +0000 From: Khalid Rafiq Organization: Spider Software Limited Message-ID: <38971CF4.588F@spider.com> References: <874mca$9e7$1@jaydee.iway.fr> Ludovic LEMOAN wrote: > > I can't attach a second driver on ip stack because i can't set the IP Mask > and the IP Address for the specific driver i'm writen. If you've an idea to > explain the reason why, it will be really great!!! > > Sincerely, > > Ludovic LE MOAN Make sure that your END has a memory pool allocated. - -- Khalid. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: CORBA on VxWorks Date: Tue, 01 Feb 2000 18:25:08 GMT From: bill.beckwith@realtime-corba.com (Bill Beckwith) Organization: Objective Interface Message-ID: <8ECD7B325beckwbnewscom@sigma.ois.com> References: <85u217$f63$1@nnrp1.deja.com> <38963E7B.5E8A68E7@home.com> Wallace Owen wrote in <38963E7B.5E8A68E7@home.com>: >Check out TAO - The Ace Orb. It's at >http://www.cs.wustl.edu/~schmidt/TAO.html > >Free, fastest ORB, ported to VxWorks. Free definitely. Fastest? Not by a long shot. - -- Bill --------------------------- Newsgroups: comp.os.vxworks Subject: Re: A "Tracy Riddle" is spamming c.o.vxworks posters Date: Tue, 01 Feb 2000 13:06:10 -0600 From: Andrew Johnson Organization: APS, Argonne National Laboratory Message-ID: <38972EA2.F53E4BAB@aps.anl.gov> References: Ahh, Me too, makes sense now. Does anyone know if WRS are hiring at the moment? It would be a logical approach to finding experienced developers, if rather underhand... - - Andrew - -- Complexity comes for free, Simplicity you have to work for. --------------------------- Newsgroups: comp.os.vxworks Subject: fopen on VxSim Date: Tue, 1 Feb 2000 21:35:35 +0100 From: "Zakelijk" Organization: UUNET-NL (http://www.nl.uu.net) Message-ID: <877g7s$ffl$1@porthos.nl.uu.net> We got two platforms for testing a real (PC and the full-simulator) both distributed with the Tornado VxWorks development environment. Since there are more people developing we want to use both platforms. The only "strange" thing we have now is the access of files. The problem is resident within the simulator. When we want to open a file using fopen("bla", "w") everything looks fine on the shell but when we look at the console it says something like: ioCntl unknown command 0x14 I've looked the code up in the include files and is says that it is some kind of "check whether console". The other modes for opening files also act funny on the simulator. Does anyone know this problem? please ALSO reply by mail....... --------------------------- Newsgroups: comp.os.vxworks Subject: VxSim communicate to WinNT Date: Tue, 1 Feb 2000 22:22:28 +0100 From: "Mario Semo" Organization: Telenor magnet GmbH Message-ID: <877m79$hjq$1@newsmaster01.magnet.at> Hello, what possibilities are there to communuicate from VxSim (as part of Tornado 2.0) to the Host OS WinNT 4.05. ? off course i can open a file write something into the file, close it and the windows counterpart does something similar. as i remember i read, that TCP/IP is not supported. can i use SHARED memory? what else? - -- regards, Mario Semo. http://www.kirchnersoft.com --------------------------- Newsgroups: comp.os.vxworks Subject: Ethernet on PPC860 using scc1-4 Date: Tue, 1 Feb 2000 17:45:09 -0800 From: "jeff patane" Message-ID: <38978c5c$0$216@nntp1.ba.best.com> I am adding support, in my bootrom, for ethernet devices using the PPC860 scc2-4. I currently have an ethernet i/f running on scc1. I've made the appropriate additions to endDevTbl for scc2-4. I've made the appropriate mods to sysCpmEnetEnable(), sysCpmEnetCommand(), sysCpmEnetIntEnable(), sysCpmEnetIntClear(), sysCpmEnetDisable() and sysCpmEnetIntDisable(). When I select a device other than cpm0 as my boot device, ftpXfer() hangs. However, if I move any of the other entries in endDevTbl to the first position, the download succeeds. This tells me that my configuration for the PPC860 is correct. Additionally, if I select a boot device other than the one occupying the first slot in endDevTbl, endFindByName() returns a valid pointer which indeed points to the selected device's endObj data structure; i.e., if I select cpm3 as the boot device, endFindByName returns a pointer to the third data structure created by muxDevLoad() during driver initialization, and the unit number in the struct is 3, and the next and prev Node pointers point to the correct endObj structs. If anyone knows of any issues, other than modifying endDevTbl and the sysCpmEnet... functions, to get an end device to work with a PPC860 on scc2-4, I would greatly appreciate any assistance. Thanks in advance. --------------------------- Newsgroups: comp.os.vxworks Subject: Audio driver for UCB1200/UCB1300 Date: Tue, 01 Feb 2000 20:54:06 -0700 From: Marco Castellon Organization: Home Message-ID: <3897AA5E.F5388668@cadvision.com> Hello, Has anybody worked on the development of an audio driver on a StrongARM based system using either the UCB1200 or the UCB1300 Advanced Codec IC from Philips. Any response would be greatly appreciated. Thank you. Marco Castellon Hardware Engineer AudeSi Technologies castellon@audesi.com --------------------------- Newsgroups: comp.os.vxworks Subject: Is there another x86 Boot method? Date: Wed, 02 Feb 2000 13:50:58 +0800 From: Zhang Zhiqun Organization: Guoguang Electronic Group Co. Message-ID: <3897C5C1.10197AC1@guoguang.com.cn> Hi, VxWorkers I am looking for another x86 Booting method? It's booting from EEPROM. The EEPROM is located in the motherboard BIOS socket. I flash the BOOTROM_RES into my 27C020 EEPROM and insert it to the BIOS socket. There is nothing happened. Is this booting method possible? - -- Zhang Zhiqun (Bula Cheung) bula@guoguang.com.cn Guoguang Electronic Group Co. http://www.guoguang.com.cn 93 Qingtan Road, Changzhou, Jiangsu, China, 213015 Tel: +86-519-6967935 ext 2372, 6971417 Fax: +86-519-6969104 --------------------------- Newsgroups: comp.os.vxworks Subject: Real-Time CORBA Trade Study made public Date: Wed, 02 Feb 2000 05:27:46 GMT From: bill.beckwith@realtime-corba.com (Bill Beckwith) Organization: Objective Interface Message-ID: <8ECDEB278beckwbnewscom@sigma.ois.com> The U.S. Government approved for public release their study of real-time ORBs. The documents are available on: http://www.ois.com/technical/rt_diicoe/release.html The summary reads: > 3.4.4.3 Summary of Basic IDL Performance Results > > A summary of results is simple: in general, ORBexpress > products outperformed alternatives in terms of both > raw speed and predictability in most cases tested. There was no test that ORBexpress was slower than TAO. Note that the cross-network benchmarks were mostly run on a 10 Mb network. This sometimes makes the ORBs look similar in speed for large message transfers. A useful measure is to subtract the socket times from the ORB times before comparing the ORBs. This yields a measure of the ORB efficiency that does not vary much with the speed of the network. Page 17 of Volume 4 has a related point: > 5. Scenario 4 on a Fast Network > > We ran Scenario 4 on a 100BASE-T Ethernet (100Mbits per second) to > further isolate the ORB overhead from the network overhead in the > measurements of Scenario 4. As shown in Figure 15, the absolute > differences and trends between operations in ORBexpress and TAO remain > about the same. With a network that is 10 times faster than the > 10BASE-T Ethernet used in other tests, however, the ORB overhead > begins to dominate the latencies. The lower overhead of ORBexpress > shows a much greater percentage advantage in speed over TAO than was > evident when the network transfer times dominated the measurement. - -- Bill --------------------------- Newsgroups: comp.os.vxworks Subject: Re: A "Tracy Riddle" is spamming c.o.vxworks posters Date: Wed, 02 Feb 2000 08:43:29 +0200 From: Leonid Rosenboim Organization: BitBand Technologies Ltd. http://www.bitband.com Message-ID: <3897D211.57E473C4@bitband.com> References: <38972EA2.F53E4BAB@aps.anl.gov> Reply-To: leonid@bitband.com Andrew Johnson wrote: > Ahh, > > Me too, makes sense now. Does anyone know if WRS are hiring at the > moment? It would be a logical approach to finding experienced developers, > if rather underhand... > Wind River has imposed a hiring freeze shortly after the WRS & ISI merger plans where announced, and probably will remain in effect until after all parts of the organizations have been merged. On the otehr hand, this is a time when many WRS and ISI employees will be looking elsewhere for employment, and that is a gold-mine opportunity for a head hunter too. And yes, I did get an email from Tracy too, but I don't mind. Lastly, one of the official reasons for WRS to buy RouterWare, ISI, Zinc is to hire expereinced developers, but I am not sure how serious they are about keeping them around. Leonid --------------------------- Newsgroups: comp.os.vxworks Subject: Re: fopen on VxSim Date: Wed, 2 Feb 2000 08:14:44 +0100 From: "Mario Semo" Organization: Telenor magnet GmbH Message-ID: <878lj2$fmq$1@newsmaster01.magnet.at> References: <877g7s$ffl$1@porthos.nl.uu.net> this is a bug in the simulator in NT. there is a hotfix available. SPR 27918 SPR# 27918 TITLE: ntPassFsLib: creat(), and open() with O_CREAT fail FIX: Not in released product yet. FIXED ON: Not Resolved yet Using Tornado 2 FCS on Windows NT, simpc target. The open flags values which VxWorks uses differ from those used by Windows NT, and ntPassFsLib does not convert between them. For instance, the O_CREAT flag has value 0x200 in VxWorks, but value 0x100 in Windows NT. This causes both open (...,O_CREAT | ..., ...) and creat() to fail on the host: ntPassFsLib device. WORKAROUND A patch for this SPR may be downloaded via the Web from the patches section on WindSurf, or by contacting WRS Customer Support. The patch title is, "27918 ntPassFsLib file creation patch". - -- regards, Mario Semo. http://www.kirchnersoft.com Zakelijk wrote in message <877g7s$ffl$1@porthos.nl.uu.net>... > >The problem is resident within the simulator. When we want to open a file >using fopen("bla", "w") >everything looks fine on the shell but when we look at the console it says >something like: >ioCntl unknown command 0x14 >I've looked the code up in the include files and is says that it is some >kind of "check whether console". > > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: DCHP only for IP address Date: Wed, 2 Feb 2000 01:23:19 -0500 From: VijayKumar P Organization: Ye 'Ol Disorganized NNTPCache groupie Message-ID: References: <867nhk$hkc$1@nnrp1.deja.com> <3888014E.984EBEE8@cisco.com> <86pof9$b1e$1@nnrp1.deja.com> 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_01BF6DB0.494F03D2 Content-Type: text/plain Hi, You need to set the flag in the default bootline to 0x40(Specifies automatic configuration using BOOTP or DHCP). You can do this also be stopping the boot and editing the flags option. This way as soon as the boot program starts, it checks the flag and if its 0x40, initiates... regds, Vijay gonflyn@my-deja.com wrote: > Hi Vijay, > > Thanks for the reply. I am including the run time dhcp client > in the vxworks image as you outlined in choice 1. However, I > do not see where the dhcp client is ever initiating the discover > processing. Is there anything else that I have to do in order > to get "everything" taken care of by vxworks? (ie call a library > routine to tell the dhcp client to go out and get its address?) > Otherwise, it cannot attach the lan device due to not having a > valid ip address to configure. > > Thanks, > > -Randy > In article <3888014E.984EBEE8@cisco.com>, > VijayKumar P wrote: > > This is a multi-part message in MIME format. > > --------------A2A5152BB744C33568DA6612 > > Content-Type: text/plain; charset=us-ascii > > Content-Transfer-Encoding: 7bit > > > > Hi, > > > > You can do it in two ways.. > > > > 1. Include the DHCP runtime client option in your image. > > Tornado>Workspace>vxWorks>network components> > > network applications>DHCPv4 runtime client > > Here when you boot from flashdisk, the DHCP component > > will be included and 'everything' will be taken care of by vxWorks. > > > > 2.If you really want to start DHCP at a later stage, say some > > time in your application, then just use the DHCP calls provided > > by vxWorks. The complete procedure has been explained > > in detail ( with an example!) in the Network Programmer's Guide, > > Chap 5. > > > > Good luck, > > Vijay > > > > gonflyn@my-deja.com wrote: > > > > > Hi, > > > > > > I am trying to get a PC based target to boot > > > from its flash disk (both the bootrom.sys and > > > vxworks image are in flash) but I want to have > > > the target IP address assigned from a DHCP server. > > > If I configure the boot line to boot from the > > > network device, the target fails to load because > > > it didn't receive a valid boot host and file > > > name. If I configure the boot line to boot from > > > the flash disk, it never sends the DHCP request > > > to discover its ip address and vxworks fails to > > > attach the network device due to no ip address. > > > > > > How can I get it to boot from flash and only > > > request an IP address from the DHCP server? Am I > > > going to have to write some custom code for this? > > > > > > Thanks for any help. > > > > > > -Randy Smith > > > > > > Sent via Deja.com http://www.deja.com/ > > > Before you buy. > > > > --------------A2A5152BB744C33568DA6612 > > Content-Type: text/x-vcard; charset=us-ascii; > > name="vpeshkar.vcf" > > Content-Transfer-Encoding: 7bit > > Content-Description: Card for VijayKumar P > > Content-Disposition: attachment; > > filename="vpeshkar.vcf" > > > > begin:vcard > > n:Peshkar;Vijay > > tel;fax:+91-80-5537663 > > tel;work:+91-80-5530035,34,36 Extn: 3451 > > x-mozilla-html:FALSE > > org:WIPRO Technologies;Global R&D Solutions > > version:2.1 > > email;internet:vijay.peshkar@wipro.com > > title:Senior Engineer - Software > > adr;quoted-printable:;;Vijay Kumar .P, WIPRO Technologies, #8, 7th > Main, 1st Block, Koramangala=0D=0A;Bangalore;Karnataka;560 034;India > > note;quoted-printable:Seldom does an individual exceed his own > expectations=0D=0A=0D=0AWishing you a most beautiful day wherever you > may be! > > fn:Vijay Kumar.P > > end:vcard > > > > --------------A2A5152BB744C33568DA6612-- > > > > > > Sent via Deja.com http://www.deja.com/ > Before you buy. - ------ =_NextPart_000_01BF6DB0.494F03D2 Content-Type: text/x-vcard; name="vpeshkar.vcf" Content-Disposition: attachment; filename="vpeshkar.vcf" Content-Description: Card for VijayKumar P begin:vcard n:Peshkar;Vijay tel;fax:+91-80-5537663 tel;work:+91-80-5530035,34,36 Extn: 3451 x-mozilla-html:FALSE org:WIPRO Technologies;Global R&D Solutions version:2.1 email;internet:vijay.peshkar@wipro.com title:Senior Engineer - Software note:Wishing you a most beautiful day wherever you may be! adr;quoted-printable:;;Vijay Kumar .P, WIPRO Technologies, #8, 7th Main, 1st Block, Koramangala=0D=0A;Bangalore;Karnataka;560 034;India fn:Vijay Kumar.P end:vcard - ------ =_NextPart_000_01BF6DB0.494F03D2-- --------------------------- Newsgroups: comp.os.vxworks Subject: Re: DCHP only for IP address Date: Wed, 02 Feb 2000 11:53:19 +0530 From: VijayKumar P Organization: Ye 'Ol Disorganized NNTPCache groupie Message-ID: <3897CD57.38C3EC20@cisco.com> References: <867nhk$hkc$1@nnrp1.deja.com> <3888014E.984EBEE8@cisco.com> <86pof9$b1e$1@nnrp1.deja.com> This is a multi-part message in MIME format. - --------------B8DC9DE32A8783DAA92D6C75 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Hi, You need to set the flag in the default bootline to 0x40(Specifies automatic configuration using BOOTP or DHCP). You can do this also be stopping the boot and editing the flags option. This way as soon as the boot program starts, it checks the flag and if its 0x40, initiates... regds, Vijay gonflyn@my-deja.com wrote: > Hi Vijay, > > Thanks for the reply. I am including the run time dhcp client > in the vxworks image as you outlined in choice 1. However, I > do not see where the dhcp client is ever initiating the discover > processing. Is there anything else that I have to do in order > to get "everything" taken care of by vxworks? (ie call a library > routine to tell the dhcp client to go out and get its address?) > Otherwise, it cannot attach the lan device due to not having a > valid ip address to configure. > > Thanks, > > -Randy > In article <3888014E.984EBEE8@cisco.com>, > VijayKumar P wrote: > > This is a multi-part message in MIME format. > > --------------A2A5152BB744C33568DA6612 > > Content-Type: text/plain; charset=us-ascii > > Content-Transfer-Encoding: 7bit > > > > Hi, > > > > You can do it in two ways.. > > > > 1. Include the DHCP runtime client option in your image. > > Tornado>Workspace>vxWorks>network components> > > network applications>DHCPv4 runtime client > > Here when you boot from flashdisk, the DHCP component > > will be included and 'everything' will be taken care of by vxWorks. > > > > 2.If you really want to start DHCP at a later stage, say some > > time in your application, then just use the DHCP calls provided > > by vxWorks. The complete procedure has been explained > > in detail ( with an example!) in the Network Programmer's Guide, > > Chap 5. > > > > Good luck, > > Vijay > > > > gonflyn@my-deja.com wrote: > > > > > Hi, > > > > > > I am trying to get a PC based target to boot > > > from its flash disk (both the bootrom.sys and > > > vxworks image are in flash) but I want to have > > > the target IP address assigned from a DHCP server. > > > If I configure the boot line to boot from the > > > network device, the target fails to load because > > > it didn't receive a valid boot host and file > > > name. If I configure the boot line to boot from > > > the flash disk, it never sends the DHCP request > > > to discover its ip address and vxworks fails to > > > attach the network device due to no ip address. > > > > > > How can I get it to boot from flash and only > > > request an IP address from the DHCP server? Am I > > > going to have to write some custom code for this? > > > > > > Thanks for any help. > > > > > > -Randy Smith > > > > > > Sent via Deja.com http://www.deja.com/ > > > Before you buy. > > > > --------------A2A5152BB744C33568DA6612 > > Content-Type: text/x-vcard; charset=us-ascii; > > name="vpeshkar.vcf" > > Content-Transfer-Encoding: 7bit > > Content-Description: Card for VijayKumar P > > Content-Disposition: attachment; > > filename="vpeshkar.vcf" > > > > begin:vcard > > n:Peshkar;Vijay > > tel;fax:+91-80-5537663 > > tel;work:+91-80-5530035,34,36 Extn: 3451 > > x-mozilla-html:FALSE > > org:WIPRO Technologies;Global R&D Solutions > > version:2.1 > > email;internet:vijay.peshkar@wipro.com > > title:Senior Engineer - Software > > adr;quoted-printable:;;Vijay Kumar .P, WIPRO Technologies, #8, 7th > Main, 1st Block, Koramangala=0D=0A;Bangalore;Karnataka;560 034;India > > note;quoted-printable:Seldom does an individual exceed his own > expectations=0D=0A=0D=0AWishing you a most beautiful day wherever you > may be! > > fn:Vijay Kumar.P > > end:vcard > > > > --------------A2A5152BB744C33568DA6612-- > > > > > > Sent via Deja.com http://www.deja.com/ > Before you buy. - --------------B8DC9DE32A8783DAA92D6C75 Content-Type: text/x-vcard; charset=us-ascii; name="vpeshkar.vcf" Content-Transfer-Encoding: 7bit Content-Description: Card for VijayKumar P Content-Disposition: attachment; filename="vpeshkar.vcf" begin:vcard n:Peshkar;Vijay tel;fax:+91-80-5537663 tel;work:+91-80-5530035,34,36 Extn: 3451 x-mozilla-html:FALSE org:WIPRO Technologies;Global R&D Solutions version:2.1 email;internet:vijay.peshkar@wipro.com title:Senior Engineer - Software note:Wishing you a most beautiful day wherever you may be! adr;quoted-printable:;;Vijay Kumar .P, WIPRO Technologies, #8, 7th Main, 1st Block, Koramangala=0D=0A;Bangalore;Karnataka;560 034;India fn:Vijay Kumar.P end:vcard - --------------B8DC9DE32A8783DAA92D6C75-- --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Question about memory confilict Date: Wed, 19 Jan 2000 16:17:08 +0100 From: "Yves Nortier" Organization: [Posted via] Easynet France Message-ID: <864ki5$ut4$1@buggy.easynet.fr> References: <85eccb$cdh$1@nnrp1.deja.com> Personnaly, I allocate my buffers in the user reserved memory (between sysMemTop and SysPhysMemTop). In order to do that, define USER_RESERVED_MEM in config.h. Yves NORTIER zhouxiaochen@my-deja.com a écrit dans le message <85eccb$cdh$1@nnrp1.deja.com>... >Hi everybody, > >I am writing a program of AAL5 SAR driver for PPC860. My eviroment is >Tornado 1.01.(Windows NT 4.0) > >At first I should allocate some memory for buffer pool. So I can >allocate buffer to BD Link from buffer pool and release buffer of BD >Link to buffer pool. > >But when Tornado download object modules to target and the modules is >relocatable in memory and I can't control the address of bss, data and >the text segment in memory. This will induce memory conflict between >buffer pool and object files in memory. > >How can I do? > >Thanks in advance, > >Zhou Xiaochen > > >Sent via Deja.com http://www.deja.com/ >Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Task Trace in ExcHook routine? Date: Wed, 02 Feb 2000 10:40:54 +0100 From: Ton Janssen Organization: Oce Technologies B.V. Message-ID: <3897FBA6.2F55BE35@oce.nl> Hi All, We want to display a stack trace of the task which generates a Page Fault automatically, so we use the excHookAdd() routine to install this handler and then want to call something like tt( taskId) to display the stack. However, tt() cannot be called as function obviously, only from the VxWorks shell. 'INCLUDE_DEBUG' has been set, so tt interactively works, but the system reboots as a result of the Page Fault, so we want the trace immediately when the Page Fault occurs. Does any body know a way to do this? Any help is appreciated, - -- ############################################################ # Oce Technologies B.V. name: Ton Janssen (3N-38) # # P.O. Box 101 email: djan@oce.nl # # 5900 MA Venlo Private: dpc.janssen@hccnet.nl # # Tel: +31 (0)77-359 40 97 # # The Netherlands Fax: +31 (0)77-359 54 50 # ############################################################ This note does not necessarily represent the position of Oce Technoligies B.V. Therefore no liability or responsibility for whatever will be accepted. --------------------------- Newsgroups: comp.os.vxworks Subject: SNTP v. 4 Date: Wed, 19 Jan 2000 10:04:41 -0600 From: Daniel Stewart Organization: Motorola Message-ID: <3885E099.5A12EF48@email.mot.com> Does anyone know if VxWorks supports version 4 of SNTP? I am getting a VERSION_UNSUPPORTED error during the call to sntpcTimeGet(). Thanks. Daniel --------------------------- End of New-News digest ********************** From vxwexplo-errs@csg.lbl.gov Wed Feb 2 08:59:18 2000 From: "Bruce A. Wilson" Date: Wed Feb 2 08:59:21 PST 2000 Subject: Force cpci3603 using ethernet Has anybody in the VxWorks group succeeded in using a CompactPCI ethernet board along with the Force cpci3603 (3U) processor board? By default, the BSP for the 3603 only comes set up for ppp or slip, since there's no onboard ethernet. We rebuilt the bootrom.hex after setting #define INCLUDE_FRC_DC in config.h, and burned a new bootrom, but still the only boot devices it allowed were ppp and sl. How do we get it to allow a "dc" device, like for the Gespac PCILAN board? Bruce A. Wilson U.C. Lawrence Livermore National Lab P.O. Box 808 (L-491) Livermore, CA 94550 voice: (925) 422-4145 fax: (925) 423-9889 From vxwexplo-errs@csg.lbl.gov Wed Feb 2 09:17:24 2000 From: Tom Antles Date: Wed Feb 2 09:17:27 PST 2000 Subject: WDB/host memory pool vs heap usage oddities We have just noticed some very strange and unexpected behavior regarding when the WDB/host memory pool is used versus the when the system memory pool (heap) is used. I am hoping someone can shed some light on this!!! NOTE: on most processors, the term used in Windriver manuals is WDB memory pool. However, for PPC and MIPS processors, they use the term host memory pool even though they still use the WDB_POOL_SIZE macro in configAll.h to define its size. Refer to appendices a-f in vxWorks programmers guide for a memory map of the system and WDB/host memory pools for various processors. Here is the weird behavior we saw: What we expected: - anytime you dynamically load a module (*.o file), memory should be allocated from the heap (system memory pool) to contain the module. What we saw: - when we examined the location of the module in memory (using moduleShow), the module had been loaded into the WDB/host memory pool instead of the heap. - In fact, as we kept loading more and more *.o files, they all went into the WDB/host memory pool until there was no more room left, at which point they began being allocated off the heap (system memory pool) Questions: - Why would windriver use the WDB/host memory pool over the heap for loading relocatable modules? - Isn't the WDB/host memory pool explicitly for tool usage such as windview and others? What we haven't tried yet: - Will the host tools still run properly after all of the WDB/host memory pool has been allocated to code modules? Windriver, vxworks Tom Antles ---------------------- tom_antles@agilent.com Phone: (509) 921-3590 Fax: (509) 921-3991 From vxwexplo-errs@csg.lbl.gov Wed Feb 2 09:29:56 2000 From: "Darren Carlson" Date: Wed Feb 2 09:29:58 PST 2000 Subject: DOSFS2 Has VxWorks changed the rename function? It works fine when I compile it on Tornado 1, but it doesn't do anything on T2. Yes, rename is platform dependant and I'm running it on DSOFS. Has anyone else encountered this? Thank you From vxwexplo-errs@csg.lbl.gov Wed Feb 2 10:15:28 2000 From: David Laight Date: Wed Feb 2 10:15:30 PST 2000 Subject: RtX-Windows gamma correction tables Anyone any idea how to stop RtX-Windows spending 30 seconds creating gamma lookup tables during its startup? These tables involve calculating x**y (pow(x,y)) some 37000 times - a little slow with only software fp routines on my ARM system. (and they are non-integral powers, and not 1/2 either). I recall there being a configuration option (vxworks, rtx-windows or maybe the jvm) to used a fixed colour table to speed things up. I'm sure I enabled this once - and don't remember disabling it - but can't offhand think how to check. Any ideas? David ---------------------------------------------------------------- David Laight email: dsl@tadpole.co.uk Tadpole Technology plc phone: +44 1223 428 232 Cambridge, UK fax: +44 1223 428 201 From vxwexplo-errs@csg.lbl.gov Wed Feb 2 11:54:22 2000 From: brian.r.walker@exgate.tek.com Date: Wed Feb 2 11:54:25 PST 2000 Subject: RE: WDB/host memory pool vs heap usage oddities VxWorks, Tornado > -----Original Message----- > Sent: Wednesday, February 02, 2000 9:32 AM > Submitted-by vxwexplo-errs@csg.lbl.gov Wed Feb 2 09:17:24 2000 > Submitted-by: Tom Antles > > > We have just noticed some very strange and unexpected > behavior regarding when the WDB/host memory pool is used > versus the when the system memory pool (heap) is used. > I am hoping someone can shed some light on this!!! > > > Here is the weird behavior we saw: This is not what I expect: > What we expected: > > - anytime you dynamically load a module (*.o file), memory should be > allocated from the heap (system memory pool) to contain the module. This is what I expected: > What we saw: > > - when we examined the location of the module in memory > (using moduleShow), > the module had been loaded into the WDB/host memory pool > instead of the > heap. > > - In fact, as we kept loading more and more *.o files, they > all went into > the WDB/host memory pool until there was no more room > left, at which > point they began being allocated off the heap (system memory pool) > > > Questions: > > - Why would windriver use the WDB/host memory pool over the heap for > loading relocatable modules? It cuts down on traffic between the target and host. In order to allocate memory from the heap, the target server must call memPartAlloc() on the target. The target server manages memory in the WDB memory pool so it can allocate the memory on the host and then write the memory directly on the target. > - Isn't the WDB/host memory pool explicitly for tool usage > such as windview and others? The memory is managed by the target server. Some of the tools load object modules. > What we haven't tried yet: > > - Will the host tools still run properly after all of the > WDB/host memory pool has been allocated to code modules? You have already seen the results. When the WDB pool runs out of memory the WDB task adds memory to the pool by allocating more from the system memory pool ("the Heap"). Coincidentally, using the target server to load object modules is one way to thwart the PPC 26 bit addressing problem during development. Since VxWorks initially reserves the WDB memory pool immediately after the kernel, relocatable code that is loaded in the pool will link with the kernel. However, if you exceed the size of the pool, the target allocates memory from the end of first available memory block which is usually at the end of memory. -- Brian Walker Software Design Services Brian.R.Walker@tek.com Central Engineering, Tektronix, Inc. From vxwexplo-errs@csg.lbl.gov Wed Feb 2 12:43:13 2000 From: Tom Antles Date: Wed Feb 2 12:43:21 PST 2000 Subject: RE[2]: WDB/host memory pool vs heap usage oddities Brian, This makes perfect sense! Thanks for the reply. Interestingly enough, the way we stumbled accross this was exaclty as you described in the last paragraph, as we were working to thwart the +-32Meg issue during development. We saw exactly what you described. Small modules would successfully ld and link dynamically while large modules would NOT! Which is how we observed this behavior. This then leads us to the 2nd part of our problem. This solution works great for development, but it looks like it will NOT work if the target will be loading and dynamically linking its own modules during run time. (not that we do this today, but have discussed this as a future feature) At this point I am not sure what happens to the WDB/host memory pool when you are not using target server tools (i.e. product shipped to customer). Does this memory get added into the heap? Or is it wasted? Tom If... we decide we also want to do dynamic linking at run time other than development, brian.r.walker@exgate.tek.com wrote > > > > - Why would windriver use the WDB/host memory pool over the heap for > > loading relocatable modules? > > It cuts down on traffic between the target and host. In order to allocate > memory from the heap, the target server must call memPartAlloc() on the > target. The target server manages memory in the WDB memory pool so it can > allocate the memory on the host and then write the memory directly on the > target. > > > - Isn't the WDB/host memory pool explicitly for tool usage > > such as windview and others? > > The memory is managed by the target server. Some of the tools load object > modules. > > > What we haven't tried yet: > > > > - Will the host tools still run properly after all of the > > WDB/host memory pool has been allocated to code modules? > > You have already seen the results. When the WDB pool runs out of memory the > WDB task adds memory to the pool by allocating more from the system memory > pool ("the Heap"). > > Coincidentally, using the target server to load object modules is one way to > thwart the PPC 26 bit addressing problem during development. Since VxWorks > initially reserves the WDB memory pool immediately after the kernel, > relocatable code that is loaded in the pool will link with the kernel. > However, if you exceed the size of the pool, the target allocates memory > from the end of first available memory block which is usually at the end of > memory. > -- > Brian Walker Software Design Services > Brian.R.Walker@tek.com Central Engineering, Tektronix, Inc. Tom Antles ---------------------- tom_antles@agilent.com Phone: (509) 921-3590 Fax: (509) 921-3991 From vxwexplo-errs@csg.lbl.gov Wed Feb 2 15:13:34 2000 From: "Kevin Dodd [Healdsburg]" Date: Wed Feb 2 15:13:36 PST 2000 Subject: vxworks and LongLong Int I am trying to get a longlong math pack going, and I keep running into difficulties. Does anyone have surce code for a working one whtat is all in C. I have a i960 and would prefer to have it naturally, but it is not there. Thanks Kevin E. Dodd SW Engineer Wescam-Healdsburg Fax: 707-433-7110 Email: kdodd@versatron.wescam.com Direct Phone:707-473-1114 Those who say a thing cannot be done..... Should not hamper those who are doing it From vxwexplo-errs@csg.lbl.gov Wed Feb 2 18:15:14 2000 From: brian.r.walker@exgate.tek.com Date: Wed Feb 2 18:15:17 PST 2000 Subject: RE: WDB/host memory pool vs heap usage oddities VxWorks, Tornado > -----Original Message----- > From: Tom Antles [mailto:antles@spk.hp.com] > Sent: Wednesday, February 02, 2000 12:41 PM > > > Brian, > > This makes perfect sense! Thanks for the reply. Interestingly > enough, the way we stumbled accross this was exaclty as you > described in the last paragraph, as we were working to thwart > the +-32Meg issue during development. > > We saw exactly what you described. Small modules would > successfully ld and link dynamically while large modules would NOT! > Which is how we observed this behavior. > > This then leads us to the 2nd part of our problem. This > solution works great for development, but it looks like it will NOT > work if the target will be loading and dynamically linking its own > modules during run time. (not that we do this today, but have > discussed this as a future feature) > > At this point I am not sure what happens to the WDB/host memory pool > when you are not using target server tools (i.e. product shipped to > customer). Does this memory get added into the heap? Or is > it wasted? The WDB memory pool is separate from the system memory pool so it will not be used in a standalone product. The function usrInit() determines the allocation of memory between the WDB pool and the system memory pool. For the 26-bit addressing problem, one solution is use only the first 32MB for the system memory pool. You can add rest of the memory to the system pool using memPartAddToPool() after you have loaded all of your object modules. Or, you can create a second memory pool and use the rest of memory for data. For extra fun, you can do the opposite and reserve a section of low memory, do your own memory allocation and use loadModuleAt() instead of ld(). Alternatively, you can ask Wind River for their compiler patch for 32 bit branches. The down side is that it converts single branch instructions into 3 or 4 instructions. Wind River could also implement a few tricks in their loader. First, they could provide a way to override the memory partition that loadLib uses. That would make it easier to create a memory partition close to the kernel image for object modules. Second, they could fix elfLoadLib so that if it would try to turn relative branches into absolute branches. That would only be a 98% cure but it would allow object files to link to the kernel code which is usually located in the first 32MB of memory. -- Brian Walker Software Design Services Brian.R.Walker@tek.com Central Engineering, Tektronix, Inc. From vxwexplo-errs@csg.lbl.gov Wed Feb 2 19:46:57 2000 From: Fred Roeber Date: Wed Feb 2 19:47:00 PST 2000 Subject: Re: serial line problem On Wed, 2 Feb 2000, Herve Mathieu wrote: > I'm using MVME167 board with VxWorks 5.3. > A sensor is connected to a rs232 line, without XON/XOFF (hard or soft), at > 9600 bauds. > > The problem is that some characters are loose. > The quantity of data is less than the input buffer size (about 800 chars). VxWorks uses a ring buffer to store the characters that are received prior to their being read. The size of the ring buffer defaults to 512 bytes. If you don't read the characters and the ring buffer fills up, data will be lost. If you think this is your problem (due to losing a number of sequential characters) then you should just boost the size of the ring buffer the serial driver uses. Based on which version of VxWorks you are using, there are different ways to change the size. Since you are using a pre "Tornado 2 project tool" version I suggest just going into the config/all/usrConfig.c file and changing the arguments to ttyDevCreate to boost the size of the read buffer (third parameter) from 512 to 1024. You can also use the FIORBUFSET ioctl (see tyLib man page). If you are only losing characters once in a while rather than in blocks, you could be having "noisy" line problems. Most of the serial drivers are coded to just discard characters that are received with things like framing and parity errors. If you suspect this is your problem, you could try putting some debugging code in the serial driver read interrupt routine to bump a global counter when it discards a character due to an error. VxWorks comes with source code for most serial drivers. The one you want is target/src/drv/sio/cd2400Sio.c. Hope this helps. Fred | Fred J Roeber, BBN Systems & Technologies | | 4 John Clarke Road Middletown, RI 02842-5202 | | froeber@bbn.com 401-848-3548 | | TraceMaker product manager | | http://www.tracemaker.bbn.com | From vxwexplo-errs@csg.lbl.gov Thu Feb 3 04:03:09 2000 From: Vxworks Exploder Date: Thu Feb 3 04:03:12 PST 2000 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Thu Feb 3 04:03:06 PST 2000 Subject: [Q] How should I pronounce 'vxWorks'? Subject: Re: Problems with Booting from Network Subject: Re: serial line problem Subject: DBG problem Subject: Building vxworks_rom in T2 Subject: auto configuration on boot from ROM? Subject: Re: DBG problem Subject: dosFs2.0 Code to mount unknown number of partitions? Subject: Re: heap management Subject: Re: Debugger Problem Subject: Re: [Q] How should I pronounce 'vxWorks'? Subject: Re: reply to relocation overflow Subject: Re: dosFs2.0 Code to mount unknown number of partitions? ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: [Q] How should I pronounce 'vxWorks'? Date: Tue, 1 Feb 2000 20:17:17 +0900 From: "Kim, Yong-Woon" Organization: Electronics and Telecommunications Research Institute (ETRI) Message-ID: <876fa6$q5n$1@news.etri.re.kr> Hi! Today is the first day that I just entered into the 'vxWorks' world. I've navigated a lot of pages related to it, including the old FAQ and an unofficial VxWorks & Tornado FAQ'. No pages have been given information about pronunciation of 'vxWorks'. Can I pronounce it, just 'v', 'x', and 'works'? As you know, the Netscape co. wanted to call 'Netscape' 'Mozilla'. But it failed at least in our country, Republic of Korea. Thanks in advance. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Problems with Booting from Network Date: Wed, 02 Feb 2000 10:34:01 -0600 From: Bob Feng Message-ID: <38985C79.5C9992E6@csun.edu> References: <85vh0h$g17$1@nnrp1.deja.com> <389017B3.84249549@email.sps.mot.com> The problem was solved by discovering the SMC Ultra card's jumper cannot be set to "soft configuration". Moreover it has to be in Jumper3 position particularly in my target computer. I changed IRQ/IO setting correspondingly in "config.h" and then everything seems work fine. Sincerely thank those of you who generously offered help. - ---bob Wilson Lo wrote: > Hi, > > The error code "0xd0003" means that the board can't from the image file. > U may check if the path of FTP is correct. I use the "Mircosoft Peer web > Service" > to act as a FTP server. It is Ok. and I think the problem is not due to the "/". > > One thing to remain is that U need to setup the path name of directory in the > Web Service. > > regards > Wilson > > bfeng@noran.com wrote: > > > I am trying to boot a Pentium PC via a network using a default factory > > Pentium BSP boot disk. I managed to change the booting parameters after > > "[boot rom]" prompt showed up. A ftp server "wftpd.exe" from > > "Tornado2.0" CD is always launched on my host NT computer. The network > > connection was tested OK while booting the Pentium PC into Win95. The > > FTP login user name and password were also tested OK in Win95. However > > everytime I try to boot it with Vxworks, I have the following error > > messages. I appreciate anyone's hint. What does error code "0xd0003" > > actually mean? > > > > boot device : ultra > > unit number : 0 > > processor number : 0 > > host name : bfeng2.YP.solsys > > file nmae : /tornado/target/config/bobPentium/vxworks.st > > inet on ethernet (e) : x.x.x.x:ffffff00 > > host inet (h) : z.z.z.z > > user(u) : bfeng > > ftp password(pw) : xxxxx > > flags (f) : 0x0 > > > > Attaching TCP/IP interface to ultra0. > > Attaching network interface lo0... done. > > Loading... > > Error loading file: errno = 0xd0003. > > dosFsDevInit failed. > > userFdConfig failed. > > > > Sent via Deja.com http://www.deja.com/ > > Before you buy. > > -- > ======================================== > Wilson Lo > Software Application Engineer > Asia Pacific Operation > Networking and Communications Systems Division > Motorola Semiconductor Products Sector > ======================================== --------------------------- Newsgroups: comp.os.vxworks Subject: Re: serial line problem Date: Wed, 02 Feb 2000 18:29:21 +0100 From: Leif Thuresson Organization: TSL Message-ID: <38986971.9579401@tsl.uu.se> References: <3895A9E9.BAB2F855@inrialpes.fr> Reply-To: leif.thuresson@tsl.uu.se Herve Mathieu wrote: > I'm using MVME167 board with VxWorks 5.3. > A sensor is connected to a rs232 line, without XON/XOFF (hard or soft), at > 9600 bauds. > > The problem is that some characters are loose. > The quantity of data is less than the input buffer size (about 800 chars). > > So I'm very surprise to loose characters, 68040 should be fast enough > to get all characters! > > Has anybody such a problem, or something better, a solution ? > Hervé Mathieu > A 68040 should not have any problem reading characters at 9600 baud. Are you sure the input buffer is big enough ? On all my 68xxx BSP the default size is 512 byte (you can change this in config/all/usrConfig.c, the call to ttyDevCreate() ) If the buffer is smaller then the receive message size your application must be reading characters from the buffer while your sensor is writing them otherwise the buffer will of course overflow. If this isn't the problem the only other thing I can think of is that your application is disabling interrupts for long periods, or some other hardware device is generating higher priority interrupts (higher than the serial port) at very high rate. regards, - --leif --------------------------- Newsgroups: comp.os.vxworks Subject: DBG problem Date: Wed, 02 Feb 2000 18:06:12 GMT From: "Jeff Byrd" Organization: AT&T Worldnet Message-ID: Has anyone seen a problem with the debugger not being able to find a function in the c source if the function name begins with a capital L? Any other character (upper or lower case) works just fine. Any clues? Jeff --------------------------- Newsgroups: comp.os.vxworks Subject: Building vxworks_rom in T2 Date: Wed, 02 Feb 2000 10:31:25 -0500 From: Paul Livingston Message-ID: <38984DCD.776EAE9B@triton.npt.nuwc.navy.mil> I am attempting a very simple test using a Moto 2400 BSP and T2. I am trying to build a custom vxworks romable image in T2. I have made changes to the image and am able download and execute it on the target, no problems. Now I want to flash it on the target. I have changed the rule from vxworks to vxworks_rom in the build tab. When I build the image I receive "romInit.o(.text+0x31c): undefined reference to `sysHawkParamConfig'". This image is strictly based on the BSP. I have added no code to it, nor have I changed any of the options in the vxWorks tab. What am I missing ? Thank you for help. Paul Livingston --------------------------- Newsgroups: comp.os.vxworks Subject: auto configuration on boot from ROM? Date: Wed, 02 Feb 2000 20:38:37 +0000 From: Matthew Barratt Message-ID: <389895CD.641FACF5@skat.demon.co.uk> Hello all, I've got some PowerPC cards booting from ROM and using the VME id pins so that they can autoconfigure ip addresses etc. as appropriate to their slot. Trouble is I'm sure that there's a better way to do it than my ignorant hacking through the BSP's source code. Can anyone tell me how it's supposed to be done? Radstone cards, version 1.6 of their BSP. Thanks in advance, Matthew --------------------------- Newsgroups: comp.os.vxworks Subject: Re: DBG problem Date: Wed, 02 Feb 2000 22:33:40 GMT From: john_94501@my-deja.com Organization: Deja.com - Before you buy. Message-ID: <87abc5$5mn$1@nnrp1.deja.com> References: Jeff, What architecture? Which host? Which version of Tornado? John... Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: dosFs2.0 Code to mount unknown number of partitions? Date: 2 Feb 2000 23:34:20 GMT From: cjohnson@oresis.com Organization: NewsOne.Net - Free Usenet News via the Web - http://newsone.net/ Message-ID: <87aets$a8c$1@news.netmar.com> With the new disk partition manageger which is part of the dosFs2.0 release, is there an elegant means to mount all the valid partitions ?? Since the current code does not actually do the mount until the first access, none of the calls (dpartPartGet() and dosFsDevCreate() for example.) return errors even if the partition you are mounting really doesn't exist. You have to access it first. (I'm doing an opendir() on the root directory.) Anyone have any better suggestions ?? Charles Johnson Oresis Communications, Inc. Beaverton, OR cjohnson@oresis.com ----- Posted via NewsOne.Net: Free Usenet News via the Web ----- ----- http://newsone.net/ -- Discussions on every subject. ----- NewsOne.Net prohibits users from posting spam. If this or other posts made through NewsOne.Net violate posting guidelines, email abuse@newsone.net --------------------------- Newsgroups: comp.os.vxworks Subject: Re: heap management Date: Wed, 02 Feb 2000 23:48:24 GMT From: dixon@crd.ge.com Organization: Deja.com - Before you buy. Message-ID: <87afo7$8ta$1@nnrp1.deja.com> References: <874dih$ogg$1@nnrp1.deja.com> In article <874dih$ogg$1@nnrp1.deja.com>, dixon@crd.ge.com wrote: > I am working with a colleague to resolve a performance problem that we > believe to be caused by heap fragmentation. There have been several > posts discussing this problem. We've tried to replace the standard heap > management code with Doug Lea's implementation that we found at > www.pso.com. There's a comment in this code that it is not compatible > with vxworks partition management. We initially ignored this warning, > and replaced the standard routines with Doug's. When the system boots we > are getting errors that appear to be related to the comment. When I look > at the replacement code, it really doesn't create partitions. The > comment goes on to suggest renaming the replacement routines and using > the new names in our code. This solution doesn't seem to be an option, > because we don't have source for what we believe to be the biggest > culprit. Is the solution as simple as really creating partitions? I've > found one reference to a replacment memPartLib (www.seaweed.com), but > it's not available for x86 at least according to their web page. Does > anyone have any suggestions? > > TIA > Standard disclaimers apply > > Sent via Deja.com http://www.deja.com/ > Before you buy. > Since this post I have spoken with bob schulman at seaweed and he assures me that their product does support x86. Sorry for the confusion. Standard disclaimers apply Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Debugger Problem Date: Tue, 1 Feb 2000 08:26:38 -0500 From: Charles H. Chapman Organization: Dynacs Engineering Company Message-ID: <876mue$jhf@niven.ksc.nasa.gov> References: Reply-To: chc@nasa2.ksc.nasa.gov Jeff Byrd wrote: > I am working with Tornado 1.0.1 and have a problem with the debugger. When I > place my source files in different directories, the debugger cannot seem to > find them. When all the source files are in the same directory, no problem. You can use the DIR command at the Crosswind prompt to add directories to the search path for source files. I'm not sure if there is a way to have that done automatically (like an environment variable or something). Chuck --------------------------- Newsgroups: comp.os.vxworks Subject: Re: [Q] How should I pronounce 'vxWorks'? Date: Wed, 2 Feb 2000 19:38:18 -0500 From: David Cooper Organization: Prodigy Internet http://www.prodigy.com Message-ID: References: <876fa6$q5n$1@news.etri.re.kr> In article <876fa6$q5n$1@news.etri.re.kr>, qkim@pec.etri.re.kr says... > Hi! > > Today is the first day that I just entered into the 'vxWorks' world. > I've navigated a lot of pages related to it, including the old FAQ and > an unofficial VxWorks & Tornado FAQ'. No pages have been given > information about pronunciation of 'vxWorks'. > > Can I pronounce it, just 'v', 'x', and 'works'? > > As you know, the Netscape co. wanted to call 'Netscape' 'Mozilla'. > But it failed at least in our country, Republic of Korea. > > Thanks in advance. > > > > The only way I have ever heard it pronounced (by Wind River folks and others) is V X works. - -- +--+--+--+--+--+--+--+--+--+--+--+--+--+--+--+--+--+--+ ô¿ô Genealogy Utilities & Family Research Researching: ALLEN,BERZONER,COHEN,COOPER,KUPFERZMIDT, POSNER (POIZNER), PRITIKIN, STARKMAN Dachau Camp Entry List, Genealogy Utilities http://pages.prodigy.net/dcooper000/index.htm --------------------------- Newsgroups: comp.os.vxworks Subject: Re: reply to relocation overflow Date: Wed, 2 Feb 2000 21:47:32 -0800 From: "who?" Organization: Global Network Services - Remote Access Mail & News Services Message-ID: <3899171e_4@news1.prserv.net> References: <1000202165154A8584106@pyungtaik.lge.co.kr> I think that particular error message indicates either the linker is trying to resolve a symbol that hasn't been defined or possibly a jump target address is too far away from the JAL instruction. (MIPS R2000/R3000/R6000 share this. I dunno about R4650 but suspect it is similar) Good luck! - Glenn Chuljin Ahn wrote in message <1000202165154A8584106@pyungtaik.lge.co.kr>... >Subject: relocation overflow >Submitted-by: "Raymund Christopher A. Toribio" > > >Hello, > > Does anyone know about relocation overflow? > > Im trying to make an object file and during compilation I >got that error. > I am using ccmips and my architecture is MIPS with a cputype of >R4650. > But when I tried using gcc or ccsimso I did not get that error. >The >ROM_SIZE is set to its maximum. Is there anything that I need to >alter/configure in VxWorks? in the configuration files? > > Thank you very much. > >Best Regards, >rc > >mailto:rtoribio@tspi.com.ph >----------------------------------------- >Dear Mr. Toribio, > >If ccmips generates an object image in COFF format, >the "--split-by-reloc" option of $(LD) might be of help. >In my case, ldsh, linker for SH processor, generated >relocation overflow error because COFF couldn't have >65535 relocations in a single section. With the option, >the error disappeared. Check the $(LD) manual about the >option. > >I wish it might be useful to you. > >Chuljin Ahn > >------------------------------------------------------------------ >Ahn, Chuljin (ahnmace@lge.co.kr) >Engineer >IK Business Team, Digital Venture Business, >LG Electronics, Inc. >Jinwee Myun, Pyongtaek City, Kyungki Do, Rep. of Korea, 451-713 >Phone:+82-333-610-5330, FAX:+82-333-610-5352, PCS:+82-19-223-6599 >------------------------------------------------------------------ --------------------------- Newsgroups: comp.os.vxworks Subject: Re: dosFs2.0 Code to mount unknown number of partitions? Date: Thu, 03 Feb 2000 08:52:27 +0200 From: Leonid Rosenboim Organization: BitBand Technologies Ltd. http://www.bitband.com Message-ID: <389925AA.A8DBABF7@bitband.com> References: <87aets$a8c$1@news.netmar.com> Reply-To: leonid@bitband.com DosFs 2.0 requires you to specify the maximum number of partition you EXPECT to see on the disk, and to provide each of these expected partition a logical device name. Sinc VxWorks does not have (never had) an automatic mechanism for assigning logical names to devices (a'la A:, B:, C:, ... in MessDos), and a logical name is a necesity to be able to access the device after initialization. Alternatively, the number of expected partitions can be large, since it is not an error if there are less partitions on the actual disk then expected - the exraneous partitions will have logical device names associated with them, but will yield an error on any access attempt. Moreover, the number of expected partitions must accomodate any removable disk which could be inserted into a particular drive. The partition mamager module "dpartCbio" has an arbitrary limitation of 8 partitions, that is supposed to be sufficient for most applications. Many operating systems have a similar limitation (SunOS, Solaris have max 8 partitions per disk drive) which are seldom reached. It could be possible by slight modification of dpartCbio source code to get more partitions, but my feeling is that there is something strange if you need more then 8 partitions on a disk, perhaps your design considerations should be revised... Leonid cjohnson@oresis.com wrote: > With the new disk partition manageger which is part of the dosFs2.0 release, > is there an elegant means to mount all the valid partitions ?? Since the > current code does not actually do the mount until the first access, none of > the calls (dpartPartGet() and dosFsDevCreate() for example.) return errors > even if the partition you are mounting really doesn't exist. You have to > access it first. (I'm doing an opendir() on the root directory.) > > Anyone have any better suggestions ?? > > Charles Johnson > Oresis Communications, Inc. > Beaverton, OR > cjohnson@oresis.com > > ----- Posted via NewsOne.Net: Free Usenet News via the Web ----- > ----- http://newsone.net/ -- Discussions on every subject. ----- > NewsOne.Net prohibits users from posting spam. If this or other posts > made through NewsOne.Net violate posting guidelines, email abuse@newsone.net --------------------------- End of New-News digest ********************** From vxwexplo-errs@csg.lbl.gov Thu Feb 3 04:05:40 2000 From: David Anderson Date: Thu Feb 3 04:05:42 PST 2000 Subject: EST backend and VisionIce. Tornado. Hello vxWorkers, Has anyone out there got experience of the EST backend for the Coldfire & Tornado 1.0.1 ? Do you know how to resolve back-end problems with WDB_TGT_INFO errors in target server window ? ( these do not appear using wdbrpc ( same BSP ) ). If so, how do I resolve an error return from tgtGetInfoMem_c() during tgtConnect() ? Messages suggest that the TGT_ADDR_T value is incorrect which is used to set up 'memBase' but my BSP configuration sets it correctly I believe ( at 0x00000000 ) ? Thanks in anticipation David Anderson. mailto:danderson@dtrack.demon.co.uk 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 Thu Feb 3 10:02:36 2000 From: Yaron Rotstein Date: Thu Feb 3 10:02:38 PST 2000 Subject: Debugger attach problem I have a problem viewing source code after attaching to my target. Here is the routine my task is pending on: Void WaitHere(void) { int keepWaiting = 1; While (keepWaiting) { taskDelay(100); } } I start the system _without_ the debugger and wait for my task to PEND. Then I start the debugger and attach to pending task. I get a disassembly window of something that looks like system code (as expected since the task is pending). However, I cannot step back on the stack to my code to see who jumped to WaitHere(). In addition, the viewing options (source, assembly, mixed) are disabled (grayed out). This is very strange because I remember using this method in another project, but can't remember the tools version. Here is some additional information: 1) The code is compiled with -g and no optimization. 2) I'm using Tornado 1.0.1 for Hitachi SH and dve7043 BSP 3) Using the Run menu option works just fine. I'm working on this for few days now. I appreciate your help. Thanks. From vxwexplo-errs@csg.lbl.gov Thu Feb 3 10:18:42 2000 From: Dennis Daniels Date: Thu Feb 3 10:18:44 PST 2000 Subject: MPC603e Assembly I'm relatively new to VxWorks and am adding some low-level, power-on diagnostics (in MPC603e assembly, GNU ToolKit) to be called from romInit.s. In my assembly file I have added some tables as part of the .text section and I have also added initialized arrays in a *.c module (declared as consts) which is linked into the boot image. When I attempt to load the address of any of any of these tables (using the HIADJ and LO macros) I get an address located in RAM vs. the actual table location in ROM. As a check I put a branch to a table label and the address is correctly resolved. I've looked through the docs, but haven't found an answer so far. Can anyone shed some light on this? Thanks in advance for any help. Dennis VxWorks, Tornado. From vxwexplo-errs@csg.lbl.gov Thu Feb 3 13:53:38 2000 From: "Shah, Ranjana" Date: Thu Feb 3 13:53:40 PST 2000 Subject: RE: Debugger attach problem VxWorks, Tornado The best way I have found to attach debug to a task is to put a taskSuspend (0) as the first line in the task to be debugged. Thus when you attach the debugger to it, you would get the disassembly window with the code for taskResume. Once you have that, you may close that window and click on Finish from the debug toolbar to jump out of the taskSuspend routine. Now the program counter should point to a line in your code. At that point you can click on the "up" button on the toolbar to go up the stack, which would take you to the caller. Even without taskSuspend(0), in your case you may be able to use the same technique to view your source code with the program counter pointer. Ranjana From vxwexplo-errs@csg.lbl.gov Fri Feb 4 02:43:17 2000 From: David Laight Date: Fri Feb 4 02:43:20 PST 2000 Subject: Re: MPC603e Assembly Dennis, I suspect that the boot process is more complicated than you realise :-) On my (ARM) system, loading a compressed rom image to low memory, it goes as follows: - romInit.s code starts executing at 0x10000 - the first instruction is an absolute jump to the rom alias at 0x4010000 (ROM_BASE_ADDRS - or more probably ROM_TEXT_ADRS) - This code, while only doing relative jumps, initialises DRAM at address 0, then jumps to the C code romStart. - romStart copies all the rom (except the compressed vxWorks image) into ram at the location it was linked at. For my system this is RAM_HIGH_ADRS (for a normal bootrom it is RAM_LOW_ADRS). - The code then jumps into the copy in RAM (by calling via an absolute jump). - The vxworks image is then uncompressed from ROM into RAM at RAM_LOW_ADRS. - The vxworks image is then entered at RAM_LOW_ADRS, all the code at RAM_HIGH_ADRS is never executed again - indeed RAM_HIGH_ADRS itself isn't used. The effect of this is that the code in romInit.s isn't running at the address it was linked at. So absolute references to memory items will fail. PC relative ones will be ok. The branch references the correct location because it is PC relative. My romInit.s contains constructs like: .long ROM_TEXT_ADRS + _romStart - _romInit in order to get the correct address of the rom copy of the boot code. This also show the two bounds for RAM_HIGH_ADRS: - it must be high enough that the uncompressed vxWorks image won't overwrite it. - it must be low enough that the entire bootrom can be copied to RAM_HIGH_ADRS without hitting the top of physical RAM. The first limit is the one that seems to give most problems, but I have hit the second when trying to avoid the first... David > Subject: MPC603e Assembly > Submitted-by: Dennis Daniels > > I'm relatively new to VxWorks and am adding some low-level, power-on > diagnostics (in MPC603e assembly, GNU ToolKit) to be called from romInit.s. > In my assembly file I have added some tables as part of the .text section > and I have also added initialized arrays in a *.c module (declared as consts) > which is linked into the boot image. When I attempt to load the address of > any of any of these tables (using the HIADJ and LO macros) I get an address > located in RAM vs. the actual table location in ROM. As a check I put a > branch to a table label and the address is correctly resolved. I've looked > through the docs, but haven't found an answer so far. Can anyone shed some > light on this? > Thanks in advance for any help. > > Dennis > > VxWorks, Tornado. ---------------------------------------------------------------- David Laight email: dsl@tadpole.co.uk Tadpole Technology plc phone: +44 1223 428 232 Cambridge, UK fax: +44 1223 428 201 From vxwexplo-errs@csg.lbl.gov Fri Feb 4 04:03:11 2000 From: Vxworks Exploder Date: Fri Feb 4 04:03:13 PST 2000 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Fri Feb 4 04:03:08 PST 2000 Subject: etherInputHookAdd breaks PPP Subject: Build problem Subject: Re: Build problem Subject: Re: Task Trace in ExcHook routine? Subject: unusual IP addresses Subject: How to process form POST's with Wind Web Subject: Download application from Tornado Subject: re: serial line problem Subject: Re: Build problem Subject: Re: etherInputHookAdd breaks PPP Subject: VxWorks and Zinc - ZafDiskFileSystem Subject: Re: iterate over all files in a directory Subject: Looking for BusNet implementation Subject: Re: excHookAdd Subject: Re: QED RM7000 PROM Monitor Source Package Subject: iterate over all files in a directory Subject: Re: unusual IP addresses Subject: munching c-code-->Alloc failure Subject: builds in T2 Subject: Template Instantiation and Munching Subject: Re: WDB/host memory pool vs heap usage oddities ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: etherInputHookAdd breaks PPP Date: Thu, 03 Feb 2000 14:03:49 GMT From: rfisher@iee.org Organization: Deja.com - Before you buy. Message-ID: <87c1s4$aqh$1@nnrp1.deja.com> I have found that the presence of a hook created by etherInputHookAdd() causes PPP to stop working properly. Packets sent from a vxWorks machine to an NT machine via PPP arrive ok. But packets sent from the NT machine to vxWorks are never seem to make it onto the TCP/IP stack. They are seen by the PPP software (the ip packet count displayed by pppstatShow() goes up) and they are seen by the hook routine, but if I send UDP packets, for example, the count displayed by udpstatShow() does not get incremented. The hook routine currently does nothing except return FALSE, so packets received should continue to be processed normally. This is true for ethernet but does not seem to be true for PPP. Has anyone else seen this, or does anyone have any ideas as to what might be happening. (Note - I'm using Tornado 1.0.1). Rob. Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Build problem Date: Thu, 03 Feb 2000 19:45:06 +0530 From: Mahesh Kurse Organization: Lucent Technologies Message-ID: <38998D69.3FD69B16@lucent.com> Hi, I am trying to compile a C source for MIPS target on Solaris host. I am getting the following error - $ ccmips -c foo.c cpp: Invalid oprion '-lang-c' What could be the cause of this error? Thanks, Mahesh. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Build problem Date: Thu, 03 Feb 2000 13:13:44 -0600 From: Andrew Johnson Organization: APS, Argonne National Laboratory Message-ID: <3899D368.47B4BFD9@aps.anl.gov> References: <38998D69.3FD69B16@lucent.com> Mahesh Kurse wrote: > > I am trying to compile a C source for MIPS target on Solaris host. > I am getting the following error - > > $ ccmips -c foo.c > cpp: Invalid oprion '-lang-c' > > What could be the cause of this error? Do you have the correct setting for your GCC_EXEC_PREFIX environment variable, or a -B option giving the same path? Check the instructions for the required switches and settings in Appendix E of your vxWorks programmers guide. - - Andrew - -- Complexity comes for free, Simplicity you have to work for. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Task Trace in ExcHook routine? Date: Thu, 03 Feb 2000 15:18:50 -0500 From: Tim Sohacki Organization: Nortel Networks Message-ID: <3899E2AA.7523F387@americasm01.nt.com> References: <3897FBA6.2F55BE35@oce.nl> Ton, You could ask WRS for the source for tt() ... and then call the appropriate lower level function. This might be the easiest solution. It would be nice if more of the "shell" commands were callable programmatically, e.g., returning or filling a buffer. I/O redirection is sometimes a solution, but not always. Cheers, Tim. Ton Janssen wrote: > > We want to display a stack trace of the task which > generates a Page Fault automatically, so we use > the excHookAdd() routine to install this handler > and then want to call something like tt( taskId) > to display the stack. > > However, tt() cannot be called as function obviously, > only from the VxWorks shell. 'INCLUDE_DEBUG' has been > set, so tt interactively works, but the system reboots > as a result of the Page Fault, so we want the trace > immediately when the Page Fault occurs. > > Does any body know a way to do this? --------------------------- Newsgroups: comp.os.vxworks Subject: unusual IP addresses Date: Thu, 3 Feb 2000 17:03:46 -0500 From: Sakir Yucel Organization: University of Delaware, EE/CIS Lab Message-ID: Hello all, I am new to VxWorks. I would like to assign IP addresses of the form 0.0.0.11 and 0.0.0.20 to some machines that run VxWorks within an isolated IP network. Could a VxWorks guru help me by telling me the things that I should do both with respect to IP configuration as well as modifying the VxWorks source code? The reason for assigning these unusual addresses is to isolate the network from outer world. Any other way of doing this in VxWorks? Any help will be greatly appreciated. Sakir Yucel CIS Graduate Student University of Delaware --------------------------- Newsgroups: comp.os.vxworks Subject: How to process form POST's with Wind Web Date: Thu, 03 Feb 2000 17:58:16 -0500 From: Thomas Davis Organization: Interpath Communications, Inc. Message-ID: <389A0808.C829CF73@galois.net> Hi, I been working on the vxWorks kernels for a couple years and I just got WindWeb. I have most of it working well, but I found no documentation discribing POSTed forms. So, How is an rpm suppost to process POST'ed forms? Any help is greatly appreciated! :-) Thomas - -- Galois, Inc. http://www.galois.net We color the world! --------------------------- Newsgroups: comp.os.vxworks Subject: Download application from Tornado Date: Thu, 03 Feb 2000 19:27:03 -0800 From: Peter Johansson Organization: SBC Internet Services Message-ID: Hello. I am starting work on a MUX (END) driver for VxWorks and am pretty new to the development environment (Tornado, etc.). I'm trying to build, download and run a mindlessly simple application so that I know I'm using Tornado and the tools correctly. The application does nothing more than printf("Hello"). I compile and build successfully. I boot the target board with a VxWorks image built from the appropriate BSP. I launch the target server, which connects successfully with the agent on the target board. I can initiate browser, shell and GDB tools from within Tornado: they connect correctly to the target server. My problem arises when I attempt to download the executable application, hello,out in this particular example. Within the workspace window, I right-click on the desired project and select "Download 'hello.out'". A downloading files window appears and file transfer proceeds to 71% (always the same point) at which an error dialog box pops up to inform me that WTX Error 0x100de (AGENT_COMMUNICATION_ERROR) has ocurred---and that's the end of my download! I have no doubt but what I'm doing something wrong (likely stupid) and would very much appreciate advice on how to proceed. Thanks in advance, Peter Johansson --------------------------- Newsgroups: comp.os.vxworks Subject: re: serial line problem Date: Thu, 3 Feb 2000 21:05:38 -0700 From: Rich Ireland Organization: XMission http://www.xmission.com/ Message-ID: References: <87avdv$7h5@overload.lbl.gov> In article <87avdv$7h5@overload.lbl.gov>, Fred Roeber wrote: > On Wed, 2 Feb 2000, Herve Mathieu wrote: > > > I'm using MVME167 board with VxWorks 5.3. > > A sensor is connected to a rs232 line, without XON/XOFF (hard or soft), at > > 9600 bauds. > > > > The problem is that some characters are loose. > > The quantity of data is less than the input buffer size (about 800 chars). > > VxWorks uses a ring buffer to store the characters that are received prior > to their being read. The size of the ring buffer defaults to 512 bytes. > If you don't read the characters and the ring buffer fills up, data > will be lost. If you think this is your problem (due to losing a number > of sequential characters) then you should just boost the size of the > ring buffer the serial driver uses. Based on which version of VxWorks > you are using, there are different ways to change the size. Since you are > using a pre "Tornado 2 project tool" version I suggest just going into > the config/all/usrConfig.c file and changing the arguments to ttyDevCreate > to boost the size of the read buffer (third parameter) from 512 to 1024. > You can also use the FIORBUFSET ioctl (see tyLib man page). Note that if you use the FIORBUFSET (or FIOWBUFSET) method, you will incur additional heap allocations. For tight jobs, you shouldn't do that too often. HTH - -- Rich Ireland mailto:r.ireland@computer.org Embedded & Real Time Engineering --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Build problem Date: Fri, 04 Feb 2000 09:46:50 +0530 From: Mahesh Kurse Organization: Lucent Technologies Message-ID: <389A52B1.3DC3504E@lucent.com> References: <38998D69.3FD69B16@lucent.com> <3899D368.47B4BFD9@aps.anl.gov> Yes, I had set the variable GCC_EXEC_PREFIX to appropriate value (ofcourse, with the trailing '/' !!). Now, I tried compiling with -B option, but now also I am getting the same error :-( Andrew Johnson wrote: > Mahesh Kurse wrote: > > > > I am trying to compile a C source for MIPS target on Solaris host. > > I am getting the following error - > > > > $ ccmips -c foo.c > > cpp: Invalid oprion '-lang-c' > > > > What could be the cause of this error? > > Do you have the correct setting for your GCC_EXEC_PREFIX environment > variable, or a -B option giving the same path? Check the instructions for > the required switches and settings in Appendix E of your vxWorks > programmers guide. > > - Andrew > -- > Complexity comes for free, Simplicity you have to work for. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: etherInputHookAdd breaks PPP Date: Thu, 03 Feb 2000 15:31:34 GMT From: rfisher@iee.org Organization: Deja.com - Before you buy. Message-ID: <87c70n$edq$1@nnrp1.deja.com> References: <87c1s4$aqh$1@nnrp1.deja.com> In article <87c1s4$aqh$1@nnrp1.deja.com>, rfisher@iee.org wrote: > I have found that the presence of a hook created by > etherInputHookAdd() > causes PPP to stop working properly. I have just learnt that this is a FAQ - apparently there it is an os bug for which a patch is available. Rob. Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: VxWorks and Zinc - ZafDiskFileSystem Date: Thu, 3 Feb 2000 14:27:44 +0200 From: "Steven Venter" Message-ID: <389973c3$0$35284@helios.is.co.za> Hi all I'm trying to read all the files with a certain extension from a HD and store those names in an array for later use. Someone told me that the ZafFileDialog object doesn't work under VxWorks, so now I have to try and perform that function with ZafDiskFileSystem but can't get it to work. Thanks Steven - -- sventer@grintek.com sventer@icon.co.za --------------------------- Newsgroups: comp.os.vxworks Subject: Re: iterate over all files in a directory Date: Thu, 03 Feb 2000 13:14:36 +0200 From: Leonid Rosenboim Organization: BitBand Technologies Ltd. http://www.bitband.com Message-ID: <3899631B.355FCBA1@bitband.com> References: <87bigt$okp$1@newsmaster01.magnet.at> Reply-To: leonid@bitband.com Tkae a look at $WIND_BASE/target/src/usr/usrLib.c or usrIoLib.c for source of the ls() function you are using. Mario Semo wrote: > Hello, > > how can i iterate over all files in a directory? we tried opendir(), but: > > when we call opendir() an output is generated by the system to the console: > > ... > opendir... > testdir: not a plain file. > ... > > Help says: > > WARNING > For remote file systems mounted over netDrv, opendir( ) fails, > because the netDrv implementation strategy does not provide a > way to distinguish directories from plain files. To permit use > of opendir( ) on remote files, use NFS rather than netDrv. > > this is correct. we use ftp-mount and not NFS. > > but , obvious, i can type "ls" at the target machine at the console and i > see a list of all files. > > how can i get a list of files in a given directory in this scenario? > > -- > regards, Mario Semo. http://www.kirchnersoft.com --------------------------- Newsgroups: comp.os.vxworks Subject: Looking for BusNet implementation Date: Thu, 3 Feb 2000 08:48:27 +0100 From: Jerome.SOUQUIERES@dematic-postal.com Organization: Deja.com - Before you buy Message-ID: Hello, Does anyone know about a BusNet implementation for VxWorks ? Thanks, - -- ... Jess ... Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: excHookAdd Date: Fri, 28 Jan 2000 08:46:55 -0500 From: Jeff Daly Organization: Sanders A Lockheed Martin Company Message-ID: <38919DCE.6AB2D751@lmco.com> References: <86r1k5$n841@overload.lbl.gov> Reply-To: jeffrey.j.daly@lmco.com excHookAdd changes the pointer excExcepHook to point to the function you provide. there is no chaining. "Ferrara, Bob" wrote: > This message is in MIME format. Since your mail reader does not understand > this format, some or all of this message may not be legible. > > ------_=_NextPart_001_01BF693C.BDEEB91A > Content-Type: text/plain > > VxWorks/Tornado users - > > Can someone tell me if excHookAdd() can be called more than once, and have > the routines specified on the excHookAdd calls ALL be invoked when an > exception > occurs? > > It's behaving like it can only be called one time. It is called in > bootConfig.c to add > a boot exception handler (standard VxWorks) and I have called it a 2nd time > to add > my own specific things to do when an exception occurs. My exception routine > only > gets invoked when I comment out the first call to excHookAdd in bootConfig.c > > Thanks, > Bob Ferrara > bferrara@unispheresolutions.com > > ------_=_NextPart_001_01BF693C.BDEEB91A > Content-Type: text/html > > > > > > > excHookAdd > > >
> >

VxWorks/Tornado users - >

>
> >

Can someone tell me if excHookAdd() can be called more than once, and have >
the routines specified on the excHookAdd calls ALL be invoked when an exception >
occurs?  >

> >

It's behaving like it can only be called one time.  It is called in bootConfig.c to add >
a boot exception handler (standard VxWorks) and I have called it a 2nd time to add >
my own specific things to do when an exception occurs.  My exception routine only >
gets invoked when I comment out the first call to excHookAdd in bootConfig.c >

> >

Thanks, >
Bob Ferrara >
bferrara@unispheresolutions.com >

> > > > ------_=_NextPart_001_01BF693C.BDEEB91A-- --------------------------- Newsgroups: comp.os.vxworks Subject: Re: QED RM7000 PROM Monitor Source Package Date: Fri, 28 Jan 2000 08:42:32 -0500 From: Jeff Daly Organization: Sanders A Lockheed Martin Company Message-ID: <38919CC7.C21A9351@lmco.com> References: <3890F7AA.27692BE1@acc.com> Reply-To: jeffrey.j.daly@lmco.com stefan, i've used Algorithmics' PMON package for years and it's been great. they have full support for the RM7K (which i've been using for the past year) and a great compiler toolchain for pretty cheap. they have support for several ethernet, RS-232, and many flash devices. the PMON package itself is totally free, unlike some other ones i know of, and they also offer a source code package for diagnostics for a (slightly more expensive) fee that is pretty easy to port to custom hardware. http://www.algor.co.uk Stefan Dinev wrote: > Hi all: > > I am looking for RM7000 PROM Monitor Source Package with functions for > memory/register access, Ethernet/RS-232 file download, flash memory > support. > Any information? > > Thanks. > Stefan Dinev --------------------------- Newsgroups: comp.os.vxworks Subject: iterate over all files in a directory Date: Thu, 3 Feb 2000 10:42:20 +0100 From: "Mario Semo" Organization: Nextra Telekom GmbH Message-ID: <87bigt$okp$1@newsmaster01.magnet.at> Hello, how can i iterate over all files in a directory? we tried opendir(), but: when we call opendir() an output is generated by the system to the console: ... opendir... testdir: not a plain file. ... Help says: WARNING For remote file systems mounted over netDrv, opendir( ) fails, because the netDrv implementation strategy does not provide a way to distinguish directories from plain files. To permit use of opendir( ) on remote files, use NFS rather than netDrv. this is correct. we use ftp-mount and not NFS. but , obvious, i can type "ls" at the target machine at the console and i see a list of all files. how can i get a list of files in a given directory in this scenario? - -- regards, Mario Semo. http://www.kirchnersoft.com --------------------------- Newsgroups: comp.os.vxworks Subject: Re: unusual IP addresses Date: Thu, 03 Feb 2000 21:58:24 -0800 From: Peter Johansson Organization: SBC Internet Services Message-ID: <7dqk9ss64sl708i0k92qg2ldsjuub05k27@4ax.com> References: On Thu, 3 Feb 2000 17:03:46 -0500, Sakir Yucel wrote: > >Hello all, > >I am new to VxWorks. I would like to assign IP addresses of the form >0.0.0.11 and 0.0.0.20 to some machines that run VxWorks within an isolated >IP network. Could a VxWorks guru help me by telling me the things that I >should do both with respect to IP configuration as well as modifying the >VxWorks source code? Sakir, I am no VxWorks guru and my answer is applicable only to your development environment---if you need another solution for a production situation, I hope others answer. When a target board with a suitable ROM is powered on or reset, there is a dialog (presented on the serial port) that permits you to specify the IP address of the target. In my configuration, this is saved in NVRAM somewhere, since it is available through subsequent power cycles. I'm sure this is not the whole answer you're seeking, but I hope it helps... Peter Johansson --------------------------- Newsgroups: comp.os.vxworks Subject: munching c-code-->Alloc failure Date: Thu, 3 Feb 2000 17:40:16 +0100 From: "Elke Hendon" Organization: STN ATLAS Elektronik GmbH Message-ID: <87caq7$8j04@dsit03.atlas.de> Hi! I have noted a peculiar behavior: If I munch c-code and then load the resulting object, the VxWorks loader results in an Error: ld < zdl_lib.o 0x3fb56a4 (tShell): memPartAlloc: block too big - 1281316959 in partition 0x1b24 e0. ld error: error reading file (errno = 0x110001). value = 0 = 0x0 The source file is actually c-code, and is compiled with g++386 (on a Sun-solaris) and I attempt to load it onto a PC target. Please don't say, well why munch it.... because the error has on occasion appeared after loading the "raw" object file. Can anyone guide me around this error? What could also cause this error? Thank you! Elke --------------------------- Newsgroups: comp.os.vxworks Subject: builds in T2 Date: Fri, 4 Feb 2000 08:24:19 +0100 From: "Achim Zimmer" Organization: Colt Telecom GmbH Message-ID: <87dupr$egk$1@crusher.de.colt.net> I am desperate. I try to change the build properties of a project called A (downloadable appl) in workspace 1 so, that it 1. invoke a build of a project B and C in the same workspace before its own build 2. link them all together at the end Remember, that the builds are placed in subdirectories of the project directories. Are there any possibilities in Tornado2 ??? Thanks very much - -- Achim Zimmer R&D DICA Technologies Have a nice day ! --------------------------- Newsgroups: comp.os.vxworks Subject: Template Instantiation and Munching Date: Thu, 03 Feb 2000 09:50:38 GMT From: paulvroberts@my-deja.com Organization: Deja.com - Before you buy. Message-ID: <87bj1e$143$1@nnrp1.deja.com> Hi, I am just converting over our makefiles from Tornado 1 to Tornado 2 and hoping to use templates with the -frepo option. I read that the link stage must be done using the cross compiler. Previously we would compile all source code, then link it all together, munch it (to generate static ctor/dtor calls), compile the munch file, and then relink. My question is - does anyone know if you are supposed to munch and link after the link step to instantiate templates OR do you instatiate templates and then munch and relink? Thanks, Paul Roberts. Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: WDB/host memory pool vs heap usage oddities Date: Thu, 03 Feb 2000 13:09:19 -0600 From: Andrew Johnson Organization: APS, Argonne National Laboratory Message-ID: <3899D25F.5442F7D6@aps.anl.gov> References: <87aocv$7g6@overload.lbl.gov> brian.r.walker@exgate.tek.com wrote: > > Wind River could also implement a few tricks in their loader. First, they > could provide a way to override the memory partition that loadLib uses. > That would make it easier to create a memory partition close to the kernel > image for object modules. Second, they could fix elfLoadLib so that if it > would try to turn relative branches into absolute branches. That would only > be a 98% cure but it would allow object files to link to the kernel code > which is usually located in the first 32MB of memory. Third, on discovering relative branches that don't fit into 26 bits, the loader could create a jump-table for the module it's loading (using absolute branches) and resolve the branches through that. - - Andrew - -- Complexity comes for free, Simplicity you have to work for. --------------------------- End of New-News digest ********************** From vxwexplo-errs@csg.lbl.gov Fri Feb 4 04:27:14 2000 From: Bob Date: Fri Feb 4 04:27:16 PST 2000 Subject: Initialisusinf DosFs I am trying to initialise a DOS filing system onto a SCSI drive using example code from the vxworks programmers guide. However the dosFsInit call always returns a 0. Can someone please suggest what is missing from the following code. The pointer to the BLK_DEV appears to get set up correctly by the scsiBlkDevCreate call. //Pointer to SCSI Controller SCSI_CTRL *pScsiCtrl = pSysScsiCtrl; //pSysScsiCtrl is created by vxworks //Pointer SCSI device structure SCSI_PHYS_DEV *pSPD; //Pointer I/O block device structure BLK_DEV *pBD; //See if LUN has been created already pSPD = scsiPhysDevIdGet( pScsiCtrl, busId, LUN ); if (pSPD != 0) { printf(" SCSI device busdId %d LUN %d already exist. \r\n", busId, LUN ); } else { //Create device pSPD = scsiPhysDevCreate( pScsiCtrl,0,0,0,NONE,0,0,0 ); } //if created then create dosFs block device if (pSPD == NULL) { printf(" Unable to create SCSI device busdId %d LUN %d \r\n", busId, LUN); return; } else { //assume total size of 20000 blocks if ((pBD = scsiBlkDevCreate (pSPD, NUMDISKBLOCKS, 0)) == NULL) { printf(" scsiBlkDevCreate failed busdId %d LUN %d \r\n", busId, LUN ); return ; } // initialize dosFs file system if (dosFsDevInit ("/sd1/", pBD, NULL) == NULL) { printf("dosFs initialize failed \r\n"); return ; } } } TIA Robert Baker From vxwexplo-errs@csg.lbl.gov Fri Feb 4 05:11:13 2000 From: mmenge@dspt.com Date: Fri Feb 4 05:11:15 PST 2000 Subject: boot parameters Mark Menge@DSPT 02/04/2000 08:10 AM Hi there vxworkers. Has anyone implemented a MS Windows program that allows the user to input his boot parameters into a dialog and programs the vxWorks boot parameters via the serial port? tornado vxworks From vxwexplo-errs@csg.lbl.gov Fri Feb 4 06:44:11 2000 From: Ian Willats Date: Fri Feb 4 06:44:13 PST 2000 Subject: RE: Build problem > $ ccmips -c foo.c > cpp: Invalid oprion '-lang-c' > > What could be the cause of this error? I think another possible cause for this (apart from GCC_EXEC_PREFIX) is that you're picking up the wrong cpp, i.e. not the one supplied by WRS. Worth checking your PATH environment variable, anyway. Ian --------------------------------------------------------------- Ian Willats mailto:ian.willats@telelogic.com Telelogic UK Ltd. direct: +44 (0) 121 234 6637 (formerly Real Time Products Ltd) Chancery House, tel: +44 (0) 121 234 6600 8 Edward Street, Birmingham. fax: +44 (0) 121 234 6611 B1 2RX. England. web: http://www.telelogic.com --------------------------------------------------------------- VxWorks From vxwexplo-errs@csg.lbl.gov Fri Feb 4 10:08:02 2000 From: Bruce Wilson Date: Fri Feb 4 10:08:04 PST 2000 Subject: Re: Force cpci3603 using ethernet Thanks to all the VxWorks users who responded. It was a very simple fix. We needed to specify "frcDc" instead of "dc". (We were accustomed to using "dc" with the Gespac PCIPPC-1 / PCILAN-1 board combo.) Thanks again for all the help. --Bruce Wilson >Submitted-by vxwexplo-errs@csg.lbl.gov Wed Feb 2 08:59:18 2000 >Submitted-by: "Bruce A. Wilson" > >Has anybody in the VxWorks group succeeded in using a CompactPCI >ethernet board along with the Force cpci3603 (3U) processor board? >By default, the BSP for the 3603 only comes set up for ppp or slip, >since there's no onboard ethernet. > >We rebuilt the bootrom.hex after setting #define INCLUDE_FRC_DC in >config.h, and burned a new bootrom, but still the only boot devices >it allowed were ppp and sl. How do we get it to allow a "dc" device, >like for the Gespac PCILAN board? > >Bruce A. Wilson >U.C. Lawrence Livermore National Lab >P.O. Box 808 (L-491) >Livermore, CA 94550 >voice: (925) 422-4145 >fax: (925) 423-9889 From vxwexplo-errs@csg.lbl.gov Fri Feb 4 13:22:16 2000 From: "Kok-Hwa Lee" Date: Fri Feb 4 13:22:18 PST 2000 Subject: Re: Initialisusinf DosFs Hi, Are you using a DOS formatted SCSI drive? If so, that explains why dosFsInit fails. vxWorks dosFS does not parse the partition table on a dos formatted drive. So, if your SCSI drive's system is initialized by formatting it under DOS, then it will not work. There are two workarounds. 1. Make a new file system on the drive by using dosMkfs(not recommended...since the drive will only be readable under vxWorks) 2. Download dosPartLibScsi.c from Wind River. This utility parses the partition table on dos formatted drive and mounts each partition as seperate device under vxWorks. This way, you get to preserve the partition table and thus makes the drive portable between vxWorks and windows. lee lee@echotek.com vxWorks, Tornado ----- Original Message ----- From: the vxWorks Users Group Exploder To: Sent: Friday, February 04, 2000 6:32 AM Subject: Initialisusinf DosFs > Submitted-by vxwexplo-errs@csg.lbl.gov Fri Feb 4 04:27:14 2000 > Submitted-by: Bob > > I am trying to initialise a DOS filing system onto a SCSI drive using > example code from the vxworks programmers guide. However the dosFsInit > call always returns a 0. Can someone please suggest what is missing from > the following code. The pointer to the BLK_DEV appears to get set up > correctly by the scsiBlkDevCreate call. > > file://Pointer to SCSI Controller > SCSI_CTRL *pScsiCtrl = pSysScsiCtrl; file://pSysScsiCtrl is created by > vxworks > > file://Pointer SCSI device structure > SCSI_PHYS_DEV *pSPD; > > file://Pointer I/O block device structure > BLK_DEV *pBD; > > file://See if LUN has been created already > pSPD = scsiPhysDevIdGet( pScsiCtrl, busId, LUN ); > if (pSPD != 0) > { > printf(" SCSI device busdId %d LUN %d already exist. \r\n", busId, LUN > ); > } > else > { > file://Create device > pSPD = scsiPhysDevCreate( pScsiCtrl,0,0,0,NONE,0,0,0 ); > } > > file://if created then create dosFs block device > if (pSPD == NULL) > { > printf(" Unable to create SCSI device busdId %d LUN %d \r\n", busId, > LUN); > return; > } > else > { > file://assume total size of 20000 blocks > if ((pBD = scsiBlkDevCreate (pSPD, NUMDISKBLOCKS, 0)) == NULL) > { > printf(" scsiBlkDevCreate failed busdId %d LUN %d \r\n", busId, LUN ); > return ; > } > > // initialize dosFs file system > if (dosFsDevInit ("/sd1/", pBD, NULL) == NULL) > { > printf("dosFs initialize failed \r\n"); > return ; > } > } > } > > > TIA > > Robert Baker From vxwexplo-errs@csg.lbl.gov Sat Feb 5 04:03:10 2000 From: Vxworks Exploder Date: Sat Feb 5 04:03:13 PST 2000 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Sat Feb 5 04:03:07 PST 2000 Subject: Watchdog timer question.. Subject: Re: Debugger attach problem Subject: Byte Alignment Subject: Embedded assembler for i386 code.. Subject: Re: Booting from MVME147? Subject: Re: heap management Subject: Re: Debugger Problem Subject: Re: Embedded assembler for i386 code.. Subject: Virtual Functions & Class Heirarchies. Subject: Re: Virtual Functions & Class Heirarchies. Subject: Re: Virtual Functions & Class Heirarchies. Subject: Re: virtual memory exhausted Subject: Re: builds in T2 Subject: Re: unusual IP addresses Subject: VxWorks 5.3.1/5.4 SENS question Subject: Re: Virtual Functions & Class Heirarchies. Subject: HELP! sun2ce(sparc) -Device driver causes reboot Subject: Re: boot parameters Subject: Re: Initialisusinf DosFs Subject: Re: iterate over all files in a directory Subject: Crosswind/GDB getting wrong source file and line ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Watchdog timer question.. Date: Fri, 4 Feb 2000 07:19:43 -0500 From: Burt Bicksler Organization: BBB Message-ID: Hi, I'm working on porting an existing application from a homegrown Kernel to vxWorks. Virtually all of the work is completed and I am now working on issues related to the Watchdog timer code. We are booting from a floppy image of a compressed bootrom which then loads the actual vxWorks runtime image. I've installed code in the vxWorks runtime that keeps the WD timer alive and this works well. I've added the same code to bootrom and if I wait for the auto boot countdown to start I can physically enable the WD timer and that works fine, including the physical boot to the runtime image. The problem is that the current WD timer has to be "petted" at least once a second and it takes too long in the booting process before the WD handler code gets started, so the system resets due to the WD. I'm trying to get more details on the WD timer hardware on this card, but wondered if anyone else has run into a similar problem, and how you worked around it. I've tried moving my WD handler code in bootrom further up the "food chain" but then it doesn't appear to work at all, since the Kernel isn't fully initialized at that point. I'm using wdCreate/Start to handle this. Thanks, Burt Bicksler P.S. Remove the obvious spam bait if replyling by e-mail. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Debugger attach problem Date: Fri, 4 Feb 2000 09:06:09 -0500 From: Charles H. Chapman Organization: Dynacs Engineering Company Message-ID: <87emd2$l7r@niven.ksc.nasa.gov> References: <87cgkv$7g7@overload.lbl.gov> Reply-To: chc@nasa2.ksc.nasa.gov Yaron Rotstein wrote: > 1) The code is compiled with -g and no optimization. > 2) I'm using Tornado 1.0.1 for Hitachi SH and dve7043 BSP > 3) Using the Run menu option works just fine. > The debugger has to be able to find the source code in order to perform symbolic debugging. That means the .c file either has to be in the current directory, or in one of the directories specified with the debugger DIR command. Chuck --------------------------- Newsgroups: comp.os.vxworks Subject: Byte Alignment Date: Fri, 4 Feb 2000 16:12:39 +0200 From: =?iso-8859-1?Q?Morn=E9_Hartman?= Message-ID: <389ade41$0$35286@helios.is.co.za> Hi all, Does anyone here know anything about byte alignment. I have an application that communicates between a VxWorks board and an NT Box. I am sending a message from a structure to the remote system. The message contains various data types including Char, Double, Float, Int etc. In my application in NT I use byte alignment so that the message is received in the correct format on the remote end. I believe that I need to use byte alignment on the VxWorks side aswell, but this did not work I initialise byte alignment as follows : #pragma options -a1 struct MyStruct .... #pragma options -a4 If I communicate between two NT boxes there is no problem. When I try the same thing between an NT and VxWorks box the received message is not identical to the sent message. Does anyone have a solution to this problem Morné Hartman mhartman@grintek.com --------------------------- Newsgroups: comp.os.vxworks Subject: Embedded assembler for i386 code.. Date: Fri, 4 Feb 2000 09:12:11 -0500 From: Burt Bicksler Organization: BBB Message-ID: Hi, I'm trying to get an embedded assembler instruction to compile and haven't quite figured out the syntax. __asm__ volatile ( "outw %0,%1" : : "ax" (arg1), "dx" (arg2) ) ; Where arg1 and arg2 are 0x200 and 0x800 respectively. It whines about no 386 instructions match... If I change "dx" to "N" and arg2 to a constant from 0 - 255 it compiles just fine. The docs on using GNU CC __asm__ are skimpy, at best, and only show real examples for the 68881 CPU. Has anyone else gotten this to work? Where is there some decent documentation on __asm__ for the Intel iX86 family of processors? Remove the obvious spam bait from the e-mail address if responding via e- mail. Thanks, Burt Bicksler --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Booting from MVME147? Date: Mon, 31 Jan 2000 13:46:29 -0600 From: Andrew Johnson Organization: APS, Argonne National Laboratory Message-ID: <3895E695.1926BE77@aps.anl.gov> References: Andreas Kabel wrote: > > I recently got 3 Motorola MVME147 boards cheaply. They have "VxWorks" > boot EPROMS on them. Obvioulsy, they can boot an image off a remote > FTP server. Can anyone point me in the right direction for information > about format, parameters, ... of that boot image? I'd like to be able > to boot netBSD on them. I think you just need an MC68020 a.out format object file ('file vxWorks' returns "impure mc68020 executable"), which IIRC will be loaded into RAM at 0x1000 and executed from that same address. It might be a good idea to confirm that load address using the 'l' and 'd' commands that the bootrom provides ('?' gives help). BTW, it can also boot using rsh providing you have a .rhosts file that permits access - the switch for this is to omit the password in the boot parameters. Some versions can also use tftp or bootp, check the info about the flags, given by the '?' command. - - Andrew - -- Complexity comes for free, Simplicity you have to work for. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: heap management Date: Tue, 01 Feb 2000 04:53:17 GMT From: john_94501@my-deja.com Organization: Deja.com - Before you buy. Message-ID: <875ort$p79$1@nnrp1.deja.com> References: <874dih$ogg$1@nnrp1.deja.com> Hi, > When the system boots we > are getting errors that appear to be related to the comment. What are the errors? I don't believe that there is much use of partitions in VxWorks itself, mostly it just uses the standard system memory partition (either directly, or through malloc), and that should work just fine with the replacement code. If you are using a VxWorks component that uses its own memory partition(s), then you may find that you simply need to augment the library from pso.com to create partitions too, or at least pretends to do so. The sequence of use is generally to allocate memory from the system memory partition, then call memPartCreate() to create a new partition from that memory. By making your memPartCreate() simply free the memory back to the pool, and then return a valid ID (ie non-NULL) you should be just fine. You'll also need to handle memAddToPool() in the same way (at least one VxWorks library uses this to extend its private partition when necessary). AFAICS, only Streams, DosFS and UGL are making use of their own partitions, but I could be wrong there... Hope that helps a little, I'd be interested to hear how you get on! John... Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Debugger Problem Date: Tue, 1 Feb 2000 13:56 +0000 (GMT Standard Time) From: rew@cp.cix.co.uk (Richard Wenham) Organization: CIX - Compulink Information eXchange Message-ID: References: Reply-To: rew@cp.cix.co.uk Have you added all the directories to the Source Search Path (Debug menu)? Works OK here. Richard In article , jeffbyrd@worldnet.att.net (Jeff Byrd) wrote: > I am working with Tornado 1.0.1 and have a problem with the debugger. > When I > place my source files in different directories, the debugger cannot > seem to > find them. When all the source files are in the same directory, no > problem. > > Any ideas? > > Thanks --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Embedded assembler for i386 code.. Date: Fri, 4 Feb 2000 12:05:32 -0500 From: Burt Bicksler Organization: BBB Message-ID: References: Answering myself.. Finally figured it out.. __asm__ volatile ( "outw %0,%1" : : "a" (arg1), "d" (arg2) ) ; With arg1 an int and arg2 a short. --------------------------- Newsgroups: comp.os.vxworks Subject: Virtual Functions & Class Heirarchies. Date: Fri, 4 Feb 2000 11:39:18 -0500 From: "Pankaj Datta" Message-ID: <3EDm4.934$rH3.454@newsfeed.slurp.net> Consider two heirarchies: H1: class A {public: virtual void f(){cout << "A" << endl;}}; class B: public A{public: void f() {cout << "B" << endl;}}; class C: public B{public: void f() {cout << "C" << endl;}}; H2: class A {public: virtual void f(){cout << "A" << endl;}}; class B: public A{public: virtual void f() {cout << "B" << endl;}}; class C: public B{public: virtual void f() {cout << "C" << endl;}}; Is there any difference between H1 & H2 in 1) behavior 2) code size 3) execution time If not, which is preferable? Thanks Pankaj --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Virtual Functions & Class Heirarchies. Date: Fri, 04 Feb 2000 11:28:40 -0800 From: Harvey Taylor Organization: Organization? What organization?! Message-ID: <389B2868.7CDE@despam.pangea.ca> References: <3EDm4.934$rH3.454@newsfeed.slurp.net> Hi Pankaj, This is a perfect question for WALP [Write A Little Program]. ie. it is testable. However I believe the virtual is only required once in the inheritance tree. ie. answers to 1),2) & 3) are No. Programs are written as much for people, as the compiler so I have always suggested H2. ie providing as much info as possible in any one class of a hierarchy. [If you don't, someone will not check the whole hierarchy & get suprising behaviour]. - -het Pankaj Datta wrote: > Consider two heirarchies: > > H1: > class A {public: virtual void f(){cout << "A" << endl;}}; > class B: public A{public: void f() {cout << "B" << endl;}}; > class C: public B{public: void f() {cout << "C" << endl;}}; > > H2: > class A {public: virtual void f(){cout << "A" << endl;}}; > class B: public A{public: virtual void f() {cout << "B" << endl;}}; > class C: public B{public: virtual void f() {cout << "C" << endl;}}; > > Is there any difference between H1 & H2 in > 1) behavior > 2) code size > 3) execution time > > If not, which is preferable? > > Thanks > > Pankaj - -- "The road to wisdom? Well it’s plain and simple to express: Err and err and err again, but less and less and less." -Piet Hein Harvey Taylor Internet: het@despam.portal.ca --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Virtual Functions & Class Heirarchies. Date: Fri, 4 Feb 2000 12:46:19 -0500 From: "Pankaj Datta" Message-ID: References: <3EDm4.934$rH3.454@newsfeed.slurp.net> <389B2868.7CDE@despam.pangea.ca> Thankyou so much Harvey. I did WALP before I sent the mail. The post was intended to uncover anything which my LP (Little Program) or MC++ (Medium C++) understanding did not! Thanks again. Harvey Taylor wrote in message news:389B2868.7CDE@despam.pangea.ca... > Hi Pankaj, > This is a perfect question for WALP [Write A Little Program]. > ie. it is testable. > However I believe the virtual is only required once in the > inheritance tree. ie. answers to 1),2) & 3) are No. > Programs are written as much for people, as the compiler > so I have always suggested H2. ie providing as much info as > possible in any one class of a hierarchy. [If you don't, > someone will not check the whole hierarchy & get suprising > behaviour]. > > -het > > > Pankaj Datta wrote: > > Consider two heirarchies: > > > > H1: > > class A {public: virtual void f(){cout << "A" << endl;}}; > > class B: public A{public: void f() {cout << "B" << endl;}}; > > class C: public B{public: void f() {cout << "C" << endl;}}; > > > > H2: > > class A {public: virtual void f(){cout << "A" << endl;}}; > > class B: public A{public: virtual void f() {cout << "B" << endl;}}; > > class C: public B{public: virtual void f() {cout << "C" << endl;}}; > > > > Is there any difference between H1 & H2 in > > 1) behavior > > 2) code size > > 3) execution time > > > > If not, which is preferable? > > > > Thanks > > > > Pankaj > > > > -- > "The road to wisdom? > Well it's plain and simple to express: > Err and err and err again, > but less and less and less." -Piet Hein > > Harvey Taylor Internet: het@despam.portal.ca --------------------------- Newsgroups: comp.os.vxworks Subject: Re: virtual memory exhausted Date: 4 Feb 2000 19:38:12 GMT From: johan.borkhuis@peektraffic.nl Organization: XS4ALL Internet BV Message-ID: <87f9r4$pc5$1@news1.xs4all.nl> References: <3898b398$0$217@nntp1.ba.best.com> Yao (Jonathan) Lin wrote: : We are running Tornado 2.0 on NT4.0. : I am using ccsimpc to build my project. There is no problem if I don't turn : on : any optimization. But when I used "-O1", it failed with "virtual memory : exhausted". : I have 1.6GB for virtual memory and judging from the disk operation, I would : say : it had used less than 5% of that amount. Did you turn on virtualk memory? Groeten, Johan - -- o o o o o o o . . . ___________________________________ o _____ || Johan Borkhuis | .][__n_n_|DD[ ====_____ | johan@borksoft.xs4all.nl | >(________|__|_[_________]_|________________________________| _/oo OOOOO oo` ooo ooo 'o!o!o o!o!o` === VxWorks page: http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html === --------------------------- Newsgroups: comp.os.vxworks Subject: Re: builds in T2 Date: 4 Feb 2000 19:45:44 GMT From: johan.borkhuis@peektraffic.nl Organization: XS4ALL Internet BV Message-ID: <87fa98$pc5$2@news1.xs4all.nl> References: <87dupr$egk$1@crusher.de.colt.net> Achim Zimmer wrote: : I am desperate. : I try to change the build properties of a project called A (downloadable : appl) in workspace 1 so, that it : 1. invoke a build of a project B and C in the same workspace before its own : build : 2. link them all together at the end : Remember, that the builds are placed in subdirectories of the project : directories. : Are there any possibilities in Tornado2 ??? Take a look at my VxWorks page. There is an example descibing this. Groeten, Johan - -- o o o o o o o . . . ___________________________________ o _____ || Johan Borkhuis | .][__n_n_|DD[ ====_____ | johan@borksoft.xs4all.nl | >(________|__|_[_________]_|________________________________| _/oo OOOOO oo` ooo ooo 'o!o!o o!o!o` === VxWorks page: http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html === --------------------------- Newsgroups: comp.os.vxworks Subject: Re: unusual IP addresses Date: Fri, 04 Feb 2000 20:23:07 GMT From: Curt McDowell Organization: Broadcom Corp. Message-ID: <389B3529.921BF668@broadcom.com> References: Reply-To: csm@broadcom.com Sakir Yucel wrote: > > I am new to VxWorks. I would like to assign IP addresses of the form > 0.0.0.11 and 0.0.0.20 to some machines that run VxWorks within an isolated > IP network. Could a VxWorks guru help me by telling me the things that I > should do both with respect to IP configuration as well as modifying the > VxWorks source code? > > The reason for assigning these unusual addresses is to isolate the network > from outer world. Any other way of doing this in VxWorks? Sakir, You should consider choosing a class C address of the form 192.168.x.x with a netmask of ffffff00, a class B address in the range 172.16.x.x through 172.31.x.x with a netmask of ffff0000, or a class A address of the form 10.x.x.x with a netmask of ff000000. These addresses are specifically reserved for the purpose of creating isolated networks. You shouldn't have to modify any source code, just set the address and netmask. If you use illegal IP addresses, you may create a lot of trouble for yourself. Curt McDowell Broadcom Corp. --------------------------- Newsgroups: comp.os.vxworks Subject: VxWorks 5.3.1/5.4 SENS question Date: Fri, 04 Feb 2000 21:43:20 GMT From: newsmine@my-deja.com Organization: Deja.com - Before you buy. Message-ID: <87fh5n$r4d$1@nnrp1.deja.com> Hi, I was wondering if socket options IP_RECVDSTADDR and IP_RECVIF are supported by the SENS stack. In my application, I need to know which network interface recieved the datagram delivered to my UDP server. If these options are not supported, is there any other way of knowing the interface? Thanks. Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Virtual Functions & Class Heirarchies. Date: Fri, 04 Feb 2000 16:06:14 -0800 From: Harvey Taylor Organization: Organization? What organization?! Message-ID: <389B6976.51F2@despam.pangea.ca> References: <3EDm4.934$rH3.454@newsfeed.slurp.net> <389B2868.7CDE@despam.pangea.ca> Hi Again Pankaj, Just for the heck of it, I checked the standard. Section 10.3 says: - --- [class.virtual] 10.3 Virtual functions [...] 2 If a virtual member function vf is declared in a class Base and in a class Derived, derived directly or indirectly from Base, a member function vf with the same name and same parameter list as Base::vf is declared, then Derived::vf is also virtual (whether or not it is so declared) and it overrides Base::vf. [...] - --- The thing is it is generally not a good idea to argue from compiler behaviour. I have seen a lot of _bad_ code written "because the compiler lets me". However, what with the slow progress of compilers crawling toward the ISO C++ Standard, it seems a necessary evil. I think we can safely say this horse is dead. - -het - -- "The road to wisdom? Well it’s plain and simple to express: Err and err and err again, but less and less and less." -Piet Hein Harvey Taylor Internet: het@despam.portal.ca --------------------------- Newsgroups: comp.os.vxworks,comp.unix.solaris Subject: HELP! sun2ce(sparc) -Device driver causes reboot Date: Fri, 04 Feb 2000 18:52:07 -0500 From: Al Johnston Organization: MindSpring Enterprises Message-ID: <389B6627.8E5ABCF4@mindspring.com> I am having a little trouble with a driver I wrote for my sun2ce card. It looks like anytime there is a high density of interrupts from the vme bus to the cpu (sun2ce/sparc) the damn card is likely to reboot without ANY diagnostic msg to the screen. My os is vxWorks 5.2... but the hardware is the same as a "sun4" (sparcstation2)...i think... Is there any sun4/sun2ce/sparc guru's left? I need some help!!! thanks (I hope...) - -al --------------------------- Newsgroups: comp.os.vxworks Subject: Re: boot parameters Date: Fri, 04 Feb 2000 20:31:24 -0500 From: Joe Durusau Message-ID: <389B7D6C.42D2C969@bellsouth.net> References: <87ejee$7g9@overload.lbl.gov> I don't know about a port to windows, but the freeware program expect has the capabilites you want. It should be easy to write an expect script to get the job done. Speaking only for myself, Joe Durusau mmenge@dspt.com wrote: > > Mark Menge@DSPT > 02/04/2000 08:10 AM > Hi there vxworkers. Has anyone implemented a MS Windows program that allows > the user to input his boot parameters into a dialog and programs the > vxWorks boot parameters via the serial port? > > tornado vxworks --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Initialisusinf DosFs Date: 5 Feb 2000 01:49:31 GMT From: cjohnson@oresis.com Organization: NewsOne.Net - Free Usenet News via the Web - http://newsone.net/ Message-ID: <87fvjb$nbn$1@news.netmar.com> Wind River has released an update to the dos filesystem which they call dosFs2.0. It now has a partition manager. dosFs2.0 supports VFAT, FAT32, FAT16, FAT12. Call your local Wind River FAE and they should be able to help you. Charles Johnson Oresis Communications, Inc. Beaverton, OR cjohnson@oresis.com ----- Posted via NewsOne.Net: Free Usenet News via the Web ----- ----- http://newsone.net/ -- Discussions on every subject. ----- NewsOne.Net prohibits users from posting spam. If this or other posts made through NewsOne.Net violate posting guidelines, email abuse@newsone.net --------------------------- Newsgroups: comp.os.vxworks Subject: Re: iterate over all files in a directory Date: Sat, 05 Feb 2000 02:47:16 -0500 From: Al Johnston Organization: MindSpring Enterprises Message-ID: <389BD584.D06C3CD7@mindspring.com> References: <87bigt$okp$1@newsmaster01.magnet.at> The "is it a file or is it a directory" problem is one of the most asinine defects in vxworks. There does not seem to be any decent approach to determine this for all file systems (net, nfs,dos)... the best I came up with was to a long (300 lines of c code) algorithm which include using .... to determine if ".../foo" is a directory see if you can create ".../foo/bar" as a normal file. If you can do this then foo is a directory. Anyway.. I have an algo'tm to do this... and it seems to work.. but i junked it because it was just too much of a kludge... If you want I can dig it out for you... WRS's responce when I ask them how to this was basicly that it was not possable. luck! --------------------------- Newsgroups: comp.os.vxworks Subject: Crosswind/GDB getting wrong source file and line Date: Sat, 05 Feb 2000 08:31:57 GMT From: tttt9999@my-deja.com Organization: Deja.com - Before you buy. Message-ID: <87gn5t$m0r$1@nnrp1.deja.com> I am using VxSim on Solaris, and developing C++ code. I have a relocatable object created by linking a main.o with a list of archive libraries. I then dynamically loaded it to the VxSim, and then started the Crosswind debugger. For a small number of function names in my main.o, the source files and line number for them are listed correctly. However, for most of the stuff in the archive libraries, I was not able to display to the correct file nor line number. "info address " would display the correct address "info line " displayed a totally bogus file and location. I then did a add-symbol to manually load some object files in some of the archive libraries, and they displayed the correct line within the files; this implies that the object files contains the correct stab info. I also dumped the stab section of my relocatable object, and they seems to be okay. Does anyone has similar problems ? What is the different between "info address" and "info line", ie. what does gdb use to find the information for "info line" ? I have been working with WindRiver for the last 3 weeks, and to NO solutions. They were not helpful at all. Thx. in advance. Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- End of New-News digest ********************** From vxwexplo-errs@csg.lbl.gov Sat Feb 5 17:41:01 2000 From: Fred Roeber Date: Sat Feb 5 17:41:05 PST 2000 Subject: Re: Byte alignment On Sat, 5 Feb 2000, wrote: > Does anyone here know anything about byte alignment. > > I have an application that communicates between a VxWorks board and an > NT Box. I am sending a message from a structure to the remote system. > The message contains various data types including Char, Double, Float, > Int etc. In my application in NT I use byte alignment so that the > message is received in the correct format on the remote end. I believe > that I need to use byte alignment on the VxWorks side aswell, but this > did not work > > I initialise byte alignment as follows : > > #pragma options -a1 > struct MyStruct .... > #pragma options -a4 Morn, I hate to make things tougher but your only problem isn't alignment. You could also be having a problem with data ordering (ie. an "endian" issue). When passing data between two C programs there is no "sure fire" way to guarantee data portability within the language. You can try things like the pragma's you are using but they are not "guaranteed" to work in any particular fashion. In fact, pragma's can be silently ignored by a compiler that is behaving legally. You don't say what your target machine is but many are "big endian" whereas the NT box is "little endian". This means that integers and floats have opposite byte orders on the two machines. If you want to pass data between different machines in a guaranteed portable fashion, you really need to use additional software to handle the task. In particular, you should look at using software to "marshall" the data (i.e. order it in a known fashion). There are various standard ways of doing this including the XDR protocol and the ANS-1 protocol. VxWorks uses XDR protocol for sending debug data between hosts and targets. If you want to do the job yourself, without the trouble of using something like XDR, I would suggest padding everything out to 4 byte boundaries and then using the "htons" and "htonl" functions to handle the endian issues. This usually works pretty well (unless you have a target that requires 64 bit alignment). Good luck. Fred | Fred J Roeber, BBN Systems & Technologies | | 4 John Clarke Road Middletown, RI 02842-5202 | | froeber@bbn.com 401-848-3548 | | TraceMaker product manager | | http://www.tracemaker.bbn.com | From vxwexplo-errs@csg.lbl.gov Mon Feb 7 01:50:49 2000 From: "Dipl.-Phys. S. Richter, IKP, Tel. 16-5152" Date: Mon Feb 7 01:50:51 PST 2000 Subject: Help on etherOutput and etherOutputHook Hi! I've a couple question concerning the use of etherOutput and the etherOutputHook routines. I need to implement a DECnet like communication in VxWorks. I succeeded to listen for incomming DECnet messages (tuFilterSetup). But now I've to fake the source address when sending the packet out because the other side is only listening for DECnet addresses. I thought that I could grap the ethernetpacket just before it's going to be sent via the etherOutoutHook - manipulate the source address in the ether_header structure and then just return FALSE to send it out onto the ethernet. As the DECnet uses an extended 802 header i also set the ether_type field to the total lenght of that packet. Unfortunatly the other side doesn't receive anything. Now the question is can I really do that? And concerning the type/length field - do I have to swap the bytes? If I look any IP-Packets I get as the type a 0x00-08 but the protocol definition says it should be 0x0800. (I tried the length field swapped and not swapped already) We're using VxWorks 5.31 for an AXPvme SBC (Alpha processor) Any help is greatly appreciated! Simone Dipl.-Phys. Simone Richter simone@ikp.tu-darmstadt.de Institut f. Kernphysik Schlossgartenstr. 9 Tel: +49 (0) 6151 / 16-5152, 16-3323 TU Darmstadt 64289 Darmstadt Fax: +49 (0) 6151 / 16-4321 From vxwexplo-errs@csg.lbl.gov Mon Feb 7 05:26:50 2000 From: Henrik Bergstrom Date: Mon Feb 7 05:26:52 PST 2000 Subject: Ctrl-C terminates gdb when running from CygWin Dear VxWorks users! I am no fan of gui:s and since I am already doing most of my Windows NT work in CygWin I have tried to "port" the VxWorks tools as well. Everything works fine except for SIGINT (Ctrl-C) in gdb. When I send a SIGINT by pressing Ctrl-C in the CygWin terminal running gdb, the debugger exits instead of breaking the task. Should one only use breakpoints while debugging VxWorks code or is there a way to make SIGINT work like it does in normal Unix debugging? Please let me know in that case since it is highly frustrating to be thrown out of gdb each time I type Ctrl-C out of habit. Regards, Henrik PS. I noticed that the 'handle' command is disabled in gdb for VxWorks so I guess the chance of making this work is pretty small. - -- Henrik Bergstrom +46-8-628 28 28 (phone) Intertex Data AB Software Engineer +46-8-628 64 14 (fax) Rissnelden 45 Intertex Data AB henrik.bergstrom@intertex.se SE-174 44 SUNDBYBERG SWEDEN From vxwexplo-errs@csg.lbl.gov Mon Feb 7 08:48:29 2000 From: Bob Date: Mon Feb 7 08:48:31 PST 2000 Subject: initialising DosFs I am intending to use a local scsi hard drive connected to a MVME2603. I was hoping to initialise the disk by the calls. dosFsMkfs("/sd0/", pBD) fd = open ("/sd0/", O_WRONLY,0); status = ioctl (fd, FIODISKFORMAT, 0); status = ioctl (fd, FIODISKINIT, 0); And the access files by dosFsDevInit("/sd0/", NULL) fd = open ("/sd0/test.dat", O_CREAT | O_RDWR, 0); etc THe SCSI drive will not be connected to a PC. As the SCSI drive is 9GBytes would it be more sensible to use the rt11 file system. TIA Bob >Hi, > > Are you using a DOS formatted SCSI drive? If so, that explains why >dosFsInit fails. >vxWorks dosFS does not parse the partition table on a dos formatted drive. >So, if your >SCSI drive's system is initialized by formatting it under DOS, then it will >not work. >There are two workarounds. >1. Make a new file system on the drive by using dosMkfs(not >recommended...since the drive > will only be readable under vxWorks) > ....... From vxwexplo-errs@csg.lbl.gov Mon Feb 7 20:50:34 2000 From: "Tyler, Paul A (DBY)" Date: Mon Feb 7 20:50:36 PST 2000 Subject: Links from VxWorks target to Allen-Bradley PLCs > Dear VxWorkers, > > I am looking for drivers to allow target VxWorks systems (pc486 based) to > communicate with Allen-Bradley PLC-5 and SLC-500 series PLCs. I have used > AB6008SV scanners using remote I/O links in the past, but this is limited > to eight words. I would like to be able to use other types of link such as > Ethernet and DH-485 data highway. > > We use RSLinx to communicate with NT hosts, but this is not available for > VxWorks. > > Something which could provide a common API between VxWorks and Windows NT > platforms would be ideal, but then this is not an ideal world... > > All suggestions welcome. > > Thanks > > > VxWorks, Tornado > > Paul Tyler > Measurement Systems Specialist > Electronics & Measurement Systems Dept. > Rolls-Royce plc. > > e-mail Paul.A.Tyler@Rolls-Royce.com > Tel. +44 (1332) 247808 > Fax. +44 (1332) 247928 > > From vxwexplo-errs@csg.lbl.gov Tue Feb 8 02:03:32 2000 From: Johan van der Werf Date: Tue Feb 8 02:03:35 PST 2000 Subject: Re: CPU util measurement? / how to measure 'idle' time? >Leonid wrote: > >Bruce, VxWorks does not have an IDLE task! Instead it has got a special >kernel state for "Idle" so that when the first task becomes ready, >it only has to do "half" a context switch. > This of course is great. Unfortunately these 'half' context switches don't show up in handlers installed with taskSwitchHookAdd(). In our experience (VxWorks 5.3.1/M68K) only 'full' context switches are passed to taskSwitchHookAdd() handlers. If the handler would be called with a NULL pointer for the missing half of the context switch it would be useful to measure idle time (exactly). As the kernelIsIdle flag does not have a hook mechanisme to monitor its state changes we're left with no good way to measure idle time. Apart from the "busy loop at lowest priority" task, which is an effective way of killing the `half' context switches. Any ideas? B.t.w. does anyone know what is the use of the undocumented functions taskSwapHookDelete(), taskSwapHookShow(), taskSwapHookAttach(), taskSwapHookAdd(), taskSwapHookDetach(). Regards, -- |Johan van der Werf | tel: ++31 (0)40 2304280 |$B%O%s%9(B | |ASM Lithography | fax: ++31 (0)40 2302224 |$B%t%!%s(B | |Dept: Basic Software | email: ujw@asml.nl |$B%G%k(B | |Room: 7A3.012 | vdwerf@xs4all.nl |$B%&%(%k%U(B| From vxwexplo-errs@csg.lbl.gov Tue Feb 8 02:13:06 2000 From: Leonid Rosenboim Date: Tue Feb 8 02:13:08 PST 2000 Subject: Re: CPU util measurement? / how to measure 'idle' time? The taskSwapHook* is exactly the machnism you are looking for to handle half context switches (task-> idle and idle->task). They are undocumented for an unknown, and long foregotten reason. Perhaps WRS support will let yuou have some instructions on using it, or you can remain with the idla task approach which is usually good enough. : Leonid Johan van der Werf wrote: > >Leonid wrote: > > > >Bruce, VxWorks does not have an IDLE task! Instead it has got a special > >kernel state for "Idle" so that when the first task becomes ready, > >it only has to do "half" a context switch. > > > > This of course is great. Unfortunately these 'half' context switches > don't show up in handlers installed with taskSwitchHookAdd(). In our > experience (VxWorks 5.3.1/M68K) only 'full' context switches are passed > to taskSwitchHookAdd() handlers. If the handler would be called with a > NULL pointer for the missing half of the context switch it would be > useful to measure idle time (exactly). > > As the kernelIsIdle flag does not have a hook mechanisme to monitor its > state changes we're left with no good way to measure idle time. Apart > from the "busy loop at lowest priority" task, which is an effective way > of killing the `half' context switches. > > Any ideas? > > B.t.w. does anyone know what is the use of the undocumented functions > taskSwapHookDelete(), taskSwapHookShow(), taskSwapHookAttach(), > taskSwapHookAdd(), taskSwapHookDetach(). > > Regards, > > -- > |Johan van der Werf | tel: ++31 (0)40 2304280 |$B%O%s%9(B | > |ASM Lithography | fax: ++31 (0)40 2302224 |$B%t%!%s(B | > |Dept: Basic Software | email: ujw@asml.nl |$B%G%k(B | > |Room: 7A3.012 | vdwerf@xs4all.nl |$B%&%(%k%U(B| From vxwexplo-errs@csg.lbl.gov Tue Feb 8 11:04:18 2000 From: Yaron Rotstein Date: Tue Feb 8 11:04:20 PST 2000 Subject: Debugger attach problem I'd like to thank all the list members who helped me resolve this issue. I believe the problem is in Tornado for SH2. Identical source code and setup worked just fine on other systems. My debugger version simply refused to recognize the stack structure. Fortunately, single stepping through the assembly code eventually returned to my file. No other solution worked (including "run to end of routine", setting up breakpoints, source code path, etc.). Thanks. ----- original message follows ---- I have a problem viewing source code after attaching to my target. Here is the routine my task is pending on: Void WaitHere(void) { int keepWaiting = 1; While (keepWaiting) { taskDelay(100); } } I start the system _without_ the debugger and wait for my task to PEND. Then I start the debugger and attach to pending task. I get a disassembly window of something that looks like system code (as expected since the task is pending). However, I cannot step back on the stack to my code to see who jumped to WaitHere(). In addition, the viewing options (source, assembly, mixed) are disabled (grayed out). This is very strange because I remember using this method in another project, but can't remember the tools version. Here is some additional information: 1) The code is compiled with -g and no optimization. 2) I'm using Tornado 1.0.1 for Hitachi SH and dve7043 BSP 3) Using the Run menu option works just fine. I'm working on this for few days now. I appreciate your help. Thanks. From vxwexplo-errs@csg.lbl.gov Tue Feb 8 19:12:07 2000 From: Nagaraj G Date: Tue Feb 8 19:12:09 PST 2000 Subject: VxWorks: Ping Limitation Problem Hi VxWorkers, I have a few problems with the "ping". When I try ping command with odd number of bytes (for example, 61 or 63 bytes), it does not get the ping response. I do this manipulation from the shell by changing the "_pingTxLen" global variable. If I keep it 62 or 64 or any multiple of 2 it just works without any problem. Why is this? Another problem is, when I ping with "_pingTxLen = 4000", it sends 4000 bytes and receives 4000 (ofcourse 8 bytes less). But when I try with "_pingTxLen=5000", it sends and receives only 4476. Why is this? Another problem is, if I send ping request from a non-vxWorks machine (Windows), with a ping length of 30,000 bytes, vxWorks responds properly. But, if I try with something more, say 35,000 bytes, it just halts. Forget responding, it even freezes the vxWorks itself. I have to do a reboot. The BSP I am using is pcPentium with Tornado II. The strange thing is that WindRiver does not believe this problem. They are not accepting this as a bug. Can anyone please help me in getting a work-around at least, if not a solution? If there is a solution for this it is really great. Thanks in advance, Best Regards, Nagaraj G EnThink Inc www.EnThink.com From vxwexplo-errs@csg.lbl.gov Tue Feb 8 19:18:15 2000 From: Nagaraj G Date: Tue Feb 8 19:18:17 PST 2000 Subject: VxWorks: Memory alignment problem Hi VxWorkers, I have a problem with the memory alignment in VxWorks. I am using pcPentium BSP with Tornado II. I used ping to test my driver. When I use small ping lengths, for example: up to 100 bytes, I don't face any problem. I always get quadlet-aligned address in the MBLK_ID's mData field. But, if I increase the length, I start getting byte-aligned, I mean non-quadlet-aligned, memory address. My hardware does not support this. What should be done for this? I also tried the routing. I placed two cards in a VxWorks machine, one "ene0" and other my hardware. I tried doing a ping from a Windows host on the "ene0" network and tried routing it to the other interface. In this case, I always get the memory non-quadlet-aligned. Any solution for this. I saw a SPR with a similar problem. But, it states on x86 it should just work fine. I am using pcPentium BSP which is a x86 type. Does anyone have a solution or a work-around for this problem? Thanks in advance. Best Regards, Nagaraj G EnThink Inc www.EnThink.com From vxwexplo-errs@csg.lbl.gov Tue Feb 8 22:33:25 2000 From: Nagaraj G Date: Tue Feb 8 22:33:27 PST 2000 Subject: VxWorks: Another problem with Ping Hi VxWorkers, I have another problem with the "ping" in VxWorks. I am using pcPentium BSP with Tornado II. I tried doing a IP Broadcast to the broadcast address with the "ping" command. If I try ping with a "pingTxLen" up to 1480 bytes, it works fine. But, if I try anything more than that, it just fails. It does not even send anything on the LAN, which it does when the length is <=1480 bytes. Does anyone have a solution or a work-around for this problem? Thanks in advance. Best Regards, Nagaraj G EnThink Inc www.EnThink.com From vxwexplo-errs@csg.lbl.gov Wed Feb 9 01:06:39 2000 From: "Sandeep Chakravarthy" Date: Wed Feb 9 01:06:42 PST 2000 This is a multi-part message in MIME format. ------=_NextPart_000_0001_01BF730C.4993FBA0 Content-Type: multipart/alternative; boundary="----=_NextPart_001_0002_01BF730C.4993FBA0" ------=_NextPart_001_0002_01BF730C.4993FBA0 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: 7bit Hi , I would like to subscribe to your mailing list to recieve any mails that are sent on VxWorks . Sandeep ************************************************************** Sandeep Chakravarthy Software Engineer - Global R & D , Wipro Technologies , Embedded Systems Business Unit , 2nd Floor , Sri Chamundi Complex , 26 , Hosur Main Road , Bommanahalli , Bangalore - 560068 Call me at : 91-80-5722293/96 Extn.-5330 Fax : 91-80-5722696 E-mail id : mailto:sandeep.chakravarthy@wipro.com Web site : http://www.wipro.com ************************************************************* ------=_NextPart_001_0002_01BF730C.4993FBA0 Content-Type: text/html; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable
Hi ,=20
      I would like to = subscribe to=20 your mailing list to recieve any mails that are sent on VxWorks .=20
 
Sandeep

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

Sandeep Chakravarthy
Software = Engineer -=20 Global R & D ,
Wipro Technologies ,
Embedded Systems Business = Unit=20 ,
2nd Floor , Sri Chamundi Complex ,
26 , Hosur Main Road=20 ,
Bommanahalli ,
Bangalore - 560068

Call me at : = 91-80-5722293/96=20 Extn.-5330
Fax          = :=20 91-80-5722696
E-mail id  : mailto:sandeep.chakravarthy@wipro.com
Web=20 site  : http://www.wipro.com

*************************************************************

  ------=_NextPart_001_0002_01BF730C.4993FBA0-- ------=_NextPart_000_0001_01BF730C.4993FBA0 Content-Type: application/octet-stream; name="Ivy.gif" Content-Transfer-Encoding: base64 Content-ID: <290161309@09022000-0939> R0lGODlh/wNdAPf/AP///4SEhIyMjJSUlJycnKWlpa2trbW1tb29vcbGxs7OztbW1t7e3ufn5+/v 7/f3987GxtbOzt7W1r21ta2lpbWtrca9vZyUlKWcnMa9td7WztbOxr21rc7Gvefezt7Wxt7Wve/v 5/f37///987OxtbWzt7e1ufn3r29ta2tpbW1rcbGvZSUjJyclKWlnIyMhN7ezufn1u/v3tbWxr29 rcbGtc7OvbW1pf//562tnPf33qWllO/v1t7exufnztbWvc7Otb29pcbGrf//3vf31u/vzufnxt7e vdbWtc7Orf//1u/vxufnvff/zvf/1u/3zufvxt7nvff/3u/31ufvztbevc7WtcbOrd7nxr3Gpff/ 5+/33ufv1tbexs7WvcbOtbW9pa21nO//zt7vvdbntd7nzr3Gre//1uf3ztbnvd7vxufv3sbOvaWt nPf/79bezs7WxrW9ra21pe//3uf31s7evcbWtd7vztbnxr3Ord7n1r3Gtef33sbWvd7v1s7extbn zr3OtbXGrefv5+/378bOxs7Wztbe1t7n3rW9taWtpa21rb3GvYyUjJSclJylnISMhM7ezsbWxrXG tb3OvcbOzq21tZScnMbGzq2lrcDAwAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAACH5BAEAAJoALAAAAAD/A10A QAj/AGmYMUND4BcgVuxc2cMwkJk9NFAgQFCghQ0fT6BQUYMFS5cub94kIEASwYIGmlKqXMmypcuX MGPKnEmzps2bOHPq3Mmzp8+fQIMKHUq0qNGjSJMqXcq0qdOnUKNKVUowzp44cWhgRaGoxQABYAMI 6NEDCZQoUWCs4cKjTBc4IBe5oNGlRxm4f+rAgRNpBpxDChY4eDC1sOHDiBMrXsy4sePHkCNLnky5 MlA5KGhARKGCBgEBYr1EwVKlx48jU7z48NLFRxcsM/T4wFIGBgy/Nmb8gFGmDAkXAw4oQGm5uPHj yJMrX868ufPn0KPbHGgGjBw5i1Sk2H5ATxnTXqr8/wBZhocWPTO8eJnh9sePHh/hrN/b5UsCBIOl 69/Pv7///wAGKOCAAgbwQgCY7FEDG/IF0kcNM9jWQxUiLLhCCSSQsAJooAUQACQE6AEDFx/N0IV7 cCSQAAP5EejiizDGKOOMNNZo42Qx+OADDjx44AEMJpjAAAMw6HECCV/UEMYNYIBhxhVmCAFRIioo UgABGKigggEUuFDAlwYskkAhCUTAAAkLEHbjmmy26eabcMYpZ2VTOEHFFDwUceIRpVUBRRV1ePFH aV3M8EUcTtbwBSM1oOAoCgeg8IUQVwiR5ENm1KCgAg+oOeenoIYq6qikllqjiD3wBoNdM7AxA3sx xP9QYkh7laAAko5qWsMKbKywhyAoMIICAV8NQNIBCAw3WKemNuvss9BGK+20RcH2BRWj+TVDHyEV kkIOZnzhhbgM9WpGIo8yMmVWKiRCwCVgJsuAp9TWa++9+OarL5twwIAFWVV0UQcbLuzgQg5MslFC CSbEICIMJXhhw2444DDHEE44MQUVHPtwh3d6rKGHCQ0Qt+/JKKes8sosJ1aAlSkkUogJM5BAQxhx uEBsIwas4N6/VJTRURVOSEEFGlPcAcgfMOAByBtdlAGIW288fYiQiNDb8tZcd+311y0TkAICAJgQ RpJJ1lBDhnr8UIIBBHg1QAF74GFEFUVA0UUVVeD/UaghZBYwwOAEfFkCHCWUAXVIJWgN9uOQRy75 5DYKYoYdA+Ux0B5CeGHHVXFomYgLixzASJB2lcFFRz/4JUEij8hhRm4wFDqDCS1SrvvuvPfue3NV gVGQlma40KEAA0x4xA91qJEGFGf9O4Wsb/haA1w9eARXJCG9lWEhCiBgrAEJnOT47+inr/767O8k Rw4uPJKCClelQGwLH6m3txFWeEEFD1ywjV9i4C+AvccvcIGDJBjxCAM04Hzti6AEJ0hB9mVKCANh AxCS8AWGfMEG3+kCHjxSBSqQpVAksAFcbMAF77AmVR+BQUhIcB8HVPCGOMyhDsHGIQscAF4pMAAC /xhhKCDIEAby8QsCUsCCFrDAESwYXCPAQkUBEGAFNHuDbeBQCAsIBoI7DKMYx0hGOXXkCdgCGhX4 9hG+1S5IXajUF7IwEAyaISsoWAENhAAEzqnNUSswX6cckLsyGvKQiExkgMRDhSIUwQh1mQEexgOf LnxgBn+Awx9YUwJDfMEOXxjIFfpghzyAUlFf2OShQreI+c0vEchKVgMKqcha2vKWuGRMXbLHt4Dp pmY2YEQFqOgIF7SBSQLZw0MiwoYvCAIrcWjUHlQApgMk4AS0zKU2t8nNbirFD1zAwxv00AUghWQF o5PDDaxghS984Qqh3EMiSGC6A1hCiGdaAQr2EP++Axjgn8gaDrO8SdCCGvSgNfGSAdgwzpqxYQAu kIMNCACcFHwhkl74gQ8C9gMcnEEKU9jCHbBABzrcgQ5U8IFGylCHGcwzTQiNqUxnys0CbMcAKagA ChBXAkWogAN5hMMbhNYRKnjkB2UoIRpGAwWP0KYOAesCa0zEBUBMrXvgmxdNt8rVrt6QBnQUSJSS 1K9LKeghjtJOAeCABY00FT548MEM3kACYxHAml/0ql73ytcICsKdX2gmG9QT2A4yBAU1SMQeUJCI FUCgBFPwwQ+40ANazcBWBnBBChiBoVkOtK+gDa1owSYIOuaBBpZTZiLk8IgWtIAAHhJLARLgIx3/ VKEMGPFBql4FhxTMJRAqRGBIJIGIB4JxtMhNrnJNJbys0CAIBYkDCtrQAg8J4BFA4NN7mNARI/SA DlzgwhuE6q0atC6jvNWWfAzBXgjcBwEMeOBy50vf+rYpK2YIQlb2ayCw2IAMfAtPFZZQBSz4gA49 8AIB9VADMwjKL1Z9QwtrI9XwAdSaCVBAfLNp3w57+MP6cYELblCQKX3mBQIwQhT49gMrPGEGNlij D47Qg9hwAbcU/kgI1WMDwuUVxEAOspClQxA5KCIFbXDUy7bDHthITDx22cIUZEgCOACCsv+qwqtM VCiQsGEiHB6ymMdM5sdY6p0D0QoNtEQDt1RB/z1DM7BRocCGH0gMPlgABBu6AIUY3AFqQv0yfo5b 5kIb+tBLIbEZgLCCGgAhXIP9AhxExBqPRA2Jb5jBBlYglgOUkwtCvQMcTACxINmghoRGtKpXzWqd WJciyHuEChjBCInZYM8nklgKSYACDoXF1wNQgYmqpsW90PAkrU62spdNEwxcaQEmOEABMnGAcb3z CmmY2cJKsAckywFRYBAEB+LQSmIZ60skKYABbKAHPZBgBvKyIbPnTe9WP6GEVOgCFXRTmjLEykha nIEQxJoFStVAMwdQwSL++QMbQAQMN0AWhhWgAEwkwBAa/my9N85xIWtECUNoAhurcAeOfSR7vf95 VSECC4S0EaQqujIDEoAghIIPfLEokG/Hd85zIHvkTj6wjaU/kOAZJBgOfikB0+BAgj6YwQpmsNyu 9oCACdSgD55zp68YIt3FMkLDs+y52McuWo1y4Q5luINU1VM1cgo1EnAJyWXlw5AVrIARvMJ7HRvM WIXj9J/qPoA1A2Ncshv+8DP9s9TKwKcud+EPIXGLekhQCEaMqRIkUIAh2KApNjDk8wpaLCwDfxJ5 I/70qI8pwOCTnkjOwBCUWAELGjGAFqRAeJr6vCCSyRAaJCkOES/clwIa5tQb//i2xAJc8KCHqs3n VSmQnR0896Cz3tFRCEhEArJSAzwe4ALCF2L/xpFP/vLbUnUnF1EJFpCAR+wgB3IQ3uxMAO2FIf2S PcgYxpxAhClsrDwt9AbsF2/mV4AGGEYDYAA/4AclADElUAiP4AJhYAYKE15Cs0ZaxhpSZSdPMAUd 2IEbgwYlhwVosAIGUADmc4AquIIS9CWLsAIkkGlM1wIjpjMCcCAFwBtRo1sbVQVS8AQdKIJeQIL/ Qhsg8RHjVAKH4Fks2IROuDteogg4lQIoYAgLwFgpkB0qgAAMo29thQXvwRocgwVjGBISlj35g4Th 1W4mABgs8oRwGIdc8yWZsAIgsBpIYAOaYnclEAE+4F1ogAQ1RhpHwCcm5zfjZWwL4xchAQNL/9MH egEIIGEI2JRqcniJmNgsVmADV6ApdWQFlZIFJAADOLclKVAAj1AANeADbjEHWJAGXcAGmwQX3mJX BWAJE7EA8fUAItCLg5SJwBiM0bIH1edO0+cFfeBObKAZXCd6CZAIXrAFFzgbvIUhJ2gAKLBhliiM 3NiNoeJ7fRAIg/UgeaAeTsd1WKECB5csvSE0d3AHeIBAJkACp/iC41UCs2R63riP/AgqvvcFDmFH HUQl2bEIcfMZtncCJ6ADWOBdZ8AF+RYShpBZ0WQHJWIiSjgI+tiPHNmRNTIQcWAGWQBNBOECjkAA jjAAYhEAB/ADVXBv+cYD/pdvhaIHqwUG4v+iZXBQB5u0HoZgA0NiMh45lEQZIDQABnEgCBAROhmA AinQAh3iISpwBCtmBIXoB2UAXgTEGm8QBzmwB3BRO2mAdLXjFyRQAhBQCbqoc0XZlm7pHOkYXdGl Ap+xknvDN1bQVGPQVFTABVsgXkiEAnGQG5bGF+P1B2+hSYWwmAnwTxpWfG8ZmZLZGGqGlFlwAyoQ B8ejCFbQS0eQBkRwBG1FBTDgBWsAJIwCIW8BCH7QfLUBAy1FeSNROP8kL4U3mbiZm4vRGQeHR3Fg AB2SAj2wYkhQBUBgBKShb1NQY28QJFVgA+rhBXUAEjPgA33pPTglHLoYX561jbr5neBpFKT/YxVa gRWwBRZVQGMuOYRU4AVl4AUAgwUmsAZBwxu9oWPxYQMHYEUr4p3h+Z8AWhSK0AaalQiJUBDAiTxY sGLrUWDPyTEpVQav8prZkx4mgkA2sAIHQAAO5J8B+qEgyhPWkQME4FPZpwiPQBIf8RF2hgUZ5QM8 4BqHA0Kz0QWAMCGF8gNwUQIzwAgFcABsGaJCOqRAASVemQNtEAcHcIpaMlelmVExVB5bEBIM8gau gaN+wRqZFGgJcAAbSaRgGqY2oR4HB5KCGTpxsCoJZgOrR4ZBYwNdAJ0BQxZwITRl8AcwJpv4IaZ8 2qcyIRBCwE4NRhARQQM2QGo/QxZYoAZU/3BJ5pVCrVEb+qM4taEHh0BDe+qnmrqpmnBmD1ED8GQH irIHo+hmM1YFRiChh8MgkcIG7Rgh9qmEh7AAGpIAX8qpuCqkA3BkGmJeNjApA1FZSSU0KyAxhrAX hZAIKskhB0ACXMAHMaAbLFoFKYIAYZer2CqkHUIRikADDBIIimIDJfAGWAAEbsBjulYAK8khYsEG JFAGa4RUWCCAG1A+QZqt+AqeK4kAmWBF6pYAdmcFQAA1t5FENLQIVYRiVSQWKsAgXDCuSJehlaBV +Vqxukk4GlAClvAlKnBrBwEEl6UHx1ozBNoG8YcdcpACVvIIg2NXhYMChZA4GFIIJgGZFv97s/0Y S/SXeYWwAoYiBIFqBaNGV4wAf0sSkgMnmAv3FbR5JSRBAAhQCM1pCCRgEkKJs1jbkUtgBEYQo2X5 BogQKwrZADPwcGAQBFlAA5UiEGu2JQXgAiqAAhkABr71TwaALAlgAV9nAlZ4q1n7t8I4nBmRPSnl ogHzLz3gMOjxA1dQKQwRJXiEAhxwAO1UppkBKSuSjyVTMh4KuJ6rgnsJhFHAJ1HAAyR0oVpUAmvA ozVgB0IAXVFCEIsFEZ+qKNLkKBSrCZ37ubxbgGrwBBnzBKTRS3hgVL1UYJeFJFcArgvyR4+yT51j KZmSZo2yAlfbu9grhzfWSGMIG5U0V+3/cVlewKPc5itfIFY4NwH71E6GZVgRARGBxITZO79N+B1l MAXfMa+Ztirp8QaQ9wc6+gYL8wZeUI4PsQcdZHeMYAGB0kziiFpnylgSUQiCRL8WXIAm0jH7ViJw waJS1S+F8ge24gULwgZ5xAZ6CEqBkAcfixWLcAPboQiKgC4TkXnye8E4nHowQEA7rCpJVE5loAd7 ATUfUQjHagihtwe0tja0hgKeFwctSzg/miwnkY8al8NYbHj5pqgrKlUwZlS9gUSJuRfHSgKToMR7 UAhs4HmgByk6c24ZxiK7m8V0zGymUYQlkqMTMx6LgiwroMbvZgg1EwjlYgaZIV0FkUfS/6Zu8HWv dfzIZHci32siXiBUXmB3XyEALBBRWKFYg8qMDAEsJaYCKRovmGCzkJzK9VYGeMAaMICneFrJXQBL tRdRz0QDffCrDJF7+6QriXADMDw3w1c+fqvKxlxvT2oXfvAWeHoiJKBO7zQufaBMipKUNIAAKGAB NHBwmaECBzAAF/AlFCAvc3zM5kxmWnSfUdN8bwA4BwMGntNyVyCqBEEljJAIFYCLJhAHjJCZFKFu tUnM5XzOBA1kPIAeb5CxJiDAlGdMwMxOQFAp4NorClAIEsyFDRACbSgkJmArCBBLmTvQBT3S9aVv 7xkwW/SUIqZOA3E7J5BFJVBj7zEEOP+QMWIgBmhkVHggNXrwNLYSXyQd1KrmAgwiiT69AjSYAlnB bjJwY6vDouPxA2iUMVTNf0eDBnSgEUFDArso1F5daIVTCOiBRDMYUV8AAzLAilhAYzNQB3YmHl4w BGhwBnfwBGhw12iwEVAwQncwLATQn18d2GKmbiqgAA0YMWxAgzcgB58BFgZwQj9QhHa2f3QwBVzw B/CKY1jgN1MDErZSzIId2soFJgZgAUJFedF3A8TCAqCBAnszIVSgYnwCvEiTNG4hNH3Tjj7dnPQH 1KL92/PlJTe1AoeAIQShCPFDABTAWYWCqlAASeLhBGj0BO94F4DgiP6rY+oMAycwq4T+BNzgPVpf sh0quwKGTY8GkJlx8G61MRsmpF0vaVSLqgYjpDgyxMGYTQdl4AeA0IZWeL3hHeBbZVPeDCEXqiGe pzA9UAQdAQV/SBqVxDF4MOGFcmsfDBJ/IImJOANPEwllYIVXLOAijlCFY1H/Mim70igokEImFAVI MAOOsAJH4AUjVEJrrWVlAHdsgHELsAIqwi0fTB+pAgOHANojfuTb5ChxQAhmsAgDYbukinSQcoIp Wjg1cJdo0BHhMYtcZAAtW+LkExgOsAaHEBKIgORoflBPMnCUsmju5AWGgAID90eM4E9v2wFadAer 0xHx0QWFADeFgywbNgJpXugzFRAAOw== ------=_NextPart_000_0001_01BF730C.4993FBA0-- From vxwexplo-errs@csg.lbl.gov Wed Feb 9 02:22:26 2000 From: David Laight Date: Wed Feb 9 02:22:28 PST 2000 Subject: Re: VxWorks: Ping Limitation Problem > Subject: VxWorks: Ping Limitation Problem > > Submitted-by: Nagaraj G > > Hi VxWorkers, > > I have a few problems with the "ping". > > When I try ping command with odd number of bytes (for example, 61 or 63 > bytes), it does not get the ping response. I do this manipulation from > the shell by changing the "_pingTxLen" global variable. If I keep it 62 > or 64 or any multiple of 2 it just works without any problem. Why is > this? My system behaves the same way. The problem is that the data checksum of the transmitted ping packet is incorrect. The first byte is ok, the second wrong, I suspect - but have not checked - that it is including the byte after the end of the packet into the sum. > > Another problem is, when I ping with "_pingTxLen = 4000", it sends 4000 > bytes and receives 4000 (ofcourse 8 bytes less). But when I try with > "_pingTxLen=5000", it sends and receives only 4476. Why is this? This looks like a restriction of the shell, it trucates _pingTxLen to 0x1000 limiting the packet size to 4076 (not 5576). I suspect that is the size of the buffer used. > > Another problem is, if I send ping request from a non-vxWorks machine > (Windows), with a ping length of 30,000 bytes, vxWorks responds > properly. But, if I try with something more, say 35,000 bytes, it just > halts. Forget responding, it even freezes the vxWorks itself. I have to > do a reboot. That seems like a problem with your ethernet hardware/device driver. To correctly respond to a 35k ping the ethernet card must pick up 23 back to back packets. If one is missed the rest are held by IP awaiting reassembly, the system then runs out of recieve buffers very quickly. > > The BSP I am using is pcPentium with Tornado II. Which ethernet card/driver - it is important but I don't know about any of them! > > The strange thing is that WindRiver does not believe this problem. They > are not accepting this as a bug. The odd length packet problem MIGHT be more pervasive as ping calls the same cksum routine as the rest of the IP stack. However I haven't seem any other problems. > > Can anyone please help me in getting a work-around at least, if not a > solution? If there is a solution for this it is really great. I don't see why ping problems are a big issue - unless you are seeing the same problems elsewhere and this is just an easy way to reproduce the same fault. > > Thanks in advance, > > Best Regards, > > Nagaraj G > EnThink Inc > www.EnThink.com > David ---------------------------------------------------------------- David Laight email: dsl@tadpole.co.uk Tadpole Technology plc phone: +44 1223 428 232 Cambridge, UK fax: +44 1223 428 201 From vxwexplo-errs@csg.lbl.gov Wed Feb 9 02:28:42 2000 From: David Laight Date: Wed Feb 9 02:28:44 PST 2000 Subject: Re: VxWorks: Memory alignment problem > Subject: VxWorks: Memory alignment problem > > Submitted-by: Nagaraj G > > Hi VxWorkers, > > I have a problem with the memory alignment in VxWorks. I am using > pcPentium BSP with Tornado II. > > I used ping to test my driver. When I use small ping lengths, for > example: up to 100 bytes, I don't face any problem. I always get > quadlet-aligned address in the MBLK_ID's mData field. But, if I increase > the length, I start getting byte-aligned, I mean non-quadlet-aligned, > memory address. My hardware does not support this. What should be done > for this? Fix your driver. You cannot assume any fragmentation/alignment rules for transmitted packets. It is perfectly reasonable for a 1500 byte packet to be in 1500 message blocks. I have seen VERY fragmented packets on Unix Streams, vxWorks not generate them as often - indeed your system may never generate really bad cases - but they can happen. > > Any solution for this. I saw a SPR with a similar problem. But, it > states on x86 it should just work fine. I am using pcPentium BSP which > is a x86 type. The 'similar' problem is entirely different. Any relates to the fact that many other cpus can't do misaligned memory accesses (for a very good reason - the complications of bus faults on the second memory cycle on what should be an atomic action...). > > Does anyone have a solution or a work-around for this problem Fix your driver! > > Thanks in advance. > > Best Regards, > > Nagaraj G > EnThink Inc > www.EnThink.com > David ---------------------------------------------------------------- David Laight email: dsl@tadpole.co.uk Tadpole Technology plc phone: +44 1223 428 232 Cambridge, UK fax: +44 1223 428 201 From vxwexplo-errs@csg.lbl.gov Wed Feb 9 02:32:59 2000 From: David Laight Date: Wed Feb 9 02:33:01 PST 2000 Subject: Re: VxWorks: Another problem with Ping > Subject: VxWorks: Another problem with Ping > > Submitted-by: Nagaraj G > > Hi VxWorkers, > > I have another problem with the "ping" in VxWorks. I am using pcPentium > BSP with Tornado II. > > I tried doing a IP Broadcast to the broadcast address with the "ping" > command. If I try ping with a "pingTxLen" up to 1480 bytes, it works > fine. But, if I try anything more than that, it just fails. It does not > even send anything on the LAN, which it does when the length is <=1480 > bytes. I suspect, but don't have the relevant RFC to hand, that IP broadcasts cannot be fragmented (transmitted as multiple packets). Thus the maximum size of an IP broadcast depends on the MTU of the network - 1500 for ethernet. The RFCs only require that the MTU be >= 576 - limiting the size of IP broadcast that can be sent. > Does anyone have a solution or a work-around for this problem? > Not a problem! > Thanks in advance. > > Best Regards, > > Nagaraj G > EnThink Inc > www.EnThink.com > David ---------------------------------------------------------------- David Laight email: dsl@tadpole.co.uk Tadpole Technology plc phone: +44 1223 428 232 Cambridge, UK fax: +44 1223 428 201 From vxwexplo-errs@csg.lbl.gov Wed Feb 9 04:03:10 2000 From: Vxworks Exploder Date: Wed Feb 9 04:03:12 PST 2000 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Wed Feb 9 04:03:06 PST 2000 Subject: Re: nanosleep problem Subject: Re: iterate over all files in a directory Subject: Embedded developers using VxWorks, Linux, NT or CE wanted Subject: quick question about dhcp function in T2 Subject: Ctrl-X doesn't work Subject: Forcing an application to stop and return to the debugger. Subject: Re: initialising DosFs Subject: Re: dosFs2.0 Code to mount unknown number of partitions? Subject: Bootp Question Subject: Re: MPC603e Assembly Subject: Re: Bootp Question Subject: 16 bit A/D VME board recommendations Subject: Re: Links from VxWorks target to Allen-Bradley PLCs Subject: Re: Bootp Question Subject: Re: Byte Alignment Subject: dhcpcLibInit and DHCPC_TIMEOUT_OFFER Subject: What does "chksum: out of data" mean Subject: Re: boot parameters Subject: Re: How to handle FP exceptions on PowerPC/VxWorks Subject: Re: Using PMC cards with Extended VME Memory Model Subject: Fast serial driver Subject: Fast serial driver ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: nanosleep problem Date: Tue, 08 Feb 2000 16:57:03 +0100 From: mario sangalli Organization: Ye 'Enter S.R.L. Message-ID: <38A03CCF.8976792D@txt.it> References: In effect the 'nanosleep' has the same resolution of a system tick. Jeff Byrd wrote: > Has anyone used the nanosleep function? We seem to get a much longer time > than the value we pass to the nanosleep function. > > Any clues? > > Thanks, Jeff --------------------------- Newsgroups: comp.os.vxworks Subject: Re: iterate over all files in a directory Date: Sat, 05 Feb 2000 10:49:26 -0500 From: Joe Durusau Message-ID: <389C4686.4E825FB5@bellsouth.net> References: <87bigt$okp$1@newsmaster01.magnet.at> <389BD584.D06C3CD7@mindspring.com> If your system is hosted on unix, the find utility can do this sort of thing. There are some cases of unix tools beign ported to windows. Look, for instance, for djgpp and its cousins. It's just possible that someone has done a port of find to windows, but I don't know of any personally. Speaking only for myself, Joe Durusau Al Johnston wrote: > > The "is it a file or is it a directory" problem is one of the most > asinine defects in vxworks. There does not seem to be any > decent approach to determine this for all file systems (net, > nfs,dos)... the best I came up with was to a long (300 lines > of c code) algorithm which include using > > .... to determine > if ".../foo" is a directory see if you can create ".../foo/bar" > as a normal file. If you can do this then foo is a directory. > > Anyway.. I have an algo'tm to do this... and it seems to work.. > but i junked it because it was just too much of a kludge... > If you want I can dig it out for you... WRS's responce > when I ask them how to this was basicly that it was not > possable. > > luck! --------------------------- Newsgroups: comp.os.vxworks Subject: Embedded developers using VxWorks, Linux, NT or CE wanted Date: Tue, 8 Feb 2000 09:31:34 -0800 From: "Nick Berliner" Message-ID: GoAhead Software is offering a chance to beta test remote upgrading capability in your networked appliance or device. GoAhead FieldUpgrader provides manufacturers of networked devices with a comprehensive solution for automatically upgrading their products in the field. GoAhead FieldUpgrader allows you to upgrade software, firmware, O/Ss or data using a device-initiated approach. These upgrades can provide products with new features, reliability enhancements, performance improvements, updated applications, security patches and bug fixes. GoAhead FieldUpgrader removes the end-customer burden of manually upgrading their product, and reduces the manufacturer's expenditures for costly support visits and product recalls. If you are developing in an NT or Linux environment and your device runs Embedded NT, Embedded Linux, VxWorks, CE, NT, or Linux, you may qualify for participation. If you are interested in trying GoAhead FieldUpgrader out in your device, visit: http://www.goahead.com/fieldupgrader/fubeta.htm or email beta@goahead.com. GoAhead FieldUpgrader beta program will run through February 2000. Please sign up right away if you are interested in participating!! Best regards, Nick Berliner ______________________________________________________ Nick Berliner Product Manager/GoAhead Software nickb@goahead.com www.goahead.com phone 425-468-5404 --------------------------- Newsgroups: comp.os.vxworks Subject: quick question about dhcp function in T2 Date: Tue, 8 Feb 2000 11:40:59 -0600 From: "Tim Berger" Organization: Motorola Message-ID: <87pkc7$q912@nntp.cig.mot.com> The function dhcpcBind() has a boolean argument "syncFlag" that is supposed to specify synchronous or asynchronous execution. Conveniently, the documentation is not clear on which boolean value is which. I don't have the time to mess around with it yet, so would anybody know off-hand what boolean value synchronous corresponds to? Thanks! --------------------------- Newsgroups: comp.os.vxworks Subject: Ctrl-X doesn't work Date: Tue, 08 Feb 2000 10:31:57 -0800 From: mreed Organization: http://www.remarq.com: The World's Usenet/Discussions Start Here Message-ID: <032967d8.7c727136@usw-ex0108-061.remarq.com> Does anyone know how you turn on the rebooting of VxWorks by using CTRL-X? Is there some part of the operating system that has to be included/excluded? Thanks, Mike Reed * Sent from AltaVista http://www.altavista.com Where you can also find related Web Pages, Images, Audios, Videos, News, and Shopping. Smart is Beautiful --------------------------- Newsgroups: comp.os.vxworks Subject: Forcing an application to stop and return to the debugger. Date: Tue, 8 Feb 2000 21:05:59 +0200 From: "Boaz Kahana" Organization: RADLAN Message-ID: <38a06964.0@news.barak.net.il> Reply-To: "Boaz Kahana" Working with Tornado Cross debugger to a PowerPC platform: I want to force the code to return to the debugger, to simulate an abort in real application. What is the assembly line that will cause that ? Is that some kind of a trap command ? --------------------------- Newsgroups: comp.os.vxworks Subject: Re: initialising DosFs Date: 8 Feb 2000 19:26:54 GMT From: johan.borkhuis@peektraffic.nl Organization: XS4ALL Internet BV Message-ID: <87pqlu$d1$1@news1.xs4all.nl> References: <87pffu$guf2@overload.lbl.gov> Bob wrote: : THe SCSI drive will not be connected to a PC. As the SCSI drive is 9GBytes : would it be more sensible to use the rt11 file system. First try to get the new DosFS2.0 from VxWorks. This has a lot of extras. Groeten, Johan - -- o o o o o o o . . . ___________________________________ o _____ || Johan Borkhuis | .][__n_n_|DD[ ====_____ | johan@borksoft.xs4all.nl | >(________|__|_[_________]_|________________________________| _/oo OOOOO oo` ooo ooo 'o!o!o o!o!o` === VxWorks page: http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html === --------------------------- Newsgroups: comp.os.vxworks Subject: Re: dosFs2.0 Code to mount unknown number of partitions? Date: Tue, 08 Feb 2000 13:26:16 -0800 From: Todd Organization: http://www.remarq.com: The World's Usenet/Discussions Start Here Message-ID: <04d80ac4.e4d8ed52@usw-ex0103-024.remarq.com> References: <87aets$a8c$1@news.netmar.com> I was fighting that exact same problem. It is easy to just create the maximum number of devices, but not very elegant. Who wants a bunch of inaccessible devices in the device list... I looked at the code in usrFdiskPartShow routine. It parses the MBR and finds what partitions exist on the disk and prints out a bunch of info on each one. If the type is non-zero, the partition must be valid and the device can be added to the device list. I just copied a section of that code and usd it to validate the partitions on the disk. Hope this helps. Todd * Sent from RemarQ http://www.remarq.com The Internet's Discussion Network * The fastest and easiest way to search and participate in Usenet - Free! --------------------------- Newsgroups: comp.os.vxworks Subject: Bootp Question Date: 8 Feb 2000 22:09:00 GMT From: pbennett@volta.ee.calpoly.edu (Paul Curtis Bennett) Organization: Cal Poly, San Luis Obispo Message-ID: <87q45s$hev$1@cscnews.csc.calpoly.edu> I need to include some vendor specific information in the bootp bootrequest message (the message originated from the bootp client that asks for service). Can anyone tell me an easy way of doing that? (I don't have access to the VxWorks sources). The idea is to include a model specific tag, so as a particular unit boots, the server can send it a "generic" bootfile name specific to that model. That way, we don't have to keep track of which unit has which MAC address. I realize I'll have to write my own server... Thanks for any help! (Note that I'm posting this from my school account, but if you wish to email me, use the address below, despammed of course...) - -- Paul Bennett Software Engineer Applied Signal Technology pcb@.appsig.com.whitehouse.gov (I'm not the president, so you work out the address) --------------------------- Newsgroups: comp.os.vxworks Subject: Re: MPC603e Assembly Date: Fri, 04 Feb 2000 21:13:51 GMT From: John Fusco Organization: Deja.com - Before you buy. Message-ID: <87ffe9$ppl$1@nnrp1.deja.com> References: <4.1.20000203131603.00b196c0@mill.cisco.com> This one got me wondering because I know I have put code in ROM and it works, but I wasn't sure how, so I checked. VxWorks links the non-resident bootrom (e.g. bootrom_uncmp) to start at RAM_HIGH_ADRS. So all global symbols in bootrom_uncmp are located in RAM, even though the text and data resides initially in ROM. See for yourself, type "nmppc bootrom_uncmp". Code that runs from ROM can call other code in ROM because all subroutine calls are done using relative branching. So even though the linker thinks your code resides at RAM_HIGH_ADRS+something, subroutine calls are relative, not absolute. Since the text images are identical (i.e. ROM and RAM), the relative branch is the same in both cases. Since you are loading a symbol via HIADJ, this is going to resolve to an absolulte address, which this is always going to point to RAM. I am not sure what the "right" way to work around this is, but one trick I have seen in romInit.s is to put the data in the middle of your assembly code like so, bl skip mydata: .long a, b, c, d skip: /* Link register now contains a pointer to "mydata" * copy it to another register and you can use it. */ mfspr r21,LR I used to wonder why they did this, but now it makes sense. In article <4.1.20000203131603.00b196c0@mill.cisco.com>, Dennis Daniels wrote: > I'm relatively new to VxWorks and am adding some low-level, power-on diagnostics (in MPC603e assembly, GNU ToolKit) to be called from romInit.s. In my assembly file I have added some tables as part of the .text section and I have also added initialized arrays in a *.c module (declared as consts) which is linked into the boot image. When I attempt to load the address of any of any of these tables (using the HIADJ and LO macros) I get an address located in RAM vs. the actual table location in ROM. As a check > I put a branch to a table label and the address is correctly resolved. I've looked through the docs, but haven't found an answer so far. Can anyone shed some light on this? > Thanks in advance for any help. > > Dennis > > VxWorks, Tornado. > Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Bootp Question Date: 8 Feb 2000 22:42:27 GMT From: pbennett@volta.ee.calpoly.edu (Paul Curtis Bennett) Organization: Cal Poly, San Luis Obispo Message-ID: <87q64j$ib5$1@cscnews.csc.calpoly.edu> References: <87q45s$hev$1@cscnews.csc.calpoly.edu> I found the bootpParamsGet() and bootpMsgSend() functions, but am I mistaken in believing that to utilize these at boot time, I'd have to modify (ie, burn new) roms? Yuck. Suggestions welcome! Paul Curtis Bennett (pbennett@volta.ee.calpoly.edu) wrote: : I need to include some vendor specific information : in the bootp bootrequest message (the message originated : from the bootp client that asks for service). Can : anyone tell me an easy way of doing that? : (I don't have access to the VxWorks sources). --------------------------- Newsgroups: comp.dsp,comp.os.vxworks,comp.realtime,comp.arch.bus.vmebus Subject: 16 bit A/D VME board recommendations Date: 9 Feb 2000 00:43:13 GMT From: dima@chaparral.phys.unm.edu (Dmitri A. Sergatskov) Organization: University of New Mexico, Albuquerque Keywords: VxWorks,VME,A/D Message-ID: I am looking for 16-bit A/D and D/A (12-bit D/A is OK) board for VME bus with support for VxWorks. May be a dumb board or DSP based. So far I looked at RADSTONE and ACROMAG boards. I have not dealt with those companies before, so if anyone has any comments and other recommendations I would appreciate it very much. Sincerely, Dmitri. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Links from VxWorks target to Allen-Bradley PLCs Date: Wed, 09 Feb 2000 02:48:42 GMT From: "H1" Organization: AT&T Worldnet Message-ID: References: <87pffs$gue1@overload.lbl.gov> I'f seen somthing for Linux. Maybe there is the sourcecode incuded. You can search on www.freshmeat.net Tyler, Paul A (DBY) wrote in message news:87pffs$gue1@overload.lbl.gov... > > Dear VxWorkers, > > > > I am looking for drivers to allow target VxWorks systems (pc486 based) to > > communicate with Allen-Bradley PLC-5 and SLC-500 series PLCs. I have used > > AB6008SV scanners using remote I/O links in the past, but this is limited > > to eight words. I would like to be able to use other types of link such as > > Ethernet and DH-485 data highway. > > > > We use RSLinx to communicate with NT hosts, but this is not available for > > VxWorks. > > > > Something which could provide a common API between VxWorks and Windows NT > > platforms would be ideal, but then this is not an ideal world... > > > > All suggestions welcome. > > > > Thanks > > > > > > VxWorks, Tornado > > > > Paul Tyler > > Measurement Systems Specialist > > Electronics & Measurement Systems Dept. > > Rolls-Royce plc. > > > > e-mail Paul.A.Tyler@Rolls-Royce.com > > Tel. +44 (1332) 247808 > > Fax. +44 (1332) 247928 > > > > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Bootp Question Date: Tue, 08 Feb 2000 18:14:12 -0500 From: Thomas Davis Organization: Interpath Communications, Inc. Message-ID: <38A0A344.6B12AFE9@galois.net> References: <87q45s$hev$1@cscnews.csc.calpoly.edu> Paul Curtis Bennett wrote: > > I need to include some vendor specific information > in the bootp bootrequest message (the message originated > from the bootp client that asks for service). Can > anyone tell me an easy way of doing that? > (I don't have access to the VxWorks sources). > > The idea is to include a model specific tag, so as a > particular unit boots, the server can send it a "generic" > bootfile name specific to that model. That way, we don't > have to keep track of which unit has which MAC address. > I realize I'll have to write my own server... > > Thanks for any help! > > (Note that I'm posting this from my school account, but > if you wish to email me, use the address below, despammed of course...) > > -- > Paul Bennett > Software Engineer > Applied Signal Technology > pcb@.appsig.com.whitehouse.gov > (I'm not the president, so you work out the address) I have been there, and mostly I just don't remember but I do know that you are on the right track. Yes you will use, bootpMsgSend() and yes you will have to update the BIOS/ROM if you are booting over the network. I also remember that I had to change the BSP code for config/all/bootConfig.c to accomodate it. Lastly, I remember that it will probably require two BOOTP's on at the BIOS and once after the kernel is loaded to setup the interface. I did not use the vendor specific tags but I did utilize more fields that the default provides. I have read two RFC's one for BOOTP and one for the Vendor Specific Options (sorry I do not remember the numbers). Seems bad but it is pretty easy. It is just hacking through the Wind River boot code that is the real pain in the rear. You will not need to reimplement the BOOTP protocol as bootpMsgSend() is a good client, you will not use bootpParamsGet as it does not do what you want and happily you will not need to write a bootp server as the standard public server (bootpd) is fully capable of vendor options. Of course if you use a Microsoft OS for the server (silly idea) you might need to do more work as Windows is anything but a server OS. Hope this helps, Thomas Davis - -- Galois, Inc. http://www.galois.net We color the world! --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Byte Alignment Date: Mon, 7 Feb 2000 07:45:10 +0200 From: =?iso-8859-1?Q?Morn=E9_Hartman?= Message-ID: <389e5bd4$0$35288@helios.is.co.za> References: <389ade41$0$35286@helios.is.co.za> Thanks for the replies and help... To answer the questions : I am using a Intel Pentium II 400 on the NT box and a Intel Pentium II 266 on the VxWorks VME Board. I am using the Borland C++ Builder 3 compiler for the NT Box and the Wind River Tornado 2 complier for the VxWorks board. I will try the byte ordering techniques mensioned, thanks Morné Morné Hartman wrote in message news:389ade41$0$35286@helios.is.co.za... > Hi all, > > Does anyone here know anything about byte alignment. > > I have an application that communicates between a VxWorks board and an > NT Box. I am sending a message from a structure to the remote system. > The message contains various data types including Char, Double, Float, > Int etc. In my application in NT I use byte alignment so that the > message is received in the correct format on the remote end. I believe > that I need to use byte alignment on the VxWorks side aswell, but this > did not work > > I initialise byte alignment as follows : > > #pragma options -a1 > struct MyStruct .... > #pragma options -a4 > > If I communicate between two NT boxes there is no problem. When I try > the same thing between an NT and VxWorks box the received message is > not identical to the sent message. > > Does anyone have a solution to this problem > > Morné Hartman > mhartman@grintek.com > > > --------------------------- Newsgroups: comp.os.vxworks Subject: dhcpcLibInit and DHCPC_TIMEOUT_OFFER Date: Wed, 09 Feb 2000 08:00:39 GMT From: nunya Organization: Nunya Business Message-ID: <38A11EA0.3EFA78E@nunya.com> I'm trying to modify the offer timeout for the DHCP client code. If my device is started on a network without a DHCP server, and it is set to use the DHCP client, it can take over a minute for the DHCP client code to return failure. I've tried modifying the value for the DHCPC_TIMEOUT_OFFER input to dhcpcLibInit(), but that does not seem to help. Anything else I can tweak to get this timing lower? TIA JLM (I'm really j l m @ d e s s e r t f i r s t . c o m, no spam, please...) --------------------------- Newsgroups: comp.os.vxworks Subject: What does "chksum: out of data" mean Date: Mon, 7 Feb 2000 09:37:51 -0000 From: "Eddie Parker" Message-ID: <949915463.20318.0.nnrp-13.9e98192c@news.demon.co.uk> Hi, I'm running VxWorks 5.3.1 on x86 platforms. I have 3 Ethernet boards in each node on the network. I am using a mixture of UDP and TCP. I have recently started getting the message "chksum: out of data" from the kernel and this is often followed by a page fault from my network handling task. What does this message mean and where is it created? Thanks in advance for any help. Eddie Parker --------------------------- Newsgroups: comp.os.vxworks Subject: Re: boot parameters Date: Wed, 09 Feb 2000 15:53:30 +0800 From: Wilson Lo Organization: Motorola Semiconductor Products Sector Message-ID: <38A11CFA.2C7D7B44@email.sps.mot.com> References: <87ejee$7g9@overload.lbl.gov> Reply-To: Wilson.Lo@motorola.com Hi, U may use the HyperTerminal to input the parameter through the serial port. Before setting up the connection, U need to setup to 9600bps. The program is supported by the Windows95/NT. : ) Wilson mmenge@dspt.com wrote: > Mark Menge@DSPT > 02/04/2000 08:10 AM > Hi there vxworkers. Has anyone implemented a MS Windows program that allows > the user to input his boot parameters into a dialog and programs the > vxWorks boot parameters via the serial port? > > tornado vxworks - -- ======================================== Wilson Lo Software Application Engineer Asia Pacific Operation Networking and Communications Systems Division Motorola Semiconductor Products Sector ======================================== --------------------------- Newsgroups: comp.sys.powerpc.tech,comp.os.vxworks Subject: Re: How to handle FP exceptions on PowerPC/VxWorks Date: Wed, 09 Feb 2000 10:34:57 GMT From: "Paul, Nam" Organization: Dacom Message-ID: References: <86i0pb$m0s$1@nnrp1.deja.com> <86ii7a$qqc$1@user2.teleport.com> I'd like to write an FP exception handler for an 8260 running Custom OS. Can you give me the real sample code for that? How to clear sticky bits of FPSCR ? Thanks for your advance. Andrew Klossner ÀÌ(°¡) <86ii7a$qqc$1@user2.teleport.com> ¸Þ½ÃÁö¿¡¼­ ÀÛ¼ºÇÏ¿´½À´Ï´Ù... >I wrote an FP exception handler for an 8240 running VxWorks. I found >that it's necessary to AND the FPSCR register contents with 0x0007f0ff >so that the FP exception doesn't fire all over again when the next FP >instruction is executed. You don't have to disable exceptions, just >clear the outstanding reports (the status and "sticky" bits.) > >Also, when you enter the exception handler, SRR0 (the exception-time >program counter) will point to the instruction that caused the >exception. You need to add 4 to it before returning, unless your >version of VxWorks does that for you. > > -=- Andrew Klossner (andrew@teleport.com) --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Using PMC cards with Extended VME Memory Model Date: Tue, 08 Feb 2000 09:12:59 GMT From: rplavery@my-deja.com Organization: Deja.com - Before you buy. Message-ID: <87ommo$5ah$1@nnrp1.deja.com> References: <389F1E96.6836863E@boeing.com> In article <389F1E96.6836863E@boeing.com>, kirk.d.ellett@boeing.com wrote: > Hello, > > I am in the unenviable position of integrating several different PMC > cards for use (not all at the same time) on a Motorola MV2308 board. > Since the Motorola board isn't PCI aware, which sucks, I must manually > configure the PCI registers and base addresses for each card. I had > successfully done this using the Pseudo-PREP memory model, by writing > PCI memory addresses to the base address registers and using CPU > addresses that looked like this. > > 0xC0000000 | (PCI address) > > I now have to integrate the PMC cards using an Extended VME memory > model. Since my PMC cards fit fine in a 16MB space, I kept the same PCI > addresses that I used on Pseudo-Prep, and the CPU addresses became... > > 0xFD000000 | (PCI address) > > This was an utter failure for memory reads and writes. I have tried a > number of different combinations of CPU addresses vs. PCI addresses, but > none of them work. Can anyone give me or point me to a short rundown of > how to map PMC cards to CPU memory using Extended VME model? > I have been through this wonderous experience recently - it is possible and I am still learning. Dont forget that the PCI I/O space changes from 0xfe000000 to 0x80000000 when you go from PReP to Extended VME. Also, as well as the Processor's memory map changing, the PCI memory map also changes. By using pciDeviceShow and pciHeaderShow you can see which devices are on the PCI and where the are. I suggest you do this with your PReP and Extended VME kernels. As usual the documentation never really spells things out for you. However the windsurf site does have some stuff if you search for the right things - I suggest you look at WTN 47 and the target.txt file for your BSP. Also, I think you should be using pciDevConfig, and that your PCI addresses will have to change. hope this helps a bit. Richard Lavery Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Fast serial driver Date: Wed, 9 Feb 2000 13:22:50 +0200 From: "Tomer Baz" Organization: Terayon Communication Systems Message-ID: <87riap$ku6$1@news2.inter.net.il> Hi group, I want to download files using the XModem protocol over a serial line in my application. VxWorks' serial driver supports only 38,400 baud but I need a driver that can reach 115,200. Does anyone know where to find such driver? Thanks, Tomer. --------------------------- Newsgroups: comp.os.vxworks Subject: Fast serial driver Date: Wed, 9 Feb 2000 13:26:50 +0200 From: "Tomer Baz" Organization: Terayon Communication Systems Message-ID: <87rii9$l4a$1@news2.inter.net.il> Hi group, My application needs to download files using the XModem protocol over a serial line. VxWorks' serial driver supports only 38,400 baud (actually it works only up to 19,200 according to my experience), but I need a driver that can reach 115,200. Does anyone know where I can find such driver? Thanks, Tomer Baz - Software Engineer Terayon Communication Systems Integrated DSL Systems BU (Formerly Radwiz Ltd.) Tel: +972-3-7658971 Fax:+972-3-6458517 Email: tomerb@radwiz.co.il http://www.radwiz.com http://www.terayon.com --------------------------- End of New-News digest ********************** From vxwexplo-errs@csg.lbl.gov Wed Feb 9 07:45:20 2000 From: stevens Date: Wed Feb 9 07:45:22 PST 2000 Subject: MPC850 SMC is slowly Hi VxWorkers, we use a custom BSP with a MPC850 and Tornado2. A serial device (19.200 baud) realized with the Serial Management Controller needs 0.7 ms to send a byte. The SMC is in idle. The time is measured from setting the ready-bit in the buffer descriptor to the occurance of the byte on the serial line. How can we get a faster reaction ? Regards, Torsten -- ************************************************************************** fraunhofer institute of microelectronic circuits and systems department sat name torsten stevens mailto:stevens@ims.fhg.de address finkenstrasse 61 phone +0049/203/3783-238 47057 duisburg fax +0049/203/3783-266 germany http://www.fhg.de/depts/ims-du-e.html ************************************************************************** From vxwexplo-errs@csg.lbl.gov Wed Feb 9 07:58:31 2000 From: Dennis Nicklaus Date: Wed Feb 9 07:58:33 PST 2000 Subject: MVME2400 FLASH programming from VxWorks Hello, Can anyone tell me how to writet to the FLASH memory on a motorola 2400 series Power PC board, from within vxWorks. I'm able write to the FLASH from moto's ROMBUG and pflash with no problem. But when I am running vxWorks, I don't know how to write to it. I have motorola 2300 boards and can program the flash in them with no problems, but the 2400 uses different Flash memory chips. I am able to read the FLASH, apparently, since d 0xff000100 shows me the Wind River Copyright stuff that comes at the start of the vxWorks kernel which is stored there. thanks for any help or clues! Dennis Nicklaus nicklaus@fnal.gov From vxwexplo-errs@csg.lbl.gov Wed Feb 9 10:35:51 2000 From: "Larry Sinn" Date: Wed Feb 9 10:35:53 PST 2000 Subject: Re: initialising DosFs --------------10B7539E703691C977475BC2 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Bob: If my memory serves me and I don't show my age too much, I think DECs (now Compaq) RT-11 could only handle disks upto 256Mbytes (8 logical partitions and 32Mbytes each). I don't know if WRS has extended this. Also RT-11 can't be mounted from your host like DOS can. I agree with Johan that DosFs2.0 is the way to go but you need T2 (a major pain in upgrading from T1). The documentation says you can format it under DOS and use it on VxWorks, and the largest single file you can have is 2Gbytes. Larry. Bob wrote: : THe SCSI drive will not be connected to a PC. As the SCSI drive is 9GBytes : would it be more sensible to use the rt11 file system. johan@borksoft.xs4all.nl wrote: First try to get the new DosFS2.0 from VxWorks. This has a lot of extras. -- Larry Sinn Spectral Dynamics (408)474-1746 voice 1983 Concourse Drive (408)474-1780 fax San Jose, Ca. 95131-1708 sinnl@sd-star.com --------------10B7539E703691C977475BC2 Content-Type: text/html; charset=us-ascii Content-Transfer-Encoding: 7bit Bob:
If my memory serves me and I don't show my age too much, I think DECs (now Compaq) RT-11 could only handle disks upto 256Mbytes (8 logical partitions and 32Mbytes each). I don't know if WRS has extended this. Also RT-11 can't be mounted from your host like DOS can.
I agree with Johan that DosFs2.0 is the way to go but you need T2 (a major pain in upgrading from T1). The documentation says you can format it under DOS and use it on VxWorks, and the largest single file you can have is 2Gbytes.

Larry.
 

Bob <Bob@dskti.com> wrote:
: THe SCSI drive will not be connected to a PC.  As the SCSI drive is 9GBytes
: would it be more sensible to use the rt11 file system.

johan@borksoft.xs4all.nl wrote:
First try to get the new DosFS2.0 from VxWorks. This has a lot of extras.
  

-- 
Larry Sinn
Spectral Dynamics               (408)474-1746  voice
1983 Concourse Drive            (408)474-1780  fax
San Jose, Ca. 95131-1708        sinnl@sd-star.com
  --------------10B7539E703691C977475BC2-- From vxwexplo-errs@csg.lbl.gov Wed Feb 9 11:14:35 2000 From: Andrew Booth Date: Wed Feb 9 11:14:37 PST 2000 Subject: ethernet PMC cards for VxWorks Hi, I want to use an ethernet PMC card on a Motorola MVME2307. Has anyone come across a good supplier? Andrew Booth From vxwexplo-errs@csg.lbl.gov Wed Feb 9 12:52:15 2000 From: John_W_Cosgrove@res.raytheon.com Date: Wed Feb 9 12:52:17 PST 2000 Subject: Re: ethernet PMC cards for VxWorks vxWorks vxWorks vxWorks Andrew, I won't comment about good/bad, but a supplier we use for our PMC ethernet's is RAMix. The card we are using is the PMC664 and they have both MUX and non-MUX drivers. I am using the non-MUX driver with only a little startup problems. The card supports 10/100 auto/manual negotiation and such. It can be configured as a boot device also... John Cosgrove Raytheon Electronic Systems (401)842-4167 From vxwexplo-errs@csg.lbl.gov Fri Feb 11 04:03:15 2000 From: Vxworks Exploder Date: Fri Feb 11 04:03:18 PST 2000 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Fri Feb 11 04:03:11 PST 2000 Subject: Printer driver questions Subject: Re: 16 bit A/D VME board recommendations Subject: Re: Printer driver questions Subject: Re: Fast serial driver Subject: Re: builds in T2 Subject: Re: malloc headers? Subject: Re: 16 bit A/D VME board recommendations Subject: Re: quick question about dhcp function in T2 Subject: Re: Ctrl-X doesn't work Subject: Visual Studio Integration with T2 Subject: Standalone Boot Problem Subject: Re: Power Arch. interrupt question Subject: Re: quick question about dhcp function in T2 Subject: Re: Maximum Number of Open File Descriptors Subject: Using PMC cards with Extended VME Memory Model Subject: Employment Opportunities Subject: Re: Can't reboot w/o floppy Subject: etherOutput returns exception Subject: VxWorks vs. Linux Subject: VxWorks and Streaming Video Subject: How to setup 5 serial port(16C554) in x86 Board(K6-266+16MB RAM) Subject: Re: Increasing telnet Security Subject: Re: iterate over all files in a directory Subject: How to utilize more than 5 serial port in x86 Board Subject: Re: iterate over all files in a directory Subject: Need custom board 68360 BSP developed Subject: Re: Target shell fails to recognise commands Subject: ncr810 SCSI driver on MCP750 - question Subject: symbolTableSize Subject: Re: Ctrl-X doesn't work Subject: PowerPC860 Board and VME bus Subject: FAQ for this newsgroup? Subject: Re: FAQ for this newsgroup? Subject: intro to vx-works Subject: Re: intro to vx-works Subject: VLAN (IEEE 802.1q) Subject: Re: intro to vx-works Subject: Re: VxWorks vs. Linux Subject: Re: Ctrl-X doesn't work Subject: Re: Standalone Boot Problem Subject: benchmarks for VxWorks on x86 Subject: Re: VxWorks vs. Linux Subject: Re: Target shell fails to recognise commands ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Printer driver questions Date: Wed, 9 Feb 2000 13:23:14 +0100 From: "Serge Wenger" Organization: MATISA SA Message-ID: <87rm6b$7b3$1@pollux.ip-plus.net> Reply-To: "Serge Wenger" I need to print graphics files (PCL or PS) to a TCP/IP networks HP LaserJet Printer. Has somebody already done something like this. How to transform GDI like command to PCL? Which file transfer protocol must I use? Which Eternet Port must I use? Help on one or many point will be appreciated - -- /Serge Wenger In my e-mail address, replace nobody by matisa --------------------------- Newsgroups: comp.dsp,comp.os.vxworks,comp.realtime,comp.arch.bus.vmebus Subject: Re: 16 bit A/D VME board recommendations Date: Wed, 09 Feb 2000 08:49:23 -0500 From: Ted Marz Organization: The Software Engineering Institute Message-ID: <38A17063.82C282BC@sei.cmu.edu> References: In my previous position, we used a lot of VMIC A/D, D/A and discrete I/O products, and they worked very well (mostly 12 bit, but they have 16 bit products available). Very easy to interface, fairly reliable. Look under I/O products on http://www.vmic.com Ted Marz --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Printer driver questions Date: Wed, 09 Feb 2000 07:42:12 -0600 From: Frank Minich Organization: NCP Message-ID: <38A16EB4.785E@ti.com> References: <87rm6b$7b3$1@pollux.ip-plus.net> Reply-To: fminich@ti.com Serge Wenger wrote: > I need to print graphics files (PCL or PS) to a TCP/IP networks HP > LaserJet Printer. Has somebody already done something like this. I assume you are trying to initiate the printing from a VxWorks machine. Two approaches: implement lpr, or implement smb/netbios/tcp. The first approach is much easier, if the printer implements lpd. See RFC 1179. > How to transform GDI like command to PCL? Get a copy of HP's PCL spec, then write the code. Too bad you can't use the unidriver in the Windows DDK. > Which file transfer protocol must I use? I would use lpr if I were you. > Which Eternet Port must I use? As specified in RFC 1179. - ------------------------ Good luck, Frank. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Fast serial driver Date: Wed, 09 Feb 2000 09:03:16 -0500 From: Jeff Daly Organization: Sanders A Lockheed Martin Company Message-ID: <38A173A4.708353DF@lmco.com> References: <87riap$ku6$1@news2.inter.net.il> Reply-To: jeffrey.j.daly@lmco.com assuming your device can handle it, change the driver yourself (should be part of your bsp, or call your local FAE) Tomer Baz wrote: > Hi group, > > I want to download files using the XModem protocol over a serial line in my > application. VxWorks' serial driver supports only 38,400 baud but I need a > driver that can reach 115,200. Does anyone know where to find such driver? > > Thanks, > > Tomer. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: builds in T2 Date: Mon, 7 Feb 2000 15:58:37 +0100 From: "Achim Zimmer" Organization: Colt Telecom GmbH Message-ID: <87mmhs$r8k$1@crusher.de.colt.net> References: <87dupr$egk$1@crusher.de.colt.net> <87fa98$pc5$2@news1.xs4all.nl> It is not the easiest way, but it seems to work. Other solutions would be appreciated.. Thanks schrieb in im Newsbeitrag: 87fa98$pc5$2@news1.xs4all.nl... > Achim Zimmer wrote: > : I am desperate. > : I try to change the build properties of a project called A (downloadable > : appl) in workspace 1 so, that it > : 1. invoke a build of a project B and C in the same workspace before its own > : build > : 2. link them all together at the end > > : Remember, that the builds are placed in subdirectories of the project > : directories. > > : Are there any possibilities in Tornado2 ??? > > Take a look at my VxWorks page. There is an example descibing this. > > Groeten, > Johan > > -- > o o o o o o o . . . ___________________________________ > o _____ || Johan Borkhuis | > .][__n_n_|DD[ ====_____ | johan@borksoft.xs4all.nl | > >(________|__|_[_________]_|________________________________| > _/oo OOOOO oo` ooo ooo 'o!o!o o!o!o` > === VxWorks page: http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html === --------------------------- Newsgroups: comp.os.vxworks Subject: Re: malloc headers? Date: Wed, 09 Feb 2000 08:43:17 -0500 From: Jeff Daly Organization: Sanders A Lockheed Martin Company Message-ID: <38A16EF5.70E279C0@lmco.com> References: <87pbon$4a4$1@newsmaster01.nextra.at> Reply-To: jeffrey.j.daly@lmco.com h/private/memPartLibP.h the header for each allocated block contains a pointer to the previous header, a size (LSB indicates whether block is free) word and possible pad bytes, depending on the value of _ALLOC_ALIGN_SIZE for your architecture. Mario Semo wrote: > Hello, > > is the header for malloc somewhere documented? > has it someone analyzed? > i just want it for debug/analysis purposes. > > what i have seen so far: > > a) pointers decreases (so the first allocated pointer has the highest value) > b) when i look at the 4 bytes before pointer, i see something related to the > size of the data i have allocated. > e.g. i allocate 1 or 2 .. 8 bytes the value there is 8 > e.g. i allocate 16bytes, the value is 12 (0xc) > e.g. i allocate 32bytes, the value is 20 (0x14) > c) when i look at the 4 bytes before the size, this is a pointer. ALL the > values changes, when i allocate a new pointer. > > any ideas? > > -- > regards, Mario Semo. http://www.kirchnersoft.com --------------------------- Newsgroups: comp.dsp,comp.os.vxworks,comp.realtime,comp.arch.bus.vmebus Subject: Re: 16 bit A/D VME board recommendations Date: Wed, 9 Feb 2000 15:34:47 -0000 From: "Ralph Weir" Message-ID: <950110331.22358.0.nnrp-10.c2de8e24@news.demon.co.uk> References: I'd propose one of our boards. They are modular - start with a VME carrier, then add ADC and DAC modules, plus maybe a DSP to control it. VxWorks drivers are available, along with most other common OSs. What you need to define before going further is speed and number of channels. For example - we've got 12-bit ADCs up to 40MHz today (watch this space for >100MHz 12-bit...), but you don't get huge speeds and huge channel counts at the same time! Email or post some details, and I'll put together a system proposal for you. Best regards, Ralph Weir Hunt Engineering http://www.hunteng.co.uk Dmitri A. Sergatskov wrote in message ... >I am looking for 16-bit A/D and D/A (12-bit D/A is OK) board >for VME bus with support for VxWorks. >May be a dumb board or DSP based. So far I looked at >RADSTONE and ACROMAG boards. I have not dealt with those companies >before, so if anyone has any comments and other recommendations I >would appreciate it very much. > >Sincerely, > >Dmitri. > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: quick question about dhcp function in T2 Date: Wed, 09 Feb 2000 19:15:42 +0300 From: "Arthur V. Grigorian" Organization: Auriga, Inc. Message-ID: <38A192AE.CB922101@auriga.ru> References: <87pkc7$q912@nntp.cig.mot.com> Reply-To: archy@auriga.ru Hi, > The function dhcpcBind() has a boolean argument "syncFlag" that is supposed > to specify synchronous or asynchronous execution. > > Conveniently, the documentation is not clear on which boolean value is > which. > > I don't have the time to mess around with it yet, so would anybody know > off-hand what boolean value synchronous corresponds to? Hmm that's strange, but the documentation was clear even for me with my rather poor english. In the asynchronous mode dhcpBind() returns control immediately, so you have to catch its result by a hook function which can be added by dhcpcEventHookAdd(). In this hook function you can use the dhcpcParamsGet() call to save the result. In the synchronous mode dhcpBind() returns control only after it receives a result (the delay is more than a minute), so you'll probably have to design your DHCP client code as a separate task to avoid such a long delay in your code. You can contact me if my description isn't clear enough. Regards, __ R2 --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Ctrl-X doesn't work Date: Wed, 09 Feb 2000 18:05:14 GMT From: Ted Dennison Organization: Deja.com - Before you buy. Message-ID: <87sa8l$mi6$1@nnrp1.deja.com> References: <032967d8.7c727136@usw-ex0108-061.remarq.com> In article <032967d8.7c727136@usw-ex0108-061.remarq.com>, mreed wrote: > Does anyone know how you turn on the rebooting of VxWorks > by using CTRL-X? Is there some part of the operating > system that has to be included/excluded? That's a function of the target shell, so I'm pretty sure you need that included. - -- T.E.D. http://www.telepath.com/~dennison/Ted/TED.html Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Visual Studio Integration with T2 Date: Wed, 09 Feb 2000 18:22:34 GMT From: "Peter Luscher" Organization: @Home Network Canada Message-ID: Hi, We would like to integrate Visual Studio to be the editor that T2 uses. (Tools --> options --> External Editor). The editor comes up but we are unable to double click on a build error output line and launch Visual Studio so that it opens the file with the cursor set to the correct line (ie. the line with the error). Does anyone know what the parameters are to Visual Studio so that we can open the file and automatically go the a given line number? (Note: Tornado 2 will supply the $lineno macro, we just need to know how to pass it to Visual Studio) Thanks! Peter. --------------------------- Newsgroups: comp.os.vxworks Subject: Standalone Boot Problem Date: 09 Feb 2000 18:33:48 GMT From: bagdazian@aol.com (BAGDAZIAN) Organization: AOL http://www.aol.com Message-ID: <20000209133348.01018.00000686@ng-cg1.aol.com> I need some help related to a problem I am having. Here is the description of the problem. I have two Cetia Vmpc4a SBC's running vxworks and communicating via backplane networking. I have been successful in configuring my vxWorks image to support the master board booting from a PC host via an ethernet connection. I have also been successful in configuring the second card to boot via the ethernet using the first processor board as a gateway to the ethernet. I now need to create standalone vxWorks images to allow booting from a pcmcia flash memory card mounted on the first processor card. I have been able to successfully build the vxWorks.st image using the same configuation files as I used for the host support image. I also included STANDALONE_NET in the appropriate include files. HOWEVER, when I boot the system using this standalone vxWorks.st image, the backplane networking does not startup and the second card is unable to connect to the first card. I get an error message on the first card that says something to the effect that "pcmcia network device not recognized" at the point that the system is attempting to bring up the backplane network. How do I get the backplane networking facility to startup in the same way that it did when I used the lnPci driver as the boot device ? Thanks in advance. Rich Bagdazian bagdazian@aol.com --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Power Arch. interrupt question Date: 9 Feb 2000 19:38:28 GMT From: johan.borkhuis@peektraffic.nl Organization: XS4ALL Internet BV Message-ID: <87sfnk$ji$1@news1.xs4all.nl> References: <20000208161858.01177.00000043@ng-ft1.aol.com> MaggieJohn wrote: : The power architecture has several interrupts: system reset, machine check, : data storage, instruction storage, alignment, program, external, FP unavailable : and supervisor call. : What does vxWorks do if one of these interrupts occurs? Suspend the task? : Write a message? You can connect a routine to a interrupt using intConnect. This function will be called when the interrupt occures. : If it tries to write a message to a UART, does it matter if the UART is powered : off? Depends on your implementation of the device driver. If your device driver checks the UART to see if a byte has been written it will detect the failure, otherwise it will just write it. Groeten, Johan - -- o o o o o o o . . . ___________________________________ o _____ || Johan Borkhuis | .][__n_n_|DD[ ====_____ | johan@borksoft.xs4all.nl | >(________|__|_[_________]_|________________________________| _/oo OOOOO oo` ooo ooo 'o!o!o o!o!o` === VxWorks page: http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html === --------------------------- Newsgroups: comp.os.vxworks Subject: Re: quick question about dhcp function in T2 Date: Wed, 9 Feb 2000 14:27:03 -0600 From: "Tim Berger" Organization: Motorola Message-ID: <87sifh$j697@nntp.cig.mot.com> References: <87pkc7$q912@nntp.cig.mot.com> <38A192AE.CB922101@auriga.ru> Arthur V. Grigorian wrote in message news:38A192AE.CB922101@auriga.ru... > > The function dhcpcBind() has a boolean argument "syncFlag" that is supposed > > to specify synchronous or asynchronous execution. > > > > Conveniently, the documentation is not clear on which boolean value is > > which. > > > > I don't have the time to mess around with it yet, so would anybody know > > off-hand what boolean value synchronous corresponds to? > > Hmm that's strange, but the documentation was clear even for me with my rather > poor english. > > In the asynchronous mode dhcpBind() returns control immediately, so you have to > catch its result by a hook function which can be added by dhcpcEventHookAdd(). > In this hook function you can use the dhcpcParamsGet() call to save the result. > > In the synchronous mode dhcpBind() returns control only after it receives a > result (the delay is more than a minute), so you'll probably have to design > your DHCP client code as a separate task to avoid such a long delay in your > code. No, no, I meant whether TRUE corresponded to synchronous and FALSE to asynchronous (or the other way around). syncFlag=true does indeed mean synchronous... It is blatantly obvious from the name "syncFlag," I guess. I realized that just after I posted the original question. Please forgive any other dumb questions I may ask in the future :) Tim --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Maximum Number of Open File Descriptors Date: Mon, 07 Feb 2000 13:35:45 -0600 From: Peter Marqui Organization: Verio Message-ID: <389F1E91.413897D8@rauland.com> References: <874u13$n833@overload.lbl.gov> <38964583.996F5558@aps.anl.gov> - --------------467C787B2A9555DB50DBD5BE Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Increasing NUM_FILES to 100, only got us to about 79 file descriptors then we had to change to following #defines as well /*Changed SJP 2/1/00 all entries were 64 and then bumbed to 128*/ /*Since the NUM_FILES in configAll.h was changed to 256 we had to increase the size of the clusters that are used when opening sockets*/ #define NUM_SYS_64 256 /* 64 byte system clusters */ #define NUM_SYS_128 256 /* 128 byte system clusters */ #define NUM_SYS_256 256 /* 256 byte system clusters */ #define NUM_SYS_512 256 /* 512 byte system clusters */ Andrew Johnson wrote: > "Blake, Don" wrote: > > > > What is the maximum number of open file descriptors in VxWorks? Is it > > configurable? > > Yes, see config/all/configAll.h: > #define NUM_FILES 50 /* max 50 files open simultaneously */ > > - Andrew > -- > Complexity comes for free, Simplicity you have to work for. - --------------467C787B2A9555DB50DBD5BE Content-Type: text/html; charset=us-ascii Content-Transfer-Encoding: 7bit Increasing  NUM_FILES to 100, only got us to about 79  file descriptors
then we had to change to following #defines as well
 
/*Changed SJP 2/1/00 all entries were 64 and then bumbed to 128*/
/*Since the NUM_FILES in configAll.h was changed to 256 we had to increase the size of the clusters that are used when opening sockets*/
#define NUM_SYS_64   256 /* 64 byte system clusters */
#define NUM_SYS_128   256 /* 128 byte system clusters */
#define NUM_SYS_256   256 /* 256 byte system clusters */
#define NUM_SYS_512   256 /* 512 byte system clusters */

Andrew Johnson wrote:

"Blake, Don" wrote:
>
> What is the maximum number of open file descriptors in VxWorks?  Is it
> configurable?

Yes, see config/all/configAll.h:
        #define NUM_FILES 50 /* max 50 files open simultaneously */

- Andrew
--
Complexity comes for free, Simplicity you have to work for.

  - --------------467C787B2A9555DB50DBD5BE-- --------------------------- Newsgroups: comp.os.vxworks Subject: Using PMC cards with Extended VME Memory Model Date: Mon, 7 Feb 2000 19:35:50 GMT From: Kirk Ellett Organization: The Boeing Co. Message-ID: <389F1E96.6836863E@boeing.com> Reply-To: kirk.d.ellett@boeing.com Sender: nntp@news.boeing.com (Boeing NNTP News Access) Hello, I am in the unenviable position of integrating several different PMC cards for use (not all at the same time) on a Motorola MV2308 board. Since the Motorola board isn't PCI aware, which sucks, I must manually configure the PCI registers and base addresses for each card. I had successfully done this using the Pseudo-PREP memory model, by writing PCI memory addresses to the base address registers and using CPU addresses that looked like this. 0xC0000000 | (PCI address) I now have to integrate the PMC cards using an Extended VME memory model. Since my PMC cards fit fine in a 16MB space, I kept the same PCI addresses that I used on Pseudo-Prep, and the CPU addresses became... 0xFD000000 | (PCI address) This was an utter failure for memory reads and writes. I have tried a number of different combinations of CPU addresses vs. PCI addresses, but none of them work. Can anyone give me or point me to a short rundown of how to map PMC cards to CPU memory using Extended VME model? Kirk Ellett --------------------------- Newsgroups: comp.os.vxworks Subject: Employment Opportunities Date: Mon, 7 Feb 2000 14:19:29 -0800 From: "Yury" Message-ID: Luxcom provides fiber optic communications equipment for Metropolitan Area Networks. We are now developing the next generation high-speed data/voice networking equipment for public as well as private MANs. This will enable both new and incumbent service providers to offer new LAN services, premium quality Internet services, and more cost- effective traditional access services. This is an opportunity to get in on the ground floor as a member of a new R&D development team. Luxcom is currently developing the next-generation fiber optic, multi-service access solution employing Packet over SONET technology. The Series 3000 MetroStackT will enable both new and incumbent service providers to offer new LAN services, premium quality Internet services, and more cost-effective traditional access services. Luxcom offers equity and profit participation, competitive salaries and benefits, and excellent opportunities for growth. Qualified individuals are invited to contact us at (510) 770-3300. You may mail, fax or e-mail your resume in confidence to: Luxcom, Inc., 3249 Laurelview Court, Fremont, CA 94538, FAX # (510) 770-3399, e-mail to resume@luxcom.com. http://www.luxcom.com/ Title: Senior Software/Firmware Engineer Description: Experienced senior software/firmware engineer with a good understanding of real-time multi-tasking OS architectures to develop real-time embedded software/firmware for our new Integrated SONET/SDH Router. Requirements: · BS/MS in CS/EE with 5+ years related experience in embedded communication software. · Experience with complex software architectures such as device drivers, ISRs, multi-tasking OS and kernels, mailboxes, semaphores, memory management is required. · Experience with development of packet processing software/firmware for bridges, switches and/or routers is highly desirable. · Experience in Motorola 603/604 processors and architecture is required. · Experience in C/C++ and assembly languages is required. · Practical knowledge of data networks, principles of digital communication and understanding of T1/E1, SONET/SDH and ATM systems are highly desirable. · Familiarity with applicable Bellcore, ANSI and ITU standards for wide area network interfaces is a plus. · Experience in low-level communication protocols such as HDLC, BCP, PPP is a plus. · Must be experienced with structured methods of software development. Title: Senior Software Engineer/System Architect Description: Experienced senior software engineer/system architect, with a good understanding of computer networks and IP routing, to design and develop networking real-time embedded software for our new Integrated SONET/SDH Router. Requirements: · BS/MS in CS/EE with 5+ years related experience in embedded communication software. · A strong understanding of computer networks, TCP/IP, IP routing, OSPF, BGP protocols, etc. as well as their implementation. · A minimum of 5 years experience developing networking products (switches, routers, adapter cards) a must. · Must also be familiar with multi-tasking real-time operating systems. Knowledge of VxWorks, pSOS, Nucleus is a plus. · Must be experienced with structured methods of software development and be capable of preparing specifications, product functional and operational documents and related written materials. Title: Software/Firmware Engineer (Senior - OK) Description: Experienced senior software/firmware engineer with a good understanding of real-time multi-tasking OS architectures to develop real-time embedded software/firmware for our new Integrated SONET/SDH Router. Requirements: · BS/MS in CS/EE with 3+ years related experience in embedded communication software. · Experience with complex software architectures such as device drivers, ISRs, real-time multi-tasking OS and kernels required. · Experience in Motorola PPC processors and architecture is required. · Experience in C/C++ and Motorola assembly languages is required. · Practical knowledge of data networks, principles of digital communication and understanding of T1/E1, SONET/SDH and ATM systems are highly desirable. · Experience in low-level communication protocols such as HDLC, LAPD, BCP, PPP and ML-PPP, TCP/IP and OSI is a plus. · Must be experienced with structured methods of software development. Title: Embedded Network Management Software Engineer Description: Embedded Network Management Software Engineer with a good understanding of Network Management concepts to develop embedded management software for our new Packet over SONET multi-service access product. Requirements: · BS/MS in CS/EE with 3+ years related experience in embedded communication software. · Experience with Embedded Management Systems such as SNMP, TL1 is required, CORBA - preferred. · Strong C/C++ programming experience is required · Experience with real-time OS such as VxWorks, pSos, Nucleus is required. · Team work experience required. · Practical knowledge of data networks, principles of digital communication and understanding of the T1/E1, Ethernet, SONET/SDH and ATM systems is highly desirable. Title: Software Engineer Description: Software engineer to develop real-time embedded software for our new Integrated SONET/SDH Router. Requirements: BS/MS in CS/EE with 3+ years industry experience. · Experience in programming for real-time embedded systems in C/C++ languages is required, · Must also be familiar with finite state machines and multi-tasking real-time operating systems. Knowledge of VxWorks and/or pSOS is a plus. · Practical knowledge of data networks, principles of digital communication and understanding of the T1/E1, Ethernet, SONET/SDH and ATM systems is required. · Familiarity with applicable Bellcore, ANSI and ITU standards for wide area network interfaces desired. · Must be experienced with structured methods of software development and be capable of preparing requirement specifications, product functional descriptions and related written materials. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Can't reboot w/o floppy Date: Thu, 10 Feb 2000 01:04:18 GMT From: snaphook@southwesternbell.net (Bruce) Organization: SBC Internet Services Message-ID: <38a20e20.1483623@news.swbell.net> References: <87sget$rm2$1@nnrp1.deja.com> In comp.os.vxworks Ted Dennison wrote: >We are using the PC486 BSP on a more-or-less standard PC. We'd like >to be able to initiate a reboot from software under certian >circumstances. What we have found is that if we try to perform a >"reboot" w/o a floppy in the floppy drive, we get the following error >message: > dosFsDevInit failed > usrFdConfig failed > value = -1 ....... > >The only way to reboot seems to be to hit the reset button or the power >switch. But if there is a boot floppy in the floppy drive, it works >(even though we don't boot that way). > >Does anyone have an idea how to fix this? What is your default boot device in your bios? There is also a global that determines the type of reboot. Look for something like SYS_WARM_BOOT. How are you rebooting it anyway? Using the keyboard controller or triple faulting it? --------------------------- Newsgroups: comp.os.vxworks Subject: etherOutput returns exception Date: Mon, 07 Feb 2000 23:15:03 GMT From: Van Hoang Organization: Deja.com - Before you buy. Message-ID: <87njli$c2a$1@nnrp1.deja.com> Since I upgraded Radstone Single Board Computer model PPC1A-604e to PPC4-750, my program ran into exception at "etherOutput" function call to write a packet to "lnPci0" interface. What could be the cause or solution? Any help will be greatly appreciated. Thanks and regards, Van Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: VxWorks vs. Linux Date: Thu, 10 Feb 2000 10:46:05 +0900 From: "ChaeHo You" Organization: Korea Telecom Message-ID: <87t4tl$55u$1@news2.kornet.net> Reply-To: "ChaeHo You" Hello, VxWorks users... I'm concerned about Linux development. And I'm interested about the price of VxWorks. Our company sales the Dy4 systems' products. That company makes the VME board for military/commercial version. And the main RTOS of the boards is VxWorks. That company determined a joining a Linux company,AMIRIX. If RTLinux would be ported to Dy4 boards, there will be many interesting changes in RTOS market. I think the price of the VxWorks is too high. The price of Tornado II per 1 user is about 50000 us dollars, isn't it? The problem is the capability of Linux. The developing speed of Linux is amazing. Maybe in a near future, RTLinux will compete VxWorks. Best Regards... --------------------------- Newsgroups: comp.os.vxworks Subject: VxWorks and Streaming Video Date: Mon, 7 Feb 2000 21:20:27 -0800 From: "Robin Kletke" Message-ID: <87o990$s68$1@q.seanet.com> Does anyone know if VxWorks web server will support live video streams from an mpeg encoder? I need to be able to view a live camera feed from a VxWorks system. Thanks. Robin Kletke Superior Software, Inc. rkletke@seanet.com --------------------------- Newsgroups: comp.os.vxworks Subject: How to setup 5 serial port(16C554) in x86 Board(K6-266+16MB RAM) Date: Tue, 8 Feb 2000 14:56:38 +0900 From: "VxWorker" Organization: Korea Telecom Message-ID: <87ob77$bg9$1@news2.kornet.net> Hi ! I'm using a board HS4500 Ver1.3(Boser company http://www.boser.com.tw) This board has a 4 built-in serial ports which have a 16C550 compatibility. Then I added a new PC-104 board which have 4 serial ports using TI chip (TL16C554FN). Then I modified my source like following ######### 1. change no. of serial port in pc.h as 5 //before #define N_UART_CHANNELS 2 //before #define N_UART_CHANNELS 5 ######## 2. enter information of newly added serial ports // before /* serial ports (COM1,COM2) */ #define COM1_BASE_ADR 0x3f8 #define COM2_BASE_ADR 0x2f8 #define COM1_INT_LVL 0x04 #define COM2_INT_LVL 0x03 #define UART_REG_ADDR_INTERVAL 1 /* address diff of adjacent regs. */ //after /* serial ports (COM1,COM2 through 5,6) */ #define COM1_BASE_ADR 0x3f8 #define COM2_BASE_ADR 0x2f8 #define COM1_INT_LVL 0x04 #define COM2_INT_LVL 0x03 #define UART_REG_ADDR_INTERVAL 1 /* address diff of adjacent regs. */ #define COM3_BASE_ADR 0x3e8 #define COM4_BASE_ADR 0x2e8 #define COM3_INT_LVL 0x04 #define COM4_INT_LVL 0x03 /* IRQ jumbers in KCI's Host board should be checked before changing following lines */ #define COM5_BASE_ADR 0x100 #define COM5_INT_LVL 0x07 /* I undefined LPT and also disabled LPT in BIOS SETUP */ ####### 3. add COM3_INT_VEC,,, macros in ..\pcpentium\config.h //before #define COM1_INT_VEC (INT_VEC_GET (COM1_INT_LVL)) #define COM2_INT_VEC (INT_VEC_GET (COM2_INT_LVL)) //after #else /* SYMMETRIC_IO_MODE */ ...... #define COM1_INT_VEC (INT_VEC_GET (COM1_INT_LVL)) #define COM2_INT_VEC (INT_VEC_GET (COM2_INT_LVL)) #define COM3_INT_VEC (INT_VEC_GET (COM3_INT_LVL)) #define COM4_INT_VEC (INT_VEC_GET (COM4_INT_LVL)) #define COM5_INT_VEC (INT_VEC_GET (COM5_INT_LVL)) #define RTC_INT_VEC (INT_VEC_GET (RTC_INT_LVL)) ####### 4. modify ...\pcPentium\sysSerial.c // before static I8250_CHAN i8250Chan[N_UART_CHANNELS]; static I8250_CHAN_PARAS devParas[] = { {COM1_INT_VEC,COM1_BASE_ADR,UART_REG_ADDR_INTERVAL,COM1_INT_LVL}, {COM2_INT_VEC,COM2_BASE_ADR,UART_REG_ADDR_INTERVAL,COM2_INT_LVL} }; // after static I8250_CHAN i8250Chan[N_UART_CHANNELS]; static I8250_CHAN_PARAS devParas[] = { {COM1_INT_VEC,COM1_BASE_ADR,UART_REG_ADDR_INTERVAL,COM1_INT_LVL}, {COM2_INT_VEC,COM2_BASE_ADR,UART_REG_ADDR_INTERVAL,COM2_INT_LVL}, {COM3_INT_VEC,COM3_BASE_ADR,UART_REG_ADDR_INTERVAL,COM3_INT_LVL}, {COM4_INT_VEC,COM4_BASE_ADR,UART_REG_ADDR_INTERVAL,COM4_INT_LVL}, {COM5_INT_VEC,COM5_BASE_ADR,UART_REG_ADDR_INTERVAL,COM5_INT_LVL} }; ###### 5. build bootrom_uncmp and vxWorks files each. In Tornado shell in my host computer, I can see 5 serial ports is installed using devs command. But 5 th port does not work when I used following command. - ->fd=open("/tyCo/4", 2) - ->write(fd,"Hello", 5) - ->close(fd) Pls, help me Thanks in advance. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Increasing telnet Security Date: Tue, 08 Feb 2000 08:39:32 +0200 From: Leonid Rosenboim Organization: BitBand Technologies Ltd. http://www.bitband.com Message-ID: <389FBA24.89F47265@bitband.com> References: <389F08F1.267D155F@americasm01.nt.com> Reply-To: leonid@bitband.com I'll answer your question with another question: How much more secure do you want this to be ? You can replace the systems loginLib with your own that maintains userid/password pairs in any way you like, but the fact will remain that the password as well as the entire session is going to travel the network in clear text. In security terms, you must clearly define what security risks you need to protect for, in order of importance, but also the "price" you are willing to pay for each of these levels of protection. Of course a reasonably good solution would be to have SSH server implemented on VxWorks, but I dont know of anyone who done that (yet). Leonid Tim Sohacki wrote: > Folks, > > For lab purposes, I have a static username/password for telnet > access to my targets, only to prevent completely accidental > access. For customer deployment, this is perceived as weak > security (correctly, I would say). I have "secure" access via > a serial port, but for remote debug purposes, I'd prefer full > (and fast) access via telnet. > > Any thoughts? Have you increased telnet security? > > Thanks, > Tim. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: iterate over all files in a directory Date: Mon, 7 Feb 2000 09:10:25 +0100 From: "Mario Semo" Organization: Nextra Telekom GmbH Message-ID: <87odfm$kof$1@newsmaster01.nextra.at> References: <87bigt$okp$1@newsmaster01.magnet.at> <389BD584.D06C3CD7@mindspring.com> <389C4686.4E825FB5@bellsouth.net> <389CD13A.DE5B9B6E@aracnet.com> Hello, first i am confused. as my developer told me, he run into the problem as he used openDir. from my ´first jmail: >when we call opendir() an output is generated by the system to the console: > .>.. > opendir... >testdir: not a plain file. >... >Help says: > WARNING > For remote file systems mounted over netDrv, opendir( ) fails, > because the netDrv implementation strategy does not provide a > way to distinguish directories from plain files. To permit use > of opendir( ) on remote files, use NFS rather than netDrv. regarding what we built: we have a IEC1131 based development system (e.g. for graphical languages) and we generate C code which requires a runtime system. we have this runtime system in C++. and we have it ported to a number of Operating systems. now we port it to vxWorks, since some of our customers want support vxWorks as target too. - -- >I believe he means from within vxWorks. Using stat() on a NFS >drive in vxWorks reports directories as files (it also reports >files as files.) The "feature" is documented by Wind River; >why their NFS client cannot tell the difference, I don't know. > >The way I check (for NFS) is try to do a opendir() on a questionable >file; opendir() DOES fail for things that ain't directories. Don't >know about net filesystems; dosfs doesn't have the stat() problem >to begin with. > >What are you trying to build? You might be better off using Linux >or Lynx or QNX or some other full-fledged UNIX than using vxWorks. > --------------------------- Newsgroups: comp.os.vxworks Subject: How to utilize more than 5 serial port in x86 Board Date: Tue, 8 Feb 2000 15:50:43 +0900 From: "vxWorker" Organization: Korea Telecom Message-ID: <87oecj$fjf$1@news2.kornet.net> References: <10gp79m$7u$2@fe2.cs.interbusiness.it> <38947605.C1D25D1D@inxpress.net> Hi ! I'm using a board HS4500 Ver1.3(Boser company http://www.boser.com.tw) This board has a 4 built-in serial ports which have a 16C550 compatibility. Then I added a new PC-104 board which have 4 serial ports using TI chip (TL16C554FN). Then I modified my source like following ######### 1. change no. of serial port in pc.h as 5 //before #define N_UART_CHANNELS 2 //before #define N_UART_CHANNELS 5 ######## 2. enter information of newly added serial ports // before /* serial ports (COM1,COM2) */ #define COM1_BASE_ADR 0x3f8 #define COM2_BASE_ADR 0x2f8 #define COM1_INT_LVL 0x04 #define COM2_INT_LVL 0x03 #define UART_REG_ADDR_INTERVAL 1 /* address diff of adjacent regs. */ //after /* serial ports (COM1,COM2 through 5,6) */ #define COM1_BASE_ADR 0x3f8 #define COM2_BASE_ADR 0x2f8 #define COM1_INT_LVL 0x04 #define COM2_INT_LVL 0x03 #define UART_REG_ADDR_INTERVAL 1 /* address diff of adjacent regs. */ #define COM3_BASE_ADR 0x3e8 #define COM4_BASE_ADR 0x2e8 #define COM3_INT_LVL 0x04 #define COM4_INT_LVL 0x03 /* IRQ jumbers in KCI's Host board should be checked before changing following lines */ #define COM5_BASE_ADR 0x100 #define COM5_INT_LVL 0x07 /* I undefined LPT and also disabled LPT in BIOS SETUP */ ####### 3. add COM3_INT_VEC,,, macros in ..\pcpentium\config.h //before #define COM1_INT_VEC (INT_VEC_GET (COM1_INT_LVL)) #define COM2_INT_VEC (INT_VEC_GET (COM2_INT_LVL)) //after #else /* SYMMETRIC_IO_MODE */ ...... #define COM1_INT_VEC (INT_VEC_GET (COM1_INT_LVL)) #define COM2_INT_VEC (INT_VEC_GET (COM2_INT_LVL)) #define COM3_INT_VEC (INT_VEC_GET (COM3_INT_LVL)) #define COM4_INT_VEC (INT_VEC_GET (COM4_INT_LVL)) #define COM5_INT_VEC (INT_VEC_GET (COM5_INT_LVL)) #define RTC_INT_VEC (INT_VEC_GET (RTC_INT_LVL)) ####### 4. modify ...\pcPentium\sysSerial.c // before static I8250_CHAN i8250Chan[N_UART_CHANNELS]; static I8250_CHAN_PARAS devParas[] = { {COM1_INT_VEC,COM1_BASE_ADR,UART_REG_ADDR_INTERVAL,COM1_INT_LVL}, {COM2_INT_VEC,COM2_BASE_ADR,UART_REG_ADDR_INTERVAL,COM2_INT_LVL} }; // after static I8250_CHAN i8250Chan[N_UART_CHANNELS]; static I8250_CHAN_PARAS devParas[] = { {COM1_INT_VEC,COM1_BASE_ADR,UART_REG_ADDR_INTERVAL,COM1_INT_LVL}, {COM2_INT_VEC,COM2_BASE_ADR,UART_REG_ADDR_INTERVAL,COM2_INT_LVL}, {COM3_INT_VEC,COM3_BASE_ADR,UART_REG_ADDR_INTERVAL,COM3_INT_LVL}, {COM4_INT_VEC,COM4_BASE_ADR,UART_REG_ADDR_INTERVAL,COM4_INT_LVL}, {COM5_INT_VEC,COM5_BASE_ADR,UART_REG_ADDR_INTERVAL,COM5_INT_LVL} }; ###### 5. build bootrom_uncmp and vxWorks files each. In Tornado shell in my host computer, I can see 5 serial ports is installed using devs command. But 5 th port does not work when I used following command. - ->fd=open("/tyCo/4", 2) - ->write(fd,"Hello", 5) - ->close(fd) Pls, help me Thanks in advance. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: iterate over all files in a directory Date: Tue, 8 Feb 2000 09:01:04 +0100 From: "Mario Semo" Organization: Nextra Telekom GmbH Message-ID: <87ojcf$pi8$1@newsmaster01.nextra.at> References: <87bigt$okp$1@newsmaster01.magnet.at> <389BD584.D06C3CD7@mindspring.com> <389C4686.4E825FB5@bellsouth.net> <389CD13A.DE5B9B6E@aracnet.com> Hello, one more comment. NFS dont seem to be a problem. but my question regards netdrv and not nfs. - -- regards, Mario Semo. http://www.kirchnersoft.com engineer_scotty wrote in message <389CD13A.DE5B9B6E@aracnet.com>... >Joe Durusau wrote: >I believe he means from within vxWorks. Using stat() on a NFS >drive in vxWorks reports directories as files (it also reports >files as files.) The "feature" is documented by Wind River; >why their NFS client cannot tell the difference, I don't know. > >UNIX NFS clients don't have this problem. When WRS says that it >is "not possible", that means "we're can't be bothered to implement >that feature", not "it is not possible for a client to tell if an >NFS filesystem element is a directory or not." > --------------------------- Newsgroups: comp.os.vxworks Subject: Need custom board 68360 BSP developed Date: Thu, 10 Feb 2000 12:50:20 -0800 From: "Bernard Littau" Organization: Alternate Access Inc. Affordable, Reliable Access to the InterNet Message-ID: <38a324d1$1_2@huge.aa.net> Hi, I am a newbie to VxWorks, and I just got a partial BSP that needs to be completed dropped into my lap. I have a in-house designed 68360 board that currently runs VxWorks based on a modified ADS360 BSP. The board has one RS232 serial port working correctly(used for WDB). The second RS232 serial is flaky. I need the rest of the RS232 and RS485 serial ports working, and it would be nice to get the Ethernet working as well. Naturally, I need all this working yesterday, so I am interested in who out there can help me out and what kind of deal we can make. I am in Redmond WA (near Seattle :-). You can post here or reach me via email at bernardl@acumenassociates.com Best, Bernard Littau --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Target shell fails to recognise commands Date: Thu, 10 Feb 2000 20:54:06 GMT From: Ted Dennison Organization: Deja.com - Before you buy. Message-ID: <87v8hb$t0c$1@nnrp1.deja.com> References: <38A2DB87.EB293596@cadence.com> In article <38A2DB87.EB293596@cadence.com>, Gill Bisset wrote: > Hi - > > After defining INCLUDE_SHELL in configAll.h, the target shell banner > appears. However, the shell does not respond to any commands, such as > help, or any C statements eg: > > ->x = 5 > can't add 'x' to system symbol table - error = 0x3d0001 > > Does anyone know if I need to include anything else before the target > shell will respond to commands? I've gotten errors like that when I didn't include any symbols with the kernel. You either need to build a stand-alone kernel (vxworks.st) or place the vxworks.sym file in the same location as your vxworks kernel file. - -- T.E.D. http://www.telepath.com/~dennison/Ted/TED.html Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: ncr810 SCSI driver on MCP750 - question Date: Thu, 10 Feb 2000 14:58:22 -0600 From: Jennifer Askew Organization: ADC Telecommunications Message-ID: <38A3266E.E75FAB6@adc.com> I posted this information to WRS support, but was hoping someone here might be able to help a little quicker. Ok, I am having a problem getting the ncr810 (WRS supplied) driver running on our setup (we are using the SYM53C825A chip). We are using the MCP106 bridge chip. Additionally, we have a DEC21143 working on the PCI bus as well and a Serial Driver off the ISA bus. Also, when another OS is loaded onto our target platform (namely Linux), the SYM53C825A and SCSI harddrive work as expected. Here is the Debug information (the code I wrote is attached below). Any Ideas??? - ------------------------------------ TARGET OUTPUT - ---------------------------------- 0x7fffdf8 (tRootTask): ***** ENTRY: sysScsiInit() 0x7fffdf8 (tRootTask): ***** ENTRY: swapScript() 0x7fffdf8 (tRootTask): Found Primary SCSI Ctrl - Dev/Ven ID = 31000 0x7fffdf8 (tRootTask): busNo 0, devNo 2, funcNo 0, devType 3 0x7fffdf8 (tRootTask): Status/Command 2000007 0x7fffdf8 (tRootTask): Class Code/Rev ID 1000014 0x7fffdf8 (tRootTask): Base Address I/O 1000001 0x7fffdf8 (tRootTask): Base Address Mem 80000000 0x7fffdf8 (tRootTask): iobaseCsr 81000000 0x7fffdf8 (tRootTask): irq a 0x7fffdf8 (tRootTask): ncr810CtrlCreate completed successfully 0x7fffdf8 (tRootTask): intConnect completed successfully 0x7fffdf8 (tRootTask): intEnable completed successfully 0x7fffdf8 (tRootTask): ncr810CtrlInit completed successfully 0x7fffdf8 (tRootTask): ***** ENTRY: sysScsiConfig() 0x7fffdf8 (tRootTask): Configuring SCSI Hard Disk to Bus 2 LUN 0 0x7fffdf8 (tRootTask): scsiReqSense: 0x7fffdf8 (tRootTask): scsi2Transact: 0x7ff2f38 (tScsiTask): scsiMgrRequestProc: client request: 0 0x7ff2f38 (tScsiTask): scsiMgrActivateRequest: thread 0x07ff1688 0x7ff2f38 (tScsiTask): ncr810ThreadActivate: thread 0x07ff1688: activating 0x7ff2f38 (tScsiTask): ncr810XferParamsCvt: converted to: offset = 0, period = 0 xferParams = 0, clockDivide = 3 0x7ff2f38 (tScsiTask): ncr810ThreadStateSet: thread 0x07ff1688: 1 -> 6 0x7ff2f38 (tScsiTask): scsiMgrTimeoutProc: thread 0x07ff1688 (state = 6) timed out 0x7ff2f38 (tScsiTask): ncr810ThreadAbort: thread 0x07ff1688 (state = 6) aborting 0x7ff2f38 (tScsiTask): ncr810ThreadStateSet: thread 0x07ff1688: 6 -> 9 - ------------------------------------------------------------------------------------------------- Snippits of the code to setup the SCSI (note the ncr810 driver is WRS supplied) STATUS sysScsiInit () { UINT pciBusNo; UINT pciDevNo; UINT pciFuncNo; UINT devId; UINT16 devType; UINT findClass; static BOOL firstTime = TRUE; UINT32 iobaseCsr; /* base address 0 */ char irq; /* IRQ level */ UINT temp; /* Local structure(s) with a prefill for ncr825SetHwRegister */ static NCR810_HW_REGS hwRegs = SYS_REG_SIOP_HW_REGS; #ifdef sysScsiDebug SCSI_PRINT_DEBUG_MSG("***** ENTRY: sysScsiInit()\n",0,0,0,0,0,0); #endif /* Set CPU <-> PCI address offset */ ncr810PciMemOffset = PCI2DRAM_BASE_ADRS; /* * Since the PCI bus is little endian, and the PPC CPU's are big endian, * the ncr8xx scripts need to be byte swapped in memory. However, they * should only be swapped the first time that sysScsiInit() is invoked. */ if (firstTime) { swapScript ((UINT32 *) &ncr810Wait, (UINT32 *) ((UINT32) &ncr810Wait + (UINT32)(NCR810_INSTRUCTIONS * 8))); firstTime = FALSE; } /* Try to automatically configure the correct type of NCR8XX controller. */ findClass = (UINT)((PCI_CLASS_MASS_STORAGE << 16) | (PCI_SUBCLASS_00 << 8) | (0 << 0) ); if (pciFindClass (findClass, 0, &pciBusNo, &pciDevNo, &pciFuncNo) == OK) { /* Get Device and Vendor ID */ pciConfigInLong (pciBusNo, pciDevNo, pciFuncNo, PCI_CFG_VENDOR_ID, &devId); SCSI_PRINT_DEBUG_MSG ( "Found Primary SCSI Ctrl - Dev/Ven ID = %x \n",devId,0,0,0,0,0); devType = (devId >> 16) & 0x0000FFFF; } else { SCSI_PRINT_DEBUG_MSG ("SCSI controller not found\n", 0, 0, 0, 0, 0, 0); return (ERROR); } #ifndef SCSI_WIDE_ENABLE devType = NCR810_DEVICE_ID; #endif #ifdef sysScsiDebug SCSI_PRINT_DEBUG_MSG ("busNo %x, devNo %x, funcNo %x, devType %x\n", pciBusNo, pciDevNo, pciFuncNo, devType, 0, 0); #endif ncr810DelayCount = NCR810_DELAY_MULT; #ifdef sysScsiDebug /* for debug purposes, read some of the other config registers */ pciConfigInLong(pciBusNo, pciDevNo, pciFuncNo, PCI_CFG_COMMAND, &temp); SCSI_PRINT_DEBUG_MSG("Status/Command %x\n", temp,0,0,0,0,0); pciConfigInLong(pciBusNo, pciDevNo, pciFuncNo, PCI_CFG_REVISION, &temp); SCSI_PRINT_DEBUG_MSG("Class Code/Rev ID %x\n", temp,0,0,0,0,0); pciConfigInLong(pciBusNo, pciDevNo, pciFuncNo, PCI_CFG_BASE_ADDRESS_0, &temp); SCSI_PRINT_DEBUG_MSG("Base Address I/O %x\n", temp,0,0,0,0,0); pciConfigInLong(pciBusNo, pciDevNo, pciFuncNo, PCI_CFG_BASE_ADDRESS_1, &temp); SCSI_PRINT_DEBUG_MSG("Base Address Mem %x\n", temp,0,0,0,0,0); #endif /* Set the IRQ Number and Base Address */ irq = (UINT) INT_VEC_SCSI; pciConfigInLong (pciBusNo, pciDevNo, pciFuncNo, PCI_CFG_BASE_ADDRESS_0, &iobaseCsr); iobaseCsr = ((iobaseCsr & 0xfffffffe) - PCI_MSTR_IO_BUS) + PCI_MSTR_IO_LOCAL; #ifdef sysScsiDebug SCSI_PRINT_DEBUG_MSG("iobaseCsr %x\n", iobaseCsr,0,0,0,0,0); SCSI_PRINT_DEBUG_MSG("irq %x\n", irq,0,0,0,0,0); #endif /* Create the SCSI controller */ if ((pSysScsiCtrl = (SCSI_CTRL *) ncr810CtrlCreate ( (UINT8 *) iobaseCsr, (UINT) NCR810_40MHZ, devType )) == NULL) { return (ERROR); } #ifdef sysScsiDebug SCSI_PRINT_DEBUG_MSG("ncr810CtrlCreate completed successfully\n",0,0,0,0,0,0); #endif /* connect the SCSI controller's interrupt service routine */ if (intConnect (INUM_TO_IVEC ((int)irq), ncr810Intr, (int) pSysScsiCtrl) == ERROR) { return (ERROR); } #ifdef sysScsiDebug SCSI_PRINT_DEBUG_MSG("intConnect completed successfully\n",0,0,0,0,0,0); #endif /* Enable SCSI interrupts */ intEnable (irq); #ifdef sysScsiDebug if (intIsEnabled(irq)) SCSI_PRINT_DEBUG_MSG("intEnable completed successfully\n",0,0,0,0,0,0); #endif /* initialise SCSI controller with default parameters (user tuneable) */ if (ncr810CtrlInit ((NCR_810_SCSI_CTRL *)pSysScsiCtrl, SCSI_DEF_CTRL_BUS_ID) == ERROR) return (ERROR); #ifdef sysScsiDebug SCSI_PRINT_DEBUG_MSG("ncr810CtrlInit completed successfully\n",0,0,0,0,0,0); #endif #if (USER_D_CACHE_MODE & CACHE_SNOOP_ENABLE) scsiCacheSnoopEnable ((SCSI_CTRL *) pSysScsiCtrl); #else scsiCacheSnoopDisable ((SCSI_CTRL *) pSysScsiCtrl); #endif /* * Set the good value in the registers of the SIOP coupled * with the hardware implementation */ if (ncr810SetHwRegister ((NCR_810_SCSI_CTRL *)pSysScsiCtrl, &hwRegs) == ERROR) return(ERROR); /* Include tape support if configured in config.h */ #ifdef INCLUDE_TAPEFS tapeFsInit (); /* initialise tapeFs */ #endif /* INCLUDE_TAPEFS */ scsiDebug = TRUE; /* enable SCSI debugging output */ scsiIntsDebug = TRUE; /* enable SCSI interrupt debugging output */ scsiErrors = TRUE; /* enable SCSI error logging */ return (OK); } STATUS sysScsiConfig (void) { #ifdef sysScsiDebug SCSI_PRINT_DEBUG_MSG("***** ENTRY: sysScsiConfig()\n",0,0,0,0,0,0); #endif /* * NOTE: Either of the following global variables may be set or reset * from the VxWorks shell. Under 5.0, they should NOT both be set at the * same time, or output will be interleaved and hard to read!! They are * intended as an aid to trouble-shooting SCSI problems. */ scsiDebug = TRUE; /* enable SCSI debugging output */ scsiIntsDebug = TRUE; /* enable SCSI interrupt debugging output */ scsiErrors = TRUE; /* enable SCSI error logging */ /* * The following section of code provides sample configurations within * VxWorks of SCSI peripheral devices and VxWorks file systems. It * should however be noted that the actual parameters provided to * scsiPhysDevCreate(), scsiBlkDevCreate(), dosFsDevInit() etc., are * highly dependent upon the user environment and should therefore be * modified accordingly. */ /* * HARD DRIVE CONFIGURATION * * In order to configure a hard disk and initialise both dosFs and rt11Fs * file systems, the following initialisation code will serve as an * example. */ #ifdef sysScsiDebug SCSI_PRINT_DEBUG_MSG("Configuring SCSI Hard Disk to Bus 2 LUN 0\n",0,0,0,0,0,0 ); #endif /* configure a SCSI hard disk at busId = 2, LUN = 0 */ if ((pSpd20 = scsiPhysDevCreate (pSysScsiCtrl, 2, 0, 0, NCR810_TYPE, 0, 0, 0)) == (SCSI_PHYS_DEV *) NULL) { printErr ("usrScsiConfig: scsiPhysDevCreate failed.\n", 0, 0, 0, 0, 0, 0); } else { #ifdef sysScsiDebug SCSI_PRINT_DEBUG_MSG("Creating Block Devices \n",0,0,0,0,0,0); #endif /* create block devices */ if (((pSbd0 = scsiBlkDevCreate (pSpd20, 0x10000, 0)) == NULL) || ((pSbd1 = scsiBlkDevCreate (pSpd20, 0x10000, 0x10000)) == NULL) || ((pSbd2 = scsiBlkDevCreate (pSpd20, 0, 0x20000)) == NULL)) { return (ERROR); } if ((dosFsDevInit ("/sd0/", pSbd0, NULL) == NULL) ) { return (ERROR); } return (OK); } - -Jennifer --------------------------- Newsgroups: comp.os.vxworks Subject: symbolTableSize Date: Tue, 8 Feb 2000 11:50:45 +0100 From: "Mario Semo" Organization: Nextra Telekom GmbH Message-ID: <87osde$4gb$1@newsmaster01.nextra.at> Hello, Tornado 2, C++ code. target Vxw86, PPC, (VxSim) we use a lot of templates, and instantiate them, we have seen that long names are generated and that nearly half of the size of our object is the symbolTable. The templates are in fact just internal functions, public functions are just via virtual abstract functions, so just function tables are public. The means, that except for some functions in the loadable units, nearly everything is internal. what can we do to reduce the size of the symboltable? I assume that the symbols in my objects are inserted 1:1 into the symboltable of the target, when i load my object with ld < object. - -- regards, Mario Semo. http://www.kirchnersoft.com --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Ctrl-X doesn't work Date: Thu, 10 Feb 2000 13:24:21 -0800 From: mreed Organization: http://www.remarq.com: The World's Usenet/Discussions Start Here Message-ID: <15efa2ac.98af8f76@usw-ex0108-061.remarq.com> References: <032967d8.7c727136@usw-ex0108-061.remarq.com> <87sa8l$mi6$1@nnrp1.deja.com> I tried adding the target shell. Ctrl-X still doesn't work. * Sent from AltaVista http://www.altavista.com Where you can also find related Web Pages, Images, Audios, Videos, News, and Shopping. Smart is Beautiful --------------------------- Newsgroups: comp.os.vxworks Subject: PowerPC860 Board and VME bus Date: Thu, 10 Feb 2000 22:10:31 GMT From: Jan Graczyk Organization: DRS Technologies Canada Inc. Message-ID: <38A3374A.87D0A8AF@drs.ca> This is a multi-part message in MIME format. - --------------74A330A977A1A0F888F3E947 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Hello, I am trying to change the ADS860 BSP to support VME bus using Tundra chips such as QSpan and Universe II. Any help in this matter would be very much appreciated. Jan Graczyk DRS Technologies Canada Inc. Tel. (613) 727-4142 - --------------74A330A977A1A0F888F3E947 Content-Type: text/x-vcard; charset=us-ascii; name="jgraczyk.vcf" Content-Transfer-Encoding: 7bit Content-Description: Card for Jan Graczyk Content-Disposition: attachment; filename="jgraczyk.vcf" begin:vcard n:Graczyk;Jan tel;work:(613) 727-4142 x-mozilla-html:FALSE adr:;;;;;; version:2.1 email;internet:jgraczyk@drs.ca fn:Graczyk end:vcard - --------------74A330A977A1A0F888F3E947-- --------------------------- Newsgroups: comp.os.vxworks Subject: FAQ for this newsgroup? Date: Thu, 10 Feb 2000 17:23:34 -0500 From: Mohanram BR Organization: ITLInfosys Message-ID: <38A33A66.74E36475@ntl.inf.com> Reply-To: mohan@ntl.inf.com This is a multi-part message in MIME format. - --------------D60CDDF05DCAF49DD1248F82 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Can anyone point me to a FAQ related to this newsgroup Thanx, Mohan - --------------D60CDDF05DCAF49DD1248F82 Content-Type: text/x-vcard; charset=us-ascii; name="mohan.vcf" Content-Transfer-Encoding: 7bit Content-Description: Card for Mohanram BR Content-Disposition: attachment; filename="mohan.vcf" begin:vcard n:BR;Mohanram x-mozilla-html:FALSE org:ITLInfosys;TNPC adr:;;;;;; version:2.1 email;internet:mohan@ntl.inf.com title:Senior Systems Analyst x-mozilla-cpt:;0 fn:Mohanram BR end:vcard - --------------D60CDDF05DCAF49DD1248F82-- --------------------------- Newsgroups: comp.os.vxworks Subject: Re: FAQ for this newsgroup? Date: Thu, 10 Feb 2000 19:01:03 -0800 From: Harvey Taylor Organization: Organization? What organization?! Message-ID: <38A37B6F.5698@despam.pangea.ca> References: <38A33A66.74E36475@ntl.inf.com> Mohanram BR wrote: > Can anyone point me to a FAQ related to this newsgroup > These might be useful. A VxWorks FAQ http://www.pso.com/vxfaq.html Tutorial http://www.digital.com/oem/library/docs/vxintro/vx0001.htm Old VxWorks FAQ http://www.faqs.org/faqs/vxworks-faq/part1/index.html Tornado 2 FAQ http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html - -het - --- "Assembly of Japanese bicycle require great peace of mind." - Instruction manual mentioned by R. Pirsig Harvey Taylor mailto:het@despam.portal.ca http://www.pangea.ca/~het --------------------------- Newsgroups: comp.os.vxworks Subject: intro to vx-works Date: Fri, 11 Feb 2000 01:04:40 GMT From: Ephraim Gadsby Message-ID: Does anyone know of an online introduction to vxworks? I am starting with a company that uses it and I would like to hit the ground running, but I don't yet have access to wind river documentation. Thanks --------------------------- Newsgroups: comp.os.vxworks Subject: Re: intro to vx-works Date: Thu, 10 Feb 2000 17:43:42 -0800 From: Robert Organization: http://www.remarq.com: The World's Usenet/Discussions Start Here Message-ID: <3ae4afc3.77067a6a@usw-ex0108-063.remarq.com> References: You can download the manuals at http://www.wrs.com/products/html/manual-form.html * Sent from RemarQ http://www.remarq.com The Internet's Discussion Network * The fastest and easiest way to search and participate in Usenet - Free! --------------------------- Newsgroups: comp.os.vxworks Subject: VLAN (IEEE 802.1q) Date: Thu, 10 Feb 2000 07:46:19 GMT From: Yusuf Motiwala Organization: Deja.com - Before you buy. Message-ID: <87tqcb$qa1$1@nnrp1.deja.com> Hi, We are looking for VLAN (IEEE 802.1Q) implemetation on VxWorks. Has anyone implemeted this. Can you point me to some vendors for the same. Regards, Yusuf Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: intro to vx-works Date: Thu, 10 Feb 2000 17:53:19 -0800 From: "Michael R. Kesti" Organization: MK Associates Message-ID: <38A36B8F.670B@gv.net> References: Reply-To: mkesti@gv.net Ephraim Gadsby wrote: >Does anyone know of an online introduction to vxworks? > >I am starting with a company that uses it and I would like to hit the >ground running, but I don't yet have access to wind river documentation. Ask your new employer for their WRS license number, and register for access to WindSurf on http://www.wrs.com. The complete manual set is there in HTML and PDF. - -- ======================================================================== Michael Kesti | "And like, one and one don't make | two, one and one make one." mkesti@gv.net | - The Who, Bargain --------------------------- Newsgroups: comp.os.vxworks Subject: Re: VxWorks vs. Linux Date: Thu, 10 Feb 2000 20:49:45 -0600 From: "Ed & Lisa Brown" Organization: MindSpring Enterprises Message-ID: <87vt22$oac$1@nntp2.atl.mindspring.net> References: <87t4tl$55u$1@news2.kornet.net> <38a22765.0@news.pacifier.com> Where would one find information on RTEMS or Linux RT? DuckE wrote in message <38a22765.0@news.pacifier.com>... >A closer competitor would be RTEMS, which is rapidly developing a set of >features to compete directly with VxWorks. > . --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Ctrl-X doesn't work Date: Fri, 11 Feb 2000 04:40:01 GMT From: john_94501@my-deja.com Organization: Deja.com - Before you buy. Message-ID: <8803r0$gg4$1@nnrp1.deja.com> References: <032967d8.7c727136@usw-ex0108-061.remarq.com> <87sa8l$mi6$1@nnrp1.deja.com> <15efa2ac.98af8f76@usw-ex0108-061.remarq.com> In article <15efa2ac.98af8f76@usw-ex0108-061.remarq.com>, mreed wrote: > I tried adding the target shell. Ctrl-X still doesn't work. Make sure you have tExcTask running... John... Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Standalone Boot Problem Date: Fri, 11 Feb 2000 05:00:29 GMT From: john_94501@my-deja.com Organization: Deja.com - Before you buy. Message-ID: <88051b$hav$1@nnrp1.deja.com> References: <20000209133348.01018.00000686@ng-cg1.aol.com> Hi, Your problem has become a little different with your change of configuration, if I understood your description that is... What you had before was simply the master board acting as a gateway to your host, which was acting as the boot server. With your new configuration you want the the master to act as the boot server for the slave card. So, you need to configure the master to boot its image, start its backplane network, and then it needs to be running some kind of boot server application for the second card's bootrom to communicate with (eg an FTP server - see INCLUDE_FTP_SERVER). You could also write a simple rsh-like server which listened for connections from the slave card over the backplane network (on the rsh port), then accepted the bootrom's 'cat bootfile' command and simply returned the data from flash (in this case make sure your bootrom is configured for rsh booting and not ftp). The latter is probably simpler if your flash is not already a filesystem! Good luck, John... In article <20000209133348.01018.00000686@ng-cg1.aol.com>, bagdazian@aol.com (BAGDAZIAN) wrote: > I need some help related to a problem I am having. Here is the description of > the problem. > > I have two Cetia Vmpc4a SBC's running vxworks and communicating via backplane > networking. I have been successful in configuring my vxWorks image to support > the master board booting from a PC host via an ethernet connection. I have also > been successful in configuring the second card to boot via the ethernet using > the first processor board as a gateway to the ethernet. I now need to create > standalone vxWorks images to allow booting from a pcmcia flash memory card > mounted on the first processor card. I have been able to successfully build the > vxWorks.st image using the same configuation files as I used for the host > support image. I also included STANDALONE_NET in the appropriate include files. > HOWEVER, when I boot the system using this standalone vxWorks.st image, the > backplane networking does not startup and the second card is unable to connect > to the first card. I get an error message on the first card that says something > to the effect that "pcmcia network device not recognized" at the point that the > system is attempting to bring up the backplane network. How do I get the > backplane networking facility to startup in the same way that it did when I > used the lnPci driver as the boot device ? > > Thanks in advance. > Rich Bagdazian > bagdazian@aol.com > Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: benchmarks for VxWorks on x86 Date: Thu, 10 Feb 2000 11:12:45 -0500 From: "Keo S. Kelly" Organization: Raytheon Message-ID: <38A2E37C.540E4E38@fallschurch.esys.com> All: Anyone has or know of information concerning benchmarks for VxWorks on x86 processor? A comparison benchmarks between Linux and VxWorks would be perfect. Thanks. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: VxWorks vs. Linux Date: Fri, 11 Feb 2000 07:20:15 GMT From: thoth2487@my-deja.com Organization: Deja.com - Before you buy. Message-ID: <880d7d$mlg$1@nnrp1.deja.com> References: <87t4tl$55u$1@news2.kornet.net> <38a22765.0@news.pacifier.com> <87vt22$oac$1@nntp2.atl.mindspring.net> "Ed & Lisa Brown" wrote: > Where would one find information on RTEMS or Linux RT? All regarding RTEMS is present in www.oarcorp.com or www.rtems.com. RTEMS is a full-free REALTIME OS with support various microprocessors and BSP. It is available in source code, and it uses GNU tool chain like VxWorks. In next days should be available the new versione, I think 4.5, in that will be present more BPSs, CPU families, RPM for Linux e Windows (with CygWin), ITRON layer, and even more..... Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Target shell fails to recognise commands Date: Fri, 11 Feb 2000 09:41:36 GMT From: "Philippe LE SAUZE" Organization: KPNQwest customer news service Message-ID: References: <38A2DB87.EB293596@cadence.com> Gill Bisset a écrit dans le message <38A2DB87.EB293596@cadence.com>... >Hi - > >After defining INCLUDE_SHELL in configAll.h, the target shell banner >appears. However, the shell does not respond to any commands, such as >help, or any C statements eg: > >->x = 5 >can't add 'x' to system symbol table - error = 0x3d0001 > >Does anyone know if I need to include anything else before the target >shell will respond to commands? > >I'm running VxWorks 5.3.1 on an ARM7TDMI architecture. The host machine >is a Sun workstation running Solaris2. > >Thanks for your help, > >Gill Hi Gill You have to load the symbol table in your target. of course you can build a vxWorks.st or you can download the vxWorks.sym from your host This is done automaticaly, if you include some macros let you look at symbol table components in your vxWorks GUI Good luck Phil --------------------------- End of New-News digest ********************** From vxwexplo-errs@csg.lbl.gov Fri Feb 11 05:20:47 2000 From: Guy Kronental Date: Fri Feb 11 05:20:49 PST 2000 Subject: C++ inline problems Hi We have a big problem with inlines in our code : When using inlines within a class that has a virtual function (one or more ) the inlines are not accepted and become function calls. We asked wind river about it and after admitting its a problem they retreated and the answer we got was : what the compiler is said to do is not against the appropriate C++ standards This seems like avoiding the problem. We are working on a Solaris host and a INTEL or IDT cpu as the target cpu . We tested the problem both with vxWorks 5.3 , 5.3.1 and 5.4 using cc960 (gnu version 2.6-95q2 , and cygnus-2-7-2 , cygnus-2-7-2 -960126 egcs 971225 tornado 2 ) and g++ and they all have the same problem , we also tried different optimization flags , but nothing helped A guy in wind river check it as well before the "wired" answer and said that : " I did some tests with inlines and they are broken on all architectures mainly: g++ (egcs 1.1.2) (linux) g++ (egcs 1.0.2) (sun, pc, powerpc, i960) It only worked with cygwin for NT (egcs 1.1 using the -O option) " Any suggenstions ??? a sample code for example class X { private : int x; public: X(); virtual ~X(); inline get_x () {get x; } }; then the following calls int main() { X my_obj; int x = my_obj->get_x(); // become a function call . . . } ======================================================= Guy Kronenthal Email : gkronent@lucent.com Lucent ISR (LANNET) R&D Voice : 972-3-6457542 Fax : 972-3-6487146 Atidim Technology Park, Bldg. 3, Tel-Aviv 61131, Israel. ======================================================= From vxwexplo-errs@csg.lbl.gov Sat Feb 12 04:03:09 2000 From: Vxworks Exploder Date: Sat Feb 12 04:03:11 PST 2000 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Sat Feb 12 04:03:05 PST 2000 Subject: Image Processing on a PC/104 Subject: Re: Ctrl-X doesn't work Subject: Re: VxWorks vs. Linux Subject: Re: Target shell fails to recognise commands Subject: want to load an object module into memory, while using rom resident vxWorks Subject: Re: VxWorks vs. Linux Subject: Re: VxWorks vs. Linux Subject: Re: dosFs2.0 Code to mount unknown number of partitions? Subject: Re: Power Arch. interrupt question Subject: Re: C++ inline problems Subject: Re: VxWorks vs. Linux Subject: TOD clock on an MVME2700 Subject: Re: VxWorks vs. Linux Subject: VxWorks and Windows Subject: how I can use inflate in my own function ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Image Processing on a PC/104 Date: Fri, 11 Feb 2000 13:37:57 +0100 From: Rafael Garcia Organization: CESCA News service. Catalunya Message-ID: <38A402A5.783BBF48@eia.udg.es> We are planning to put a "Matrox Meteor II/4" frame grabber running on a PC/104 under VxWorks. Does anyone know if there is a driver for this card for VxWorks? Or do you know of any other PC/104 vision card which for VxWorks? Thanks. Rafael Garcia Computer Vision and Robotics Group University of Girona rafa@eia.udg.es --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Ctrl-X doesn't work Date: Fri, 11 Feb 2000 13:42:46 +0100 From: Gerhard Juen Organization: 'FH Gelsenkirchen, Abt. Bocholt' Message-ID: <38A403C6.795D23FC@bocholt.fh-gelsenkirchen.de> References: <032967d8.7c727136@usw-ex0108-061.remarq.com> <87sa8l$mi6$1@nnrp1.deja.com> no it works also without target shell Ted Dennison schrieb: > In article <032967d8.7c727136@usw-ex0108-061.remarq.com>, > mreed wrote: > > Does anyone know how you turn on the rebooting of VxWorks > > by using CTRL-X? Is there some part of the operating > > system that has to be included/excluded? > > That's a function of the target shell, so I'm pretty sure you need that > included. > > -- > T.E.D. > > http://www.telepath.com/~dennison/Ted/TED.html > > Sent via Deja.com http://www.deja.com/ > Before you buy. - -- - ----------------------------------------------------------------------- Prof. Dr. Gerhard Juen FH Gelsenkirchen, Abt. Bocholt Münsterstr. 265 D-46397 Bocholt email: juen@bocholt.fh-gelsenkirchen.de web: http://www-et.bocholt.fh-gelsenkirchen.de/~juen Fon: 02871-2155 830 Fax: 02871-2155 800 --------------------------- Newsgroups: comp.os.vxworks Subject: Re: VxWorks vs. Linux Date: Fri, 11 Feb 2000 13:47:57 +0100 From: Gerhard Juen Organization: 'FH Gelsenkirchen, Abt. Bocholt' Message-ID: <38A404FD.68FE5A8D@bocholt.fh-gelsenkirchen.de> References: <87t4tl$55u$1@news2.kornet.net> <38a22765.0@news.pacifier.com> <87vt22$oac$1@nntp2.atl.mindspring.net> embedded linux try here Embedded Linux try this link-list http://www-et.bocholt.fh-ge.de/~juen/ak_tele/ak_tele4.html there exists a headline "embedded linux" with some releated links gerhard Ed & Lisa Brown schrieb: > Where would one find information on RTEMS or Linux RT? > > DuckE wrote in message <38a22765.0@news.pacifier.com>... > >A closer competitor would be RTEMS, which is rapidly developing a set of > >features to compete directly with VxWorks. > > > . - -- - ----------------------------------------------------------------------- Prof. Dr. Gerhard Juen FH Gelsenkirchen, Abt. Bocholt Münsterstr. 265 D-46397 Bocholt email: juen@bocholt.fh-gelsenkirchen.de web: http://www-et.bocholt.fh-gelsenkirchen.de/~juen Fon: 02871-2155 830 Fax: 02871-2155 800 --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Target shell fails to recognise commands Date: Fri, 11 Feb 2000 13:52:14 +0100 From: Gerhard Juen Organization: 'FH Gelsenkirchen, Abt. Bocholt' Message-ID: <38A405FE.F2414FDB@bocholt.fh-gelsenkirchen.de> References: <38A2DB87.EB293596@cadence.com> i had similar problems included shells, symbol tables etc, nothing solved the problem. in the end i found out that in the bsp configuration i had to activate INCLUDE_CONFIGURATION_5_2 then it worked Philippe LE SAUZE schrieb: > Gill Bisset a écrit dans le message <38A2DB87.EB293596@cadence.com>... > >Hi - > > > >After defining INCLUDE_SHELL in configAll.h, the target shell banner > >appears. However, the shell does not respond to any commands, such as > >help, or any C statements eg: > > > >->x = 5 > >can't add 'x' to system symbol table - error = 0x3d0001 > > > >Does anyone know if I need to include anything else before the target > >shell will respond to commands? > > > >I'm running VxWorks 5.3.1 on an ARM7TDMI architecture. The host machine > >is a Sun workstation running Solaris2. > > > >Thanks for your help, > > > >Gill > > Hi Gill > > You have to load the symbol table in your target. > > of course you can build a vxWorks.st > > or you can download the vxWorks.sym from your host > > This is done automaticaly, if you include some macros > let you look at symbol table components in your vxWorks GUI > > Good luck > Phil - -- - ----------------------------------------------------------------------- Prof. Dr. Gerhard Juen FH Gelsenkirchen, Abt. Bocholt Münsterstr. 265 D-46397 Bocholt email: juen@bocholt.fh-gelsenkirchen.de web: http://www-et.bocholt.fh-gelsenkirchen.de/~juen Fon: 02871-2155 830 Fax: 02871-2155 800 --------------------------- Newsgroups: comp.os.vxworks Subject: want to load an object module into memory, while using rom resident vxWorks Date: Fri, 11 Feb 2000 15:50:47 GMT From: findik@my-deja.com Organization: Deja.com - Before you buy. Message-ID: <881b4l$chr$1@nnrp1.deja.com> I am using Tornado II, I have downloaded vxWorks.res_rom_res_low build to my target, I can use Tornado II, open a shell for this target, I want to load an object module into memory, I got the following errors: form CONSOLE: - -> ld (1) ld < viaFlash.o Loading D:/Via/vxproje/FlashProg/PPC860gnu/viaFlash.o | WTX Error 0x10034 (LOADER_RELOCATION_OFFSET_TOO_LARGE) value = -1 = 0xffffffff Thanks for your valuable comments, best regards, Reyhan Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: VxWorks vs. Linux Date: Fri, 11 Feb 2000 09:24:43 -0600 From: john hagen Organization: Rockwell International Message-ID: <38A429BB.D413BCA9@collins.rockwell.com> References: <87t4tl$55u$1@news2.kornet.net> <38a22765.0@news.pacifier.com> <87vt22$oac$1@nntp2.atl.mindspring.net> Ed & Lisa Brown wrote: > Where would one find information on RTEMS or Linux RT? > > DuckE wrote in message <38a22765.0@news.pacifier.com>... > >A closer competitor would be RTEMS, which is rapidly developing a set of > >features to compete directly with VxWorks. > > > . http://www.rtlinux.org/~rtlinux/ --------------------------- Newsgroups: comp.os.vxworks Subject: Re: VxWorks vs. Linux Date: Fri, 11 Feb 2000 17:10:19 +0100 From: Peter Fenzke Organization: Siemens AG,(Hofmannstr) Munich-Germany-Europe. Message-ID: <38A4346B.D1F20FC5@icn.siemens.de> References: <87t4tl$55u$1@news2.kornet.net> checkout http://www.linuxdevices.com/ Embedded Linux compete mainly with QNX, LynxOS, and WinCE. The same market - the same features ... Today you can download/order about 20 different embedded linux distributions. Which embLinux and which linux support is the best choise for your customer ? That is the question ? regards peter --------------------------- Newsgroups: comp.os.vxworks Subject: Re: dosFs2.0 Code to mount unknown number of partitions? Date: Fri, 11 Feb 2000 18:30:15 +0200 From: Leonid Rosenboim Organization: BitBand Technologies Ltd. http://www.bitband.com Message-ID: <38A43917.6A21319@bitband.com> References: <87aets$a8c$1@news.netmar.com> <04d80ac4.e4d8ed52@usw-ex0103-024.remarq.com> Reply-To: leonid@bitband.com Sounds reasonable. But what you hadf to do to make it work, is devise a scheme (formula) to generate logical device names automatically for each partition. This can be done a la MessDos, devname[0] = 'C' + parNum ; or a la UNIX, e.g. sprintf( devName, "/disk%x/$02x", drive, partNum); or something similar. Problem is, any of these schemes is specific to someone's taste, and no one particular scheme is appealing to everyone equally, hence it is hard to support any number of partition in VxWorks without disapointing some people's taste in logical device names (i.e. keep OS neutrality). Hope this explains things, Leonid Todd wrote: > I was fighting that exact same problem. It is easy to just create > the maximum number of devices, but not very elegant. Who wants a > bunch of inaccessible devices in the device list... > > I looked at the code in usrFdiskPartShow routine. It parses the > MBR and finds what partitions exist on the disk and prints out a > bunch of info on each one. If the type is non-zero, the partition > must be valid and the device can be added to the device list. > > I just copied a section of that code and usd it to validate the > partitions on the disk. > > Hope this helps. > > Todd > > * Sent from RemarQ http://www.remarq.com The Internet's Discussion Network * > The fastest and easiest way to search and participate in Usenet - Free! --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Power Arch. interrupt question Date: Fri, 11 Feb 2000 16:27:17 GMT From: gold@sdsu.edu Organization: Deja.com - Before you buy. Message-ID: <881d8u$ecu$1@nnrp1.deja.com> References: <20000208161858.01177.00000043@ng-ft1.aol.com> <87sfnk$ji$1@news1.xs4all.nl> In article <87sfnk$ji$1@news1.xs4all.nl>, johan.borkhuis@peektraffic.nl wrote: > MaggieJohn wrote: > : The power architecture has several interrupts: system reset, machine check, > : data storage, instruction storage, alignment, program, external, FP unavailable > : and supervisor call. > > : What does vxWorks do if one of these interrupts occurs? Suspend the task? > : Write a message? > > You can connect a routine to a interrupt using intConnect. This function > will be called when the interrupt occures. vxWorks (in a "default" configuration) does different things depending on the exception. For DSI and ISI, it loads these handlers if you enable the MMU. You'll never see anything that tells you these were called. There is no reason to tell you they were called. It handles the DSI/ISI and returns to the program. For program error, alignment, machine check (and I think FP via the program exception), it prints a message out the console saying which type of error, task, some relevant task state information, and suspends the task. External exceptions (0x500) are a different case. Since the PPC has only 1 external INT line (at least on the 403, 860, 8260, and 603/4/750), vxWorks loads their own handler that knows how to "de-multiplex" various external interrupts. When you do an intConnect(), it keeps the address for the vector in a table. When the external interrupt occurs, it reads a vector register and then knows the particular handler to call, which was set by you with intConnect(). Whatever your handler does (i.e. print, reset, whatever) whill be done. For SMI, I can't remember if they load a default handler (i.e. rfi) or not. Of course you can change all of the above, but vxWorks does have the assembly "glue" at those locations to save the important things before calling your handler. > : If it tries to write a message to a UART, does it matter if the UART is powered > : off? > > Depends on your implementation of the device driver. If your device > driver checks the UART to see if a byte has been written it will detect > the failure, otherwise it will just write it. If your driver works in the general 'tty driver' way, i.e. waiting for the Tx complete interrupt to load the next character, if the UART is off this could be a problem :-) And the stock vxWorks tty drivers do work that way. hth, Dan Gold gold@sdsu.edu Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: C++ inline problems Date: Fri, 11 Feb 2000 10:56:27 -0800 From: Harvey Taylor Organization: Organization? What organization?! Message-ID: <38A45B5B.7C1@despam.pangea.ca> References: <8815is$lth1@overload.lbl.gov> Guy Kronental wrote: > We have a big problem with inlines in our code : > When using inlines within a class that has a virtual function (one or more ) > the inlines are not accepted and become function calls. > We asked wind river about it and after admitting its a problem they > retreated and the answer we got was : > what the compiler is said to do is not against the appropriate C++ > standards > > This seems like avoiding the problem. > It might seem that way to you because you want the code to be inline, however it is no. I refer you to the ISO C++ Standard section 7.1.2 Function specifiers where you will read: - --- 2 A function declaration (8.3.5, 9.3, 11.4) with an inline specifier declares an inline function. The inline specifier indicates to the implementation that inline substitution of the function body at the point of call is to be preferred to the usual function call mechanism. An implementation is not required to perform this inline substitution at the point of call; however, even if this inline substitution is omitted, the other rules for inline functions defined by 7.1.2 shall still be respected. - --- It is like the keyword register. It is a hint to the compiler not an ironclad guarantee. You will have to solve your problem in another way. - -het - --- "Assembly of Japanese bicycle require great peace of mind." - Instruction manual mentioned by R. Pirsig Harvey Taylor mailto:het@despam.portal.ca http://www.pangea.ca/~het --------------------------- Newsgroups: comp.os.vxworks Subject: Re: VxWorks vs. Linux Date: Fri, 11 Feb 2000 13:27:30 -0600 From: "Rosimildo daSilva" Organization: Illuminati Online Message-ID: <881o3l$6bc$1@hiram.io.com> References: <87t4tl$55u$1@news2.kornet.net> <38a22765.0@news.pacifier.com> <87vt22$oac$1@nntp2.atl.mindspring.net> Ed & Lisa Brown wrote in message <87vt22$oac$1@nntp2.atl.mindspring.net>... >Where would one find information on RTEMS or Linux RT? > RT Linux: http://luz.cs.nmt.edu/~rtlinux/ RTEMS: http://www.oarcorp.com Rosimildo. --------------------------- Newsgroups: comp.os.vxworks Subject: TOD clock on an MVME2700 Date: Fri, 11 Feb 2000 15:03:30 -0500 From: Dan Lucas Organization: Applied Research Lab Penn State Univ. Message-ID: <38A46B12.1664B2AD@psu.edu> I would like to read the "Time of Day" clock on an MVME2700, has anyone done this, if so, could you give me some pointers ? - -- Daniel K. Lucas email: dlucas@psu.edu Applied Research Lab phone: (814) 865-2907 Penn State Univ. fax: (814) 863-7270 P.O. Box 30 State College, Pa. 16804 --------------------------- Newsgroups: comp.os.vxworks Subject: Re: VxWorks vs. Linux Date: Fri, 11 Feb 2000 23:57:19 +0100 From: "Mario Semo" Organization: Nextra Telekom GmbH Message-ID: <8826i1$3tk$1@newsmaster01.nextra.at> References: <87t4tl$55u$1@news2.kornet.net> <38a22765.0@news.pacifier.com> <87vt22$oac$1@nntp2.atl.mindspring.net> regarding RTLinux: http://www.thinkingnerds.com/fsmlabs/index.html and there is an email newsletter: http://linuxplc.org/mailman/listinfo/linuxplc - -- regards, Mario Semo. http://www.kirchnersoft.com Ed & Lisa Brown wrote in message <87vt22$oac$1@nntp2.atl.mindspring.net>... >Where would one find information on RTEMS or Linux RT? --------------------------- Newsgroups: comp.os.vxworks Subject: VxWorks and Windows Date: Sat, 12 Feb 2000 01:59:27 GMT From: "Harald" Organization: @Home Network Message-ID: <383p4.6728$pN1.99950@news1.rdc1.nj.home.com> I believe to have heard that VxWorks is available on Intel X86 PCs in a version that shares the same CPU with Windows running concurrently. Unfortunatelly I do not find information on this in the FAQ or on the Web. Can anybody provide information on whether this is true? What version of Windows are supported? Can anybody comment on how realtime-capable this combination would be? Please no general comments on how good or bad Windows performs, we can't really change that is is there. Thanks in advance - -Harald --------------------------- Newsgroups: comp.os.vxworks Subject: how I can use inflate in my own function Date: Sat, 12 Feb 2000 11:30:56 +0800 From: "Zha Junhai" Organization: Alcatel Message-ID: <38a5b6c7.0@dnews.sbell.com.cn> Hi, all Could you tell me how I can use inflate in my own function, whatever I defined extern inflate(), and download to target, it always tell me Undefined symbols: inflate, how can I solve it, any answer will be appreciated, Thanks. Best Regards Zha Junhai --------------------------- End of New-News digest ********************** From vxwexplo-errs@csg.lbl.gov Sun Feb 13 04:03:04 2000 From: Vxworks Exploder Date: Sun Feb 13 04:03:07 PST 2000 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Sun Feb 13 04:03:02 PST 2000 Subject: TCAP over IP on VxWorks Subject: Re: VxWorks: Memory alignment problem ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: TCAP over IP on VxWorks Date: Sat, 12 Feb 2000 22:50:33 GMT From: balaji narayana Organization: AT&T Worldnet Message-ID: <38A5E418.5157FFBC@worldnet.att.net> Reply-To: bbalaji@worldnet.att.net Hi, Has anybody used Trillium's TCAP over IP solution on VxWorks (Intel platform)? If so: 1. How expensive is the implementation of the ITU-T Invoke Timers? 2. Any benchmarks for timer and thread resources required for different throughput considerations? 3. Any other (recommended) TCAP/IP implementation on VxWorks? Please reply to balaji_narayana@yahoo.com Thanks, Balaji --------------------------- Newsgroups: comp.os.vxworks Subject: Re: VxWorks: Memory alignment problem Date: Sun, 13 Feb 2000 11:28:26 +0200 From: Leonid Rosenboim Organization: BitBand Technologies Ltd. http://www.bitband.com Message-ID: <38A6793A.1EC89C7C@bitband.com> References: <87qpls$guf7@overload.lbl.gov> Reply-To: leonid@bitband.com First, it would be nice to know what type of network hardware you are writing your driver for. As to your question, you can not assume alignment of received mbuf's, so if your hardware is stupid enough to require strict alignment, consider replacing the hardware, or suffer the consequences. The only workaround is to copy the received mbufs into a buffer's own buffer, which is aluigned appropriately, and transmit from there. Many driver work this way, some work well too. There could be a number of reasons to copy a packet into driver's own buffer, alignment is one, anotehr is when packet is fragmented into "too many" small buffers, or if the packet is simply a small one, costs you less to copy it rather then transmit from mbufs directly and then worry about freeing the mbufs etc. Good luck, Leonid Nagaraj G wrote: > Hi VxWorkers, > > I have a problem with the memory alignment in VxWorks. I am using > pcPentium BSP with Tornado II. > > I used ping to test my driver. When I use small ping lengths, for > example: up to 100 bytes, I don't face any problem. I always get > quadlet-aligned address in the MBLK_ID's mData field. But, if I increase > the length, I start getting byte-aligned, I mean non-quadlet-aligned, > memory address. My hardware does not support this. What should be done > for this? > > I also tried the routing. I placed two cards in a VxWorks machine, one > "ene0" and other my hardware. I tried doing a ping from a Windows host > on the "ene0" network and tried routing it to the other interface. In > this case, I always get the memory non-quadlet-aligned. > > Any solution for this. I saw a SPR with a similar problem. But, it > states on x86 it should just work fine. I am using pcPentium BSP which > is a x86 type. > > Does anyone have a solution or a work-around for this problem? > > Thanks in advance. > > Best Regards, > > Nagaraj G > EnThink Inc > www.EnThink.com --------------------------- End of New-News digest ********************** From vxwexplo-errs@csg.lbl.gov Mon Feb 14 00:41:47 2000 From: "Michael Lawnick" Date: Mon Feb 14 00:41:49 PST 2000 Subject: Using PMC cards with Extended VME Memory Model vxWorks > Hello, > > I am in the unenviable position of integrating several different PMC > cards for use (not all at the same time) on a Motorola MV2308 board. > Since the Motorola board isn't PCI aware, which sucks, I must manually > configure the PCI registers and base addresses for each card. I had > successfully done this using the Pseudo-PREP memory model, by writing > PCI memory addresses to the base address registers and using CPU > addresses that looked like this. > > 0xC0000000 | (PCI address) > > I now have to integrate the PMC cards using an Extended VME memory > model. Since my PMC cards fit fine in a 16MB space, I kept the same PCI > addresses that I used on Pseudo-Prep, and the CPU addresses became... > > 0xFD000000 | (PCI address) > > This was an utter failure for memory reads and writes. I have tried a > number of different combinations of CPU addresses vs. PCI addresses, but > none of them work. Can anyone give me or point me to a short rundown of > how to map PMC cards to CPU memory using Extended VME model? > > Kirk Ellett > Hi Kirk, have you taken a look to your sysPhysMemDesc[] in sysLib.c ? For every address range you want to use there must be an entry. HTH MfG Lawnick, SOFTEC GmbH ============================================== SOFTEC GmbH Tel +49-731-96600-0 Promenade 17 Fax +49-731-96600-23 D-89073 Ulm Michael Lawnick Germany lawnick@softec.de ============================================== From vxwexplo-errs@csg.lbl.gov Mon Feb 14 04:03:08 2000 From: Vxworks Exploder Date: Mon Feb 14 04:03:11 PST 2000 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Mon Feb 14 04:03:04 PST 2000 Subject: Re: FAQ for this newsgroup? Subject: Adding SSL to Wind Web Server Subject: Re: FAQ for this newsgroup? Subject: Re: intro to vx-works Subject: Re: want to load an object module into memory, while using rom resident vxWorks Subject: Re: A "Tracy Riddle" is spamming c.o.vxworks posters Subject: objcopy and binary files Subject: Re: benchmarks for VxWorks on x86 Subject: C++ - CTORS Subject: Re: C++ - CTORS Subject: Re: FAQ for this newsgroup? Subject: Re: A "Tracy Riddle" is spamming c.o.vxworks posters Subject: Re: routing code Subject: GDB problems Subject: PPP-Callback (RFC 1570) ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: FAQ for this newsgroup? Date: 13 Feb 2000 13:24:11 GMT From: johan.borkhuis@peektraffic.nl Organization: XS4ALL Internet BV Message-ID: <886b9r$cpc$1@news1.xs4all.nl> References: <38A33A66.74E36475@ntl.inf.com> <38A37B6F.5698@despam.pangea.ca> Harvey Taylor wrote: : Tutorial http://www.digital.com/oem/library/docs/vxintro/vx0001.htm This page has been moved to: http://www.smartmodulartech.com/systems/library/docs/vxworks/vxintro/VXINTRO.HTM But this page does not work correctly. They use capitals as names for the pages, but the pages are named using normal letters. For example the file vx001.htm is referred to from the intro page, but the filename is VX001.HTM. I already reported this to the webmaster, but it has not changed yet. Groeten, Johan - -- o o o o o o o . . . ___________________________________ o _____ || Johan Borkhuis | .][__n_n_|DD[ ====_____ | johan@borksoft.xs4all.nl | >(________|__|_[_________]_|________________________________| _/oo OOOOO oo` ooo ooo 'o!o!o o!o!o` === VxWorks page: http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html === --------------------------- Newsgroups: comp.os.vxworks Subject: Adding SSL to Wind Web Server Date: 13 Feb 2000 16:11:36 GMT From: kbeckley@aol.com (KBeckley) Organization: AOL http://www.aol.com Message-ID: <20000213111136.22425.00003204@ng-fh1.aol.com> We have the Wind Web Server for the 68k target and now need to add SSL to the kernel. Has anyone done this before and can you explain it in detail. Thank you, Keith Beckley --------------------------- Newsgroups: comp.os.vxworks Subject: Re: FAQ for this newsgroup? Date: Sun, 13 Feb 2000 10:17:19 -0800 From: Harvey Taylor Organization: Organization? What organization?! Message-ID: <38A6F52F.10A4@despam.pangea.ca> References: <38A33A66.74E36475@ntl.inf.com> <38A37B6F.5698@despam.pangea.ca> <886b9r$cpc$1@news1.xs4all.nl> johan.borkhuis@peektraffic.nl wrote: > Harvey Taylor wrote: > This page has been moved to: > http://www.smartmodulartech.com/systems/library/docs/vxworks/vxintro/VXINTRO.HTM > Hmmm. I see what you mean. Hopefully they will fix those pages. [One wonders VMS -> UNIX move?] Anyway thanks for the note Johan. I'll update my links. - -het - --- "Assembly of Japanese bicycle require great peace of mind." - Instruction manual mentioned by R. Pirsig Harvey Taylor mailto:het@despam.portal.ca http://www.pangea.ca/~het --------------------------- Newsgroups: comp.os.vxworks Subject: Re: intro to vx-works Date: Sun, 13 Feb 2000 16:47:56 GMT From: Ephraim Gadsby Message-ID: <8vndas4mlktdae4qinr9ug4kudaakoaa9b@4ax.com> References: thanks for your help --------------------------- Newsgroups: comp.os.vxworks Subject: Re: want to load an object module into memory, while using rom resident vxWorks Date: 13 Feb 2000 21:13:33 GMT From: johan.borkhuis@peektraffic.nl Organization: XS4ALL Internet BV Message-ID: <8876pt$kqn$1@news1.xs4all.nl> References: <881b4l$chr$1@nnrp1.deja.com> findik@my-deja.com wrote: : I am using Tornado II, What system are you using (processor, amount of memory, etc.) Groeten, Johan - -- o o o o o o o . . . ___________________________________ o _____ || Johan Borkhuis | .][__n_n_|DD[ ====_____ | johan@borksoft.xs4all.nl | >(________|__|_[_________]_|________________________________| _/oo OOOOO oo` ooo ooo 'o!o!o o!o!o` === VxWorks page: http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html === --------------------------- Newsgroups: comp.os.vxworks Subject: Re: A "Tracy Riddle" is spamming c.o.vxworks posters Date: Sun, 13 Feb 2000 22:47:55 GMT From: pwhicker@home.net (Paul Whicker) Organization: @Home Network Message-ID: References: <38972EA2.F53E4BAB@aps.anl.gov> <3897D211.57E473C4@bitband.com> First, I DO NOT speak for my employer. In article <3897D211.57E473C4@bitband.com>, leonid@bitband.com wrote: >Wind River has imposed a hiring freeze shortly after the WRS & ISI merger >plans where >announced, and probably will remain in effect until after all parts of the >organizations have >been merged. On the otehr hand, this is a time when many WRS and ISI employees >will be >looking elsewhere for employment, and that is a gold-mine opportunity for a >head hunter too. With respect to the hiring freeze, this isn't quite accurate but is not too far from the truth, the merger information actually precludes them from hiring, this is a standard defensive measure to prevent the management of the merging companies from loading highly paid staff. It is also unfair of you to suggest we are all running for the hills. Wind River is undervalued and not well known to the general public, however we are by far the top embedded o/s and tools company which is a great place to be in this expanding market. I can honestly say that I have heard no negative comments about the merger from within Wind River. I hope ISI have a similar attitude because together we can do even greater things. I look forward to meeting and working with the rest of our people. --------------------------- Newsgroups: comp.os.vxworks Subject: objcopy and binary files Date: 13 Feb 2000 22:49:11 GMT From: defaultuser@domain.com Organization: http://www.msfc.nasa.gov/ Message-ID: <38A734C4.7F3033D@domain.com> Hello, I am attempting to convert a s-record file to binary using objcopymips, but the data segment is filled with '0' in the resulting binary file. Is there any way that I can coax objcopy to do the conversion without losing the data segment? Thanks --------------------------- Newsgroups: comp.os.vxworks Subject: Re: benchmarks for VxWorks on x86 Date: Sun, 13 Feb 2000 22:49:31 GMT From: pwhicker@home.net (Paul Whicker) Organization: @Home Network Message-ID: <%xGp4.31756$Fb1.732808@news1.frmt1.sfba.home.com> References: <38A2E37C.540E4E38@fallschurch.esys.com> Ask your Wind River FAE, they have all this information to hand. Paul In article <38A2E37C.540E4E38@fallschurch.esys.com>, "Keo S. Kelly" wrote: >All: > >Anyone has or know of information concerning benchmarks for VxWorks >on x86 processor? A comparison benchmarks between Linux and VxWorks >would be perfect. > >Thanks. > > > --------------------------- Newsgroups: comp.os.vxworks Subject: C++ - CTORS Date: Mon, 14 Feb 2000 03:51:26 GMT From: Yves Boudreault Organization: Sympatico Message-ID: <38A77BD9.4FE05E7C@sympatico.ca> Hi All, Target: MVME-2604 (PPC-604) Dev: WinNT, Tornado II, GCC I seem to be having with the static constructors. When set in "manual" mode, the object file loads correctly, but fails when I manualy start the static constructors. The object file consists of a number of object files and the Wind Object Library (including RogueWave stuff). Any suggestions to what appears to be a most common problem ? I am also concerned about the size of the image (~ 7MB), it feels much bigger than it should be ... Thanks Yves yves.boudreault2@sympatico.ca --------------------------- Newsgroups: comp.os.vxworks Subject: Re: C++ - CTORS Date: Sun, 13 Feb 2000 21:40:32 -0800 From: Pierre Frigon Organization: Wind River Systems Message-ID: <38A79550.8C32EACA@wrs.com> References: <38A77BD9.4FE05E7C@sympatico.ca> Bonjour Yves, Your image looks pretty big. Normally it should be around 500-700K. Here I assume that you have 2 projects in the Tornado project facility; one that you use to build vxWorks and another that you use to build your application. If your application is linked with the vxWorks image then the size you are getting is probably mostly your application. To fix your constructor initialization problem, stay in automatic constructor initilization mode and download the file with a ".out" extension. This file is built to ensure that any static objects are properly initialized. Also if you are using Tools.h++ you need to call rwEnable for each task using the tools.h collection. In this case you may need to call this routine form the first constructor being initialized. The order of constructor initialization is determined by the link order. To set the link order go to the project facility and select the builds tab. Open the build specification for your project and select the link order tab. Hope this helps, Pierre Yves Boudreault wrote: > Hi All, > > Target: MVME-2604 (PPC-604) > Dev: WinNT, Tornado II, GCC > > I seem to be having with the static constructors. When set in "manual" > mode, the object file loads correctly, but fails when I manualy start > the static constructors. > > The object file consists of a number of object files and the Wind Object > Library (including RogueWave stuff). > > Any suggestions to what appears to be a most common problem ? > > I am also concerned about the size of the image (~ 7MB), it feels much > bigger than it should be ... > > Thanks > > Yves > yves.boudreault2@sympatico.ca --------------------------- Newsgroups: comp.os.vxworks Subject: Re: FAQ for this newsgroup? Date: Mon, 14 Feb 2000 09:12:10 +0200 From: Leonid Rosenboim Organization: BitBand Technologies Ltd. http://www.bitband.com Message-ID: <38A7AAC9.C776A1D7@bitband.com> References: <38A33A66.74E36475@ntl.inf.com> <38A37B6F.5698@despam.pangea.ca> <886b9r$cpc$1@news1.xs4all.nl> Reply-To: leonid@bitband.com Off the subject: the case sensitivity of Web server is a known issue, Windows based Web servers are cese-insensitive for file names, thus such a web site would have worked perfectly on an NT machine, but not on Unix. Perhaps Apache has a case-sensitivity option for file names. First time I encountered this when generating an HTML version of a PowerPoint presentation, the image files had a .GIF extention in the HTML files, but the files themselves had a .gif extention. Leonid johan.borkhuis@peektraffic.nl wrote: > Harvey Taylor wrote: > : Tutorial http://www.digital.com/oem/library/docs/vxintro/vx0001.htm > > This page has been moved to: > http://www.smartmodulartech.com/systems/library/docs/vxworks/vxintro/VXINTRO.HTM > > But this page does not work correctly. They use capitals as names for the > pages, but the pages are named using normal letters. For example the file > vx001.htm is referred to from the intro page, but the filename is > VX001.HTM. I already reported this to the webmaster, but it has not > changed yet. > > Groeten, > Johan > > -- > o o o o o o o . . . ___________________________________ > o _____ || Johan Borkhuis | > .][__n_n_|DD[ ====_____ | johan@borksoft.xs4all.nl | > >(________|__|_[_________]_|________________________________| > _/oo OOOOO oo` ooo ooo 'o!o!o o!o!o` > === VxWorks page: http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html === --------------------------- Newsgroups: comp.os.vxworks Subject: Re: A "Tracy Riddle" is spamming c.o.vxworks posters Date: Mon, 14 Feb 2000 09:19:21 +0200 From: Leonid Rosenboim Organization: BitBand Technologies Ltd. http://www.bitband.com Message-ID: <38A7AC79.DB6D0EA2@bitband.com> References: <38972EA2.F53E4BAB@aps.anl.gov> <3897D211.57E473C4@bitband.com> Reply-To: leonid@bitband.com Well, thanks for the honest response, and accept my appologies for any inacuracies, after all I am not a WRS employee for over a year now, and I get my information second-hand. Anyway, tomorrow, Feb 15, is the sock holders meeting which will most probably vote to approve the merger, and then the merger is final, and then we shall see what comes out. I am however happy for you and that you feel happy at Wind River, perhaps the last year's changes in management have helped their situation. I do wonder however how does the customer community feel about this merger, although I am sure that gutsy memos from customers won't change a thing. Leonid Paul Whicker wrote: > First, I DO NOT speak for my employer. > > In article <3897D211.57E473C4@bitband.com>, leonid@bitband.com wrote: > >Wind River has imposed a hiring freeze shortly after the WRS & ISI merger > >plans where > >announced, and probably will remain in effect until after all parts of the > >organizations have > >been merged. On the otehr hand, this is a time when many WRS and ISI employees > >will be > >looking elsewhere for employment, and that is a gold-mine opportunity for a > >head hunter too. > > With respect to the hiring freeze, this isn't quite accurate but is not too > far from the truth, the merger information actually precludes them from > hiring, this is a standard defensive measure to prevent the management of the > merging companies from loading highly paid staff. > > It is also unfair of you to suggest we are all running for the hills. Wind > River is undervalued and not well known to the general public, however we are > by far the top embedded o/s and tools company which is a great place to be in > this expanding market. > > I can honestly say that I have heard no negative comments about the merger > from within Wind River. I hope ISI have a similar attitude because together we > can do even greater things. I look forward to meeting and working with the > rest of our people. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: routing code Date: Mon, 14 Feb 2000 08:04:57 GMT From: cxj@gsta.com Organization: Deja.com - Before you buy. Message-ID: <888cv8$2tu$1@nnrp1.deja.com> References: <3891E2FE.512785C4@vitesse.com> Hello,Ravi As far as I know, RouterWare has these Routing protocols software to sell. But I don't know how about their performance.The website is: www.routerware.com . Good luck. In article <3891E2FE.512785C4@vitesse.com>, ravi wrote: > Hello Everyone, > > We are trying to build a L2/L3 switch. I was just wondering if anyone is > aware of companies > that sell the code for implementation along with support of Routing > protocols like RIP,OSPF,BGP > and bridging over ethernet. I would appreciate if u could mail me the > names of those companies > along with their contacts if possible. > > Thanks > Ravi > ravi@vitesse.com > > Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: GDB problems Date: Mon, 14 Feb 2000 10:50:34 +0100 From: Andy Winton Organization: Swisscom IP+ (post doesn't reflect views of Swisscom) Message-ID: <38A7CFEA.857A2B76@stest.ch> Hi, In summary, gdb doesn't stop when a breakpoint is hit. This used to be a lot worse with the graphical front end, but we managed to have a work around by using just the command line 'gdbi86' (using the wtx protocol). Now that appears to be broken too. :-( -- We are using: GDB 4.12 on CrossWind 3.1 on VxWorks 5.3.1 for the PC, Wind version 2.5. Sometimes it works in comand line mode, sometimes we get - Proceedure - (1) - Attach Tornado to board (gdb needs the wtx) so that the checksums don't complain... License request... authorized on host 'zeus'. Attaching backend... succeeded. Connecting to target agent... succeeded. Attaching C++ interface... succeeded. Attaching a.out OMF reader... succeeded. Added target_modules to target-server.....done Added radius.lib to target-server.....done (2) - Load the program on vxWorks (3) - Run gdb and attach to board. My t204 is mapped to 'target wtx wd5tu204'. lyra:a_winton> gdbi86 GDB is free software and you are welcome to distribute copies of it under certain conditions; type "show copying" to see the conditions. There is absolutely no warranty for GDB; type "show warranty" for details. +------------------------------------------------------------------+ | CrossWind 3.1 - Copyright (C) 1990-1996 Wind River Systems, Inc. | | This is a modified version of GDB by Wind River Systems Inc. | | Based on code from the original GDB version shown below: | +------------------------------------------------------------------+ GDB 4.12 (sparc-sun-solaris2.4 --target i386-wrs-vxworks), Copyright 1994 Free Software Foundation, Inc. (gdb) t204 Connecting to target server... C++ Constructors/Destructors Strategy is AUTOMATIC Connected to wd5tu204@lyra vxWorks.st.bb586: (no debugging symbols found)...ok tesa.lib: tesa.lib: No such file or directory. tcpip.lib: ok forth.lib: forth.lib: No such file or directory. bcha.lib: ok xntpd: xntpd: No such file or directory. tddb.lib: tddb.lib: No such file or directory. putzfrau.lib: putzfrau.lib: No such file or directory. casd.lib: warning: could not lookup symbol: SYMTBL_SYMBOL_NOT_FOUND ok hdlc.lib: ok foex.lib: foex.lib: No such file or directory. super.lib: warning: could not lookup symbol: SYMTBL_SYMBOL_NOT_FOUND ok atmc.lib: atmc.lib: No such file or directory. sftc.lib: warning: could not lookup symbol: SYMTBL_SYMBOL_NOT_FOUND ok dvm.lib: dvm.lib: No such file or directory. pmi.lib: pmi.lib: No such file or directory. admo.lib: admo.lib: No such file or directory. hi32.lib: ok resu.lib: resu.lib: No such file or directory. strt.lib: ok tmsl.lib: tmsl.lib: No such file or directory. layer1.lib: ok spr.lib: spr.lib: No such file or directory. tdad.lib: tdad.lib: No such file or directory. retd.lib: ok lay2.lib: ok radius.lib: ok Done. (gdb) (gdb) b radius_main Breakpoint 1 at 0xeb8c957: file radiusd.c, line 299. (gdb) r radius_main Starting program: radius_main ^C (4) Just as I do the run, but before I do the Ctrl-C, I see on the console - - -> Unknown breakpoint at 0xeb89253 (5) Gdb does not stop at radius_main, but keeps running. I had to stop this with Ctrl-C. When I did this I get on the console - ... 0xeb8daf8 (ä¸): dbgTrace: invalid case -1 pc=0x11cad0 0xeb8daf8 (ä¸): dbgTrace: invalid case -1 pc=0x1435dc 0xeb8daf8 (ä¸): dbgTrace: invalid case -1 pc=0x11cad0 0xeb8daf8 (ä¸): dbgTrace: invalid case -1 pc=0x11cad0 0xeb8daf8 (ä¸): dbgTrace: invalid case -1 pc=0x11cad0 0xeb8daf8 (ä¸): dbgTrace: invalid case -1 pc=0x11cad0 ... This is in a nutshell one of the problems with GDB. Thanks for your time, andy - -- Andy Winton, andy.winton@stest.ch --------------------------- Newsgroups: comp.os.vxworks Subject: PPP-Callback (RFC 1570) Date: 14 Feb 2000 11:03:46 GMT From: johan.borkhuis@peektraffic.nl Organization: XS4ALL Internet BV Message-ID: <888nei$gpe$1@news1.xs4all.nl> Hello to all, In RFC 1570 (PPP LCP Extensions) a Callback mechanism is defined. We are looking at using this mechanism, but could not find an implementation of this function. Does anyone know of an implementation of this extension or someone who can provide this? Groeten, Johan - -- o o o o o o o . . . ___________________________________ o _____ || Johan Borkhuis | .][__n_n_|DD[ ====_____ | johan@borksoft.xs4all.nl | >(________|__|_[_________]_|________________________________| _/oo OOOOO oo` ooo ooo 'o!o!o o!o!o` === VxWorks page: http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html === --------------------------- End of New-News digest ********************** From vxwexplo-errs@csg.lbl.gov Mon Feb 14 04:32:56 2000 From: stevens Date: Mon Feb 14 04:32:58 PST 2000 Subject: PPP Beta for VxWorks Hi, is anybody out there using the new PPP beta (PPP Framework) from WindRiver Systems? I've the problem, that we have not enough documentation at this time, so I'll need more documentation or demo code for initializing and starting the PPP stack. Torsten -- ************************************************************************** fraunhofer institute of microelectronic circuits and systems department sat name torsten stevens mailto:stevens@ims.fhg.de address finkenstrasse 61 phone +0049/203/3783-238 47057 duisburg fax +0049/203/3783-266 germany http://www.fhg.de/depts/ims-du-e.html ************************************************************************** From vxwexplo-errs@csg.lbl.gov Mon Feb 14 08:39:33 2000 From: "Darnold, Joel R" Date: Mon Feb 14 08:39:35 PST 2000 Subject: Using PMC cards with Extended VME Memory Model 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_01BF7709.EC25CED8 Content-Type: text/plain; charset=iso-8859-1 Content-Transfer-Encoding: 7bit Hello Kirk, I have recently felt your pain integrating PMCs on a MVME2307 board but have successfully added 2 different boards to our system. One of the PMCs is a custom design of ours and another is a Pentek 7110. Both boards use a PLX chip as the PCI bridge. It looks like you are correct with your addressing, the Processor's PCI memory base starts at 0xfd000000 but something you might not have figured out is that it is translated up to 0xfd000000 on the PCI bus. In other words in the VxWorks extended model, the lowest PCI address is 0xfd000000. So.....the address that you would load into your PMC device's PCI base address register would be 0xfd000000 + (PCI address)which would also be the address to use from the Processor's point of view. I've included a utility which will scan thru the PCI bus config and find all the devices on the PCI. It will display all of the configuration registers for anything on the bus. I found it useful when I was setting up my PMCs. Feel free to contact me if you want to chat about this. Joel Darnold Member of Technical Staff Sandia National Laboratories Control Subsystems Dept. 2338 Voice: 505-844-2487 Fax: 505-844-2925 ------_=_NextPart_000_01BF7709.EC25CED8 Content-Type: application/octet-stream; name=IDpciDevices.c.vx Content-Transfer-Encoding: quoted-printable Content-Disposition: attachment; filename=IDpciDevices.c.vx /* IDpciDevices */ /* Show information on PCI devices including vendor and parameters */ /* See PCIvenCodes.txt for additional information */ /* 3/25/99 JRD */ #include "vxWorks.h" #include "stdio.h" #include "sysLib.h" #include "drv/pci/pciLocalBus.h" #include "drv/pci/pciIomapLib.h" #include "MyPCI.h" STATUS pciDevConfig ( int pciBusNo, /* PCI bus number */ int pciDevNo, /* PCI device number */ int pciFuncNo, /* PCI function number */ ULONG devIoBaseAdrs, /* device IO base address */ ULONG devMemBaseAdrs, /* device memory base address = */ ULONG command /* command to issue */ ); struct ID_PCI_DEVICES { int PciVenId; int PciDevId; char VendorName[20]; char DeviceName[40]; }; /* externs */ IMPORT int pciLibInitDone; IMPORT int pciLibInitStatus; IMPORT int pciConfigMech; STATUS IDpciDevices(int busNo) { int i=3D0; int deviceNo; int devices; union { int classCode; char array[4]; } u; struct pciHeaderDevice DeviceInfo;=20 struct ID_PCI_DEVICES IDpciDevices[NUM_IDS_DEFINED]; =09 IDpciDevices[i].PciVenId =3D MOTOROLA_VEN_ID; IDpciDevices[i].PciDevId =3D RAVEN_DEV_ID; sprintf(IDpciDevices[i].VendorName,"Motorola"); sprintf(IDpciDevices[i].DeviceName,"Raven MPC/PCI Bridge"); i++; IDpciDevices[i].PciVenId =3D SYMPHONY_VEN_ID; IDpciDevices[i].PciDevId =3D W83C533_DEV_ID; sprintf(IDpciDevices[i].VendorName,"Symphony"); sprintf(IDpciDevices[i].DeviceName,"W83C533 PCI/ISA Bridge"); i++; IDpciDevices[i].PciVenId =3D TUNDRA_VEN_ID; IDpciDevices[i].PciDevId =3D CA91C042_DEV_ID; sprintf(IDpciDevices[i].VendorName,"Tundra"); sprintf(IDpciDevices[i].DeviceName,"CA91C042 Universe PCI/VME = Bridge"); i++; IDpciDevices[i].PciVenId =3D DEC_VEN_ID; IDpciDevices[i].PciDevId =3D _21140_DEV_ID; sprintf(IDpciDevices[i].VendorName,"Digital"); sprintf(IDpciDevices[i].DeviceName,"21140 Ethernet"); i++; IDpciDevices[i].PciVenId =3D PLX_VEN_ID; IDpciDevices[i].PciDevId =3D PCI9050_DEV_ID; sprintf(IDpciDevices[i].VendorName,"PLX Technology"); sprintf(IDpciDevices[i].DeviceName,"PCI9050 PCI Interface (CLIB)"); i++; IDpciDevices[i].PciVenId =3D PENTEK_VEN_ID; IDpciDevices[i].PciDevId =3D _7110_DEV_ID; sprintf(IDpciDevices[i].VendorName,"Pentek"); sprintf(IDpciDevices[i].DeviceName,"7110 C44 DSP PMC Module"); if (pciLibInitStatus !=3D OK) /* sanity check */ return (ERROR); printf ("Scanning function 0 of each PCI device on bus %d\n", busNo); printf ("Using configuration mechanism %d\n", pciConfigMech); printf ("bus dev func venID devID class description\n"); if (pciConfigMech =3D=3D PCI_MECHANISM_1) devices =3D 0x1f; else devices =3D 0x0f; for (deviceNo=3D0; deviceNo < devices; deviceNo++) { pciConfigInWord (busNo, deviceNo, 0, PCI_CFG_VENDOR_ID, = &DeviceInfo.vendorId); pciConfigInWord (busNo, deviceNo, 0, PCI_CFG_DEVICE_ID, = &DeviceInfo.deviceId); pciConfigInByte (busNo, deviceNo, 0, = PCI_CFG_PROGRAMMING_IF,&u.array[3]); pciConfigInByte (busNo, deviceNo, 0, PCI_CFG_SUBCLASS, &u.array[2]); pciConfigInByte (busNo, deviceNo, 0, PCI_CFG_CLASS, &u.array[1]); u.array[0] =3D 0; /* There are two ways to find out an empty device. * 1. check Master Abort bit after the access. * 2. check whether the read value is 0xffff. * Since I didn't see the Master Abort bit of the host/PCI bridge * changing, I use the second method. */ if ((unsigned short)DeviceInfo.vendorId !=3D 0xffff) { pciConfigInWord (busNo, deviceNo, 0, PCI_CFG_VENDOR_ID, = &DeviceInfo.vendorId); pciConfigInWord (busNo, deviceNo, 0, PCI_CFG_DEVICE_ID, = &DeviceInfo.deviceId); pciConfigInWord (busNo, deviceNo, 0, PCI_CFG_COMMAND, = &DeviceInfo.command); pciConfigInWord (busNo, deviceNo, 0, PCI_CFG_STATUS, = &DeviceInfo.status); pciConfigInByte (busNo, deviceNo, 0, PCI_CFG_REVISION, = &DeviceInfo.revisionId); pciConfigInByte (busNo, deviceNo, 0, PCI_CFG_PROGRAMMING_IF, = &DeviceInfo.classCode); pciConfigInByte (busNo, deviceNo, 0, PCI_CFG_SUBCLASS, = &DeviceInfo.progIf); pciConfigInByte (busNo, deviceNo, 0, PCI_CFG_CLASS, = &DeviceInfo.cacheLine); pciConfigInByte (busNo, deviceNo, 0, PCI_CFG_CACHE_LINE_SIZE, = &DeviceInfo.cacheLine); pciConfigInByte (busNo, deviceNo, 0, PCI_CFG_LATENCY_TIMER, = &DeviceInfo.latency); pciConfigInByte (busNo, deviceNo, 0, PCI_CFG_HEADER_TYPE, = &DeviceInfo.headerType); pciConfigInByte (busNo, deviceNo, 0, PCI_CFG_BIST, = &DeviceInfo.bist); pciConfigInLong (busNo, deviceNo, 0, PCI_CFG_BASE_ADDRESS_0, = &DeviceInfo.base0); pciConfigInLong (busNo, deviceNo, 0, PCI_CFG_BASE_ADDRESS_1, = &DeviceInfo.base1); pciConfigInLong (busNo, deviceNo, 0, PCI_CFG_BASE_ADDRESS_2, = &DeviceInfo.base2); pciConfigInLong (busNo, deviceNo, 0, PCI_CFG_BASE_ADDRESS_3, = &DeviceInfo.base3); pciConfigInLong (busNo, deviceNo, 0, PCI_CFG_BASE_ADDRESS_4, = &DeviceInfo.base4); pciConfigInLong (busNo, deviceNo, 0, PCI_CFG_BASE_ADDRESS_5, = &DeviceInfo.base5); pciConfigInLong (busNo, deviceNo, 0, PCI_CFG_CIS, &DeviceInfo.cis); pciConfigInWord (busNo, deviceNo, 0, PCI_CFG_SUB_VENDER_ID, = &DeviceInfo.subVendorId); pciConfigInWord (busNo, deviceNo, 0, PCI_CFG_SUB_SYSTEM_ID, = &DeviceInfo.subSystemId); pciConfigInLong (busNo, deviceNo, 0, PCI_CFG_EXPANSION_ROM, = &DeviceInfo.romBase); pciConfigInLong (busNo, deviceNo, 0, PCI_CFG_RESERVED_0, = &DeviceInfo.reserved0); pciConfigInLong (busNo, deviceNo, 0, PCI_CFG_RESERVED_1, = &DeviceInfo.reserved1); pciConfigInByte (busNo, deviceNo, 0, PCI_CFG_DEV_INT_LINE, = &DeviceInfo.intLine); pciConfigInByte (busNo, deviceNo, 0, PCI_CFG_DEV_INT_PIN, = &DeviceInfo.intPin); pciConfigInByte (busNo, deviceNo, 0, PCI_CFG_MIN_GRANT, = &DeviceInfo.minGrant); pciConfigInByte (busNo, deviceNo, 0, PCI_CFG_MAX_LATENCY, = &DeviceInfo.maxLatency); /* printf("Found device VendorID %x DeviceID = %x\n",DeviceInfo.vendorId,DeviceInfo.deviceId); */ for(i=3D0; i\n", busNo, deviceNo, 0,(unsigned short)DeviceInfo.vendorId, (unsigned short)DeviceInfo.deviceId, u.classCode, IDpciDevices[i].VendorName,IDpciDevices[i].DeviceName); } } printf("vendorId 0x%.5x\n",(unsigned short)DeviceInfo.vendorId); = /* vendor ID */ printf("deviceId 0x%.5x\n",(unsigned short)DeviceInfo.deviceId); = /* device ID */ printf("command 0x%.5x\n",(unsigned short)DeviceInfo.command); = /* command register */ printf("status 0x%.5x\n",(unsigned short)DeviceInfo.status); = /* status register */ printf("revisionrId 0x%.3x\n",(unsigned char = )DeviceInfo.revisionId); /* revision ID */ printf("classCode 0x%.3x\n",(unsigned char )DeviceInfo.classCode); = /* class code */ printf("subClass 0x%.3x\n",(unsigned char )DeviceInfo.subClass); = /* sub class code */ printf("progIf 0x%.3x\n",(unsigned char )DeviceInfo.progIf); = /* programming interface */ printf("cacheLine 0x%.3x\n",(unsigned char )DeviceInfo.cacheLine); = /* cache line */ printf("latency 0x%.3x\n",(unsigned char )DeviceInfo.latency); = /* latency time */ printf("headerType 0x%.3x\n",(unsigned char = )DeviceInfo.headerType); /* header type */ printf("bist 0x%.3x\n",(unsigned char )DeviceInfo.bist); /* = BIST */ printf("base0 0x%.8x\n",(unsigned int )DeviceInfo.base0); /* = base address 0 */ printf("base1 0x%.8x\n",(unsigned int )DeviceInfo.base1); /* = base address 1 */ printf("base2 0x%.8x\n",(unsigned int )DeviceInfo.base2); /* = base address 2 */ printf("base3 0x%.8x\n",(unsigned int )DeviceInfo.base3); /* = base address 3 */ printf("base4 0x%.8x\n",(unsigned int )DeviceInfo.base4); /* = base address 4 */ printf("base5 0x%.8x\n",(unsigned int )DeviceInfo.base5); /* = base address 5 */ printf("cis 0x%.8x\n",(unsigned int )DeviceInfo.cis); /* = cardBus CIS pointer */ printf("subVendorId 0x%.5x\n",(unsigned = short)DeviceInfo.subVendorId); /* sub system vendor ID */ printf("subSystemId 0x%.5x\n",(unsigned = short)DeviceInfo.subSystemId); /* sub system ID */ printf("romBase 0x%.8x\n",(unsigned int )DeviceInfo.romBase); = /* expansion ROM base address */ printf("reserved0 0x%.8x\n",(unsigned int )DeviceInfo.reserved0); = /* reserved */ printf("reserved1 0x%.8x\n",(unsigned int )DeviceInfo.reserved1); = /* reserved */ printf("intLine 0x%.3x\n",(unsigned char )DeviceInfo.intLine); = /* interrupt line */ printf("intPin 0x%.3x\n",(unsigned char )DeviceInfo.intPin); = /* interrupt pin */ printf("minGrant 0x%.3x\n",(unsigned char )DeviceInfo.minGrant); = /* min Grant */ printf("maxLatency 0x%.3x\n",(unsigned char = )DeviceInfo.maxLatency); /* max Latency */ } } return (OK); } /* This routine writes data to PLX's local configuration */ /* registers. The data is written as little endian so we do */ /* a 32 bit endian swap before we write the data */ /* Remember.... eprom data requires no data swappage */ void PciDevWriteLocCfgReg(int PCIBase,int RegAddress, int Data) { int SwapData=3D0; SwapData =3D (Data & 0xff) << 24; SwapData |=3D (Data & 0xff00) << 8; SwapData |=3D (Data & 0xff0000) >> 8; SwapData |=3D (Data & 0xff000000) >> 24; *(volatile int *)(PCIBase+RegAddress) =3D SwapData; } int PciDevReadLocCfgReg(int PCIBase, int RegAddress) { int SwapData=3D0,Data; Data =3D *(volatile int *)(PCIBase+RegAddress); SwapData =3D (Data & 0xff) << 24; SwapData |=3D (Data & 0xff00) << 8; SwapData |=3D (Data & 0xff0000) >> 8; SwapData |=3D (Data & 0xff000000) >> 24; return(SwapData); } /* The following are general purpose memory reading routines */ void Peek8(int Address) { char cData; cData =3D *(volatile char *)Address; printf("0x%x =3D 0x%x\n",Address, (0xff & (int)cData)); } void Peek16(int Address) { short sData; sData =3D *(volatile short *)Address; printf("0x%x =3D 0x%x\n",Address, (0xffff & (int)sData)); } void Peek32(int Address) { int iData; iData =3D *(volatile int *)Address; printf("0x%x =3D 0x%x\n",Address, iData); } void Poke8(int Address, int cData) { *(volatile char *)Address =3D (char)cData; printf("0x%x =3D 0x%x\n",Address, (0xff & (int)cData)); } void Poke16(int Address, int sData) { *(volatile short *)Address =3D (short)sData; printf("0x%x =3D 0x%x\n",Address, (0xffff & (int)sData)); } void Poke32(int Address, int iData) { *(volatile int *)Address =3D iData; printf("0x%x =3D 0x%x\n",Address, iData); } void Init7110(void) { int pciBusNo,pciDevNo,pciFuncNo; int iDummy,i; short sDummy; printf("\n\n"); printf("CPU_PCI_IO_ADRS 0x%x PCI_IO_ADRS = 0x%x\n",CPU_PCI_IO_ADRS,PCI_IO_ADRS); printf("CPU_PCI_MEM_ADRS 0x%x PCI_MEM_ADRS = 0x%x\n\n",CPU_PCI_MEM_ADRS,PCI_MEM_ADRS); /* printf("CLIB_PCI_FIFO_ADRS 0x%x CLIB_PCI_ALTERA_ADRS = 0x%x\n",CLIB_PCI_FIFO_ADRS,CLIB_PCI_ALTERA_ADRS); printf("CLIB_CPU_FIFO_ADRS 0x%x CLIB_CPU_ALTERA_ADRS = 0x%x\n\n",CLIB_CPU_FIFO_ADRS|CPU_PCI_MEM_ADRS,CLIB_CPU_ALTERA_ADRS|CPU_P= CI_IO_ADRS); */ if((pciFindDevice(C44_PCI_DEVICE_ID&0xffff, (C44_PCI_DEVICE_ID>>16)&0xffff, 0,&pciBusNo,&pciDevNo,&pciFuncNo) !=3D ERROR)) { /* Initialize PCI Configuration Registers enough to access Local = Configuration Registers */ pciConfigOutLong (pciBusNo, pciDevNo, pciFuncNo, = PCI_CFG_BASE_ADDRESS_0, C44_PCI_MEM0_ADRS); pciConfigInLong (pciBusNo, pciDevNo, pciFuncNo, = PCI_CFG_BASE_ADDRESS_0, &iDummy); printf("PCI_CFG_BASE_ADDRESS_0 =3D 0x%x\n",iDummy); pciConfigOutLong (pciBusNo, pciDevNo, pciFuncNo, = PCI_CFG_BASE_ADDRESS_1, C44_PCI_IO0_ADRS); pciConfigInLong (pciBusNo, pciDevNo, pciFuncNo, = PCI_CFG_BASE_ADDRESS_1, &iDummy); printf("PCI_CFG_BASE_ADDRESS_1 =3D 0x%x\n",iDummy); pciConfigOutLong (pciBusNo, pciDevNo, pciFuncNo, = PCI_CFG_BASE_ADDRESS_2, 0); pciConfigInLong (pciBusNo, pciDevNo, pciFuncNo, = PCI_CFG_BASE_ADDRESS_2, &iDummy); printf("PCI_CFG_BASE_ADDRESS_2 =3D 0x%x\n",iDummy); pciConfigOutLong (pciBusNo, pciDevNo, pciFuncNo, = PCI_CFG_BASE_ADDRESS_3, 0); pciConfigInLong (pciBusNo, pciDevNo, pciFuncNo, = PCI_CFG_BASE_ADDRESS_3, &iDummy); printf("PCI_CFG_BASE_ADDRESS_3 =3D 0x%x\n",iDummy); pciConfigOutLong (pciBusNo, pciDevNo, pciFuncNo, = PCI_CFG_BASE_ADDRESS_4, 0); pciConfigInLong (pciBusNo, pciDevNo, pciFuncNo, = PCI_CFG_BASE_ADDRESS_4, &iDummy); printf("PCI_CFG_BASE_ADDRESS_4 =3D 0x%x\n",iDummy); pciConfigOutLong (pciBusNo, pciDevNo, pciFuncNo, = PCI_CFG_BASE_ADDRESS_5, 0); pciConfigInLong (pciBusNo, pciDevNo, pciFuncNo, = PCI_CFG_BASE_ADDRESS_5, &iDummy); printf("PCI_CFG_BASE_ADDRESS_5 =3D 0x%x\n",iDummy); pciConfigOutWord (pciBusNo, pciDevNo, pciFuncNo, PCI_CFG_COMMAND, ( = PCI_CMD_MEM_ENABLE | PCI_CMD_IO_ENABLE)); pciConfigInWord (pciBusNo, pciDevNo, pciFuncNo, PCI_CFG_COMMAND, = &sDummy ); printf("PCI_CFG_COMMAND =3D 0x%x\n",sDummy); /* Now we can access Local Configuration Registers so begin to do so = */ /* Register settings to enable basic communication to PLX9080 has = been */ /* initialized from EEPROM at 7110 power on */ #ifdef SETUP_LOCAL_CONFIG_REGISTERS PciDevWriteLocCfgReg(C44_PCI_MEM0_ADRS, CNTRL, = 0x00ff7580);/*0x00ff7580*//*0x00ff3580*//*0x00fd3580*/ /* Control */ PciDevWriteLocCfgReg(C44_PCI_MEM0_ADRS, LAS0RR, 0x0ffffff0); /* Fifo = mapping */ PciDevWriteLocCfgReg(C44_PCI_MEM0_ADRS, LAS0BA, 0x000003c1); /* A0 & = A1 of Local bus aren't used */ PciDevWriteLocCfgReg(C44_PCI_MEM0_ADRS, LAS0BRD, = 0x0180003f);/*0x01800022*//*0x0180003f*//*0x0180003a*//*0x0180001a*//* = Fifo A0 =3D Local A2 */ PciDevWriteLocCfgReg(C44_PCI_MEM0_ADRS, CS0BASE, 0x000003c5); /* = Fifo A1 =3D Local A3 */ /* Shift Address bits accordingly */ PciDevWriteLocCfgReg(C44_PCI_MEM0_ADRS, LAS1RR, 0x0fffffe1); /* = Altera mapping */ PciDevWriteLocCfgReg(C44_PCI_MEM0_ADRS, LAS1BA, 0x00000141); /* A0 & = A1 of Local bus aren't used */ PciDevWriteLocCfgReg(C44_PCI_MEM0_ADRS, LAS1BRD, 0x014060c2); /* = Altera A0 =3D Local A2 */ PciDevWriteLocCfgReg(C44_PCI_MEM0_ADRS, CS1BASE, 0x00000000); /* = Altera A1 =3D Local A3 */ /* Shift Address bits accordingly */ /* Now we can enable these memory maps in the PCI Configuration = Registers */ pciConfigOutLong (pciBusNo, pciDevNo, pciFuncNo, = PCI_CFG_BASE_ADDRESS_2, C44_PCI_FIFO_ADRS); pciConfigInLong (pciBusNo, pciDevNo, pciFuncNo, = PCI_CFG_BASE_ADDRESS_2, &iDummy); printf("PCI_CFG_BASE_ADDRESS_2 =3D 0x%x\n",iDummy); pciConfigOutLong (pciBusNo, pciDevNo, pciFuncNo, = PCI_CFG_BASE_ADDRESS_3, CLIB_PCI_ALTERA_ADRS); pciConfigInLong (pciBusNo, pciDevNo, pciFuncNo, = PCI_CFG_BASE_ADDRESS_3, &iDummy); printf("PCI_CFG_BASE_ADDRESS_3 =3D 0x%x\n",iDummy); #endif } else { printf("Error initializing Pentek 7110\n"); } printf(" 7110 setup: pciBusNo =3D %d\n",pciBusNo); printf(" pciDevNo =3D %d\n",pciDevNo); printf(" pciFuncNo =3D %d\n",pciFuncNo); printf(" C44_PCI_IO0_ADRS 0x%x\n",C44_PCI_IO0_ADRS); printf(" C44_PCI_MEM0_ADRS 0x%x\n",C44_PCI_MEM0_ADRS); printf("\nLocal configuration registers\n"); for(i=3D0; i<12; i++) { iDummy =3D PciDevReadLocCfgReg(C44_PCI_MEM0_ADRS,(i*4)); printf("0x%x =3D 0x%x\n",(i*4), iDummy); } =09 for(i=3D16; i<20; i++) { iDummy =3D PciDevReadLocCfgReg(C44_PCI_MEM0_ADRS,(i*4)); printf("0x%x =3D 0x%x\n",(i*4), iDummy); } =09 for(i=3D24; i<29; i++) { iDummy =3D PciDevReadLocCfgReg(C44_PCI_MEM0_ADRS,(i*4)); printf("0x%x =3D 0x%x\n",(i*4), iDummy); } =09 } void PCIReadBack7110(void) { int pciBusNo,pciDevNo,pciFuncNo; int Bar0,Bar1,Bar2,Bar3,Bar4,Bar5; if((pciFindDevice(C44_PCI_DEVICE_ID&0xffff, (C44_PCI_DEVICE_ID>>16)&0xffff, 0,&pciBusNo,&pciDevNo,&pciFuncNo) !=3D ERROR)) { pciConfigOutLong (pciBusNo, pciDevNo, pciFuncNo, = PCI_CFG_BASE_ADDRESS_0, 0xffffffff); pciConfigOutLong (pciBusNo, pciDevNo, pciFuncNo, = PCI_CFG_BASE_ADDRESS_1, 0xffffffff); pciConfigOutLong (pciBusNo, pciDevNo, pciFuncNo, = PCI_CFG_BASE_ADDRESS_2, 0xffffffff); pciConfigOutLong (pciBusNo, pciDevNo, pciFuncNo, = PCI_CFG_BASE_ADDRESS_3, 0xffffffff); pciConfigOutLong (pciBusNo, pciDevNo, pciFuncNo, = PCI_CFG_BASE_ADDRESS_4, 0xffffffff); pciConfigOutLong (pciBusNo, pciDevNo, pciFuncNo, = PCI_CFG_BASE_ADDRESS_5, 0xffffffff); pciConfigInLong (pciBusNo, pciDevNo, pciFuncNo, = PCI_CFG_BASE_ADDRESS_0, &Bar0); pciConfigInLong (pciBusNo, pciDevNo, pciFuncNo, = PCI_CFG_BASE_ADDRESS_1, &Bar1); pciConfigInLong (pciBusNo, pciDevNo, pciFuncNo, = PCI_CFG_BASE_ADDRESS_2, &Bar2); pciConfigInLong (pciBusNo, pciDevNo, pciFuncNo, = PCI_CFG_BASE_ADDRESS_3, &Bar3); pciConfigInLong (pciBusNo, pciDevNo, pciFuncNo, = PCI_CFG_BASE_ADDRESS_4, &Bar4); pciConfigInLong (pciBusNo, pciDevNo, pciFuncNo, = PCI_CFG_BASE_ADDRESS_5, &Bar5); printf(" 7110 PCI Readback Base Address Registers\n"); printf(" PCI_CFG_BASE_ADDRESS_0 =3D 0x%x\n",Bar0); printf(" PCI_CFG_BASE_ADDRESS_1 =3D 0x%x\n",Bar1); printf(" PCI_CFG_BASE_ADDRESS_2 =3D 0x%x\n",Bar2); printf(" PCI_CFG_BASE_ADDRESS_3 =3D 0x%x\n",Bar3); printf(" PCI_CFG_BASE_ADDRESS_4 =3D 0x%x\n",Bar4); } else { printf("Error initializing CLIB\n"); } } ------_=_NextPart_000_01BF7709.EC25CED8 Content-Type: application/octet-stream; name=CLIB.h Content-Transfer-Encoding: quoted-printable Content-Disposition: attachment; filename=CLIB.h /* CLIB.h */ /* Definitions for Custom Laser Interface Board */ /***********************************************************************= ******* * LEDS * User I/O P5 Programming P1 G G Y R Reset * ------------- ----------- ----------- --- * | o o o o o o1| | o o o o o1| |7O O O O1| | X | * | o o o o o o2| | o o o o o2| |8O O O O2| | | * ------------- ----------- ------------ --- * * * Leds: Red 1 - EEPROM_CS * Red 2 - Error (ERR) * Yel 3 - Fifo Full (FIFO_FULL) * Yel 4 - Fifo Empty (FIFO_EMPTY) * Grn 5 - Fifo Write Enable (FIFO_WREN_N) * Grn 6 - Fifo Output Enable (FIFO_OEN) * Grn 7 - 10MHz Clock Select (10MHz_SEL) * Grn 8 - Running (~1Hz) (RUN) * * P5: 1 - * 2 - System tick * * SW2 (Dip Switch): SW Open Closed * 1 Reticle Wafer * 2 Stop Run * 3 33MHz 3 & 4 Control the local PCI = clock, 3 or 4 * 4 10Mhz must be closed but not both * ************************************************************************= ******/ /* #include "vxWorks.h" #include "stdio.h" #include "sysLib.h" #include "math.h" #include "drv/pci/pciLocalBus.h" #include "drv/pci/pciIomapLib.h" #include "fppLib.h" #include "intLib.h" #include "arch/ppc/ivPpc.h" */ #ifndef INCLUDE_MPIC /* These are required for getting correct = definitions from vm2600.h */ #define INCLUDE_MPIC #endif=20 #ifndef EXTENDED_VME #define EXTENDED_VME #endif #include "c:/tornado/target/config/mv2304/mv2600.h" #define PLX_VEN_ID 0x10b5 #define PCI9050_DEV_ID 0x9050 /* PLX 9050 Local Control Registers Address offsets */ #define LAS0RR 0x00 #define LAS1RR 0x04 #define LAS2RR 0x08 #define LAS3RR 0x0c #define EROMRR 0x10 #define LAS0BA 0x14 #define LAS1BA 0x18 #define LAS2BA 0x1c #define LAS3BA 0x20 #define EROMBA 0x24 #define LAS0BRD 0x28 #define LAS1BRD 0x2c #define LAS2BRD 0x30 #define LAS3BRD 0x34 #define EROMBRD 0x38 #define CS0BASE 0x3c #define CS1BASE 0x40 #define CS2BASE 0x44 #define CS3BASE 0x48 #define INTCSR 0x4c #define CNTRL 0x50 /* CLIB PCI Configuration information */ #define CLIB_PCI_DEVICE_ID (((PCI9050_DEV_ID<<16)&0xffff0000) | = PLX_VEN_ID) /* 0x905010b5 */ /* Used for pciFindDevice() */ #define CLIB_PMC_IO_SPACE 0x40100 /* CLIB Base for Local Control = Registers in PCI I/O Space */ #define CLIB_PMC_MEM_SPACE 0x40000 /* CLIB Base for Local Control = Registers in PCI Mem Space */ #define CLIB_PCI_IO0_ADRS (PCI_IO_ADRS + CLIB_PMC_IO_SPACE) /* CPU = address for Local Control Registers in PCI I/O Space */ #define CLIB_PCI_MEM0_ADRS (PCI_MEM_ADRS + CLIB_PMC_MEM_SPACE) /* CPU = address for Local Control Registers in PCI MEM Space */ #define CLIB_PCI_FIFO_ADRS (CLIB_PCI_MEM0_ADRS+0x100) /* PCI address of = CLIB Fifo */ #define CLIB_PCI_ALTERA_ADRS (CLIB_PCI_IO0_ADRS+0x100) /* PCI address = if CLIB Altera registers */ #define CLIB_CPU_FIFO_ADRS (CLIB_PCI_FIFO_ADRS) /* CPU address of CLIB = Fifo */ #define CLIB_CPU_ALTERA_ADRS (CLIB_PCI_ALTERA_ADRS) /* CPU address of = CLIB Altera registers */ /* CLIB Altera register address offsets on local bus */ #define CLIB_CONTROL 0x00 /* Register 0 */ #define CLIB_SAMP_MODULUS_LO 0x04 /* Register 1 */ #define CLIB_SAMP_MODULUS_HI 0x08 /* Register 2 */ #define CLIB_OVSAMP_MODULUS 0x0c /* Register 3 */ #define CLIB_USER_IO_DIR 0x10 /* Register 4 */ #define CLIB_UNUSED 0x14 /* Register 5 */ #define CLIB_ERROR 0x18 /* Register 6 */ #define CLIB_STATUS 0x1c /* Register 7 */ /* CLIB Alter register address on PCI bus */ #define CLIB_PCI_CONTROL (CLIB_CONTROL + CLIB_PCI_ALTERA_ADRS) #define CLIB_PCI_SAMP_MODULUS_LO (CLIB_SAMP_MODULUS_LO + = CLIB_PCI_ALTERA_ADRS) #define CLIB_PCI_SAMP_MODULUS_HI (CLIB_SAMP_MODULUS_HI + = CLIB_PCI_ALTERA_ADRS) #define CLIB_PCI_OVSAMP_MODULUS (CLIB_OVSAMP_MODULUS + = CLIB_PCI_ALTERA_ADRS) #define CLIB_PCI_USER_IO_DIR (CLIB_USER_IO_DIR + CLIB_PCI_ALTERA_ADRS) #define CLIB_PCI_UNUSED (CLIB_UNUSED + CLIB_PCI_ALTERA_ADRS) #define CLIB_PCI_ERROR (CLIB_ERROR + CLIB_PCI_ALTERA_ADRS) #define CLIB_PCI_STATUS (CLIB_STATUS + CLIB_PCI_ALTERA_ADRS) /* CLIB Alter register address viewed from Processor */ #define CLIB_CPU_CONTROL (CLIB_CONTROL + CLIB_CPU_ALTERA_ADRS) #define CLIB_CPU_SAMP_MODULUS_LO (CLIB_SAMP_MODULUS_LO + = CLIB_CPU_ALTERA_ADRS) #define CLIB_CPU_SAMP_MODULUS_HI (CLIB_SAMP_MODULUS_HI + = CLIB_CPU_ALTERA_ADRS) #define CLIB_CPU_OVSAMP_MODULUS (CLIB_OVSAMP_MODULUS + = CLIB_CPU_ALTERA_ADRS) #define CLIB_CPU_USER_IO_DIR (CLIB_USER_IO_DIR + CLIB_CPU_ALTERA_ADRS) #define CLIB_CPU_UNUSED (CLIB_UNUSED + CLIB_CPU_ALTERA_ADRS) #define CLIB_CPU_ERROR (CLIB_ERROR + CLIB_CPU_ALTERA_ADRS) #define CLIB_CPU_STATUS (CLIB_STATUS + CLIB_CPU_ALTERA_ADRS) /* CLIB register bit definitions */ /* CLIB_CONTROL register 0 */ #define CLIB_CTRL_SCLR_SAMPLE_CNT 0x01 /* 0=3Dincrement sample count = for each control sample, 1=3Dsample count stays zero , default=3D0 */ #define CLIB_IF_ENABLE 0x02 /* 0=3Dinterf. control signals (IF_AOH_N = & IF_RDEN_N) disabled, 1=3Dinterf. interface enabled,default=3D0 */ #define CLIB_FIFO_ENABLE 0x04 /* 0=3Dfifo control signal (F_WREN_N) = disabled, 1=3Dfifo interface enabled, default=3D0 */ #define CLIB_IF_POSN_RST_N 0x08 /* 0=3Dall interf. position counters = reset to 0,1=3Dall counts show position, default=3D1 */ #define CLIB_TERM_ACCESS_EN 0x10 /* 0=3Dalter will only terminate = local access that are addressed to CLIB, 1=3Dterminate all local = accesses, default=3D0 */ #define CLIB_REG_UNUSED 0x20 #define CLIB_MASTER 0x40 /* 0=3Dslave mode, 1=3Dmaster mode, = default=3D1 */ #define CLIB_RUN 0x80 /* 0=3Dstop, 1=3Drun, default=3D0 */ #define CLIB_CONTROL_RUN_MASTER_IF = (CLIB_IF_ENABLE|CLIB_FIFO_ENABLE|CLIB_IF_POSN_RST_N|CLIB_TERM_ACCESS_EN|= CLIB_MASTER|CLIB_RUN) #define CLIB_CONTROL_STOP_MASTER_IF = (CLIB_IF_ENABLE|CLIB_FIFO_ENABLE|CLIB_IF_POSN_RST_N|CLIB_TERM_ACCESS_EN|= CLIB_MASTER) #define CLIB_CONTROL_RUN_SLAVE_IF = (CLIB_IF_ENABLE|CLIB_FIFO_ENABLE|CLIB_IF_POSN_RST_N|CLIB_TERM_ACCESS_EN|= CLIB_RUN) #define CLIB_CONTROL_STOP_SLAVE_IF = (CLIB_IF_ENABLE|CLIB_FIFO_ENABLE|CLIB_IF_POSN_RST_N|CLIB_TERM_ACCESS_EN)= /* CLIB_SAMP_MODULUS(s) Registers 1-3*/ /* The master CLIB clock is 10Mhz interf. clock, determine desired = sample rate, * multiply this by number of oversamples + 1 (for servo sample), = divide this * into 10MHz, this is value written into CLIB_SAMP_MODULUS(s) * * Example * Desired servo rate =3D 7500Hz with 2 oversamples * 2*7500 =3D 22500Hz, * 100000000/22500 =3D 444.444 * (int)444.444 =3D 0x01bc (actual servo rate =3D 10Mhz/444 =3D = 22.522kHz/3 =3D 7507.5Hz * * write 0xbc to CLIB_CPU_SAMP_MODULUS_LO (lo byte of 0x01bc ); * write 0x01 to CLIB_CPU_SAMP_MODULUS_HI (hi byte of 0x01bc ); * write 0x02 to CLIB_CPU_OVSAMP_MODULUS (meaning 2 oversamples) */ /* CLIB_USER_IO_DIR Register 4*/ #define CLIB_USER_IO1_DIR 0x01 /* CLIB_ERROR Register 6 */ /* A write to CLIB_ERROR clears the error register */ /* a 1 =3D error condition */ #define CLIB_FIFO_OEN_CONFLICT 0x01 #define CLIB_FIFO_FLAG_ERROR 0x02 #define CLIB_FIFO_FULL_ERROR 0x04 #define CLIB_TERM_ACCESS_ERROR 0x08 /* bits 4-7 are unused */ /* CLIB_STATUS Register 7 */ /* A write to CLIB_STATUS register clears local interrupt (LPCI_IRQ) */ #define CLIB_FIFO_EMPTY 0x01 /* 1=3Dfifo is empty */ #define CLIB_FIFO_FULL 0x02 /* 1=3Dfifo is full */ #define CLIB_IRQ_ASSERTED 0x04 /* 0=3Dlocal interrupt is asserted */ #define CLIB_STATUS_UNUSED 0x08 #define CLIB_ALTERA_ERROR 0x10 /* the logical OR of all error register = bits */ #define CLIB_10MHZ_SELECTED 0x20 /* 1=3DCLIB is using interferometer = clock */ #define CLIB_MASTER_SELECTED 0x40 /* connected to register 0_6 = (1=3DMaster) */ #define CLIB_RUN_SELECTED 0x80 /* connected to register 0_7 (1=3DRun) = */ ------_=_NextPart_000_01BF7709.EC25CED8-- From vxwexplo-errs@csg.lbl.gov Mon Feb 14 10:15:40 2000 From: "Larry Sinn" Date: Mon Feb 14 10:15:42 PST 2000 Subject: Re: TOD clock on an MVME2700 This is a multi-part message in MIME format. --------------218262F3899371DBFC7F30B3 Content-Type: multipart/alternative; boundary="------------9F852866566238FF4040A8C2" --------------9F852866566238FF4040A8C2 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Daniel: Attached are a couple of routines that work on mvme2300 with UNIX host. This gets time from both the PPC and UNIX host. Larry. Daniel wrote: > I would like to read the "Time of Day" clock on an > MVME2700, has anyone done this, if so, could you give me > some pointers ? > Larry Sinn Spectral Dynamics (408)474-1746 voice 1983 Concourse Drive (408)474-1780 fax San Jose, Ca. 95131-1708 sinnl@sd-star.com --------------9F852866566238FF4040A8C2 Content-Type: text/html; charset=us-ascii Content-Transfer-Encoding: 7bit Daniel:
Attached are a couple of routines that work on mvme2300 with UNIX host.
This gets time from both the PPC and UNIX host.

Larry.
 
 

Daniel wrote: 

I would like to read the "Time of Day" clock on an
MVME2700, has anyone done this, if so, could you give me 
some pointers ?


Larry Sinn
Spectral Dynamics               (408)474-1746  voice
1983 Concourse Drive            (408)474-1780  fax
San Jose, Ca. 95131-1708        sinnl@sd-star.com
 
  --------------9F852866566238FF4040A8C2-- --------------218262F3899371DBFC7F30B3 Content-Type: text/plain; charset=us-ascii; name="getMotTime.c" Content-Transfer-Encoding: 7bit Content-Disposition: inline; filename="getMotTime.c" /* getMotTime.c */ static char SccsId[] UNUSED = "@(#)getMotTime.c 1.1 06/29/99 Spectral Dynamics" ; /* * Get time from Motorola PPC */ #include "stdio.h" #include "mv2600.h" #include "vxWorks.h" #include "time.h" #include "timers.h" extern UCHAR sysNvRead(ULONG); time_t getMotTime(void) { struct tm timeAndDate; unsigned char timeData[8]; int i; for (i = 0; i < 8; i++) timeData[i] = sysNvRead((ULONG)TOD_CLOCK + i); timeData[0] = 0x00; timeAndDate.tm_sec = (timeData[1] >> 4) * 10 + (timeData[1] & 0x0F); timeAndDate.tm_min = (timeData[2] >> 4) * 10 + (timeData[2] & 0x0F); timeAndDate.tm_hour = (timeData[3] >> 4) * 10 + (timeData[3] & 0x0F); timeAndDate.tm_mday = (timeData[5] >> 4) * 10 + (timeData[5] & 0x0F); timeAndDate.tm_mon = (timeData[6] >> 4) * 10 + (timeData[6] & 0x0F) - 1; timeAndDate.tm_year = (timeData[7] >> 4) * 10 + (timeData[7] & 0x0F); /* don't start introducing Y2K bugs now ... */ if (timeAndDate.tm_year < 70) timeAndDate.tm_year += 100; timeAndDate.tm_isdst = -1; timeData[0] = 0x00; return mktime(&timeAndDate); } --------------218262F3899371DBFC7F30B3 Content-Type: text/plain; charset=us-ascii; name="getTime.c" Content-Transfer-Encoding: 7bit Content-Disposition: inline; filename="getTime.c" /* getTime.c */ static char SccsId[] UNUSED = "%W% %G% Spectral Dynamics" ; #include "stdio.h" #include "stdlib.h" #include "vxWorks.h" #include "string.h" #include "time.h" #include "timers.h" #include "getTime.h" #include "boardId.h" #include "tc_tcpip.h" #include "acqLibPrototypes.h" #include "taskLib.h" #define DEBUG 0 #include "debug.h" int getTime( void ) { int status; time_t timeValue; struct timespec clockValue; char buff[80], *bP; int length = 80; timeValue = getHostTime( ) ; if( ! timeValue ) { printf("getTime() getHostTime is hosed\n") ; switch (get_board_id()) { case MOTOROLA_230X: timeValue = getMotTime(); break; default: printf("Unknown Board type for Time\n"); break; } } if( ! timeValue ) { printf("No valid time found\n"); return ERROR; } clockValue.tv_sec = timeValue ; clockValue.tv_nsec = 0 ; status = clock_settime( CLOCK_REALTIME, &clockValue ) ; if (status) { printf("Error in getTime=%d\n",status); return ERROR; } D(clock_gettime( CLOCK_REALTIME, &clockValue )); D(printf("clock_gettime sec=%d nsec=%d\n", (int)clockValue.tv_sec, (int)clockValue.tv_nsec)); ctime_r( (time_t *)&timeValue, buff, (size_t *)&length); bP = (char *)strchr( buff, 0x0a ); *bP = 0; printf("%s\n", buff); return OK; } /* * Port 13 returns local time as a string in the following format: * Sun Sep 16 01:03:52 1973\n\0 * * This routine converts this to broken-down time and then to time in seconds * to set the clock on the MDSP3 */ #define DOWINDEX 0 #define MONTHINDEX 4 #define DAYINDEX 8 #define HOURINDEX 11 #define MININDEX 14 #define SECINDEX 17 #define YEARINDEX 20 char *DayOfWeek[] = { "Sun", "Mon", "Tue", "Wed", "Thu", "Fri", "Sat", "SUN", "MON", "TUE", "WED", "THU", "FRI", "SAT" }; char *Months[] = { "Jan", "Feb", "Mar", "Apr", "May", "Jun", "Jul", "Aug", "Sep", "Oct", "Nov", "Dec", "JAN", "FEB", "MAR", "APR", "MAY", "JUN", "JUL", "AUG", "SEP", "OCT", "NOV", "DEC" }; time_t getHostTime( void ) { int numBytes, ilop, index, acp_num ; unsigned int timeValue ; char *host, *cP; SockArgs *args; char buff[80]; struct tm brokenTime; host = get_local_ip(); acp_num = get_acp_number(); args = tc_client_wait( host, 13 /* port */, 10 /* secs timeout */, acp_num ) ; timeValue = 0 ; if( args ) { numBytes = tc_read_sock( args->sock, buff, 24 ) ; tc_kill_socket( args ) ; if( numBytes < 4 ) { printf("shellGetTime() Read %d bytes\n", timeValue) ; return 0; } buff[numBytes] = 0; D(printf("getHostTime (from Workstation)= %s\n", buff)); bzero( (char *)&brokenTime, sizeof(brokenTime)); index = -1; cP = buff + MONTHINDEX; for ( ilop = 0; ilop < 24; ilop++ ) if ( ! strncmp(cP, Months[ilop], 3) ) { index = ilop; if (index > 11) index -= 12; break; } if (index == -1) return 0; brokenTime.tm_mon = index; cP = buff + DAYINDEX; brokenTime.tm_mday = atoi(cP); cP = buff + HOURINDEX; brokenTime.tm_hour = atoi(cP); cP = buff + MININDEX; brokenTime.tm_min = atoi(cP); cP = buff + SECINDEX; brokenTime.tm_sec = atoi(cP); cP = buff + YEARINDEX; brokenTime.tm_year = atoi(cP) - 1900; timeValue = mktime( &brokenTime ); } return( timeValue ) ; } #if 0 /* Port 37 gets binary GMT */ time_t getHostTime( void ) { int numBytes ; unsigned int timeValue ; char *host; SockArgs *args; D(char buff[80]); D(int length = 80); host = get_local_ip(); args = tc_client_wait( host, 37 /* port */, 10 /* secs timeout */ ) ; timeValue = 0 ; if( args ) { numBytes = tc_read_sock( args->sock, (char *)&timeValue, sizeof(timeValue) ) ; tc_kill_socket( args ) ; if( numBytes != 4 ) printf("shellGetTime() Read %d bytes\n", timeValue) ; timeValue -= JAN_1_1970 ; D(ctime_r( (time_t *)&timeValue, buff, (size_t *)&length)); D(host = (char *)strchr( buff, 0x0a )); D(*host = 0); D(printf("getHostTime= %s\n", buff)); } else { timeValue = 0 ; } return( timeValue ) ; } #endif --------------218262F3899371DBFC7F30B3-- From vxwexplo-errs@csg.lbl.gov Mon Feb 14 12:07:15 2000 From: dlee@adti.com Date: Mon Feb 14 12:07:17 PST 2000 Subject: error loading vxWorks.sym I am booting a mvm162-512 board from a Sun host over ethernet using the on board boot parameters provided to the vxWorks bootrom. The full path to the vxWorks image is provided via the boot parameters. The vxWorks image is loaded without incident but once the system attempts to load the symbol table it fails and displays message "hostname:/fullpathname .. hostname: stty status = 0x290010 Invalid argument. error opening hostname:/fullpathname. I have tried different permissions on the file thinking that could be it. The puzzling thing is that the vxWorks image loads and executes without a glitch. This is a process that has been working without any problems for a long time. Any ideas are greatly appreciated Doug Lee Applied Data Technology, Inc dlee@adti.com 619.450.9951 ext 203 From vxwexplo-errs@csg.lbl.gov Tue Feb 15 02:02:18 2000 From: stevens Date: Tue Feb 15 02:02:20 PST 2000 Subject: Porting PPP-Stack to BSD 4.4 (Tornado2) Hi VxWorkers, we had no problems using a PPP-Stack (BSD4.3 compatible) with VxWorks 5.3 (Tornado1.01). Now we changed to VxWorks 5.4 (Tornado2.0) and cannot compile the stack any longer. Problems occur compiling the file if_ppp. It is the Interface PPP <--> TCP/IP. VxWorks Network stack changed from BSD4.3 to BSD4.4. Examples of the error messages are: --------------------------------------------------------------------------- `MLEN' undeclared (first use this function) request for member `sa_family' in something not a structure or union structure has no member named `m_off' `MMINOFF' undeclared ---------------------------------------------------------------------------- Is there somebody who had the same problem and solved it ? Who can give me detailed information on the differences between BSD4.3 and BSD4.4 in VxWorks ? (especially concerning the structures and defines) Regards, Torsten -- ************************************************************************** fraunhofer institute of microelectronic circuits and systems department sat name torsten stevens mailto:stevens@ims.fhg.de address finkenstrasse 61 phone +0049/203/3783-238 47057 duisburg fax +0049/203/3783-266 germany http://www.fhg.de/depts/ims-du-e.html ************************************************************************** From vxwexplo-errs@csg.lbl.gov Tue Feb 15 04:03:11 2000 From: Vxworks Exploder Date: Tue Feb 15 04:03:13 PST 2000 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Tue Feb 15 04:03:07 PST 2000 Subject: Re: how I can use inflate in my own function Subject: Re: Links from VxWorks target to Allen-Bradley PLCs Subject: Re: VxWorks and Windows Subject: Help: can't booting from scsi disk Subject: How to recover from a bus error pb? Subject: How to recover from a bus error pb? Subject: Booting from compressed image Subject: Profibus DP support for VxWorks Subject: Re: How to recover from a bus error pb? Subject: Problem Running Post-Mortem Subject: MV2400 boot.bin Subject: ATM solution? Subject: what is difference between 1.0.1 and 2.0? Subject: Re: Help: can't booting from scsi disk Subject: Re: A "Tracy Riddle" is spamming c.o.vxworks posters Subject: Re: error loading vxWorks.sym Subject: Re: routing code Subject: Re: what is difference between 1.0.1 and 2.0? Subject: Re: VxWorks and Windows Subject: Re: switching IP addresses on vxWorks Subject: Re: TOD clock on an MVME2700 Subject: Re: Profibus DP support for VxWorks Subject: Re: How to recover from a bus error pb? Subject: Re: How to recover from a bus error pb? Subject: Re: Profibus DP support for VxWorks Subject: Re: MV2400 boot.bin Subject: Generate Tornado2 workspace from makefile Subject: Flashdriver for VxWorks Subject: Alignment in Structures ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: how I can use inflate in my own function Date: Mon, 14 Feb 2000 06:54:40 -0500 From: Joe Durusau Organization: Lockheed Martin Corporation Message-ID: <38A7ECFF.B0B4C5B7@delphi.com> References: <38a5b6c7.0@dnews.sbell.com.cn> I don't see anything that indicates that inflate is a function included with vxworks. If you have written this function yourself, the trick is to do one of the following: 1) load inflate.o (or whatever module contains inflate) first, then load the module that includes calls to it. 2). use ldxxx (xxx varies depending on your target) to link inflate and its callers to each other, then load the whole thing into the target at once. Speaking only for myself, Joe Durusau Zha Junhai wrote: > Hi, all > > Could you tell me how I can use inflate in > my own function, whatever I defined extern inflate(), and download to > target, it always tell me > Undefined symbols: inflate, how can I solve it, > any answer will be appreciated, Thanks. > > Best Regards > > Zha Junhai --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Links from VxWorks target to Allen-Bradley PLCs Date: Mon, 14 Feb 2000 09:39:26 -0500 From: "James A. Littlefield" Organization: WinStar GoodNet, Inc. Message-ID: <38A8139D.7A4F246B@alum.mit.edu> References: <87pffs$gue1@overload.lbl.gov> If your AB PLCs support Modbus/TCP perhaps that would be easy to implement on top of the vxworks tcp? Jim "Tyler, Paul A (DBY)" wrote: > > Dear VxWorkers, > > > > I am looking for drivers to allow target VxWorks systems (pc486 based) to > > communicate with Allen-Bradley PLC-5 and SLC-500 series PLCs. I have used > > AB6008SV scanners using remote I/O links in the past, but this is limited > > to eight words. I would like to be able to use other types of link such as > > Ethernet and DH-485 data highway. > > > > We use RSLinx to communicate with NT hosts, but this is not available for > > VxWorks. > > > > Something which could provide a common API between VxWorks and Windows NT > > platforms would be ideal, but then this is not an ideal world... > > > > All suggestions welcome. > > > > Thanks > > > > > > VxWorks, Tornado > > > > Paul Tyler > > Measurement Systems Specialist > > Electronics & Measurement Systems Dept. > > Rolls-Royce plc. > > > > e-mail Paul.A.Tyler@Rolls-Royce.com > > Tel. +44 (1332) 247808 > > Fax. +44 (1332) 247928 > > > > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: VxWorks and Windows Date: Mon, 14 Feb 2000 15:45 +0000 (GMT Standard Time) From: rew@cp.cix.co.uk (Richard Wenham) Organization: CIX - Compulink Information eXchange Message-ID: References: <383p4.6728$pN1.99950@news1.rdc1.nj.home.com> Reply-To: rew@cp.cix.co.uk You may be referring to VxWin, which needs some hardware assistance to achieve hard real-time performance. Have a look at http://www.trellissoftware.com/product/vxwinparts/vxwin.html Richard In article <383p4.6728$pN1.99950@news1.rdc1.nj.home.com>, HaraldHerberthRemoveThis@home.com (Harald) wrote: > I believe to have heard that VxWorks is available on Intel X86 PCs in a > version that shares the same CPU with Windows running concurrently. > Unfortunatelly I do not find information on this in the FAQ or on the > Web. > Can anybody provide information on whether this is true? What version of > Windows are supported? Can anybody comment on how realtime-capable this > combination would be? > Please no general comments on how good or bad Windows performs, we can't > really change that is is there. > > Thanks in advance > -Harald --------------------------- Newsgroups: comp.os.vxworks Subject: Help: can't booting from scsi disk Date: Mon, 14 Feb 2000 17:10:48 +0100 From: "patrick.boussard" Organization: Wanadoo, l'internet avec France Telecom Message-ID: <8899ie$sq9$1@wanadoo.fr> Hello, i've lost parameters to boot from scsi disk. I tried this: boot device: scsi;6.0 and it does'nt work. The system says: Network interface scsi;6.0 unknown Version is 5.1 P. Boussard --------------------------- Newsgroups: comp.os.vxworks Subject: How to recover from a bus error pb? Date: 14 Feb 2000 16:40:36 GMT From: pellici@ires.in2p3.fr (PELLICIOLI Michel) Organization: In2p3 Message-ID: <889b64$g4g$1@ccpntc8.in2p3.fr> Hello, I've got some bus error pb from time to time. The task involved ends up in the SUSPEND mode when it occurs (what is normal). I would like to be able to recover automatically (ie by the software itself) from that state without the need to reboot the system or at least to reboot warmly in ordre to preserve the data into memory. Does anyone know the general procedure to do so and have som e code examples? Thanks very much, Michel. --------------------------- Newsgroups: comp.os.vxworks Subject: How to recover from a bus error pb? Date: 14 Feb 2000 16:58:20 GMT From: pellici@ires.in2p3.fr (Michel Pellicioli) Organization: In2p3 Message-ID: <889c7c$g4m$1@ccpntc8.in2p3.fr> Hello, I've got some bus error pb from time to time. The task involved ends up in the SUSPEND mode when it occurs (what is normal). I would like to be able to recover automatically (ie by the software itself) from that state without the need to reboot the system or at least to reboot warmly in ordre to preserve the data into memory. Does anyone know the general procedure to do so and have some code examples? Thanks very much, Michel. --------------------------- Newsgroups: comp.os.vxworks Subject: Booting from compressed image Date: 14 Feb 2000 19:34:45 GMT From: johan.borkhuis@peektraffic.nl Organization: XS4ALL Internet BV Message-ID: <889lcl$kbu$1@news1.xs4all.nl> Hello, I am having the following problem. I am using a MBX860 board with 4 Mb Flash memory. From this flash 0.5 Mb is used as bootrom, so I have a 3.5 Mb flashdisk. My VxWorks image is 1.9 Mb (including symbols). I also need some room for my application and data, so 1.5 Mb is a bit small. I would like to have the possibility to us a compressed VxWorks image. But the image is loaded using a call to bootLoad, that just copies the file into memory. Is there a possibility to use a compressed image, or do I have to use a work-around (like defining a RAM-disk, uncompressing to this RAM-disk and loadiung from the RAM-disk)? Groeten, Johan - -- o o o o o o o . . . ___________________________________ o _____ || Johan Borkhuis | .][__n_n_|DD[ ====_____ | johan@borksoft.xs4all.nl | >(________|__|_[_________]_|________________________________| _/oo OOOOO oo` ooo ooo 'o!o!o o!o!o` === VxWorks page: http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html === --------------------------- Newsgroups: comp.os.vxworks Subject: Profibus DP support for VxWorks Date: 14 Feb 2000 19:41:04 GMT From: johan.borkhuis@peektraffic.nl Organization: XS4ALL Internet BV Message-ID: <889log$kbu$2@news1.xs4all.nl> Hello, I got a question from a colleague of mine. He is looking for an implementation of a Profibus DP protocol running under VxWorks using a normal serial port (tyCo). Is such an implementation available somewhere or can somepoint point me to a company that can provide me with such an implementation? Groeten, Johan - -- o o o o o o o . . . ___________________________________ o _____ || Johan Borkhuis | .][__n_n_|DD[ ====_____ | johan@borksoft.xs4all.nl | >(________|__|_[_________]_|________________________________| _/oo OOOOO oo` ooo ooo 'o!o!o o!o!o` === VxWorks page: http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html === --------------------------- Newsgroups: comp.os.vxworks Subject: Re: How to recover from a bus error pb? Date: Mon, 14 Feb 2000 13:43:02 -0600 From: Andrew Johnson Organization: APS, Argonne National Laboratory Message-ID: <38A85AC6.5EDB64A2@aps.anl.gov> References: <889b64$g4g$1@ccpntc8.in2p3.fr> PELLICIOLI Michel wrote: > > I've got some bus error pb from time to time. The task involved ends up in the > SUSPEND mode when it occurs (what is normal). I would like to be able to > recover automatically (ie by the software itself) from that state without the > need to reboot the system or at least to reboot warmly in ordre to preserve the > data into memory. Does anyone know the general procedure to do so and have som > e code examples? You can write a signal handler to trap these events - look at section 2.4.7 in the VxWorks Programmers Guide and the sigLib(1) entry in the VxWorks Reference Manual for details. Sorry I don't have any example code... - - Andrew - -- Complexity comes for free, Simplicity you have to work for. --------------------------- Newsgroups: comp.os.vxworks Subject: Problem Running Post-Mortem Date: Mon, 14 Feb 2000 22:48:01 GMT From: shanan8924@my-deja.com Organization: Deja.com - Before you buy. Message-ID: <88a0n0$8t9$1@nnrp1.deja.com> I am trying to run the Windview in Post-mortem mode, and I receive the following error. Error: wvLogHeaderCreate I am assuming this function is not receiving the proper memory PartId, therefore returning null. I am not sure how to correct this. Any help would be appreciated. Thanx Shanan Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: MV2400 boot.bin Date: Mon, 14 Feb 2000 15:44:12 -0800 From: Clement Wong Organization: BCnet Message-ID: If you have any advice on the following please forward! Steps I followed to create boot.bin for MV2400: 1) Set J8 to select soldered FLASH (factory configurations). 2) Created boot.bin from the latest release of MV2400 BSP. 3) Flashed boot.bin into the soldered FLASH. 4) Powered down. 5) Set J8 to select socketed FLASH. 6) Built the vxWorks file. 7) Powered up. 8) Changed the boot parameters. 9) Started the boot process by entering "@" at the [VxWorks Boot] prompt. 10) Error message that come up is "Network interface dc unknown". I then identified that I do not have "#define INCLUDE_DC". I added this to config.h, created a new boot.bin, and then I get the following message "Attaching network interface dc0...". BUT, the "done" message never arrives. Basically, we get a hang. Any ideas? Thanks in advance! Regards, Clement --------------------------- Newsgroups: comp.os.vxworks Subject: ATM solution? Date: Tue, 15 Feb 2000 09:24:19 +0900 From: "À̵¿±Ô" Organization: Korea Telecom Message-ID: <88a6d6$5h6$1@news2.kornet.net> Reply-To: "À̵¿±Ô" hi~ Has vxWorks ATM (SAR) solution of MPC860(Power QUICC)? ie, default SAR driver? --------------------------- Newsgroups: comp.os.vxworks Subject: what is difference between 1.0.1 and 2.0? Date: Tue, 15 Feb 2000 09:37:48 +0900 From: "À̵¿±Ô" Organization: Korea Telecom Message-ID: <88a76h$5ve$1@news2.kornet.net> Reply-To: "À̵¿±Ô" HI... what is difference between 1.0.1 and 2.0? --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Help: can't booting from scsi disk Date: Mon, 14 Feb 2000 17:48:35 -0800 From: "Michael R. Kesti" Organization: MK Associates Message-ID: <38A8B073.1500@gv.net> References: <8899ie$sq9$1@wanadoo.fr> Reply-To: mkesti@gv.net patrick.boussard wrote: >i've lost parameters to boot from scsi disk. I tried this: > >boot device: scsi;6.0 > >and it does'nt work. Interrupt the auto boot and, at the boot prompt type "help" and a carriage return. The last thing displayed is a list of available boot devices. - -- ======================================================================== Michael Kesti | "And like, one and one don't make | two, one and one make one." mkesti@gv.net | - The Who, Bargain --------------------------- Newsgroups: comp.os.vxworks Subject: Re: A "Tracy Riddle" is spamming c.o.vxworks posters Date: Tue, 15 Feb 2000 04:59:40 GMT From: pwhicker@home.net (Paul Whicker) Organization: @Home Network Message-ID: <035q4.32997$Fb1.760771@news1.frmt1.sfba.home.com> References: <38972EA2.F53E4BAB@aps.anl.gov> <3897D211.57E473C4@bitband.com> <38A7AC79.DB6D0EA2@bitband.com> In article <38A7AC79.DB6D0EA2@bitband.com>, leonid@bitband.com wrote: >I am however happy for you and that you feel happy at Wind River, perhaps the > last >year's changes in management have helped their situation. Remember, I do not speak for Wind River in any capacity. I also get only a small cross section of the opinion and information within Wind River so my views may be significantly inaccurate. I'm certain some customers aren't happy that they can no longer play Wind River and ISI against each other. Conversely that has to be good for both halves of the new company (I believe we are being re-incorporated?). Personally I look forward to getting my sticky little hands on SDS Single Step and the Diab Data compilers, I bought their tools in the past when I was a customer and I'm very pleased that they are with us now, I hope they enjoy the experience too. I believe Thom St Dennis is excellent though I haven't had a lot of chance to talk with him myself, he has made it clear that he wants to see significant improvements in customer satisfaction and he's made some of the first significant changes in years. He's a top class exec whereas Ron Ableman always struck me as a hatchet man who had outlived his time there and didn't have the vision for driving the company forward. For myself I think that the more we can do to make all our customers successful the better it will be for both Wind River and for our customers, that's why I sometimes intrude here. I'm very interested to see Dr Design with Wind River now, apart from anything else that makes Wind a customer for their own products and I expect them to be their most demanding customer, that should be good for everyone. I think the market for consulting is vast too; if I'd had the chance, in the past, to put off the o/s related risk to the manufacturer I would have jumped at it. Well, I hope a few of you find my honest opinion amusing. I thinks that's probably enough to get me hung :-) Paul >I do wonder however how does the customer community feel about this merger, >although I am sure that gutsy memos from customers won't change a thing. > >Leonid --------------------------- Newsgroups: comp.os.vxworks Subject: Re: error loading vxWorks.sym Date: Tue, 15 Feb 2000 05:10:07 GMT From: pwhicker@home.net (Paul Whicker) Organization: @Home Network Message-ID: References: <889npq$2093@overload.lbl.gov> How strange. There is a way round this and there are some things you might do to track down the error. First you can avoid the problem by using a built-in symbol table. Second, if you have support and you check on the keyword search for 0x290010 you'll find this:- RESOLUTION: customer was not running rsh, so needed to set up ftp password. If this doesn't help then use a built-in synbol table and investigate why you are having file access errors. With respect to the rsh comment try loking at your .rhosts file, it tells the host what targets should be allowed to connect to it without password protection. HTH Paul In article <889npq$2093@overload.lbl.gov>, dlee@adti.com wrote: >I am booting a mvm162-512 board from a Sun host over ethernet using the on >board boot parameters provided to the vxWorks bootrom. The full path to the >vxWorks image is provided via the boot parameters. The vxWorks image is >loaded without incident but once the system attempts to load the symbol >table it fails and displays message "hostname:/fullpathname .. hostname: >stty status = 0x290010 Invalid argument. error opening >hostname:/fullpathname. > >I have tried different permissions on the file thinking that could be it. >The puzzling thing is that the vxWorks image loads and executes without a >glitch. This is a process that has been working without any problems for a >long time. > >Any ideas are greatly appreciated > >Doug Lee >Applied Data Technology, Inc >dlee@adti.com >619.450.9951 ext 203 > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: routing code Date: Tue, 15 Feb 2000 05:11:24 GMT From: pwhicker@home.net (Paul Whicker) Organization: @Home Network Message-ID: <0e5q4.32999$Fb1.760771@news1.frmt1.sfba.home.com> References: <3891E2FE.512785C4@vitesse.com> <888cv8$2tu$1@nnrp1.deja.com> Part of Wind River now for what it's worth. Paul In article <888cv8$2tu$1@nnrp1.deja.com>, cxj@gsta.com wrote: >Hello,Ravi > As far as I know, RouterWare has these Routing protocols software >to sell. But I don't know how about their performance.The website is: >www.routerware.com . >Good luck. > >In article <3891E2FE.512785C4@vitesse.com>, > ravi wrote: >> Hello Everyone, >> >> We are trying to build a L2/L3 switch. I was just wondering if anyone >is >> aware of companies >> that sell the code for implementation along with support of Routing >> protocols like RIP,OSPF,BGP >> and bridging over ethernet. I would appreciate if u could mail me the >> names of those companies >> along with their contacts if possible. >> >> Thanks >> Ravi >> ravi@vitesse.com >> >> > > >Sent via Deja.com http://www.deja.com/ >Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: what is difference between 1.0.1 and 2.0? Date: Tue, 15 Feb 2000 05:14:02 GMT From: pwhicker@home.net (Paul Whicker) Organization: @Home Network Message-ID: References: <88a76h$5ve$1@news2.kornet.net> Lots and lots. Many bug fixes, very different tools, much better C++ support. Paul In article <88a76h$5ve$1@news2.kornet.net>, "À̵¿±Ô" wrote: >HI... >what is difference between 1.0.1 and 2.0? > > > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: VxWorks and Windows Date: Mon, 14 Feb 2000 08:52:45 +0100 From: "Dr. Peter Salewsky" Organization: resolut GmbH Message-ID: <38A7B44D.A475FD5F@resolut.de> References: <383p4.6728$pN1.99950@news1.rdc1.nj.home.com> Harald schrieb: > I believe to have heard that VxWorks is available on Intel X86 PCs in a > version that shares the same CPU with Windows running concurrently. [...] The product is called VxWin. Look at www.lp-elektronik.com/products/evxwin.htm Peter - -- Dr. Peter Salewsky resolut GmbH, Antonienstr. 3, D-06749 Bitterfeld, Germany phone +49 3493 743-20, fax +49 3493 743-25, web www.resolut.de --------------------------- Newsgroups: comp.os.vxworks Subject: Re: switching IP addresses on vxWorks Date: Tue, 15 Feb 2000 08:35:48 +0200 From: Leonid Rosenboim Organization: BitBand Technologies Ltd. http://www.bitband.com Message-ID: <38A8F3C4.3CD3D757@bitband.com> References: <88ag6k$jkq$1@nnrp1.deja.com> Reply-To: leonid@bitband.com Although this scheme is feasible for UDP, you wont be able to swithc the TCP connections without disruption, because the TCP Protocol Control Block, and C does not have those, unless you are going to reqrite the TCP protocol code to pass along info about each and every TCP connection in progress so C can take it over when needed. Obviously there are many more critical details you need to work out even for UDP to work, and for TCP clients to be able to re-connect quickly on restart. Please use your real name and address next time, hate the thought I am speaking to someone's panties ;-)) Leonid captain_underwear@my-deja.com wrote: > Maybe this belongs to comp.unix.networking, but I > was wondering what would happen in a vxworks > environment under the following circumstances: > > I have 3 cards A, B , C. B & C are hot standby > cards, and B is currently active, and is getting > UDP and TCP messages from A. B crashes, and C > wants to take over its role. Can C just send an > arp message to say that B's IP address is now > associated with its ethernet address ? Will the > TCP connections actually stay up if the > switchover is in a reasonable amount of time ? > > thanks in advance for any help > > Siddharth > > Sent via Deja.com http://www.deja.com/ > Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: TOD clock on an MVME2700 Date: Tue, 15 Feb 2000 08:22:38 +0100 From: "Serge Wenger" Organization: MATISA SA Message-ID: <88auql$d4k$1@pollux.ip-plus.net> References: <889gos$2092@overload.lbl.gov> Reply-To: "Serge Wenger" Larry: I think there is a little bug in the function getMotTime : before the read, you must halt the updating of the register (set READ bit (D6)) and reset this bit after the read. - -- /Serge Wenger In my e-mail address, replace nobody by matisa Larry Sinn wrote in message news:889gos$2092@overload.lbl.gov... > This is a multi-part message in MIME format. > --------------218262F3899371DBFC7F30B3 > Content-Type: multipart/alternative; boundary="------------9F852866566238FF4040A8C2" > > > --------------9F852866566238FF4040A8C2 > Content-Type: text/plain; charset=us-ascii > Content-Transfer-Encoding: 7bit > > Daniel: > Attached are a couple of routines that work on mvme2300 with UNIX host. > This gets time from both the PPC and UNIX host. > > Larry. > > > > Daniel wrote: > > > I would like to read the "Time of Day" clock on an > > MVME2700, has anyone done this, if so, could you give me > > some pointers ? > > > > Larry Sinn > Spectral Dynamics (408)474-1746 voice > 1983 Concourse Drive (408)474-1780 fax > San Jose, Ca. 95131-1708 sinnl@sd-star.com > > > > --------------9F852866566238FF4040A8C2 > Content-Type: text/html; charset=us-ascii > Content-Transfer-Encoding: 7bit > > > > Daniel: >
Attached are a couple of routines that work on mvme2300 with UNIX host. >
This gets time from both the PPC and UNIX host. >

Larry. >
  >
  >

Daniel wrote: >

> 
I would like to read the "Time of Day" clock on an
> MVME2700, has anyone done this, if so, could you give me 
> some pointers ?
>
> >


Larry Sinn >
Spectral Dynamics           &n bsp;   > (408)474-1746  voice >
1983 Concourse Drive            > (408)474-1780  fax >
San Jose, Ca. 95131-1708        > sinnl@sd-star.com >
  >
  > > --------------9F852866566238FF4040A8C2-- > > --------------218262F3899371DBFC7F30B3 > Content-Type: text/plain; charset=us-ascii; name="getMotTime.c" > Content-Transfer-Encoding: 7bit > Content-Disposition: inline; filename="getMotTime.c" > > /* getMotTime.c */ > static char SccsId[] UNUSED = "@(#)getMotTime.c 1.1 06/29/99 Spectral Dynamics" ; > > /* > * Get time from Motorola PPC > */ > > > #include "stdio.h" > #include "mv2600.h" > #include "vxWorks.h" > > #include "time.h" > #include "timers.h" > > extern UCHAR sysNvRead(ULONG); > > time_t > getMotTime(void) > { > struct tm timeAndDate; > unsigned char timeData[8]; > int i; > > for (i = 0; i < 8; i++) > timeData[i] = sysNvRead((ULONG)TOD_CLOCK + i); > > timeData[0] = 0x00; > > timeAndDate.tm_sec = (timeData[1] >> 4) * 10 + (timeData[1] & 0x0F); > timeAndDate.tm_min = (timeData[2] >> 4) * 10 + (timeData[2] & 0x0F); > timeAndDate.tm_hour = (timeData[3] >> 4) * 10 + (timeData[3] & 0x0F); > timeAndDate.tm_mday = (timeData[5] >> 4) * 10 + (timeData[5] & 0x0F); > timeAndDate.tm_mon = (timeData[6] >> 4) * 10 + (timeData[6] & 0x0F) - 1; > timeAndDate.tm_year = (timeData[7] >> 4) * 10 + (timeData[7] & 0x0F); > > /* don't start introducing Y2K bugs now ... */ > if (timeAndDate.tm_year < 70) > timeAndDate.tm_year += 100; > > timeAndDate.tm_isdst = -1; > > timeData[0] = 0x00; > > return mktime(&timeAndDate); > } > > > --------------218262F3899371DBFC7F30B3 > Content-Type: text/plain; charset=us-ascii; name="getTime.c" > Content-Transfer-Encoding: 7bit > Content-Disposition: inline; filename="getTime.c" > > /* getTime.c */ > static char SccsId[] UNUSED = "%W% %G% Spectral Dynamics" ; > #include "stdio.h" > #include "stdlib.h" > #include "vxWorks.h" > #include "string.h" > > #include "time.h" > #include "timers.h" > #include "getTime.h" > #include "boardId.h" > #include "tc_tcpip.h" > #include "acqLibPrototypes.h" > #include "taskLib.h" > > #define DEBUG 0 > #include "debug.h" > > > int > getTime( void ) > { > int status; > time_t timeValue; > struct timespec clockValue; > char buff[80], *bP; > int length = 80; > > timeValue = getHostTime( ) ; > > if( ! timeValue ) > { > printf("getTime() getHostTime is hosed\n") ; > switch (get_board_id()) > { > case MOTOROLA_230X: > timeValue = getMotTime(); > break; > default: > printf("Unknown Board type for Time\n"); > break; > } > } > > if( ! timeValue ) > { > printf("No valid time found\n"); > return ERROR; > } > > > clockValue.tv_sec = timeValue ; > clockValue.tv_nsec = 0 ; > status = clock_settime( CLOCK_REALTIME, &clockValue ) ; > if (status) > { > printf("Error in getTime=%d\n",status); > return ERROR; > } > > D(clock_gettime( CLOCK_REALTIME, &clockValue )); > D(printf("clock_gettime sec=%d nsec=%d\n", (int)clockValue.tv_sec, (int)clockValue.tv_nsec)); > > ctime_r( (time_t *)&timeValue, buff, (size_t *)&length); > bP = (char *)strchr( buff, 0x0a ); > *bP = 0; > printf("%s\n", buff); > > return OK; > } > > > /* > * Port 13 returns local time as a string in the following format: > * Sun Sep 16 01:03:52 1973\n\0 > * > * This routine converts this to broken-down time and then to time in seconds > * to set the clock on the MDSP3 > */ > > #define DOWINDEX 0 > #define MONTHINDEX 4 > #define DAYINDEX 8 > #define HOURINDEX 11 > #define MININDEX 14 > #define SECINDEX 17 > #define YEARINDEX 20 > > char *DayOfWeek[] = { "Sun", "Mon", "Tue", "Wed", "Thu", "Fri", "Sat", > "SUN", "MON", "TUE", "WED", "THU", "FRI", "SAT" }; > char *Months[] = { "Jan", "Feb", "Mar", "Apr", "May", "Jun", "Jul", "Aug", "Sep", "Oct", "Nov", "Dec", > "JAN", "FEB", "MAR", "APR", "MAY", "JUN", "JUL", "AUG", "SEP", "OCT", "NOV", "DEC" }; > time_t > getHostTime( void ) > { > int numBytes, ilop, index, acp_num ; > unsigned int timeValue ; > char *host, *cP; > SockArgs *args; > char buff[80]; > struct tm brokenTime; > > host = get_local_ip(); > acp_num = get_acp_number(); > > args = tc_client_wait( host, 13 /* port */, 10 /* secs timeout */, acp_num ) ; > > timeValue = 0 ; > > if( args ) > { > > numBytes = tc_read_sock( args->sock, buff, 24 ) ; > tc_kill_socket( args ) ; > if( numBytes < 4 ) > { > printf("shellGetTime() Read %d bytes\n", timeValue) ; > return 0; > } > buff[numBytes] = 0; > D(printf("getHostTime (from Workstation)= %s\n", buff)); > > bzero( (char *)&brokenTime, sizeof(brokenTime)); > > index = -1; > cP = buff + MONTHINDEX; > for ( ilop = 0; ilop < 24; ilop++ ) > if ( ! strncmp(cP, Months[ilop], 3) ) > { > index = ilop; > if (index > 11) > index -= 12; > break; > } > if (index == -1) > return 0; > brokenTime.tm_mon = index; > > cP = buff + DAYINDEX; > brokenTime.tm_mday = atoi(cP); > > cP = buff + HOURINDEX; > brokenTime.tm_hour = atoi(cP); > > cP = buff + MININDEX; > brokenTime.tm_min = atoi(cP); > > cP = buff + SECINDEX; > brokenTime.tm_sec = atoi(cP); > > cP = buff + YEARINDEX; > brokenTime.tm_year = atoi(cP) - 1900; > > timeValue = mktime( &brokenTime ); > } > > return( timeValue ) ; > } > > > #if 0 > > /* Port 37 gets binary GMT */ > > time_t > getHostTime( void ) > { > int numBytes ; > unsigned int timeValue ; > char *host; > SockArgs *args; > D(char buff[80]); > D(int length = 80); > > host = get_local_ip(); > > args = tc_client_wait( host, 37 /* port */, 10 /* secs timeout */ ) ; > > timeValue = 0 ; > > if( args ) > { > numBytes = tc_read_sock( args->sock, (char *)&timeValue, sizeof(timeValue) ) ; > tc_kill_socket( args ) ; > if( numBytes != 4 ) > printf("shellGetTime() Read %d bytes\n", timeValue) ; > > timeValue -= JAN_1_1970 ; > > D(ctime_r( (time_t *)&timeValue, buff, (size_t *)&length)); > D(host = (char *)strchr( buff, 0x0a )); > D(*host = 0); > D(printf("getHostTime= %s\n", buff)); > } > else > { > timeValue = 0 ; > } > > return( timeValue ) ; > } > > #endif > > --------------218262F3899371DBFC7F30B3-- > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Profibus DP support for VxWorks Date: Tue, 15 Feb 2000 08:36:24 GMT From: real@email.in.sig (Amit Raam) Organization: Unspecified Organization Message-ID: <38ab0f8b.689900744@news.ibm.net.il> References: <889log$kbu$2@news1.xs4all.nl> www.softing.com www.hilscher.com http://www.t-h.de/ johan.borkhuis@peektraffic.nl wrote: >Hello, > >I got a question from a colleague of mine. He is looking for an >implementation of a Profibus DP protocol running under VxWorks using a >normal serial port (tyCo). Is such an implementation available somewhere >or can somepoint point me to a company that can provide me with such an >implementation? > >Groeten, > Johan Amit Raam Email address: raama at indigo.co.il - -------------------------------------------------- ".sigless and proud of it" (R) (C) (tm) (look-and-feel) --------------------------- Newsgroups: comp.os.vxworks Subject: Re: How to recover from a bus error pb? Date: Tue, 15 Feb 2000 08:49:36 -0000 From: "Stephen Walsh" Message-ID: <38a91074$1@pull.gecm.com> References: <889b64$g4g$1@ccpntc8.in2p3.fr> PELLICIOLI Michel wrote in message <889b64$g4g$1@ccpntc8.in2p3.fr>... >Hello, > >I've got some bus error pb from time to time. The task involved ends up in the >SUSPEND mode when it occurs (what is normal). I would like to be able to >recover automatically (ie by the software itself) from that state without the >need to reboot the system or at least to reboot warmly in ordre to preserve the >data into memory. Does anyone know the general procedure to do so and have som >e code examples? >Thanks very much, >Michel. > > Here is a short sanatized version of some code I use for traping bus errors. You should be able to adjust it for what you want to do. Hope this helps Steve ******************************************* Stephen A Walsh Senior Software Engineer BAE Systems ******************************************* Opinions are usually mine and mine alone ******************************************* /* begin code fragment */ #inlcude #include static jmp_buf zx_context_of_call ; static unsigned long zul_check_for_error; /* Exception handler for error to be trapped */ void v_handle_exception (void) { /* do exception processing here */ /* Now exit exception handler using longjump */ longjmp(zx_context_of_call,0); } /* Now define task to trap exception for */ void v_process (void) { /* start with no error having occured this time */ zul_check_for_error = FALSE ; /* attach to interrupt vector */ signal (INTERRUPT_VECTOR_TO_TRAP, v_handle_exception); /* set point to return to when exception occurs */ setjmp(zx_context_of_call); /* possible check here for error occurred */ if (zul_check_for_error == TRUE); { /* do some processing */ /* Now restore the handler and return */ signal (INTERRUPT_VECTOR_TO_TRAP, DEFAULT_HANDLER); return ; } /* do processing that may cause error */ /* Now restore the handler */ signal (INTERRUPT_VECTOR_TO_TRAP, DEFAULT_HANDLER); } /* end code fragment */ --------------------------- Newsgroups: comp.os.vxworks Subject: Re: How to recover from a bus error pb? Date: Tue, 15 Feb 2000 08:53:25 -0000 From: "Stephen Walsh" Message-ID: <38a91158$1@pull.gecm.com> References: <889b64$g4g$1@ccpntc8.in2p3.fr> <38a91074$1@pull.gecm.com> Slight error in the exception handler should have been Hope this helps /* Exception handler for error to be trapped */ void v_handle_exception (void) { /* Indicate exception has occurred (missed this first time) */ zul_check_for_error = TRUE ; /* do exception processing here */ /* Now exit exception handler using longjump */ longjmp(zx_context_of_call,0); } Sorry, Steve ******************************************* Stephen A Walsh Senior Software Engineer BAE Systems ******************************************* Opinions are usually mine and mine alone ******************************************* > >/* begin code fragment */ > >#inlcude >#include > >static jmp_buf zx_context_of_call ; >static unsigned long zul_check_for_error; > >/* Exception handler for error to be trapped */ >void v_handle_exception (void) >{ > /* do exception processing here */ > > /* Now exit exception handler using longjump */ > longjmp(zx_context_of_call,0); >} > >/* Now define task to trap exception for */ >void v_process (void) >{ > > /* start with no error having occured this time */ > zul_check_for_error = FALSE ; > > /* attach to interrupt vector */ > signal (INTERRUPT_VECTOR_TO_TRAP, v_handle_exception); > > /* set point to return to when exception occurs */ > setjmp(zx_context_of_call); > > /* possible check here for error occurred */ > if (zul_check_for_error == TRUE); > { > /* do some processing */ > > /* Now restore the handler and return */ > signal (INTERRUPT_VECTOR_TO_TRAP, DEFAULT_HANDLER); > return ; > } > > /* do processing that may cause error */ > > /* Now restore the handler */ > signal (INTERRUPT_VECTOR_TO_TRAP, DEFAULT_HANDLER); >} > >/* end code fragment */ > > > > > > > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Profibus DP support for VxWorks Date: 15 Feb 2000 09:09:52 GMT From: johan.borkhuis@peektraffic.nl Organization: XS4ALL Internet BV Message-ID: <88b550$f20$1@news1.xs4all.nl> References: <889log$kbu$2@news1.xs4all.nl> <38ab0f8b.689900744@news.ibm.net.il> Amit Raam wrote: : www.softing.com : www.hilscher.com : http://www.t-h.de/ Thanks, that was the information I was looking for. Groeten, Johan - -- o o o o o o o . . . ___________________________________ o _____ || Johan Borkhuis | .][__n_n_|DD[ ====_____ | johan@borksoft.xs4all.nl | >(________|__|_[_________]_|________________________________| _/oo OOOOO oo` ooo ooo 'o!o!o o!o!o` === VxWorks page: http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html === --------------------------- Newsgroups: comp.os.vxworks Subject: Re: MV2400 boot.bin Date: 15 Feb 2000 09:44:04 GMT From: johan.borkhuis@peektraffic.nl Organization: XS4ALL Internet BV Message-ID: <88b754$ga5$1@news1.xs4all.nl> References: Clement Wong wrote: : 10) Error message that come up is "Network interface dc unknown". It could be that your network name is different (for example cpm)? I don't know this board, but on our board the name is cpm Groeten, Johan - -- o o o o o o o . . . ___________________________________ o _____ || Johan Borkhuis | .][__n_n_|DD[ ====_____ | johan@borksoft.xs4all.nl | >(________|__|_[_________]_|________________________________| _/oo OOOOO oo` ooo ooo 'o!o!o o!o!o` === VxWorks page: http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html === --------------------------- Newsgroups: comp.os.vxworks Subject: Generate Tornado2 workspace from makefile Date: Tue, 15 Feb 2000 10:56:22 +0100 From: "Jørgen Jakobsen" Organization: Customer at Tele Danmark Erhverv Message-ID: <88b7t0$sl5$1@news1.tele.dk> Hi I'm working on a new BSP, and wonder whether it is possible to generate a Tornado Workspace from the makefile in the BSP ? This would make it much more easy to compile a single file in the BSP. If not possible, what is the easiest way to compile/assemble a single source file in my BSP ? Jørgen Jakobsen, Cotas Computer Technology A/S --------------------------- Newsgroups: comp.os.vxworks Subject: Flashdriver for VxWorks Date: Tue, 15 Feb 2000 10:58:37 +0100 From: "Jørgen Jakobsen" Organization: Customer at Tele Danmark Erhverv Message-ID: <88b818$2r$1@news1.tele.dk> Hi On our new board, we are using the new Intel StrataFlash memory (28F128J3A). These chips support the Common Flash Interface, known as CFI, and the Scalable Command Set, SCS. Is a CFI driver that supports Intel StrataFlash available from WindRiver or from anyone else ? Eventually take a look at the http://developer.intel.com/design/builder/FLBLDR/SWB/cfi/content.htm to get an overview of CFI. Regards, Jørgen Jakobsen, Cotas Computer Technology A/S --------------------------- Newsgroups: comp.os.vxworks Subject: Alignment in Structures Date: Tue, 15 Feb 2000 12:42:34 +0100 From: "B. Lampi" Organization: Technische Universitaet Graz, Austria Message-ID: <38A93BAA.F5307BFC@hotmail.com> How can I set the alignment for a single structure? Does "@pragma align 1" not work? Thanks for any advice. Bernhard PS: Host: PC, NT 4.0, Tornado 2 target: PPC821, vxWorks GNU C compiler for PPC 2.7.2-960126 --------------------------- End of New-News digest ********************** From vxwexplo-errs@csg.lbl.gov Tue Feb 15 08:03:35 2000 From: Randy White Date: Tue Feb 15 08:03:38 PST 2000 Subject: Booting from compressed image This is a multi-part message in MIME format. --------------5A86FD9851CE404407BFBB32 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit > Hello, > > I am having the following problem. I am using a MBX860 board with 4 Mb > Flash memory. From this flash 0.5 Mb is used as bootrom, so I have a > 3.5 Mb flashdisk. My VxWorks image is 1.9 Mb (including symbols). I also > need some room for my application and data, so 1.5 Mb is a bit small. > > I would like to have the possibility to us a compressed VxWorks image. But > the image is loaded using a call to bootLoad, that just copies the file > into memory. Is there a possibility to use a compressed image, or do I > have to use a work-around (like defining a RAM-disk, uncompressing to this > RAM-disk and loadiung from the RAM-disk)? > > Groeten, > Johan We had a similar problem with our MPC8240 system. The boot loader resides in a ~200K section of FLASH. The rest of the 1 Meg flash holds a compressed vxworks.st image. The compression is done manually (or by modifying the make) to run deflate. The bootloader creates a memDrv device then feeds this to inflate(). The output of inflate goes right to the same RAM pointer that loading an uncompressed image does. The inflate routine seems to be rather iffy in it's operation, so we are exploring alternatives. We are able to comfortably fit a 1.3 Meg kernel image and the bootloader (bootrom.hex) into 1 Meg. Randy White --------------5A86FD9851CE404407BFBB32 Content-Type: text/x-vcard; charset=us-ascii; name="randy.vcf" Content-Transfer-Encoding: 7bit Content-Description: Card for Randy White Content-Disposition: attachment; filename="randy.vcf" begin:vcard n:White;Randy tel;fax:(403)492-0654 tel;work:(403)492-0321 x277 x-mozilla-html:FALSE org:Alberta Microelectronic Corp;Systems Division adr:;;#201, 8625 - 112 St.;Edmonton;AB;T6G 1K8;Canada version:2.1 email;internet:randy@amc.ab.ca title:Senior Designer x-mozilla-cpt:;-1 fn:Randy White end:vcard --------------5A86FD9851CE404407BFBB32-- From vxwexplo-errs@csg.lbl.gov Wed Feb 16 00:12:11 2000 From: "Michael Lawnick" Date: Wed Feb 16 00:12:13 PST 2000 Subject: Re: Booting from compressed image > Subject: Booting from compressed image > Date: 14 Feb 2000 19:34:45 GMT > From: johan.borkhuis@peektraffic.nl > Organization: XS4ALL Internet BV > Message-ID: <889lcl$kbu$1@news1.xs4all.nl> > > Hello, > > I am having the following problem. I am using a MBX860 board with 4 Mb > Flash memory. From this flash 0.5 Mb is used as bootrom, so I have a 3.5 > Mb flashdisk. My VxWorks image is 1.9 Mb (including symbols). I also need > some room for my application and data, so 1.5 Mb is a bit small. > > I would like to have the possibility to us a compressed VxWorks image. But > the image is loaded using a call to bootLoad, that just copies the file > into memory. Is there a possibility to use a compressed image, or do I > have to use a work-around (like defining a RAM-disk, uncompressing to this > RAM-disk and loadiung from the RAM-disk)? > > Groeten, > Johan Hi Johan, why don't you add your application to the kernel via INCLUDE_USR_APPL and MACH_EXTRA and build a vxWorks_rom ? You would save the 0.5 Mb of bootrom and your kernel + application will be compressed. MfG Lawnick, SOFTEC GmbH ============================================== SOFTEC GmbH Tel +49-731-96600-0 Promenade 17 Fax +49-731-96600-23 D-89073 Ulm Michael Lawnick Germany lawnick@softec.de ============================================== From vxwexplo-errs@csg.lbl.gov Wed Feb 16 01:39:02 2000 From: David Laight Date: Wed Feb 16 01:39:04 PST 2000 Subject: Re: Booting from compressed image > Subject: Booting from compressed image Luxury - a vxWorks image that is ONLY 1.9Mb! With RtX_Windows and the JVM mine is 4.5Mb uncompressed with a small symbol table (hand generated containing interesting symbols). I've dispensed with the bootloader. I use the compressed bootrom loader (linked high) to put a full vxWorks image straight to RAM_LOW_ADRS. A rawFs filesystem over the boot area of ROM allows the vxWorks code to copy in a new copy of itself from the host system straight into the boot ROM area. (Yes you do need a fallback position for that version that doesn't work.) 4.7Mb compresses down to 1.7Mb and sits in the 2Mb I have reserved for it. I include a small symbol table - containing all the xxxShow routines and other useful goodies so that the target Shell can be used as a diagnostic console. The full symbol table can be network loaded by request if needed. David > > Submitted-by: Randy White > > > Hello, > > > > I am having the following problem. I am using a MBX860 board with 4 Mb > > Flash memory. From this flash 0.5 Mb is used as bootrom, so I have a > > 3.5 Mb flashdisk. My VxWorks image is 1.9 Mb (including symbols). I also > > need some room for my application and data, so 1.5 Mb is a bit small. > > > > I would like to have the possibility to us a compressed VxWorks image. But > > the image is loaded using a call to bootLoad, that just copies the file > > into memory. Is there a possibility to use a compressed image, or do I > > have to use a work-around (like defining a RAM-disk, uncompressing to this > > RAM-disk and loadiung from the RAM-disk)? > > > > Groeten, > > Johan > > We had a similar problem with our MPC8240 system. The boot loader resides in a > ~200K section of FLASH. The rest of the 1 Meg flash holds a compressed > vxworks.st image. The compression is done manually (or by modifying the make) > to run deflate. The bootloader creates a memDrv device then feeds this to > inflate(). The output of inflate goes right to the same RAM pointer that > loading an uncompressed image does. > The inflate routine seems to be rather iffy in it's operation, so we are > exploring alternatives. We are able to comfortably fit a 1.3 Meg kernel image > and the bootloader (bootrom.hex) into 1 Meg. > > Randy White ---------------------------------------------------------------- David Laight email: dsl@tadpole.co.uk Tadpole Technology plc phone: +44 1223 428 232 Cambridge, UK fax: +44 1223 428 201 From vxwexplo-errs@csg.lbl.gov Wed Feb 16 04:03:09 2000 From: Vxworks Exploder Date: Wed Feb 16 04:03:12 PST 2000 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Wed Feb 16 04:03:06 PST 2000 Subject: Multi-homing Multiple IP address per physical interface Subject: Re: Alignment in Structures Subject: Time Measurement Subject: HELP!!! Need glaux.dll Subject: 3C905B Subject: Re: dosFsLib problem on Tornado 2.0 Subject: Spurious interrupt with PPC860/850 Subject: Re: Time Measurement Subject: Re: Alignment in Structures Subject: time() returns 0? Subject: Visual Studio Integration with T2 Subject: Re: time() returns 0? Subject: Re: VxSim communicate to WinNT Subject: Have you seen this funny Packet before ? Subject: tcl7.4 PPC860 blows up Subject: interbus-S support available? Subject: T2 + WSP + multiple targets Subject: VME interrups Subject: Re: GDB problems Subject: Re: Spurious interrupt with PPC860/850 Subject: Re: Booting from compressed image Subject: Re: MPC850 SMC is slowly Subject: Re: T2 + WSP + multiple targets Subject: change date and time of a file Subject: Relocatable application ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Multi-homing Multiple IP address per physical interface Date: Tue, 15 Feb 2000 07:01:48 -0600 From: "Tim Michals" Organization: Posted via Supernews, http://www.supernews.com Message-ID: Does VxWorks TCP/IP stack support multi-IP addresses per interface? If not this there a way to do this? --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Alignment in Structures Date: Tue, 15 Feb 2000 14:08:12 +0100 From: "B. Lampi" Organization: Technische Universitaet Graz, Austria Message-ID: <38A94FBB.C2467A45@hotmail.com> References: <38A93BAA.F5307BFC@hotmail.com> struct tagFoo { ... } __attribute__ ((packed)); does the job. ...but: struct tagTST { int Nr; short LevelNumber; int RefRange __attribute__ ((packed)); short LotNumber; int ApplyDate; short SampleType; }; int QCLvlSize() { struct tagTST t; t.Nr = 1; t.LevelNumber = 2; t.RefRange = 3; t.LotNumber = 4; t.ApplyDate = 5; t.SampleType = 6; HexDump( (char *)&t, 32 ); printf( "TST size: %d %d\n", sizeof(struct tagTST), sizeof(struct tagTST)==18 ? 0 : -1 ); return 0; } - -> QCLvlSize 0x21f920 0000: 00 00 00 01 00 02 00 00 ........ 0x21f928 0008: 00 03 00 04 00 00 00 05 ........ 0x21f930 0010: 00 06 EE EE EE EE EE EE ..îîîîîî <== ok 0x21f938 0018: EE EE EE EE 00 21 73 10 îîîî.!s. 0x21f940 0020: 00 21 F9 58 00 10 7C C4 .!ùX..|Ä TST size: 20 -1 <========= why 20???? and not 18? value = 0 = 0x0 - -> --------------------------- Newsgroups: comp.os.vxworks Subject: Time Measurement Date: Tue, 15 Feb 2000 16:16:00 +0200 From: =?iso-8859-1?Q?Morn=E9_Hartman?= Message-ID: <38a95f8b$0$35289@helios.is.co.za> Hi all, I would like to know how to measure the system time at a specific moment, then at another moment, only a few milliseconds later, to determine elapsed time. I have tried clock_gettime(), but this is no good. I have checked the clock resolution and it is at 60 ticks per second. Does anyone have any help in this matter. thanks in advance Morne Hartman mhartman@grintek.com --------------------------- Newsgroups: comp.os.vxworks Subject: HELP!!! Need glaux.dll Date: Tue, 15 Feb 2000 15:19:32 GMT From: hjhall@my-deja.com Organization: Deja.com - Before you buy. Message-ID: <88bqq0$ggh$1@nnrp1.deja.com> If you have it please send it to me, I am dead in the water without it. I am trying to run VxWorks Sim standalone but it requires this to run. VxWorksfield support can't find it. It was part of OpenGl. Thanks for your help! - -Howard Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: 3C905B Date: Tue, 15 Feb 2000 09:42:18 -0600 From: Frank Minich Organization: NCP Message-ID: <38A973DA.1FDD@ti.com> Reply-To: fminich@ti.com Hi VxWorksers, Has anyone seen or heard of a VxWorks driver for the 3Com 3C905B ("Etherlink XL") card? This card has a "Parallel Tasking II" chip on it (as opposed to the "Parallel Tasking" chip on the 3C905 card). The "Parallel Tasking II" chip uses a fundamentally different interaction model with the PC CPU than earlier 3Com cards (the PC CPU sets up a linked list of memory buffers, tells the 3C905B where the list starts, and the card uses its bus mastering capability to read from / write to the in-memory buffers). Any info appreciated, Frank. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: dosFsLib problem on Tornado 2.0 Date: Tue, 15 Feb 2000 16:10:26 GMT From: Mahesh Narayanaswamy Organization: @Home Network Message-ID: <38A979AC.49B9B178@home.com> References: <38713F32.890AE54F@jpl.nasa.gov> This is a multi-part message in MIME format. - --------------AB401FF8C4CC220F48F01699 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit You might try using DosFs 2.0 libraries. I had the same identicalproblem on a MIPS R5000 core. DosFs 2.0 is freely available from Wind River for all targets. Regards Mahesh "Jeffrey J. Biesiadecki" wrote: > Hi, > > I'm having a problem creating DOS file systems on our MIPS R3000 boards > running VxWorks 5.4, Tornado 2.0. I've attached a short test program > that uses ramDrv, which works fine when I boot VxWorks 5.3.1 / Tornado > 1.0 on the same board. The commercial board in use is a Heurikon V3500, > but the problem also occurs on our custom boards (also with a MIPS R3000 > processor). > > Basically, sometimes after making a new file system, trying to open() > files for writing in the file system returns with errno > VOLUME_NOT_AVAILABLE. dosFsConfigShow() prints "Cannot get > configuration data." I think it has something to do with the number of > sectors for the file system. > > Even with this error, it looks like it is writing the correct blocks for > the file system (I also tried this with block drivers for non-volatile > memory we have). After rebooting, I can do the dosFsDevInit() on the > non-volatile file systems and it works. > > I think we have the same facilities included with the T2.0 build versus > T1.0. In particular, hashLibInit() does get called before dosFsInit(). > We're using the Project tool to build the vxWorks image for T2.0. > > Below is a *test* program. Issuing > mkram "ram1/",4000,1 > results in a file system that dosFsConfigShow("ram1/") has a problem > with. But > mkram "ram2/",3900,1 > results in a file system that is fine. The problem can also occur when > using 2 sectors per cluster. Sometimes when I try calling dosFsMkfs, it > gives me an errno 0x00110001, which looks like > S_memLib_NOT_ENOUGH_MEMORY, even though there is plenty of memory (same > calls work fine on T1.0 on the same hardware). > > I'm sure I'm just doing something really stoopid... > > Many advTHANKSance for any thoughts! > Jeff Biesiadecki > Jeffrey.J.Biesiadecki@jpl.nasa.gov > -- > // mkram.cc - ramDrv test > #include > #include > #include > #include > #include > #include > > // make a DOS file system in RAM. > // must specify the name of the file system and the number of 512-byte > blocks. > // optionally, may specify number of sectors per cluster and the number > of > // sectors per copy of FAT. > int mkram(char *fsname, int nsec, int nsec_clust, int nsec_fat) > { > int blksz = 512; > > // initialize block device descriptor > char *ram_ptr = (char *)malloc(nsec*blksz); > BLK_DEV *ram_dev = ramDevCreate(ram_ptr,blksz,nsec,nsec,0); > if (ram_dev == NULL) > return 1; > > // compute number of clusters > if (nsec_clust <= 0) > nsec_clust = 2; > int nclust = nsec/nsec_clust; > > // compute number of sectors for FAT > if (nsec_fat <= 0) { > nsec_fat = (nclust > 4085) ? > ((2*nclust) + blksz - 1)/blksz : > ((3*nclust/2) + blksz - 1)/blksz; > } > > // create volume configuration > DOS_VOL_CONFIG vcfg; > if (dosFsConfigInit(&vcfg, > 0xF0, // mediaByte > nsec_clust, // secPerClust > 1, // nResrvd > 2, // nFats > nsec_fat, // secPerFat > 112, // maxRootEnts > 0, // nHidden > 0 // options > ) == ERROR) { > printf("dosFsConfigInit returned errno > 0x%08X\n",errnoGet()); > return 1; > } > > // create DOS file system > DOS_VOL_DESC *vdptr = dosFsDevInit(fsname,ram_dev,&vcfg); > if (vdptr == NULL) { > printf("dosFsDevInit returned errno > 0x%08X\n",errnoGet()); > return 1; > } > > // initialize DOS file system > int fd = open(fsname,O_WRONLY,0777); > if (fd == ERROR) { > printf("could not open %s for writing",fsname); > return 1; > } > ioctl(fd,FIODISKFORMAT,0); > ioctl(fd,FIODISKINIT,0); > close(fd); > > // all done > return 0; > } - --------------AB401FF8C4CC220F48F01699 Content-Type: text/x-vcard; charset=us-ascii; name="maheshn.vcf" Content-Transfer-Encoding: 7bit Content-Description: Card for Mahesh Narayanaswamy Content-Disposition: attachment; filename="maheshn.vcf" begin:vcard n:Narayanaswamy;Mahesh tel;cell:(510) 290-5816 tel;fax:(925) 949-5295 tel;home:(925) 735-7977 tel;work:(877) 260-0983 x-mozilla-html:FALSE adr:;;;;;; version:2.1 email;internet:maheshn@home.com fn:Mahesh Narayanaswamy end:vcard - --------------AB401FF8C4CC220F48F01699-- --------------------------- Newsgroups: comp.os.vxworks Subject: Spurious interrupt with PPC860/850 Date: Tue, 15 Feb 2000 11:28:15 -0500 From: "Jonathan Cheng" Message-ID: <24fq4.751$gi5.864@client> I am using PPC860 / 850 BSP in our custom board and I receive the following error message "uninitialized interrupt" from time to time. I check the offending interrupt and it is Level 7 interrupt. ( Interrupt #15 in vxWorks. ) As Level 7 interrupt is mask off, so I assume it is not a hardware problem. It seems to be a software problem because Level 7 interrupt is the default encoding in SIVEC register. I check the code in ppc860Intr.c and it seems OK to me. Is anyone experience any similar problem? Any idea? Thanks Jonathan Cheng --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Time Measurement Date: Tue, 15 Feb 2000 10:47:44 -0800 From: Harvey Taylor Organization: Organization? What organization?! Message-ID: <38A99F50.5C4C@despam.pangea.ca> References: <38a95f8b$0$35289@helios.is.co.za> Morné_Hartman wrote: > I would like to know how to measure the system time at a specific > moment, then at another moment, only a few milliseconds later, to > determine elapsed time. > Hi Morné, It depends what time sources you have on any given card. Do you have an RTC [real time clock]? Do you have a second clock you can use as an aux clock? If you have neither of those what processor are you using? Some CPUs have instructions which allow one to count cycles, which effectively gives you access to a microsecond clock. -het - --- "Assembly of Japanese bicycle require great peace of mind." - Instruction manual mentioned by R. Pirsig Harvey Taylor mailto:het@despam.portal.ca http://www.pangea.ca/~het --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Alignment in Structures Date: Tue, 15 Feb 2000 09:28:48 -0800 From: Art Berggreen Organization: Ericsson Data Networking and IP Services Message-ID: <38A98CD0.A050777B@acc.com> References: <38A93BAA.F5307BFC@hotmail.com> <38A94FBB.C2467A45@hotmail.com> "B. Lampi" wrote: > = > struct tagFoo > { > ... > } __attribute__ ((packed)); > = > does the job. > = > ...but: > = > struct tagTST > { > int Nr; > short LevelNumber; > int RefRange __attribute__ ((packed)); > short LotNumber; > int ApplyDate; > short SampleType; > }; > = > int QCLvlSize() > { > struct tagTST t; > t.Nr =3D 1; > t.LevelNumber =3D 2; > t.RefRange =3D 3; > t.LotNumber =3D 4; > t.ApplyDate =3D 5; > t.SampleType =3D 6; > HexDump( (char *)&t, 32 ); > = > printf( "TST size: %d %d\n", sizeof(struct tagTST), > sizeof(struct tagTST)=3D=3D18 ? 0 : -1 ); > return 0; > } > = > -> QCLvlSize > 0x21f920 0000: 00 00 00 01 00 02 00 00 ........ > 0x21f928 0008: 00 03 00 04 00 00 00 05 ........ > 0x21f930 0010: 00 06 EE EE EE EE EE EE ..=EE=EE=EE=EE=EE=EE <=3D=3D= ok > 0x21f938 0018: EE EE EE EE 00 21 73 10 =EE=EE=EE=EE.!s. > 0x21f940 0020: 00 21 F9 58 00 10 7C C4 .!=F9X..|=C4 > TST size: 20 -1 <=3D=3D=3D=3D=3D=3D=3D=3D=3D why 20???? and= not 18? > value =3D 0 =3D 0x0 > -> Because sizeof() includes padding at the end to keep both the start and end of the structure properly aligned. By definition, sizeof() of a structure returns the size occupied by the structure as an element of an array of those structures. Since, in this case, only an internal element is packed, the structure as a whole is still subject to the usual structure alignment constraints. Likely on this machine, since there is an int in the structure, the structure must be held to int alignment. Sizeof() then returns the size needed to place the next element on the same alignment. Making the entire structure packed ensures no padding, but then the code to access the structure elements has to assume no proper alignment and may be very inefficient. Hope this helps, Art --------------------------- Newsgroups: comp.os.vxworks Subject: time() returns 0? Date: Tue, 15 Feb 2000 13:36:57 -0500 From: "Dan Arsenault" Organization: ONG - Cenntenial Campus Message-ID: <88c6gp$46e@nntpb.cb.lucent.com> I thought this was previously posted, but I can't find it. Specifically, the time() function returns 0. I'm running a PPC 860. Do I need to enable/configure the time? --------------------------- Newsgroups: comp.os.vxworks Subject: Visual Studio Integration with T2 Date: Tue, 15 Feb 2000 22:11:40 GMT From: "Peter Luscher" Organization: @Home Network Canada Message-ID: Hi, We would like to integrate Visual Studio to be the editor that T2 uses. (Tools --> options --> External Editor). The editor comes up but we are unable to double click on a build error output lineand launch Visual Studio so that it opens the file with the cursor set to the correct line (ie. the line with the error). Does anyone know what the parameters are to Visual Studio so that we can open the file and automatically go the a given line number? (Note: Tornado 2 will supply the $lineno macro, we just need to know how to pass it to Visual Studio) Thanks! Peter. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: time() returns 0? Date: Tue, 15 Feb 2000 17:05:52 -0500 From: "Dan Arsenault" Organization: ONG - Cenntenial Campus Message-ID: <88ciog$646@nntpb.cb.lucent.com> References: <88c6gp$46e@nntpb.cb.lucent.com> I got it. I never set the time. thanks. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: VxSim communicate to WinNT Date: Thu, 17 Feb 2000 08:20:13 +0800 From: "joke" Organization: The news server of Changzhou TeleCom Message-ID: <88cska$2452$1@news.cz.js.cn> References: <877m79$hjq$1@newsmaster01.magnet.at> I know the full VxSim support TCP/IP to communicate with WinNT,but I don't know how to configure it. Mario Semo wrote in message news:877m79$hjq$1@newsmaster01.magnet.at... > Hello, > > what possibilities are there to communuicate from VxSim (as part of Tornado > 2.0) to the Host OS WinNT 4.05. ? off course i can open a file write > something into the file, close it and the windows counterpart does something > similar. as i remember i read, that TCP/IP is not supported. can i use > SHARED memory? what else? > > -- > regards, Mario Semo. http://www.kirchnersoft.com > > > --------------------------- Newsgroups: comp.os.vxworks Subject: Have you seen this funny Packet before ? Date: Wed, 16 Feb 2000 04:41:40 GMT From: "Jonghyeon Baik" Organization: @Home Network Message-ID: <8Upq4.11130$_p1.937558@news1.rdc1.md.home.com> Dear guys I'm working END driver (SMSC91C96). I have a BSD 4.3 and 4.4 driver for SMSC91C96. It works fine. I'm using StrongARM(SA1100). I made END driver and grasped a Ethernet Packet. This packet is that Target sends to Host machine for requesting FTP connection Target Mac Address : 00 60 0c 00 04 d9 Target IP Address : 207.196.69.93 Host PC MAC Address : 00 e0 29 41 76 dd Host IP address : 207.196.69.95 00 e0 29 41 76 dd 00 60 0c 00 04 d9 08 00 45 00 00 2c 91 e8 40 00 7f 06 3f 9e cf c4 45 5f cf c4 45 5d 00 15 04 00 logTask: 1 log messages lost. When you look at the first line, you should know Target sends packet to host. 00 e0 29 41 76 dd : destination Address (Host) 00 60 0c 00 04 d9 : Source Address (Target) Target ----> Host This is fine to me . When you look at the second and third line, probably you will be shocked. cf c4 45 5f : Source IP Address (host:207.196.69.95) cf c4 45 5d : Destination IP Address (Target:207.196.69.93) In IP layer, it looks like Host send this packet to Target. Host -------> Target. I don't think it is right. This is completely Wrong. Probably Host receives this packet, but in IP layer it will be thrown away. Because it lokks like host sends to Target. So I can't connect my host machine. I grasped whol packets, first two ICMP packet are right. After that, all packets have a same problem. Is this BUG in IP or Higher layer protocol ??????? --------------------------- Newsgroups: comp.os.vxworks Subject: tcl7.4 PPC860 blows up Date: Tue, 15 Feb 2000 22:06:03 -0800 From: "Randy Miyazaki (Go Dodgers Blue!)" Organization: Acm.Org Message-ID: <38AA3E4B.C784CA60@Acm.Org> Reply-To: Randy.Miyazaki@Acm.Org hi, i'm trying to use the sunsite tcl7.4 on a PPC860 with vxWorks 5.3/T2, some things work, but it blows up when i get a simple syntax error like expr a any suggestions ? thanx... ...randy +-------------------------------+----------------------+ | http://FreezeIt.com | fon: +1 408 859 8666 | | mailto:Randy.Miyazaki@Acm.Org | fax: +1 530 706 8359 | +-------------------------------+----------------------+ --------------------------- Newsgroups: comp.os.vxworks Subject: interbus-S support available? Date: Wed, 16 Feb 2000 08:36:48 +0100 From: "Mario Semo" Organization: Nextra Telekom GmbH Message-ID: <88doke$ndm$1@newsmaster01.nextra.at> Hello, Does anyone know of interbus-S drivers for VxWorks (from Tornado 2.0)? - -- regards, Mario Semo. http://www.kirchnersoft.com --------------------------- Newsgroups: comp.os.vxworks Subject: T2 + WSP + multiple targets Date: Wed, 16 Feb 2000 08:42:38 +0100 From: "Mario Semo" Organization: Nextra Telekom GmbH Message-ID: <88dokg$ndm$2@newsmaster01.nextra.at> Hello, assume i creates a new project in a new WSP. create DL-able app Project P10 add to NEW WSP : WSP10 (btw: can i define the default directory pathes used here for project and wsp somewhere?) a toolchain : SIMNTgnu then i added .cpp etc files now i want to compile the application for i80486 too. and later for PPC too. how can i do this? till now, i always created a P11 in WSP11 and readded and reconfigured everything. - -- regards, Mario Semo. http://www.kirchnersoft.com --------------------------- Newsgroups: comp.os.vxworks Subject: VME interrups Date: Wed, 16 Feb 2000 09:36:58 +0100 From: Tomas Larsson Organization: National Defence Research Establishment Message-ID: <38AA61A9.16FE731F@lin.foa.se> Hi all! We are using MVME2700 + a VIPC618 carrier board for IP modules in a VME cage One of the IP's DUAL/PIT is sat up to generate a PIRQ interrupt when pin H2 becomes asserted. When that happens a IRQ6 is generated on the VMEbus I call intConnect with vecnumber 0x15 as far I know all VME interrups should be connectet to that vector in the Universe But my Routine (PitISR) never get called. Can anyone explain what I'm doing wrong. intConnect(INUM_TO_IVEC(0x15),PitISR,0); I have tryed to use excIntConnect in that case my routin is called but everything else hangs when I use that approch. All help is appriciated. Tomas --------------------------- Newsgroups: comp.os.vxworks Subject: Re: GDB problems Date: Wed, 16 Feb 2000 09:57:41 +0100 From: Andy Winton Organization: Swisscom IP+ (post doesn't reflect views of Swisscom) Message-ID: <38AA6685.2912D368@stest.ch> References: <38A7CFEA.857A2B76@stest.ch> Hi again, Not even an RTFM response. :-( If there is any document or FAQ which covers debugging in detail can you please forward this to me. Go on, you know you want to help, cheers, andy Andy Winton wrote: > > Hi, > > In summary, gdb doesn't stop when a breakpoint is hit. > This used to be a lot worse with the graphical front > end, but we managed to have a work around by using just > the command line 'gdbi86' (using the wtx protocol). > Now that appears to be broken too. :-( > > -- > > We are using: GDB 4.12 on CrossWind 3.1 on VxWorks 5.3.1 > for the PC, Wind version 2.5. > > Sometimes it works in comand line mode, sometimes we > get - > > Proceedure - > > (1) - Attach Tornado to board (gdb needs the wtx) so that > the checksums don't complain... > License request... authorized on host 'zeus'. > Attaching backend... succeeded. > Connecting to target agent... succeeded. > Attaching C++ interface... succeeded. > Attaching a.out OMF reader... succeeded. > Added target_modules to target-server.....done > Added radius.lib to target-server.....done > > (2) - Load the program on vxWorks > > (3) - Run gdb and attach to board. My t204 is mapped > to 'target wtx wd5tu204'. > > lyra:a_winton> gdbi86 > GDB is free software and you are welcome to distribute copies of it > under certain conditions; type "show copying" to see the conditions. > There is absolutely no warranty for GDB; type "show warranty" for details. > > +------------------------------------------------------------------+ > | CrossWind 3.1 - Copyright (C) 1990-1996 Wind River Systems, Inc. | > | This is a modified version of GDB by Wind River Systems Inc. | > | Based on code from the original GDB version shown below: | > +------------------------------------------------------------------+ > > GDB 4.12 (sparc-sun-solaris2.4 --target i386-wrs-vxworks), > Copyright 1994 Free Software Foundation, Inc. > (gdb) t204 > Connecting to target server... > > C++ Constructors/Destructors Strategy is AUTOMATIC > > Connected to wd5tu204@lyra > vxWorks.st.bb586: (no debugging symbols found)...ok > tesa.lib: tesa.lib: No such file or directory. > tcpip.lib: ok > forth.lib: forth.lib: No such file or directory. > bcha.lib: ok > xntpd: xntpd: No such file or directory. > tddb.lib: tddb.lib: No such file or directory. > putzfrau.lib: putzfrau.lib: No such file or directory. > casd.lib: warning: could not lookup symbol: > SYMTBL_SYMBOL_NOT_FOUND > ok > hdlc.lib: ok > foex.lib: foex.lib: No such file or directory. > super.lib: warning: could not lookup symbol: > SYMTBL_SYMBOL_NOT_FOUND > ok > atmc.lib: atmc.lib: No such file or directory. > sftc.lib: warning: could not lookup symbol: > SYMTBL_SYMBOL_NOT_FOUND > ok > dvm.lib: dvm.lib: No such file or directory. > pmi.lib: pmi.lib: No such file or directory. > admo.lib: admo.lib: No such file or directory. > hi32.lib: ok > resu.lib: resu.lib: No such file or directory. > strt.lib: ok > tmsl.lib: tmsl.lib: No such file or directory. > layer1.lib: ok > spr.lib: spr.lib: No such file or directory. > tdad.lib: tdad.lib: No such file or directory. > retd.lib: ok > lay2.lib: ok > radius.lib: ok > Done. > (gdb) > (gdb) b radius_main > Breakpoint 1 at 0xeb8c957: file radiusd.c, line 299. > (gdb) r radius_main > Starting program: radius_main > > ^C > > (4) Just as I do the run, but before I do the Ctrl-C, I see on the > console - > -> > Unknown breakpoint at 0xeb89253 > > (5) Gdb does not stop at radius_main, but keeps running. I had to > stop this with Ctrl-C. When I did this I get on the console - > ... > 0xeb8daf8 (ä¸): dbgTrace: invalid case -1 pc=0x11cad0 > 0xeb8daf8 (ä¸): dbgTrace: invalid case -1 pc=0x1435dc > 0xeb8daf8 (ä¸): dbgTrace: invalid case -1 pc=0x11cad0 > 0xeb8daf8 (ä¸): dbgTrace: invalid case -1 pc=0x11cad0 > 0xeb8daf8 (ä¸): dbgTrace: invalid case -1 pc=0x11cad0 > 0xeb8daf8 (ä¸): dbgTrace: invalid case -1 pc=0x11cad0 > ... > > This is in a nutshell one of the problems with > GDB. > > Thanks for your time, > > andy > -- > Andy Winton, andy.winton@stest.ch - -- Andy Winton, andy.winton@stest.ch. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Spurious interrupt with PPC860/850 Date: Wed, 16 Feb 2000 10:02:53 +0100 From: "Richard Marsh" Organization: Siemens Inc. Message-ID: <88dp3g$831$1@news.mch.sbs.de> References: <24fq4.751$gi5.864@client> Reply-To: "Richard Marsh" > I am using PPC860 / 850 BSP in our custom board and I receive the following > error message "uninitialized interrupt" from time to time. > > I check the offending interrupt and it is Level 7 interrupt. ( Interrupt #15 > in vxWorks. ) As Level 7 interrupt is mask off, so I assume it is not a > hardware problem. It seems to be a software problem because Level 7 > interrupt is the default encoding in SIVEC register. I check the code in > ppc860Intr.c and it seems OK to me. > > Is anyone experience any similar problem? > > Any idea? > > Thanks > > Jonathan Cheng > Yes. we have exactly the same problem with our 860T's. We spent ages trying to find out what was happening, but in the end we couldn't find anything wrong in the code. We could never trap the problem using an ICE. Eventually we tracked the problem down (using good old logMsg) to a CPM error interrupt that occurs when the mask is removed. There is no mask availbale for this interrupt. The problem was made worse by the fact that some boards had the problem (perhaps happening once every couple of days), and some didn't (never). In the end we just put a dummy handler in that ignored the interrupt. We haven't noticed any side effects (yet!) and the fix has in for about 6 months. We have recently moved to the 860P, but I don't think anybody has removed the handler to see if the problem exists on that processor. and the code added to ppc860IntrInit was: intConnect (IV_ERROR, (VOIDFUNCPTR)ppc860IgnoreInterrupt, 0); and the function: LOCAL void ppc860IgnoreInterrupt (int arg) { /*logMsg("Interrupt Ignored\n",0,0,0,0,0,0)*/; } HTH Richard Marsh --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Booting from compressed image Date: 16 Feb 2000 09:14:17 GMT From: johan.borkhuis@peektraffic.nl Organization: XS4ALL Internet BV Message-ID: <88dpp9$7ir$1@news1.xs4all.nl> References: <200002160800.JAA03536@net.softec.de> Michael Lawnick wrote: : why don't you add your application to the kernel via : INCLUDE_USR_APPL and MACH_EXTRA and build a : vxWorks_rom ? You would save the 0.5 Mb of bootrom and your : kernel + application will be compressed. There is a good reason for that: we want to have one version of the kernel and several different applications. We also want to be able to replace the kernel without the risk of destroying the bootstrap. Groeten, Johan - -- o o o o o o o . . . ___________________________________ o _____ || Johan Borkhuis | .][__n_n_|DD[ ====_____ | johan@borksoft.xs4all.nl | >(________|__|_[_________]_|________________________________| _/oo OOOOO oo` ooo ooo 'o!o!o o!o!o` === VxWorks page: http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html === --------------------------- Newsgroups: comp.os.vxworks Subject: Re: MPC850 SMC is slowly Date: Wed, 16 Feb 2000 11:21:24 +0200 From: "Shuky Persky" Organization: Bezeq International Ltd. Message-ID: <38aa6bcb@news.bezeqint.net> References: <87s3ru$gue7@overload.lbl.gov> - -- Hi, This timing looks correct (see the UM explanation on how SMC operates). Your other option is (assuming you have an available SCC) is to init an SCC to UART mode, and in order to short the duration until start of transmission use the TODR (Transmit on Demand Reg.). Enjoy, Shuky - ------------------------------ email : shukyp@bigfoot.com stevens wrote in message news:87s3ru$gue7@overload.lbl.gov... > Hi VxWorkers, > > we use a custom BSP with a MPC850 and Tornado2. > A serial device (19.200 baud) realized with the Serial Management Controller needs 0.7 ms to send a byte. > The SMC is in idle. The time is measured from setting the ready-bit in the buffer descriptor to the occurance of the > byte on the serial line. > > How can we get a faster reaction ? > > Regards, > Torsten > > -- > > ************************************************************************** > fraunhofer institute of microelectronic circuits and systems > department sat > name torsten stevens mailto:stevens@ims.fhg.de > address finkenstrasse 61 phone +0049/203/3783-238 > 47057 duisburg fax +0049/203/3783-266 > germany > http://www.fhg.de/depts/ims-du-e.html > ************************************************************************** > > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: T2 + WSP + multiple targets Date: 16 Feb 2000 09:22:07 GMT From: johan.borkhuis@peektraffic.nl Organization: XS4ALL Internet BV Message-ID: <88dq7v$7ir$2@news1.xs4all.nl> References: <88dokg$ndm$2@newsmaster01.nextra.at> Mario Semo wrote: : now i want to compile the application for i80486 too. and later for PPC too. : how can i do this? When you go to the builds-tab in yout workspace window and select the project you are working on, click with the right mouse and you get a menu. One of the items is "New builds" Groeten, Johan - -- o o o o o o o . . . ___________________________________ o _____ || Johan Borkhuis | .][__n_n_|DD[ ====_____ | johan@borksoft.xs4all.nl | >(________|__|_[_________]_|________________________________| _/oo OOOOO oo` ooo ooo 'o!o!o o!o!o` === VxWorks page: http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html === --------------------------- Newsgroups: comp.os.vxworks Subject: change date and time of a file Date: Wed, 16 Feb 2000 10:54:36 GMT From: martins2@my-deja.com Organization: Deja.com - Before you buy. Message-ID: <88dvlc$1gn$1@nnrp1.deja.com> Hello everybody, is there a way I can change the date and time for a file in a dosFS filesystem in VxWorks? I am already aware of dosFsDateTimeInstall() and other functions that change the date and time for a newly. What I need is to change the data and time of a file to any value without opening it and without using dosFsDateSet(). Background: I want to copy a file from one place to another. When using copy() the new file has the date/time when it was copied. What I would like to have is the same time and date as in the original file. Thanks in advance for any help. Martin Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Relocatable application Date: Wed, 16 Feb 2000 13:33:36 +0200 From: "Tomer Baz" Organization: Terayon Communication Systems Message-ID: <88e1h7$ake$1@news2.inter.net.il> Hi all, I want to generate a relocatable application for the PowerPC (860) so I could load it anywhere in memory. I'm using the GNU compiler and I tried to compile my files with the -mrelocatable compiler option, but it didn't do the trick. Has someone experienced writing relocatable applications and can contribute some of his/her knowledge? 10x in advance, Tomer Baz Terayon Communication Systems --------------------------- End of New-News digest ********************** From vxwexplo-errs@csg.lbl.gov Wed Feb 16 05:23:13 2000 From: Leif Thuresson Date: Wed Feb 16 05:23:21 PST 2000 Subject: RE: interbus-S support available? I have used the Phonix contact IBS VME 6H CB/I-T card to control some simple I/O modules under vxWorks. This Interbus-S master card is a few years old and I don't know if it is still available, but if you are interested I can send you my driver routines. I also think there are device driver development kits available for the new VME Interbus-S cards that Phonix contact supply. Regards, -- leif --------------------------------------------------------- Leif Thuresson Tel: +46-18-4713860 Uppsala University Fax: +46-18-4713833 The Svedberg Laboratory E-mail: leif.thuresson@tsl.uu.se P.O.Box 533 S-751 21 Uppsala Sweden From vxwexplo-errs@csg.lbl.gov Thu Feb 17 04:03:09 2000 From: Vxworks Exploder Date: Thu Feb 17 04:03:11 PST 2000 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Thu Feb 17 04:03:06 PST 2000 Subject: Re: Spurious interrupt with PPC860/850 Subject: ...and the last of the RAM went where? Subject: Re: change date and time of a file Subject: Using RealMode VGA Bios to switch modes? Subject: Re: Problem Running Post-Mortem Subject: Re: time() returns 0? Subject: Re: VxWorks: Another problem with Ping Subject: sntpcTimeGet problem Subject: Re: VxWorks: Another problem with Ping Subject: T2 Shell Question Subject: Adding a 2nd END ethernet device Subject: Re: Time Measurement Subject: Re: VME interrups Subject: Re: want to load an object module into memory, while using rom resident Subject: Re: VME interrups Subject: Re: Adding a 2nd END ethernet device ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: Spurious interrupt with PPC860/850 Date: Wed, 16 Feb 2000 12:02:54 -0000 From: "Will Fookes" Organization: NDS UK Message-ID: <88e21o$v9k$1@ndsukns01.ndsuk.com> References: <24fq4.751$gi5.864@client> Jonathan, We had a similar problem with the PPC860SAR when we enabled the D-cache - turned out it was the CPM Error interrupt. We fixed it by changing a couple of lines in init.s in the bootrom... /* Initialise instruction support control register (ICTRL) */ lis r5,HIADJ(0x00000007) /* was 0x00000006 */ addi r5,r5,LO(0x00000007) /* was 0x00000006 */ mtspr ICTRL,r5 I understand that this disables the placing of certain debug information on the bus. HTH, Will Jonathan Cheng wrote in message <24fq4.751$gi5.864@client>... > >I am using PPC860 / 850 BSP in our custom board and I receive the following >error message "uninitialized interrupt" from time to time. > >I check the offending interrupt and it is Level 7 interrupt. ( Interrupt #15 >in vxWorks. ) As Level 7 interrupt is mask off, so I assume it is not a >hardware problem. It seems to be a software problem because Level 7 >interrupt is the default encoding in SIVEC register. I check the code in >ppc860Intr.c and it seems OK to me. > >Is anyone experience any similar problem? > >Any idea? > >Thanks > >Jonathan Cheng > > --------------------------- Newsgroups: comp.os.vxworks Subject: ...and the last of the RAM went where? Date: 16 Feb 2000 13:01:17 GMT From: Geir Frode Raanes Organization: UNINETT news service Message-ID: <88e72t$jpc$1@snipp.uninett.no> I have been struggeling to regain all unused DRAM on our MBX860-VxWorks setup. This PowerPC card has 4 MByte of DRAM soldered onboard. This RAM is mapped into the range of LOCAL_MEM_LOCAL_ADRS = 0x00000000 to LOCAL_MEM_SIZE = 0x00400000. This area is marked in sysLib.s as follows: PHYS_MEM_DESC sysPhysMemDesc [] = { { (void *) LOCAL_MEM_LOCAL_ADRS, (void *) LOCAL_MEM_LOCAL_ADRS, LOCAL_MEM_SIZE , VM_STATE_MASK_VALID | VM_STATE_MASK_WRITABLE | VM_STATE_MASK_CACHEABLE, VM_STATE_VALID | VM_STATE_WRITABLE | VM_STATE_CACHEABLE },.... Now, the VxWorks kernel is loaded from the network directly into RAM_LOW_ADRS = 0x00010000. The kernel is not compressed so the RAM_HIGH_ADRS = 0x000EFFFF does not really apply. Evenso, RAM_HIGH_ADRS do seem to enter the equation when calculating the value of (end) in FREE_RAM_ADRS = (end). I have stripped the VxWorks kernel to the point where Tornado2 reports a kernel size of TEXT = 454340, DATA = 10734 and BSS = 47954 with a total size of 513028 Bytes. Any less functionality than this is not acceptable. Now, the kernel is half a megabyte in size (reports 641976 Bytes downloaded,) starting at 0x10000 (64KByte.) Disregarding stack space this means there is still some 3.5 MBytes left on the machine. The following is what vxWorks reports: -> memShow status bytes blocks avg block max block ------ --------- -------- ---------- ---------- current free 2409360 168 14341 1724576 alloc 1072400 270 3971 - cumulative alloc 1072496 272 3943 - As we can see, vxWorks has allocated some 1072400 Bytes for itself, including stack space. This is far more stack space than we need - how do I adjust the size of the stack? Still there is half a megabyte missing between the top of the kernel and bottom of free space. Also, the largest allocateable block is mere 1724576 bytes, which is less than total free memory. My guess is that free mem is fragmented, whith the missing 500 KBytes inbetween. Is there an equivalent of "moduleShow" for the kernel space itself? Below is what is loaded into free space. Also, size is missing from the table. -> moduleShow MODULE NAME MODULE ID GROUP # TEXT START DATA START BSS START --------------- ---------- ---------- ---------- ---------- ---------- testDrv.exe 0x3fb8f0 1 0x251fd0 0x252fe0 0x252fe8 Next problem is that even though I've managed to free up 1.7 MBytes of continous memory, the VxWorks v5.4 target shell resident program loader refuses to load an application with no more than 700 KBytes of text segment. "Can not allocate space for text and data segments..." or something to that effect. Objectdump reports that the remaining segments totals some 200 KByte. This program should effortlessly fit into free memory, but noooo. I can not find any other macro settings in the MBX860 BSP to control the use of memory. I have not yet tried to set the FREE_RAM_ADRS explicit and I suspect that any attempt to do so will crash horrendously. So what does VxWorks use memory for and where? There is no architectual overview of VxWorks in the docs. How can I get a list of allocated RAM som that I at least can do a manual "memAddToPool" on any unclaimed memory I migth find? - -- ****************************************************** Never ever underestimate the power of human stupidity. -Robert Anson Heinlein GeirFRS@invalid.ed.ntnu.no ****************************************************** --------------------------- Newsgroups: comp.os.vxworks Subject: Re: change date and time of a file Date: Wed, 16 Feb 2000 08:47:41 -0500 From: Charles H. Chapman Organization: Dynacs Engineering Company Message-ID: <88e9q4$4kv@niven.ksc.nasa.gov> References: <88dvlc$1gn$1@nnrp1.deja.com> Reply-To: chc@nasa2.ksc.nasa.gov martins2@my-deja.com wrote: > > What I need is to change the data and time of a file to any value > without opening it and without using dosFsDateSet(). Background: I want > to copy a file from one place to another. When using copy() the new file > has the date/time when it was copied. What I would like to have is the > same time and date as in the original file. I used a combination of "stat" and "utime" to do this function. Chuck --------------------------- Newsgroups: comp.os.vxworks Subject: Using RealMode VGA Bios to switch modes? Date: Wed, 16 Feb 2000 09:18:15 -0500 From: Burt Bicksler Organization: BBB Message-ID: Hi, I have need to switch the VGA from graphics mode (used to display a splash screen while vxWorks bootrom is being expanded) to text mode. While this can be done very early in the boot process while still in real mode via int 10h this doesn't appear to be available from protected mode once the romIint process has switched to protected mode. So my question.. Is there any mechanism in vxWorks that would allow a call to the existing VGA BIOS, e.g. switch to real mode temporarily for the call and then back to protected mode? Since I haven't been able to find any reference in the docs indexes to doing this I expect that the answer is no, but wondered if anyone else has encountered a similar problem and how you dealt with it? As a hedge I'm implementing the register level code to implement the mode setting, but it is tedious and requires a fair amount of code and data to be included. If there is something documented on an easier way to do this could someone point me in the direction of that information? Thanks, Burt Bickser P.S. Remove the spam bait from my e-mail address if responding via e- mail. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Problem Running Post-Mortem Date: Wed, 16 Feb 2000 11:27:01 -0500 From: "Jonathan Cheng" Message-ID: <19Aq4.718$KQ6.1044@client> References: <88a0n0$8t9$1@nnrp1.deja.com> You may not have enough memory for the Buffer. You can check the the definition in usrWindView.c #define WV_DEFAULT_BUF_MIN 4 #define WV_DEFAULT_BUF_SIZE 0x8000 - -Jonathan Cheng shanan8924@my-deja.com wrote in message <88a0n0$8t9$1@nnrp1.deja.com>... >I am trying to run the Windview in Post-mortem >mode, and I receive the following error. > >Error: wvLogHeaderCreate > >I am assuming this function is not receiving the >proper memory PartId, therefore returning null. I >am not sure how to correct this. Any help would be >appreciated. > >Thanx > >Shanan > > >Sent via Deja.com http://www.deja.com/ >Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: time() returns 0? Date: Wed, 16 Feb 2000 18:36:46 +0100 From: "Zakelijk" Organization: UUNET-NL (http://www.nl.uu.net) Message-ID: <88encp$1d3$1@porthos.nl.uu.net> References: <88c6gp$46e@nntpb.cb.lucent.com> <88ciog$646@nntpb.cb.lucent.com> DAN... great post... but now the sollution please. We got a VxWorks running but the clock_gettime never seems to return the real bios clock but returns the time since the reboot of the system. Is there a way to set the correct time ? Dan Arsenault wrote in message news:88ciog$646@nntpb.cb.lucent.com... > I got it. I never set the time. thanks. > > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: VxWorks: Another problem with Ping Date: Wed, 16 Feb 2000 18:43:36 +0100 From: "Zakelijk" Organization: UUNET-NL (http://www.nl.uu.net) Message-ID: <88enpi$1jb$1@porthos.nl.uu.net> References: <87r47e$guf8@overload.lbl.gov> Ping problem 2? This looks a bit like a problem we are encounting. When we let the pcPentium load it's VxWorks image from a target-server (development system) everything seems to work but as soon as we move the image to the target system it self the ping seems to "fade" and doesn't do anything anymore? Anyone.. plz. Nagaraj G wrote in message news:87r47e$guf8@overload.lbl.gov... > Hi VxWorkers, > > I have another problem with the "ping" in VxWorks. I am using pcPentium > BSP with Tornado II. > > I tried doing a IP Broadcast to the broadcast address with the "ping" > command. If I try ping with a "pingTxLen" up to 1480 bytes, it works > fine. But, if I try anything more than that, it just fails. It does not > even send anything on the LAN, which it does when the length is <=1480 > bytes. > > Does anyone have a solution or a work-around for this problem? > > Thanks in advance. > > Best Regards, > > Nagaraj G > EnThink Inc > www.EnThink.com --------------------------- Newsgroups: comp.os.vxworks Subject: sntpcTimeGet problem Date: Wed, 16 Feb 2000 18:56:20 +0100 From: "Zakelijk" Organization: UUNET-NL (http://www.nl.uu.net) Message-ID: <88eohf$1vj$1@porthos.nl.uu.net> Hi all, When we build the SNTP Client into VxWorks the server seems to build without any problem but when we try to download a program which uses the sntpcTimeGet function to "retrieve" a time from a SNTP Server the download fails. We have tried to link the program into the VxWorks and the linking phase fails. We are VERY VERY sure we use the right name. Even all the parameters are right but it doesn't seem to work. If we create the VxWorks (without the program but with the SNTP Client) we can start the function (sntpcTimeGet) but this never returns any correct result. It can't seem to find the SNTP Server. We use the correct IP adress (i know this because we use more SNTP Clients in the same network !!). Thx Leendert Biemans --------------------------- Newsgroups: comp.os.vxworks Subject: Re: VxWorks: Another problem with Ping Date: 16 Feb 2000 19:24:55 GMT From: johan.borkhuis@peektraffic.nl Organization: XS4ALL Internet BV Message-ID: <88eti7$don$1@news1.xs4all.nl> References: <87r47e$guf8@overload.lbl.gov> <88enpi$1jb$1@porthos.nl.uu.net> Zakelijk wrote: : Ping problem 2? : This looks a bit like a problem we are encounting. : When we let the pcPentium load it's VxWorks image from a target-server : (development system) everything seems to work but as soon as we move the : image to the : target system it self the ping seems to "fade" and doesn't do anything : anymore? Do you startup your network? When you boot from a local device you have to staretup the network using the o (=other) boot parameter. I have this worked out on my VxWorks page (3.2). Groeten, Johan - -- o o o o o o o . . . ___________________________________ o _____ || Johan Borkhuis | .][__n_n_|DD[ ====_____ | johan@borksoft.xs4all.nl | >(________|__|_[_________]_|________________________________| _/oo OOOOO oo` ooo ooo 'o!o!o o!o!o` === VxWorks page: http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html === --------------------------- Newsgroups: comp.os.vxworks Subject: T2 Shell Question Date: Wed, 16 Feb 2000 22:55:24 GMT From: Ken Ostaszewski Organization: Boeing, Inc. Message-ID: <38AB2ADC.C55B7D0A@boeing.com> Sender: nntp@news.boeing.com (Boeing NNTP News Access) Hi There, I am having a problem with the T2 Target shell. My VxWorks config requires the target shell to be included from the project facility. However, I only want the target shell to start once my application has concluded its operation. This is due to my application requesting a start command from the user from the console. However the shell is already running and will not pass the command to the program, just to the target shell, which doesn't know what to do with it. Any ideas would be appreciated. Ken Ostaszewski --------------------------- Newsgroups: comp.os.vxworks Subject: Adding a 2nd END ethernet device Date: Wed, 16 Feb 2000 19:26:38 -0500 From: Tim Conley Organization: Raytheon Company Message-ID: <38AB403D.E9D745AA@res.raytheon.com> Hi all, I've been trying to set up a PMC ethernet card on a DY-4 SVME-178 (PowerPC) board. The Dy-4 board comes with an on board DEC 21140 ethernet chip. The PMC card I'm adding came from Radstone, but is actually a Compaq board which also uses the DEC 21140. (Note that the drivers that were supplied by Radstone are for a Radstone PPC board or a Motorola MV1604. When I called Radstone, their sales rep laughed and said that they would not supply a driver for a "competitors" board. If they had only told us that before the sale!) I'm running Tornado 2.0, on a Solaris host. The DY-4 BSP uses an END driver for the built in ethernet. This is what I've tried: I've added a 2nd entry into the endDevTbl in the BSP's configNet.h. I adjusted the parameters in the 2nd entry based on information obtained from pciDeviceShow and pciHeaderShow. The MUX initialization in usrNetwork.c appears to work for both ethernet devices. (i.e. the calls to muxDevLoad and muxDevStart). A call to muxShow has both devices, dc unit 0 and dc unit 1. The calls to ipAttach also work for both devices. The call to usrNetIfConfig works for the first device (dc0). The call for the second device (dc1) fails with a S_errno_EEXIST error. Its atually the call to ifAddrSet that fails. Each ethernet card has a unique IP address and name. I just tried using a very different IP address for the 2nd card, i.e. 100.100.100.100 and the call to ifAddrSet worked!? Can anyone provide a clue as to what I'm seeing? Any help would be greatly appreciated! TIA, Tim Conley FishEye Software, Inc. Email at client's site: Timothy_G_Conley--NR@res.raytheon.com FishEye email: tgconley@fisheye.net FishEye Web Site: http://www.fisheye.net vxworks tornado wind river --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Time Measurement Date: Wed, 16 Feb 2000 19:37:18 -0500 From: Tim Conley Organization: Raytheon Company Message-ID: <38AB42BE.2146593A@res.raytheon.com> References: <38a95f8b$0$35289@helios.is.co.za> You can change the system clock rate in $(WIND_BASE)/target/config/all/usrConfig.c. Change the sysClkRateSet call to a larger number. We are using a value of 5000, which gives us tick granularity of 200 microseconds, your mileage may vary. Tim Conley FishEye Software, Inc. Email at client's site: Timothy_G_Conley--NR@res.raytheon.com FishEye email: tgconley@fisheye.net FishEye Web Site: http://www.fisheye.net Morné Hartman wrote: > Hi all, > > I would like to know how to measure the system time at a specific > moment, then at another moment, only a few milliseconds later, to > determine elapsed time. > > I have tried clock_gettime(), but this is no good. I have checked the > clock resolution and it is at 60 ticks per second. > > Does anyone have any help in this matter. > > thanks in advance > > Morne Hartman > mhartman@grintek.com --------------------------- Newsgroups: comp.os.vxworks Subject: Re: VME interrups Date: Wed, 16 Feb 2000 16:59:46 -0800 From: Jim Pisano Organization: California Institute of Technology, Pasadena Message-ID: <38AB4802.FEA78527@nrao.edu> References: <38AA61A9.16FE731F@lin.foa.se> I had a similar problem w/ IRQ 7 & the parallel port. You need to make sure that the SuperIO chip is configured to generate the interrupt. This is a complicated chip which requires the data sheet. I have a PDF copy if you need it. I'm away from my office until next week so I can send it to you then. Also I can send code which I used to configure the chip & enable the interrupt. It was a real pain. Once the SuperIO chip is configured correctly, *then* the intConnect() function will work. - -Jim Tomas Larsson wrote: > > Hi all! > We are using MVME2700 + a > VIPC618 carrier board for IP > modules > in a VME cage > One of the IP's DUAL/PIT is > sat up to generate a > PIRQ interrupt when pin H2 > becomes asserted. When that > happens a IRQ6 is generated on > the VMEbus > I call intConnect with > vecnumber 0x15 as far I know > all VME interrups should be > connectet to > that vector in the Universe > But my Routine (PitISR) never > get called. Can anyone > explain what I'm doing wrong. > > intConnect(INUM_TO_IVEC(0x15),PitISR,0); > > I have tryed to use > excIntConnect in that case my > routin is called but > everything else hangs > when I use that approch. > > All help is appriciated. > Tomas --------------------------- Newsgroups: comp.os.vxworks Subject: Re: want to load an object module into memory, while using rom resident vxWorks Date: Thu, 17 Feb 2000 01:50:40 GMT From: nunya Organization: Nunya Business Message-ID: <38AB53ED.9563AA4F@nunya.com> References: <881b4l$chr$1@nnrp1.deja.com> Using Tornado 1.0.1 I ran into this and solved it this way: 1.) Configure a target server. 2.) Change the property of "Memory Cache Size" to a larger value. 3.) Use "Debug:Download" to download the image... Not sure about Tornado II.... findik@my-deja.com wrote: > I am using Tornado II, > > I have downloaded vxWorks.res_rom_res_low build to my target, > > I can use Tornado II, open a shell for this target, > > I want to load an object module into memory, > > I got the following errors: > > form CONSOLE: > > -> ld (1) ld error: error reading file (errno = 0xe0001). > > from SHELL: > > -> ld < viaFlash.o > Loading D:/Via/vxproje/FlashProg/PPC860gnu/viaFlash.o | > WTX Error 0x10034 (LOADER_RELOCATION_OFFSET_TOO_LARGE) > value = -1 = 0xffffffff > > Thanks for your valuable comments, > > best regards, Reyhan > > Sent via Deja.com http://www.deja.com/ > Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: VME interrups Date: Thu, 17 Feb 2000 02:46:28 GMT From: Robert Klinkhammer Organization: Road Runner Message-ID: <38AB63F1.148C9427@mediaone.net> References: <38AA61A9.16FE731F@lin.foa.se> Try only using even vector numbers. There is a known bug in the universe driver code (wrongly interpreted a universe errata) that prevents any vector with the lower bit set from working. Use vectors 10,12, ... etc. I would try to use something higher, though. Also, don't forget the sysBusEnable for the level. Tomas Larsson wrote: > > Hi all! > We are using MVME2700 + a > VIPC618 carrier board for IP > modules > in a VME cage > One of the IP's DUAL/PIT is > sat up to generate a > PIRQ interrupt when pin H2 > becomes asserted. When that > happens a IRQ6 is generated on > the VMEbus > I call intConnect with > vecnumber 0x15 as far I know > all VME interrups should be > connectet to > that vector in the Universe > But my Routine (PitISR) never > get called. Can anyone > explain what I'm doing wrong. > > intConnect(INUM_TO_IVEC(0x15),PitISR,0); > > I have tryed to use > excIntConnect in that case my > routin is called but > everything else hangs > when I use that approch. > > All help is appriciated. > Tomas --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Adding a 2nd END ethernet device Date: Thu, 17 Feb 2000 02:49:11 GMT From: Robert Klinkhammer Organization: Road Runner Message-ID: <38AB6499.A16E0757@mediaone.net> References: <38AB403D.E9D745AA@res.raytheon.com> How are you dealing with the ethernet address? (MAC address?) Tim Conley wrote: > > Hi all, > > I've been trying to set up a PMC ethernet card on a DY-4 SVME-178 > (PowerPC) board. The Dy-4 board comes with an on board DEC 21140 > ethernet chip. The PMC card I'm adding came from Radstone, but is > actually a Compaq board which also uses the DEC 21140. (Note that the > drivers that were supplied by Radstone are for a Radstone PPC board or a > > Motorola MV1604. When I called Radstone, their sales rep laughed and > said that they would not supply a driver for a "competitors" board. If > they had only told us that before the sale!) I'm running Tornado 2.0, > on a Solaris host. The DY-4 BSP uses an END driver for the built in > ethernet. > > This is what I've tried: > I've added a 2nd entry into the endDevTbl in the BSP's configNet.h. I > adjusted the parameters in the 2nd entry based on information obtained > from pciDeviceShow and pciHeaderShow. The MUX initialization in > usrNetwork.c appears to work for both ethernet devices. (i.e. the calls > to muxDevLoad and muxDevStart). A call to muxShow has both devices, > dc unit 0 and dc unit 1. The calls to ipAttach also work for > both devices. The call to usrNetIfConfig works for the first device > (dc0). The call for the second device (dc1) fails with a S_errno_EEXIST > > error. Its atually the call to ifAddrSet that fails. Each ethernet > card has a unique > IP address and name. I just tried using a very different IP address for > the 2nd card, > i.e. 100.100.100.100 and the call to ifAddrSet worked!? Can anyone > provide a clue as > to what I'm seeing? > > Any help would be greatly appreciated! > TIA, > > Tim Conley > FishEye Software, Inc. > Email at client's site: Timothy_G_Conley--NR@res.raytheon.com > FishEye email: tgconley@fisheye.net > FishEye Web Site: http://www.fisheye.net > > vxworks tornado wind river --------------------------- End of New-News digest ********************** From vxwexplo-errs@csg.lbl.gov Thu Feb 17 09:46:59 2000 From: "Phil Moy" Date: Thu Feb 17 09:47:01 PST 2000 Subject: VxWorks problem using math functions in vxWorks.st I am trying to place my application into rom on a MVME-167 and I am having problems with the cos math function: Unimplemented Opcode 1111 Program Counter: 0x000187d4 Status Register: 0x3000 11b6c _vxTaskEntry +10 : _shell (1, 0, 0, 0, 0, 0, 0, 0, 0, 0) 2f536 _shell +12e: 2f554 ([1, 0, 0, 2f364, 0]) 2f6cc _shell +2c4: _execute (7f710e) 2f7f0 _execute +ac : _yyparse ([0, 1, 0, 7f710e, 0]) 339b4 _yyparse +514: 3237e (7fea50, 7fea70) 3248a _yystart +758: _hwTest (0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0) 5040 _hwTest +15e: _vdrmMenu ([&_yyval, 3, &_hwTest, 76006ffe, 7f6ffa]) 5628 _vdrmMenu +17e: _vdrmStartupTest ([7f6f24, 1, 3, &_printf, 31000002] ) 878c _vdrmStartupTest+6 : _vdrmTestInit ([&_printf, 7f6f10, 562e, 7f6f24, 1]) 803e _vdrmTestInit +7c : _calcHanningWindow ([1, 7f6e48, 1, 1, 3a]) 86aa _calcHanningWindow+4e : _mathHardCos (3f66571b, b75d4094) 47aa0 _mathHardCos +10 : _in_arpinput (3f66571b, b75d4094) shell restarted. This does work in the non standalone VxWorks. Any help will be greatly appreciated. Thanks in advance, Phil Moy From vxwexplo-errs@csg.lbl.gov Thu Feb 17 18:33:09 2000 From: Fred Roeber Date: Thu Feb 17 18:33:13 PST 2000 Subject: Re: Time Measurement On Thu, 17 Feb 2000, Morne Hartman wrote: > I would like to know how to measure the system time at a specific > moment, then at another moment, only a few milliseconds later, to > determine elapsed time. > > I have tried clock_gettime(), but this is no good. I have checked the > clock resolution and it is at 60 ticks per second. Tim Conley answered: > You can change the system clock rate in > $(WIND_BASE)/target/config/all/usrConfig.c. > Change the sysClkRateSet call to a larger number. > We are using a value of 5000, which gives us tick granularity of 200 > microseconds, your mileage may vary. I would like to suggest a second alternative. Actually, this question comes up a fair amount and should be covered in the VxWorks FAQ. IMHO, the easiest/best way to measure detailed elapsed time is to use any "timestamp" driver that may be supplied as part of your board's BSP. The timestamp driver is a generic name for a high resolution timing factility that is intended for WindView use but can also be used by user applications. The documentation for creating/using the timestamp facility is in Appendix G of the WindView manual and the functions are defined in h/drv/timer/timerDev.h (see sysTimestamp...). Many BSPs come with timestamp driver support even if you don't buy WindView. You can add the configuration option INCLUDE_TIMESTAMP to build it into your system. You can then call sysTimestampEnable to start it up and sysTimestampLock to read detailed time values. You can call sysTimestampPeriod to find out how often the timer rolls over and sysTimestampFreq to see the rate it ticks at (which is often fast enough to provide microsecond resolution). The reason I prefer this approach to Tim's of cranking up the clock rate is that you typically get much higher resolution values without bogging down the system with all the overhead of frequent timer interrupts. As with most things in VxWorks, caveats are needed in applying a generic facility to specific boards. First, not all boards have timestamp driver support. Second, on those that do, the timestamp sometimes has a pretty small period (it rolls over frequently). For instance, on the PPC platform, the "standard" timestamp driver derives the timestamps from the decrementer timer which is also used for the system clock. This means the time value rolls over at the rate the system clock rate; making it pretty useless for measuring things with a long duration. There is an alternative timer for some PPC boards that uses the Z8536 chip. The other day, we needed to time startup code on our Radstone PPC board and I found that with a little work the PPC timestamp functions fit the bill nicely. The ppcZ8536Timer.c can be rewritten such that it only provides auxilliary clock and timestamp support (with no system clock support - that comes from ppcDecTimer.c). By using the 16 bit timers 1 and 2 in "linked" mode it was possible to get a nice 32 bit timer that rolled over every 20 minutes in our case. We left timer 3 on the chip for auxilliary clock support. The original timestamp functionality in the driver was, unfortunately, fairly well broken but not unsalvageable. Just trying to point out that in some (all?) cases you have to be careful when using VxWorks drivers. Caveats aside, people should really consider the timestamp routines. They can often provide good timing insight with minimum trouble. Fred | Fred J Roeber, BBN Systems & Technologies | | 4 John Clarke Road Middletown, RI 02842-5202 | | froeber@bbn.com 401-848-3548 | | TraceMaker product manager | | http://www.tracemaker.bbn.com | From vxwexplo-errs@csg.lbl.gov Thu Feb 17 19:51:30 2000 From: Fred Roeber Date: Thu Feb 17 19:51:32 PST 2000 Subject: Re: ...and the last of the RAM went where? On Thu, 17 Feb 2000, Geir Frode Raanes wrote: > I have been struggeling to regain all unused DRAM on our > MBX860-VxWorks setup. > Now, the VxWorks kernel is loaded from the network directly > into RAM_LOW_ADRS = 0x00010000. The kernel is not compressed > so the RAM_HIGH_ADRS = 0x000EFFFF does not really apply. > Evenso, RAM_HIGH_ADRS do seem to enter the equation when > calculating the value of (end) in FREE_RAM_ADRS = (end). > > I have stripped the VxWorks kernel to the point where > Tornado2 reports a kernel size of TEXT = 454340, DATA = 10734 > and BSS = 47954 with a total size of 513028 Bytes. Any less > functionality than this is not acceptable. If you are loading VxWorks "from the network" then RAM_HIGH_ADDRESS is involved since that is where the "boot" version of VxWorks runs that takes care of downloading the real version. However, for your problem the whole notion of RAM_HIGH_ADDRESS isn't really relavent. Once the real VxWOrks starts up, the space below 0x10000 is used for some kernel data structures and probably for some uncached areas for device access (board dependent). From 0x10000 the next 513028 bytes contains the VxWorks load image and the rest of your memory (assuming that USER_RESERVED_MEM is defined as 0) goes into the system heap. During initialization and operation, things get allocated out of the heap working from the top (higher addresses) down. > Now, the kernel is half a megabyte in size (reports 641976 > Bytes downloaded,) starting at 0x10000 (64KByte.) Disregarding > stack space this means there is still some 3.5 MBytes left on > the machine. The following is what vxWorks reports: > > -> memShow > status bytes blocks avg block max block > ------ --------- -------- ---------- ---------- > current > free 2409360 168 14341 1724576 > alloc 1072400 270 3971 - > cumulative > alloc 1072496 272 3943 - > > As we can see, vxWorks has allocated some 1072400 Bytes > for itself, including stack space. This is far more stack > space than we need - how do I adjust the size of the stack? > Still there is half a megabyte missing between the top of > the kernel and bottom of free space. Also, the largest > allocateable block is mere 1724576 bytes, which is less > than total free memory. My guess is that free mem is > fragmented, whith the missing 500 KBytes inbetween. You don't go into the details of how your application is loaded and runs but from the results, I will make some assumptions. My guess is that you are loading an application object file after startup that is about 500K bytes. I am also figuring that you haven't done anything to enable NFS support so that the loading process will use FTP (i.e. the netDrv file system) to load the object file. This file system has the nasty side effect of reading the whole file into memory when it is opened/read. Thus, your half MB object file would get stuck in an allocated memory block when it was opened and then copied to a different 1/2 MB area where it would stay for execution. Finally, the object file would get closed freeing up the 1/2 MB chunk. If you don't enable NFS, loading any object file takes more than twice the size of the object code (because the file also contains things like symbols and relocation information). You should use NFS. I must admit that the results of memShow above don't fully support my theory. In particular, the cumulative allocation amount is only 96 bytes higher than the current allocation amount. With what I propose, that number should be much higher. Then again, I don't see how there could end up being 168 free blocks when only when only 2 blocks more than the current allocation were ever allocated. That is a bit of a mystery. > Is there an equivalent of "moduleShow" for the kernel > space itself? Below is what is loaded into free space. > Also, size is missing from the table. As far as checking out memory usage, there are some things I can suggest. You can check the free list using memShow(1). This will show the blocks that are part of the free list. As far as what is allocated, I would suggest using checkStack(). It can show what stacks are allocated to what tasks and how full the stacks are. The i() command will show where the stack addresses are currently while ti() on a task will show the limits of the stack. The Tornado 2 version of WindView has a facility for instrumenting the memory allocation functions that can show when and where memory is allocated. This could be very useful to you. > Next problem is that even though I've managed to free up > 1.7 MBytes of continous memory, the VxWorks v5.4 target > shell resident program loader refuses to load an application > with no more than 700 KBytes of text segment. > "Can not allocate space for text and data segments..." > or something to that effect. Objectdump reports that the > remaining segments totals some 200 KByte. This program > should effortlessly fit into free memory, but noooo. This is the FTP issue again. Configure in NFS and make sure that you are loading off the same disk as the VxWOrks image came off of or make sure to nfsMount the disk containing the files you load. > I can not find any other macro settings in the MBX860 > BSP to control the use of memory. I have not yet tried > to set the FREE_RAM_ADRS explicit and I suspect that > any attempt to do so will crash horrendously. You shouldn't change FREE_RAM_ADRS. As it is defined, all memory from the end of the VxWorks image (in low memory) up to the end of memory is all allocated to the system heap. Making the value smaller would allow the kernel BSS data to get overwritten (bad). Moving it up would just hide memory (the opposite of what you want). You aren't going to find more memory to add to the heap. The fragmentation is a result of the specific memory allocation method that WRS uses (as has been discussed at great length in this forum). > So what does VxWorks use memory for and where? > There is no architectual overview of VxWorks in the docs. > How can I get a list of allocated RAM som that I at > least can do a manual "memAddToPool" on any unclaimed > memory I migth find? See suggestions above. Also, what gets dynamically allocated is highly architecture and application dependent. You don't say much about either. Some things that can be big on some architectures are things like MMU page tables. If you are setting up large address ranges as being accessible in sysPhysMemDesc, that can result in the need for much bigger page tables. Well, that's some suggestions. I hope it helps. Fred | Fred J Roeber, BBN Systems & Technologies | | 4 John Clarke Road Middletown, RI 02842-5202 | | froeber@bbn.com 401-848-3548 | | TraceMaker product manager | | http://www.tracemaker.bbn.com | From vxwexplo-errs@csg.lbl.gov Thu Feb 17 22:30:49 2000 From: "Basavaraj H." Date: Thu Feb 17 22:30:52 PST 2000 Subject: [Fwd: VxWorks Context Switch..] This is a multi-part message in MIME format. --------------1524CC77F742CB23BA2377BD Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit --------------1524CC77F742CB23BA2377BD Content-Type: message/rfc822 Content-Transfer-Encoding: 7bit Content-Disposition: inline X-Mozilla-Status2: 00000000 Message-ID: <3897F796.726239E9@kiwi.india.tek.com> Date: Wed, 02 Feb 2000 14:53:35 +0530 From: "Basavaraj H." Organization: Tektronix X-Mailer: Mozilla 4.6 [en] (X11; I; SunOS 5.5.1 sun4m) X-Accept-Language: en MIME-Version: 1.0 To: vxworks_users@csg.lbl.gov Subject: VxWorks Context Switch.. References: <200002020752.XAA03466@csg.lbl.gov> Content-Type: multipart/mixed; boundary="------------F71738E54F9D0CB64E5190BA" This is a multi-part message in MIME format. --------------F71738E54F9D0CB64E5190BA Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Hi All, I wanted to know how the Real Time OS Context Switch time is less than any other OS, What are the Parameters it doesn't store while Switching. Thanks in Advance, Bh --------------F71738E54F9D0CB64E5190BA Content-Type: text/x-vcard; charset=us-ascii; name="hbasava.vcf" Content-Transfer-Encoding: 7bit Content-Description: Card for Basavaraj H. Content-Disposition: attachment; filename="hbasava.vcf" begin:vcard n:Hiremath;Basavaraj x-mozilla-html:FALSE org:Tektronix Engineering Development ;GVG adr:;;;;;; version:2.1 email;internet:hbasava@kiwi.india.tek.com x-mozilla-cpt:;928 fn:Basavaraj Hiremath end:vcard --------------F71738E54F9D0CB64E5190BA-- --------------1524CC77F742CB23BA2377BD Content-Type: text/x-vcard; charset=us-ascii; name="Basavaraj.Hiremath.vcf" Content-Transfer-Encoding: 7bit Content-Description: Card for Basavaraj H. Content-Disposition: attachment; filename="Basavaraj.Hiremath.vcf" begin:vcard n:Hiremath;Basavaraj x-mozilla-html:FALSE org:Tektronix Engineering Development ;GVG adr:;;;;;; version:2.1 email;internet:hbasava@kiwi.india.tek.com x-mozilla-cpt:;928 fn:Basavaraj Hiremath end:vcard --------------1524CC77F742CB23BA2377BD-- From vxwexplo-errs@csg.lbl.gov Fri Feb 18 03:28:25 2000 From: "Ganesh.N" Date: Fri Feb 18 03:28:27 PST 2000 Hello All, I have an application that fills the SNMP PDU and enqueues to the VxWorks SNMP Agent. For some needs I want to put the source address in the SNMP pdu to be 127.0.0.1 the loop back address. First of all I want to know how VxWorks handles this packet. If it reaches the SNMP agent the Agent while forming the request will place the destination address as 127.0.0.1. Now I want to know from where I can intercept this packet and give it to my application. thanks in advance, Ganesh, From vxwexplo-errs@csg.lbl.gov Fri Feb 18 04:03:09 2000 From: Vxworks Exploder Date: Fri Feb 18 04:03:12 PST 2000 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Fri Feb 18 04:03:06 PST 2000 Subject: Re: Time Measurement Subject: Re: VxWorks vs. Linux Subject: Intel StrataFlash28F128AJ3 & TrueFFS Subject: To link Fortran 77 libraries to object module Subject: Re: T2 Shell Question Subject: VxWorks Shell via the serial interface Subject: using virtual uart to communicate to other board Subject: Re: T2 Shell Question Subject: Re: ...and the last of the RAM went where? Subject: Re: VxWorks Shell via the serial interface Subject: Re: VME interrups Subject: Re: VxWorks problem using math functions in vxWorks.st Subject: Re: Adding a 2nd END ethernet device Subject: Re: Booting from compressed image Subject: Standalone bootrom problem Subject: Re: ...and the last of the RAM went where? Subject: VME interrupt problem Subject: Re: Adding SSL to Wind Web Server Subject: Re: Generate Tornado2 workspace from makefile Subject: xircom credit card ethernet+modem driver Subject: Re: Using RealMode VGA Bios to switch modes? ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: Time Measurement Date: Thu, 17 Feb 2000 13:23:56 GMT From: martins2@my-deja.com Organization: Deja.com - Before you buy. Message-ID: <88gsp9$23k$1@nnrp1.deja.com> References: <38a95f8b$0$35289@helios.is.co.za> > I have tried clock_gettime(), but this is no good. I have checked the > clock resolution and it is at 60 ticks per second. You can increase the clock resolution by calling sysClkRateSet(). The values you can set are board/processor specific. Ofcourse performance drops the higher you set the clock rate as more context swichting/checking will occur. If this is no solution and you have a x86 compatible board you might be able to use the second timer hardware. If so, write me an eMail, I can supply you with sample code for this as I use the second timer to avoid busy waiting. Hope this helps Martin Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: VxWorks vs. Linux Date: Wed, 9 Feb 2000 18:53:02 -0800 From: "DuckE" Message-ID: <38a22765.0@news.pacifier.com> References: <87t4tl$55u$1@news2.kornet.net> Reply-To: "DuckE" A closer competitor would be RTEMS, which is rapidly developing a set of features to compete directly with VxWorks. SteveD ChaeHo You wrote in message news:87t4tl$55u$1@news2.kornet.net... > Hello, VxWorks users... > > I'm concerned about Linux development. And I'm interested about the price of > VxWorks. > Our company sales the Dy4 systems' products. That company makes the VME > board for military/commercial version. And the main RTOS of the boards is > VxWorks. > That company determined a joining a Linux company,AMIRIX. If RTLinux would > be ported to Dy4 boards, there will be many interesting changes in RTOS > market. > I think the price of the VxWorks is too high. The price of Tornado II per 1 > user > is about 50000 us dollars, isn't it? > The problem is the capability of Linux. The developing speed of Linux is > amazing. Maybe in a near future, RTLinux will compete VxWorks. > > > Best Regards... > > > > > > > > --------------------------- Newsgroups: comp.os.vxworks Subject: Intel StrataFlash28F128AJ3 & TrueFFS Date: Thu, 17 Feb 2000 15:46:50 +0100 From: "Thomas Fox" Organization: Swisscom IP+ (post doesn't reflect views of Swisscom) Message-ID: <88h1kv$427$1@pollux.ip-plus.net> Hello, has someone experience implementing Intel StrataFlash28F128AJ3 MTD for TrueFFS. We will use it with VxWorks 5.4 on a PowerPC 603 Board. Thanks, Thomas --------------------------- Newsgroups: comp.os.vxworks Subject: To link Fortran 77 libraries to object module Date: Thu, 17 Feb 2000 23:04:34 +0800 From: "Dr. Zexiang Li's Robotic Project" Organization: Hong Kong University of Science and Technology Message-ID: <38AC0E02.2781E494@ee.ust.hk> Hello, I am using Tornado1.0.1 with Mc68040. Now I would like to implement an algorithm by C-code in my system. The C-code calls many functions in 'blas' and 'lapack' of Fortran77 libraies. We have not Fortran compiler for vxWorks. I have download the 'clapack' from "http://www.netlib.org" and compiled it (using cc68k). Can I link it to my object module? Another problem, the algorithm is simulated in a HP-Covex computer. I try to compile the algorithm, but some head files cannot be found. I try to specify the directory to 'usr/include' and '~target/h', but the complier cannot work. How can I specify multi-directory in a Makefile for vxWorks? Any valuable comments are welcome. Thank you in advance. Best regards, Jiang Shilong --------------------------- Newsgroups: comp.os.vxworks Subject: Re: T2 Shell Question Date: Thu, 17 Feb 2000 10:46:34 -0500 From: Dan Lucas Organization: Applied Research Lab Penn State Univ. Message-ID: <38AC17DA.9EC3CF96@psu.edu> References: <38AB2ADC.C55B7D0A@boeing.com> Ken, I haven't tried this, but why couldn't you just suspend the shell task from within your application and then resume it once your application has received its start command ? myApp() { int tid; tid = taskNameToId( "tShell" ); taskSuspend( tid ); getStartcommand(); taskResume( tid ); . . . } Ken Ostaszewski wrote: > > Hi There, > > I am having a problem with the T2 Target shell. > > My VxWorks config requires the target shell to be included from the > project facility. > > However, I only want the target shell to start once my application has > concluded its operation. This is due to my application requesting a > start command from the user from the console. However the shell is > already running and will not pass the command to the program, just to > the target shell, which doesn't know what to do with it. > > Any ideas would be appreciated. > > Ken Ostaszewski - -- Daniel K. Lucas email: dlucas@psu.edu Applied Research Lab phone: (814) 865-2907 Penn State Univ. fax: (814) 863-7270 P.O. Box 30 State College, Pa. 16804 --------------------------- Newsgroups: comp.os.vxworks Subject: VxWorks Shell via the serial interface Date: Thu, 17 Feb 2000 18:02:21 +0200 From: Rafi Hacham Organization: RADGUARD Message-ID: <38AC1B8D.8E272F8D@radguard.com> Hi All !!! I want to be able to use the VxWorks 5.4 shell via the serial interface rather then the Ethernet interface (Tornado 2 environment) which is my default. Does anybody knows how to customize the BSP in order to use the shell via the serial interface? - -- Best Regards !!! - ---------------------------------------------------------------------------------------- Rafi Hacham Phone: +972-3-7658918 RADGUARD Ltd. Fax: +972-3-6480859 Atidim Tower mailto:rafi.hacham@radguard.com Tel-Aviv 61581 Israel http://www.radguard.com/ --------------------------- Newsgroups: comp.os.vxworks Subject: using virtual uart to communicate to other board Date: Thu, 17 Feb 2000 11:04:32 -0500 From: Jeff Daly Organization: Sanders A Lockheed Martin Company Message-ID: <38AC1C10.12B9DDEA@lmco.com> Reply-To: jeffrey.j.daly@lmco.com i have an interesting problem where i need to communicate with a cpu that doesn't have a serial connection and may not have shared memory running. i'm implementing a virtual uart through some spare registers in A16 VME space and using the mailbox interrupt capability of our board. on the one side we have an rlogind-like task that installs the virtual uart driver, creates the device and opens it. it then calls read() on the descriptor, the read function of the uart blocks until it is connected to another processor (i'm using tyLib() for the upper layer, like the normal uarts) on another processor, an rlogin-like function installs the virtual driver, creates device and opens file, an ioctl makes the connection to the waiting processor. when the waiting processor returns from the blocked read (indicating a connection has completed) it connects to the shell task by changing the shell file descriptors (shellOrigStdSet()) to the virtual uart fd and restarting the shell. the shell now functions like it has an actual serial port connected to it. a shellLogout function is installed that resets the original shell fds, closes the virtual uart (the driver just marks it as disconnected) and restarts the shell. on the other side, the rlogin-like command reads chars from stdin and sends them out to the virtual uart, another task is spawned to read from the virtual uart and spit to stdout. so, when connected the one cpu acts as the serial port to the shell on the other machine. when finished, typing 'logout' is interpreted by the 'slave shell' and does the clean disconnect. but the 'master' cpu can't tell that a close() has happened on the other side. (tyRead() blocks forever waiting on chars) i can have the 'rlogin' cmd interpret ~. as the disconnect indicator which would kill the task that reads from virtual uart and sends to stdout and close the virtual uart fds. but that seems kinda kludgey and i was hoping someone had a cleaner, more rlogin-y exit. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: T2 Shell Question Date: Thu, 17 Feb 2000 11:32:30 -0500 From: Dan Lucas Organization: Applied Research Lab Penn State Univ. Message-ID: <38AC229E.3C7C2EC0@psu.edu> References: <38AB2ADC.C55B7D0A@boeing.com> <38AC17DA.9EC3CF96@psu.edu> Sorry, I guess I should have tried it before I suggested it. suspending the shell task won't work, but changing the priority will, I tried it. taskSpawn("tMyApp",90,0,2048,myApp) ^ Note: tMyApp running with a priority of 90. myApp() { int tid; tid = taskNameToId( "tShell" ); taskPrioritySet( tid,100 ); getStartcommand(); taskPrioritySet( tid,1 ); . . . } hope this helps, - ---------------------------------------------------- Dan Lucas wrote: > > Ken, > > I haven't tried this, but why couldn't you just suspend the shell > task from within your application and then resume it once your > application has received its start command ? > > myApp() > { > int tid; > > tid = taskNameToId( "tShell" ); > taskSuspend( tid ); > > getStartcommand(); > > taskResume( tid ); > . > . > . > > } > - -- Daniel K. Lucas email: dlucas@psu.edu Applied Research Lab phone: (814) 865-2907 Penn State Univ. fax: (814) 863-7270 P.O. Box 30 State College, Pa. 16804 --------------------------- Newsgroups: comp.os.vxworks Subject: Re: ...and the last of the RAM went where? Date: Thu, 17 Feb 2000 18:03:36 +0100 From: "Richard Marsh" Organization: Siemens Inc. Message-ID: <88h9k3$pnd$1@news.mch.sbs.de> References: <88e72t$jpc$1@snipp.uninett.no> Reply-To: "Richard Marsh" are you using the MMU? If so how much address space are you mapping? Richard Marsh --------------------------- Newsgroups: comp.os.vxworks Subject: Re: VxWorks Shell via the serial interface Date: 17 Feb 2000 17:38:48 GMT From: johan.borkhuis@peektraffic.nl Organization: XS4ALL Internet BV Message-ID: <88hbn8$su2$1@news1.xs4all.nl> References: <38AC1B8D.8E272F8D@radguard.com> Rafi Hacham wrote: : Hi All !!! : I want to be able to use the VxWorks 5.4 shell via the serial interface : rather then : the Ethernet interface (Tornado 2 environment) which is my default. : Does anybody knows how to customize the BSP in order to use the shell : via the : serial interface? The Target shell is default on one of the serial lines. Check you configuration to see if it is enabled and on wich serial line it is. The host shell however (that is the one within Tornado) is using the network connection. Groeten, Johan - -- o o o o o o o . . . ___________________________________ o _____ || Johan Borkhuis | .][__n_n_|DD[ ====_____ | johan@borksoft.xs4all.nl | >(________|__|_[_________]_|________________________________| _/oo OOOOO oo` ooo ooo 'o!o!o o!o!o` === VxWorks page: http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html === --------------------------- Newsgroups: comp.os.vxworks Subject: Re: VME interrups Date: Thu, 17 Feb 2000 14:18:51 -0600 From: Andrew Johnson Organization: APS, Argonne National Laboratory Message-ID: <38AC57AB.F68AF1F7@aps.anl.gov> References: <38AA61A9.16FE731F@lin.foa.se> Tomas Larsson wrote: > > We are using MVME2700 + a > VIPC618 carrier board for IP > modules > in a VME cage > One of the IP's DUAL/PIT is > sat up to generate a > PIRQ interrupt when pin H2 > becomes asserted. When that > happens a IRQ6 is generated on > the VMEbus > intConnect(INUM_TO_IVEC(0x15),PitISR,0); You also need to call sysIntEnable(6) to tell the Universe chip to pass VME level 6 IRQ's through to the CPU. Early versions of the mv2700 BSP may not work with odd vector numbers, but later versions have fixed this. - - Andrew - -- Complexity comes for free, Simplicity you have to work for. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: VxWorks problem using math functions in vxWorks.st Date: Thu, 17 Feb 2000 14:36:04 -0600 From: Andrew Johnson Organization: APS, Argonne National Laboratory Message-ID: <38AC5BB4.8F55E06@aps.anl.gov> References: <4.1.20000217093739.00984490@mailsd1> Phil Moy wrote: > > I am trying to place my application into rom on a MVME-167 and I am having > problems with the cos math function: > > Unimplemented Opcode 1111 > 47aa0 _mathHardCos +10 : _in_arpinput (3f66571b, b75d4094) > > This does work in the non standalone VxWorks. Any help will be greatly > appreciated. Were your vxWorks and vxWorks.st files built with the same configAll.h and usrConfig.c sources? It looks to me like the vxWorks.st may not have run mathHardInit() - read the Floating Point Support section in Appendix A of the vxWorks Programmer's Guide for some stuff that might help. - - Andrew - -- Complexity comes for free, Simplicity you have to work for. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Adding a 2nd END ethernet device Date: Fri, 18 Feb 2000 02:11:20 GMT From: Curt McDowell Organization: Broadcom Corp. Message-ID: <38ACAA47.F82A2D38@broadcom.com> References: <38AB403D.E9D745AA@res.raytheon.com> Reply-To: csm@broadcom.com Tim Conley wrote: > > I just tried using a very different IP address for the 2nd card, > i.e. 100.100.100.100 and the call to ifAddrSet worked!? Can anyone > provide a clue as to what I'm seeing? We ran into this problem too. Apparently END does not like to have two devices on the same subnet. Please let us know if you find a way to get around this limitation. Curt McDowell csm@broadcom.com --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Booting from compressed image Date: Thu, 17 Feb 2000 18:20:33 -0800 From: the wiz Organization: ACC Message-ID: <38ACAC71.7B6B6447@DEacc.SPAMcomTHIS> References: <88bu3q$2095@overload.lbl.gov> Is the source code available somewhere for the deflate program? Randy White wrote: > > We had a similar problem with our MPC8240 system. The boot loader resides in a > ~200K section of FLASH. The rest of the 1 Meg flash holds a compressed > vxworks.st image. The compression is done manually (or by modifying the make) to > run deflate. The bootloader creates a memDrv device then feeds this to > inflate(). The output of inflate goes right to the same RAM pointer that loading > an uncompressed image does. > The inflate routine seems to be rather iffy in it's operation, so we are > exploring alternatives. We are able to comfortably fit a 1.3 Meg kernel image > and the bootloader (bootrom.hex) into 1 Meg. > > Randy White > --------------5A86FD9851CE404407BFBB32 > Content-Type: text/x-vcard; charset=us-ascii; > name="randy.vcf" > Content-Transfer-Encoding: 7bit > Content-Description: Card for Randy White > Content-Disposition: attachment; > filename="randy.vcf" > > begin:vcard > n:White;Randy > tel;fax:(403)492-0654 > tel;work:(403)492-0321 x277 > x-mozilla-html:FALSE > org:Alberta Microelectronic Corp;Systems Division > adr:;;#201, 8625 - 112 St.;Edmonton;AB;T6G 1K8;Canada > version:2.1 > email;internet:randy@amc.ab.ca > title:Senior Designer > x-mozilla-cpt:;-1 > fn:Randy White > end:vcard > > --------------5A86FD9851CE404407BFBB32-- - -- Ron Childs --------------------------- Newsgroups: comp.os.vxworks Subject: Standalone bootrom problem Date: Fri, 18 Feb 2000 14:45:19 +0800 From: Wilson Lo Organization: Motorola Semiconductor Products Sector Message-ID: <38ACEA7F.95A11CA3@email.sps.mot.com> Reply-To: Wilson.Lo@motorola.com Hi, I compile my application and generate the object file correctly. It is download through the debugger and executing normally. I make the "vxWorks.st_rom.hex" file and programmed in flash memory, but it can't execute. The file size is 1.4Mbyte and flash meory size is 2Mbyte and the RAM size is 8Mbyte. I use MPC860 processor. I did the same process with other application in a same board. It is OK. Can U help me ? regards Wilson Lo --------------------------- Newsgroups: comp.os.vxworks Subject: Re: ...and the last of the RAM went where? Date: Fri, 18 Feb 2000 08:43:35 +0100 From: "Richard Marsh" Organization: Siemens Inc. Message-ID: <88item$rpl$1@news.mch.sbs.de> References: <88e72t$jpc$1@snipp.uninett.no> <88h9k3$pnd$1@news.mch.sbs.de> <88he0b$aub$1@snipp.uninett.no> Reply-To: "Richard Marsh" vxWorks includes the basic MMU lib stuff. Check out the function usrMmuInit in usrConfig.c. This initialises the MMU with the table sysPhysMemDesc defined in sysLib.c. I'd check this table to see what *is* being mapped. Also, each region is broken down into 4Kbyte pages with a descriptor for each, so if your mapping 1Gbyte then thats 262144 page descriptors lurking somewhere. Richard Marsh "Geir Frode Raanes" wrote in message news:88he0b$aub$1@snipp.uninett.no... > Richard Marsh wrote: > : are you using the MMU? If so how much address space are you mapping? > > Unfortunately no. We have not bougth the "vmLib" option as we don't > really need it. But you do have a point - perhaps there is an indirect > method here in examining the dirtybits of the MMU or cache... > This I will have to look into. I really don't now how these beasts > work on the PowerPC, but perhaps now is the time to find out. > > Windriver support suggested adjust the PCI IO space assignements. > But those are mapped far out of the way so I can't really see > why that should help. I'll have to try that tomorrow. > > > -- > ****************************************************** > > Never ever underestimate the power of human stupidity. > -Robert Anson Heinlein > > GeirFRS@invalid.and.so.forth > > ****************************************************** --------------------------- Newsgroups: comp.os.vxworks Subject: VME interrupt problem Date: Fri, 18 Feb 2000 10:32:25 +0100 From: Zdenek Sebek Organization: CVUT FEL Message-ID: <38AD11A9.ECC8F5B8@fel.cvut.cz> hello, I have a PEP's VSBC32 CPU card and an other device on VME bus, which generates interrupt request level 5, vector 80. I have a test program running on VSBC32. This program runs under OS9 operating system and this interrupt works OK. So HW should be OK. Now I am not able to catch this interrupt under VxWorks. I use this sequence: sysIntEnable(5); intConnect(INUM_TO_IVEC(80), (VOIDFUNCPTR) irqHandler, 0) I tried to change vector to 31 to catch interrupt generated by button Abort on front panel of VSBC32, and it works. any help, please? I am newbie to VME. Zdenek Sebek --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Adding SSL to Wind Web Server Date: Fri, 18 Feb 2000 07:55:59 GMT From: Computer Engineer Organization: Somewhere in the Earth Message-ID: <38AC8A58.69C51CD0@netscape.net> References: <20000213111136.22425.00003204@ng-fh1.aol.com> Hello Keith, For WWS 1.0, you may download a SSL patch from WindSurf site. CE KBeckley wrote: > We have the Wind Web Server for the 68k target and now need to add SSL to the > kernel. Has anyone done this before and can you explain it in detail. > > Thank you, > Keith Beckley --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Generate Tornado2 workspace from makefile Date: Fri, 18 Feb 2000 08:02:31 GMT From: Computer Engineer Organization: Somewhere in the Earth Message-ID: <38AC8BE0.47F55B97@netscape.net> References: <88b7t0$sl5$1@news1.tele.dk> If your custom BSP is Tornado 2.0 compliant, I think you can use the project facility to build a VxWorks image. "Jørgen Jakobsen" wrote: > Hi > > I'm working on a new BSP, and wonder whether it is possible to generate a > Tornado Workspace from the makefile in the BSP ? > This would make it much more easy to compile a single file in the BSP. > > If not possible, what is the easiest way to compile/assemble a single source > file in my BSP ? > > Jørgen Jakobsen, > Cotas Computer Technology A/S --------------------------- Newsgroups: comp.os.vxworks Subject: xircom credit card ethernet+modem driver Date: Fri, 18 Feb 2000 16:01:06 +0530 From: Yusuf Motiwala Organization: Lucent Technologies, Columbus, Ohio Message-ID: <38AD1F69.2C006B61@lucent.com> Hi, Is drivers for xircom credit card ethernet+modem 56K (PCMCIA) available for VxWorks? I will appreciate any help. regards, Yusuf --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Using RealMode VGA Bios to switch modes? Date: Fri, 18 Feb 2000 06:50:36 -0800 From: "Doug Reese" Organization: Posted via Supernews, http://www.supernews.com Message-ID: References: I had the same problem, and the answer was no. I worked with our single board computer vendor who spoke with the wrs folks. Apperently, they're working to get some driver's out the most popular video controllers. My solution was the same as yours - get the datasheet and program the registers as needed for minimal required BIOS support. Good luck! Doug doug.r@indacomp.com --------------------------- End of New-News digest ********************** From vxwexplo-errs@csg.lbl.gov Fri Feb 18 09:35:05 2000 From: David Laight Date: Fri Feb 18 09:35:07 PST 2000 Subject: Re: T2 Shell Question Ken et al, The suggested: taskSusPend( tid = tasknameToId( "tShell" ) ); ... taskResume( tid ) does not work because the vxWorks shell is 'inside' the locks on the console device (where it is blocked in read). The other process will just sleep waiting for these locks. It is only possible to stop the shell by forcing "taskSuspend\n" into the keyboard input buffer - eg using a hacked serial device driver. Your code then must wait for the shell to be suspended and ensure the tty options are the desired ones. It is also necessary to check that the shells stdin is the device you need. I seem to rcall that the original problem was about system startup, this can be sorted by changing usrConfig.c - or calling shellInit() when the console is free. David ---------------------------------------------------------------- David Laight email: dsl@tadpole.co.uk Tadpole Technology plc phone: +44 1223 428 232 Cambridge, UK fax: +44 1223 428 201 From vxwexplo-errs@csg.lbl.gov Fri Feb 18 09:46:49 2000 From: David Laight Date: Fri Feb 18 09:46:51 PST 2000 Subject: Re: Adding a 2nd END ethernet device. Tim, Curt, The 'inability' to have ethernet devices on the same subnet isn't really a vxworks bug. Rather any system that LETS you have two interfaces in the same subnet is failing to police a restriction implicit in the way IP routes packets. With both cards on the same LAN you soon find that all packets are sent out of the one interface - including ARP responses for the other interface! If the cards are on different LANs then nothing will work. If you have IP forwarding enabled and are using routed you can find that every packet on the LAN is sent via the machine with two interfaces. We had a poor little 486 (probably 33MHz) busy forwarding packets back onto the same network via an 8-bit ISA ethernet card - the senders and receivers were PCI/Sbus cards. On the other hand vxWorks does seem to have a bug in the code that determines whether a route is valid. I think it is looking at stale data. Unfortunately the code is far too complicated to grog without the source. David ---------------------------------------------------------------- David Laight email: dsl@tadpole.co.uk Tadpole Technology plc phone: +44 1223 428 232 Cambridge, UK fax: +44 1223 428 201 From vxwexplo-errs@csg.lbl.gov Fri Feb 18 09:56:24 2000 From: David Laight Date: Fri Feb 18 09:56:27 PST 2000 Subject: re: xircom credit card ethernet+modem driver Yusuf, I don't believe there is a vxWorks driver freely available. We do have one (written by myself) but it isn't distributable, and uses our own pcmcia support. The card should work as a modem with any standard modem driver - its CIS is that of a modem (containing a network MAC address), not a multi-function card. The specification of the card is now on the xircom web site - with more or less enough info to write a driver. We used an earlier (internal) version of the same document obtained under an NDA. The driver is relatively straight forward - but I would say that as I can write network card drivers in my sleep. There is one issue with the card - they do consume rather a lot of power. None of the 'power save' functions seem to make much difference either. David ---------------------------------------------------------------- David Laight email: dsl@tadpole.co.uk Tadpole Technology plc phone: +44 1223 428 232 Cambridge, UK fax: +44 1223 428 201 From vxwexplo-errs@csg.lbl.gov Fri Feb 18 10:38:56 2000 From: "Reyhan ERGUN" Date: Fri Feb 18 10:38:58 PST 2000 Subject: trouble w/ T2 : cannot see anything when vxWorks tab is selected for all my bootable VxWorks image This is a multi-part message in MIME format. ------=_NextPart_000_0005_01BF7A4F.ED8A4490 Content-Type: text/plain; charset="iso-8859-9" Content-Transfer-Encoding: quoted-printable Hi, I wonder why, but I lost the view comming when I select the vxWorks tab for any = bootable VxWorks image in my T2 workspace. I created a new workspace, and a new bootable VxWorks image depending on = a BSP for which I defined a new BSP directory, i.e. I copied mbx860 directory to another location and renamed it.=20 Since I have 2 different workspace and different modifications to MBX860 = BSP, I did this way. Afterwards, I got this problem. I opened the other workspace, nothing is listed when I select vxWorks = tab for the builds included in this workspace. When I try to add a new project, I get a project creation error as = follows: ccppc.exe: Warning: `-x c' after last input file has no effect ccppc.exe: No input files Thanks for your comments Best regards, reyhan ------=_NextPart_000_0005_01BF7A4F.ED8A4490 Content-Type: text/html; charset="iso-8859-9" Content-Transfer-Encoding: quoted-printable

Hi,
I wonder why,
but I lost the view comming when I = select the=20 vxWorks tab for any bootable VxWorks image in my T2 = workspace.
I created a new workspace, and a new = bootable=20 VxWorks image depending on a BSP for which I defined a new BSP = directory,=20 i.e.
I copied mbx860 directory to another = location and=20 renamed it.
Since I have 2 different workspace and = different=20 modifications to MBX860 BSP, I did this way.
Afterwards, I got this = problem.
 
I opened the other workspace, nothing = is listed=20 when I select vxWorks tab for the builds included in this=20 workspace.
 
 
When I try to add a new project, I get = a project=20 creation error as follows:
 
ccppc.exe: Warning: `-x c' after last = input file=20 has no effect
ccppc.exe: No input files
 
Thanks for your comments
Best regards, = reyhan
------=_NextPart_000_0005_01BF7A4F.ED8A4490-- From vxwexplo-errs@csg.lbl.gov Fri Feb 18 11:48:12 2000 From: brian.r.walker@exgate.tek.com Date: Fri Feb 18 11:48:14 PST 2000 Subject: RE: Time Measurement > -----Original Message----- > From: vxwexplo@lbl.gov [mailto:vxwexplo@lbl.gov] > Sent: Thursday, February 17, 2000 6:52 PM > To: vxworks_users@csg.lbl.gov > Subject: Re: Time Measurement > > > Submitted-by vxwexplo-errs@csg.lbl.gov Thu Feb 17 18:33:09 2000 > Submitted-by: > > On Thu, 17 Feb 2000, Morne Hartman wrote: > > > I would like to know how to measure the system time at a specific > > moment, then at another moment, only a few milliseconds later, to > > determine elapsed time. > > > > I have tried clock_gettime(), but this is no good. I have > checked the > > clock resolution and it is at 60 ticks per second. > > As with most things in VxWorks, caveats are needed in applying a > generic facility to specific boards. First, not all boards have > timestamp driver support. Second, on those that do, the timestamp > sometimes has a pretty small period (it rolls over frequently). For > instance, on the PPC platform, the "standard" timestamp driver derives > the timestamps from the decrementer timer which is also used for the > system clock. This means the time value rolls over at the rate the > system clock rate; making it pretty useless for measuring things with > a long duration. It's useless only if you use it alone or you need it for longer intervals. On PowerPc, Windview uses VxWorks clock ticks for coarse timing resolution. Another option for PowerPC systems is the Timebase register. It's a 64 bit register that is driven by the same clock as the decrementer and you can sample the value at any time. You get the resolution of the timestamp driver and greater range than the system ticks. -- Brian Walker Software Design Services Brian.R.Walker@tek.com Central Engineering, Tektronix, Inc. From vxwexplo-errs@csg.lbl.gov Tue Feb 22 19:47:44 2000 From: Bharath Channakeshava Date: Tue Feb 22 19:47:46 PST 2000 Subject: UDP Broadcast/Multicast Hi, Consider the scenario below: Host "A" is sending(sendto) unicast(to "B"),multicast and broadcast packets on port N. Host "B" is listening(recvfrom and has joined the Multicast group) on the same port N. Is it possible for "B" to find out if the packets he receives on port N,has come to him due a multicast/broadcast/unicast i.e. I need to be able to say that packet "a" came to me because it a broadcast,packet "b" came to me since I had joined the multicast group,packet "c" came to me as it was unicast to me and so on. I am working on Tornado-II and x86 Target. Regards Bharath From vxwexplo-errs@csg.lbl.gov Wed Feb 23 04:03:13 2000 From: Vxworks Exploder Date: Wed Feb 23 04:03:16 PST 2000 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Wed Feb 23 04:03:10 PST 2000 Subject: Re: VME interrupt problem Subject: Re: Help: can't booting from scsi disk Subject: Information on VxWorks Subject: Re: Time Measurement Subject: Networking and loopback interface problmes Subject: Customizing tornado for tftp Subject: Cache Problems in PPC9603 Subject: Re: trigger Subject: trouble w/ T2 : cannot see anything when vxWorks tab is selected for all my bootable VxWorks image Subject: test Subject: Special Workshop on Intelligence at the Network Edge - Hotel Deadline Subject: RTOS evaluation report Subject: Re: RTOS evaluation report Subject: Re: trigger (Memory leak detection tools) Subject: trigger Subject: Re: VxWorks vs. Linux Subject: Re: how I can use inflate in my own function Subject: Re: Byte Alignment Subject: VME interrupts interrupting VME interrupts Subject: Re: Symbol table troubles Subject: Re: problem whith downloading simple C++ program to simulator Subject: NFSDaemon & MountDaemon dead Subject: Embedded Java OS by Esmertec ?? Subject: Is anyone use C++ with T2? Subject: Re: Spurious interrupt with PPC860/850 Subject: VxSim on Solaris, with SIO & END Subject: Re: Spurious interrupt with PPC860/850 Subject: Re: Information on VxWorks Subject: Re: etherInputHook with END driver Subject: Re: VxSim on Solaris, with SIO & END Subject: problem whith downloading simple C++ program to simulator Subject: Re: Is anyone use C++ with T2? Subject: Re: Question about romstart() Subject: Re: restart OS without reloading object Subject: Contractor required with VX works and C Subject: Re: Multi-homing Multiple IP address per physical interface Subject: Flash Booting ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: VME interrupt problem Date: Fri, 18 Feb 2000 13:20:47 -0000 From: "Stephen Walsh" Message-ID: <38ad446d$1@pull.gecm.com> References: <38AD11A9.ECC8F5B8@fel.cvut.cz> Zdenek Sebek wrote in message <38AD11A9.ECC8F5B8@fel.cvut.cz>... >hello, > >I have a PEP's VSBC32 CPU card and an other device on VME bus, which >generates interrupt request level 5, vector 80. I have a test program >running on VSBC32. This program runs under OS9 operating system and this >interrupt works OK. So HW should be OK. Now I am not able to catch this >interrupt under VxWorks. I use this sequence: > >sysIntEnable(5); >intConnect(INUM_TO_IVEC(80), (VOIDFUNCPTR) irqHandler, 0) > >I tried to change vector to 31 to catch interrupt generated by button >Abort on front panel of VSBC32, and it works. > >any help, please? I am newbie to VME. > >Zdenek Sebek > If you have access to a VME bus analyser or a logic analyser then this would helps as you could monitor the VME bus to see what parts of the interrupt process were being performed. The order of events should be (apologies if I get is slightly wrong) 1. Interrupter generates IRQ. 2. Handler becomes Bus Master and generates IACK. 3. 3bit code placed on address bus to indicate prioirty of interrupt. 4. DS, AS and LWORD are driven and the System Controller generates IACKIN-IACKOUT. 5. The interrupt acknowledge daisychains is way down the bus to the interrupt requester then the STATUS/ID is placed on the Data Bus (DTB) and terminates the cycle with DTACK. You could also modify the bus interrupt handler in your BSP to print out Status ID and Priority and enable all interrupts in an attempt to get more visiblity. Hope some of the above helps, Steve ******************************************* Stephen A Walsh Senior Software Engineer BAE Systems ******************************************* Opinions are usually mine and mine alone ******************************************* --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Help: can't booting from scsi disk Date: Fri, 18 Feb 2000 14:15:15 +0000 From: David Rigg Organization: GEC Marconi Avionics Ltd Message-ID: <38AD53F3.1F55@gecm.com> References: <8899ie$sq9$1@wanadoo.fr> <38A8B073.1500@gv.net> Reply-To: david.rigg@gecm.com > patrick.boussard wrote: > > >i've lost parameters to boot from scsi disk. I tried this: > > > >boot device: scsi;6.0 > > > >and it does'nt work. try scsi=6,0 cheers David Rigg \ _ / Mission Avionics Division, x__________\_(0)_/__________x BAE Systems, Edinburgh, Scotland @ @ (](_o_)[) @ @ Tel : (+44) (0)131 314 8223 --------------------------- Newsgroups: comp.os.vxworks Subject: Information on VxWorks Date: Fri, 18 Feb 2000 15:03:19 -0600 From: "Rao" Organization: Esat Net Message-ID: <38ad5e78.0@seat.s3two.ie> References: <38A3374A.87D0A8AF@drs.ca> Can anyone advise me how to get a basic information, utility and use on VxWorks...I am a novice to this environment and looking after the projects related to Drivers...Thanx a million...Rapolu --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Time Measurement Date: Thu, 17 Feb 2000 21:45:19 +0100 From: "Mario Semo" Organization: Nextra Telekom GmbH Message-ID: <88jpg1$sns$1@newsmaster01.nextra.at> References: <38a95f8b$0$35289@helios.is.co.za> <38AB42BE.2146593A@res.raytheon.com> its much easier. just open the properties in VxWorks build for the timer and recompile. hardware peripherials clock systemclock ::properties page 2: PARAMS - -- regards, Mario Semo. http://www.kirchnersoft.com Tim Conley wrote in message <38AB42BE.2146593A@res.raytheon.com>... >You can change the system clock rate in >$(WIND_BASE)/target/config/all/usrConfig.c. >Change the sysClkRateSet call to a larger number. >We are using a value of 5000, which gives us tick granularity of 200 --------------------------- Newsgroups: comp.os.vxworks Subject: Networking and loopback interface problmes Date: Fri, 18 Feb 2000 17:55:49 GMT From: reverend_ron@my-deja.com Organization: Deja.com - Before you buy. Message-ID: <88k133$9uf$1@nnrp1.deja.com> I am trying to get a standalone network running in VxWorks 5.4 to use sockets for intertask communication. I am having difficulty getting the loopback network interface "lo" up and running. I have defined INCLUDE_NETWORK, INCLUDE_NET_INIT, STANDALONE_NET, and INCLUDE_NET_SHOW. I get an error message saying that there was a problem setting inet address of "lo0" to 127.0.0.1, and the error code is EPROTONOSUPPORT. Has anyone set up a standalone network for VxWorks? Thanks Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Customizing tornado for tftp Date: 18 Feb 2000 18:34:53 GMT From: jnance@wrtph0qb.nortelnetworks.com (Jim Nance) Organization: Bell Northern Research Message-ID: <88k3cd$1be$1@bcrkh13.ca.nortel.com> Hello All, I am working on a VxWorks based system which is loaded by tftping a boot image to the system. Right now I just run tftp from the Unix command line. The Tornado target server knows the IP address of the target and the name of the core file. I would like to put a button on the launcher control panel which got this information from the target server and did the tftp itself. I can not figure out how to get this information from the target server. Has anyone done anything like this before? Thanks, Jim --------------------------- Newsgroups: comp.os.vxworks Subject: Cache Problems in PPC9603 Date: Fri, 18 Feb 2000 18:43:02 GMT From: (Jingha Bhatt) Organization: Posted via Supernews, http://www.supernews.com Message-ID: <38af9167.139105287@news.ifreedom.com> I have problems whenever I try to use the DBAT registers on a PPC 603EM (no FPU). The DBAT registers (3) are use to mark peripheral memory mapped regions to invalidated cache and gaurded-memory. The RAM is marked as cached and no-choerency (h/w bus snooping is implemented). The BSP is configured NOT to use CACHE, so the BAT and Page Table arrays are not defined. The problem is that the usrRoot task fails to activate and the processor is stuck in a tight loop somwheret in the kernel task scheduler. Any help would be appreciated? TIA JB --------------------------- Newsgroups: comp.os.vxworks Subject: Re: trigger Date: Fri, 18 Feb 2000 13:18:56 -0800 From: Harvey Taylor Organization: Organization? What organization?! Message-ID: <38ADB740.4569@despam.pangea.ca> References: <88juvb$877$1@nnrp1.deja.com> fatihozcan@my-deja.com wrote: >[...] > Can you suggest some debugging tools to find memory leakage? > design reviews. code reviews. lint. purify. modularity & unit testing. If you want a run time tool to catch leaks, talk to WRS. - -het - --- "Assembly of Japanese bicycle require great peace of mind." - Instruction manual mentioned by R. Pirsig Harvey Taylor mailto:het@despam.portal.ca http://www.pangea.ca/~het --------------------------- Newsgroups: comp.os.vxworks Subject: trouble w/ T2 : cannot see anything when vxWorks tab is selected for all my bootable VxWorks image Date: Fri, 18 Feb 2000 19:52:04 GMT From: findik@my-deja.com Organization: Deja.com - Before you buy. Message-ID: <88k7t2$eul$1@nnrp1.deja.com> Hi, I wonder why, but I lost the view comming when I select the vxWorks tab for any bootable VxWorks image in my T2 workspace. I created a new workspace, and a new bootable VxWorks image depending on a BSP for which I defined a new BSP directory, i.e. I copied mbx860 directory to another location and renamed it. Since I have 2 different workspace and different modifications to MBX860 BSP, I did this way. Afterwards, I got this problem. I opened the other workspace, nothing is listed when I select vxWorks tab for the builds included in this workspace. When I try to add a new project, I get a project creation error as follows: ccppc.exe: Warning: `-x c' after last input file has no effect ccppc.exe: No input files Thanks for your comments Best regards, reyhan Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: test Date: Fri, 18 Feb 2000 18:30:55 GMT From: (pavlov.kutta) Organization: Posted via Supernews, http://www.supernews.com Message-ID: <38ad8fce.138695547@news.ifreedom.com> --------------------------- Newsgroups: comp.arch.embedded,comp.realtime,comp.robotics,sci.engr.control,comp.home.automation,comp.os.vxworks,com.os.qnx,comp.dcom.cell-relay,comp.dcom.fax Subject: Special Workshop on Intelligence at the Network Edge - Hotel Deadline Date: Fri, 18 Feb 2000 19:58:54 GMT From: moun@usenix.org (Moun Chau) Organization: USENIX Association Keywords: Embedded Systems, QoS, SmartBox, Security, Networks, Networking, USENIX, workshops, seminar, panel, addressing, routing, infrastructure, applications, software, internet, middleware, caches, USENIX, conference, workshop, invited talks, presentation, 3Com, RSIP, servers, systems Message-ID: SPECIAL WORKSHOP ON INTELLIGENCE AT THE NETWORK EDGE Sponsored by USENIX, The Advanced Computing Systems Association with support from 3Com. March 20, 2000 Holiday Inn Fisherman's Wharf San Francisco, California, USA This event is intended to be a workshop, not a short conference or symposium. Session and panel chairs will engage the audience as active participants in the discussions. Those who are interested in identifying new opportunities in networking technology would benefit most from attending this workshop. Regardless of your level of expertise, you'll come away from this workshop with new information and practical solutions. TOPICS include: * SmartBox: An Add-on solution for Guaranteed QoS * Integrating Active Networking and Commercial Grade Routing * QoS provisioning at the network edge * RSIP: Address Sharing with End-to-End Security * Internet Addressing at the Edge * Proxy Caches Move to Middleware Pillars * Trends and Forecasts in Internet Publishing Infrastructure * Open Session on Intelligence at the Network Edge ====================================================================== Register online at: http://www.usenix.org/events/ine2000 Hotel Reservation special rates end February 25, 2000 ====================================================================== USENIX is the Advanced Computing Systems Association, an international not-for-profit society of scientists, engineers, and system administrators working on the cutting edge of systems and software. --------------------------- Newsgroups: comp.os.vxworks Subject: RTOS evaluation report Date: Fri, 18 Feb 2000 16:24:57 -0500 From: "Tanabian, Mohammad [WDLN2:2O22:EXCH]" Organization: Nortel Networks Message-ID: <38ADB8A9.D32A1BA3@americasm01.nt.com> Hello All, I was wondering if anybody out there knows of an evaluation report on RTOSes. I'm particulary interested in VxWorks, QNX and pSOS. Any help is greatly appreciated. - -- Mohammad Tanabian CDMA Wireless --------------------------- Newsgroups: comp.os.vxworks Subject: Re: RTOS evaluation report Date: Fri, 18 Feb 2000 18:33:29 -0800 From: Harvey Taylor Organization: Organization? What organization?! Message-ID: <38AE00F9.375B@despam.pangea.ca> References: <38ADB8A9.D32A1BA3@americasm01.nt.com> Tanabian, Mohammad [WDLN2:2O22:EXCH] wrote: > I was wondering if anybody out there knows of an evaluation > report on RTOSes. I'm particulary interested in VxWorks, QNX and pSOS. > There is a Nortel internal group called SmaRTOS you should check out. - -het [ex-NBWA] - --- "Assembly of Japanese bicycle require great peace of mind." - Instruction manual mentioned by R. Pirsig Harvey Taylor mailto:het@despam.portal.ca http://www.pangea.ca/~het --------------------------- Newsgroups: comp.os.vxworks Subject: Re: trigger (Memory leak detection tools) Date: Fri, 18 Feb 2000 18:01:56 -0800 From: Lori Fraleigh Organization: Real-Time Innovations Message-ID: <38ADF994.F1FA154E@rti.com> Real-Time Innovations makes a memory leak detection tool called MemScope. It, along with the rest of the ScopeTools product line, can be purchased directly from your WRS sales rep. Regards, Lori - -------- Original Message -------- Subject: trigger Date: Fri, 18 Feb 2000 17:19:45 GMT From: fatihozcan@my-deja.com Organization: Deja.com - Before you buy. Newsgroups: comp.os.vxworks In Tornado2 menu there is a triggering menu but not functional since the target can not support it? ( We use MBX MPC860 board). Can you suggest some debugging tools to find memory leakage? Sent via Deja.com http://www.deja.com/ Before you buy. - -- Lori Fraleigh Real-Time Innovations Phone: 408-734-4200 x121 Fax: 408-734-5009 http://www.rti.com --------------------------- Newsgroups: comp.os.vxworks Subject: Re: VxWorks vs. Linux Date: Sat, 19 Feb 2000 03:22:30 GMT From: lcb Organization: EarthLink Network, Inc. Message-ID: <38AE0C80.2A354436@earthlink.net> References: <87t4tl$55u$1@news2.kornet.net> $2995 gets you a seat of VxWorks. $50K buys you 10 seats of Tornado Power Professional. Good luck on every getting out a product with a free OS unless you have a group of engineers working 80 to 100 hours a week for months. ChaeHo You wrote: > Hello, VxWorks users... > > I'm concerned about Linux development. And I'm interested about the price of > VxWorks. > Our company sales the Dy4 systems' products. That company makes the VME > board for military/commercial version. And the main RTOS of the boards is > VxWorks. > That company determined a joining a Linux company,AMIRIX. If RTLinux would > be ported to Dy4 boards, there will be many interesting changes in RTOS > market. > I think the price of the VxWorks is too high. The price of Tornado II per 1 > user > is about 50000 us dollars, isn't it? > The problem is the capability of Linux. The developing speed of Linux is > amazing. Maybe in a near future, RTLinux will compete VxWorks. > > Best Regards... --------------------------- Newsgroups: comp.os.vxworks Subject: Re: how I can use inflate in my own function Date: Sat, 19 Feb 2000 13:06:32 +0800 From: "Zha Junhai" Organization: Alcatel Message-ID: <38af07b6.0@dnews.sbell.com.cn> References: <38a5b6c7.0@dnews.sbell.com.cn> <38A7ECFF.B0B4C5B7@delphi.com> <8895ta$c40$1@crusher.de.colt.net> Thanks for the answer. I can use the inflate in bootconfig.c as in bootinit.c now , but I want to use it in my own application. If I have to connect zlib's author to get the object or source file or anywhere else. Achim Zimmer wrote in message news:8895ta$c40$1@crusher.de.colt.net... > inflate is normaly only in the bootloader included > If you need it in your vxworks you have to define it and include it. > > in the bootloader code is written: > >SEE ALSO: > >inflate(), romInit(), and deflate > >AUTHOR > >The original compression software for zlib was written by Jean-loup Gailly > >and Mark Adler. See the manual pages of inflate and deflate for > >more information on their freely available compression software. > > > > Joe Durusau schrieb in im Newsbeitrag: > 38A7ECFF.B0B4C5B7@delphi.com... > > I don't see anything that indicates that inflate is a function > > included with > > vxworks. If you have written this function yourself, the trick is to do > > one > > of the following: > > > > 1) load inflate.o (or whatever module contains inflate) first, then load > > the > > module that includes calls to it. > > > > 2). use ldxxx (xxx varies depending on your target) to link inflate and > > its > > callers to each other, then load the whole thing into the target at once. > > > > Speaking only for myself, > > > > Joe Durusau > > > > > > Zha Junhai wrote: > > > > > Hi, all > > > > > > Could you tell me how I can use inflate in > > > my own function, whatever I defined extern inflate(), and download to > > > target, it always tell me > > > Undefined symbols: inflate, how can I solve it, > > > any answer will be appreciated, Thanks. > > > > > > Best Regards > > > > > > Zha Junhai > > > > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Byte Alignment Date: Sat, 19 Feb 2000 08:41:05 -0000 From: "Matthew Barratt" Message-ID: <950949666.18653.0.nnrp-12.9e987a34@news.demon.co.uk> References: <389ade41$0$35286@helios.is.co.za> Hello, I've had grief with this kind of thing too. Apart from byte alignment, I've found that some types are different sizes between different platforms (in particular, some of the time.h types on solaris/x86, solaris/sun and vxworks/PPC differ. Wasn't time_t, but can't remember exactly what it was). So I get the two programmes to print up what they think the byte offset from the beginning of a structure to a structure member: struct { int a; int b; time_t c; char achar } mystruct; printf("sizeof time_t = %i\n",sizeof(time_t)); printf("offset to achar = %i\n",(int)((char *)&mystruct.achar - (char *)&mystruct.a)); I do that for every member in the structure, and then compare results for both platforms. Tedious to do, but it worked for me. I had to typedef my own type on my vxworks/ppc programme to accomadate the size of a type sent by my programme on the sun because the sun version was longer than the vxworks equivalent. Matthew Barratt Morné Hartman wrote in message news:389ade41$0$35286@helios.is.co.za... > Hi all, > > Does anyone here know anything about byte alignment. > > I have an application that communicates between a VxWorks board and an > NT Box. I am sending a message from a structure to the remote system. > The message contains various data types including Char, Double, Float, > Int etc. In my application in NT I use byte alignment so that the > message is received in the correct format on the remote end. I believe > that I need to use byte alignment on the VxWorks side aswell, but this > did not work > > I initialise byte alignment as follows : > > #pragma options -a1 > struct MyStruct .... > #pragma options -a4 > > If I communicate between two NT boxes there is no problem. When I try > the same thing between an NT and VxWorks box the received message is > not identical to the sent message. > > Does anyone have a solution to this problem > > Morné Hartman > mhartman@grintek.com > > > --------------------------- Newsgroups: comp.os.vxworks Subject: VME interrupts interrupting VME interrupts Date: Thu, 17 Feb 2000 08:41:58 -0500 From: "Connolly, F. T. (Francis)" Organization: Ford Motor Company Message-ID: <38ABFAA6.F13EA759@mail.ford.com> - --------------C1C695F96F856CA3E32F7A26 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit This question has to do with a higher priority VME interrupt interrupting a lower priority one. Preempting here means that if a lower priority ISR (interrupt service routine) is running, the higher priority one will "interrupt" it, finish its ISR, and then the lower priority interrupt's ISR will continue. We are VxWorks "users", not "developers". Or, our application is a laboratory engine control and instrumentation system that is VME and VXI with all hand written C; we are not writing BSPs. However, we are not totally ignorant; we do know a bit about VxWorks and its inner workings. Also, the main motivation behind all this is that we have one interrupt for which we would like to keep the latency "constant", which is the case on a 68k board. We are trying to get this to work on a PPC board, on which the interrupt latency happens to not be "constant". We are using VxWorks 5.3.1 (and 5.2 and even 5.1), but not really using Tornado 1.?, since we use "5.2 mode" with the shell on the target and a startup script or whatever it's called; we do it the old fashioned way. The VME specification prioritizes the IRQ levels (and then the daisy chain prioritizes interrupters on the same level), and says that if two interrupts occur simultaneously, the higher priority one will be serviced first. I haven't found any spec. for what should happen if a higher priority interrupt is asserted while a lower priority interrupt's ISR is executing. The context of the question is the situation on a 68k board versus a PPC (604) board that uses a Universe II VME-PCI interface. On the 68K board a higher priority interrupt does preempt a lower priority one (as observed using some hardware signals), while on the particular PPC board, if the lower priority interrupt is running while the higher priority one is asserted, the higher priority one does not execute until the lower priority ISR has completed. We understand the concept of why this is so; the 68k interrupts _are_ prioritized, while on the PPC BSP the Universe II uses one PCI interrupt for all VME interrupts. Some people have said that one might be able to remap one or more of the VME interrupts to another PCI interrupt. We can see how this might happen (we have the Universe II user manual, among other info). Without enough detailed knowledge, it is hard to tell how difficult remapping an interrupt would be, and if one would even want to do it (i.e., modify the BSP for one particular situation). So, the question more or less is if anyone has any similar experience with, or comments on this situation. What I would really like to know is if anyone knows if we have any hope of readily doing this, or if it would get really ugly. I don't wish to spend a lot of (or, rather, more) time on this. The vendor is already (apparently) pondering the question as well. Thanks for any assistance or comments. - -- Francis Connolly Technical Specialist FMC-FAO-PD-R&VT-P&AE-CAPE-APPTE-CEFET-P/T ET 313-594-4081 office (1115 GTL) fconnoll@ford.com 313-317-7404 fax CDS ID: fconnoll - --------------C1C695F96F856CA3E32F7A26 Content-Type: text/html; charset=us-ascii Content-Transfer-Encoding: 7bit This question has to do with a higher priority VME interrupt interrupting a lower priority one. Preempting here means that if a lower priority ISR (interrupt service routine) is running, the higher priority one will "interrupt" it, finish its ISR, and then the lower priority interrupt's ISR will continue.

We are VxWorks "users", not "developers". Or, our application is a laboratory engine control and instrumentation system that is VME and VXI with all hand written C; we are not writing BSPs. However, we are not totally ignorant; we do know a bit about VxWorks and its inner workings. Also, the main motivation behind all this is that we have one interrupt for which we would like to keep the latency "constant", which is the case on a 68k board. We are trying to get this to work on a PPC board, on which the interrupt latency happens to not be "constant".

We are using VxWorks 5.3.1 (and 5.2 and even 5.1), but not really using Tornado 1.?, since we use "5.2 mode" with the shell on the target and a startup script or whatever it's called; we do it the old fashioned way.

The VME specification prioritizes the IRQ levels (and then the daisy chain prioritizes interrupters on the same level), and says that if two interrupts occur simultaneously, the higher priority one will be serviced first. I haven't found any spec. for what should happen if a higher priority interrupt is asserted while a lower priority interrupt's ISR is executing.

The context of the question is the situation on a 68k board versus a PPC (604) board that uses a Universe II VME-PCI interface. On the 68K board a higher priority interrupt does preempt a lower priority one (as observed using some hardware signals), while on the particular PPC board, if the lower priority interrupt is running while the higher priority one is asserted, the higher priority one does not execute until the lower priority ISR has completed.

We understand the concept of why this is so; the 68k interrupts _are_ prioritized, while on the PPC BSP the Universe II uses one PCI interrupt for all VME interrupts. Some people have said that one might be able to remap one or more of the VME interrupts to another PCI interrupt. We can see how this might happen (we have the Universe II user manual, among other info).

Without enough detailed knowledge, it is hard to tell how difficult remapping an interrupt would be, and if one would even want to do it (i.e., modify the BSP for one particular situation).

So, the question more or less is if anyone has any similar experience with, or comments on this situation. What I would really like to know is if anyone knows if we have any hope of readily doing this, or if it would get really ugly. I don't wish to spend a lot of (or, rather, more)  time on this. The vendor is already (apparently) pondering the question as well. Thanks for any assistance or comments. 

-- 
Francis Connolly            Technical Specialist
FMC-FAO-PD-R&VT-P&AE-CAPE-APPTE-CEFET-P/T ET   
313-594-4081 office (1115 GTL) fconnoll@ford.com
313-317-7404 fax                CDS ID: fconnoll
  - --------------C1C695F96F856CA3E32F7A26-- --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Symbol table troubles Date: Tue, 22 Feb 2000 08:31:24 +0100 From: "Mario Semo" Organization: Nextra Telekom GmbH Message-ID: <88v4e5$mjc$1@newsmaster01.nextra.at> References: <38b173a8.1064090260@news.inter.net.il> Hello, I had a lot of problems with T2 and symbol table synchronization till someone mentioned a bug in T2s registry with the following workaround. maybe it is of help for U2: In Tornado, go to Tools->Options. Then scroll over to the right tab to the Tornado Registry. Select remote registry instead of local registry and type in the name of your host (mariosemo). Then restart the target server and reload your modules. There is a bug in the WRS that requires this to get symSyncLib to work on Windows. Regards, Lori - -- regards, Mario Semo. http://www.kirchnersoft.com Ran Shalgi wrote in message <38b173a8.1064090260@news.inter.net.il>... >In my test setup (NT host, x86 target), I boot over the network using >a floppy disk bootrom, then startup a target server on my host and >load the main application module using "ld". All the symbols then >appear correctly in the output of "lkup" and in the debugger. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: problem whith downloading simple C++ program to simulator Date: Tue, 22 Feb 2000 08:33:22 +0100 From: "Mario Semo" Organization: Nextra Telekom GmbH Message-ID: <88v4ea$mjc$2@newsmaster01.nextra.at> References: <88qnst$8bb1@nntp.cig.mot.com> just my 1 cent guess: have youi included these functionalities to your Vxworks.exe (simulator) as you build it? - -- regards, Mario Semo. http://www.kirchnersoft.com Stas wrote in message <88qnst$8bb1@nntp.cig.mot.com>... >Dear Sirs, > >I encounter a problem when downloading a simple C++ program to simulator. >the prog creates two tasks that exchange messages, I use wrapper classes for >tasks and queues treatment. While downloading the program throws an error : > >----------------------------------------- >Unresolved symbols encountered while downloading >Errors while downloading >D:/Tornado/target/proj/QueuePrj/SIMNTgnu/QueuePrj.out: >__$_8VXWError >_myValue__7VXWMsgQ >_vxwThrow__FRC8VXWError >___8VXWErrori >----------------------------------------- --------------------------- Newsgroups: comp.os.vxworks Subject: NFSDaemon & MountDaemon dead Date: Tue, 22 Feb 2000 08:39:56 +0100 From: "Mario Semo" Organization: Nextra Telekom GmbH Message-ID: <88v4ed$mjc$3@newsmaster01.nextra.at> scenario: T2, VxWorks, NFS SERVER included in the target. we load our application we boot script (just ld < ... calls) we see an error msg (which we didnt see when NFS Server is NOT used on the target): get_myaddress::socket errno=0x37 and when we use tasktree (tt) later to see some infos e.g. tt Mountd we get: NFSDeamon (tNfsd) suspended and MountDaemon (tMountd) suspended target is X86 based. any ideas what can be the reason? - -- regards, Mario Semo. http://www.kirchnersoft.com --------------------------- Newsgroups: comp.os.vxworks Subject: Embedded Java OS by Esmertec ?? Date: Tue, 22 Feb 2000 19:51:23 -0500 From: "Eric Riseman" Organization: The Internet Access Company, Inc. Message-ID: <88v41k$3vc@news-central.tiac.net> How does the Esmertec Java RTOS differ from HP Chi or Sun's EJava ? How much sense does a hard, real time Java OS make ? Why would someone buy an RTOS based on Java instead of VxWorks or OSE ? What kind of company would be an ideal candidate for a JAVA RTOS ? Please resond to: Eric@RapidLogic.com Thanks in Advance !! Eric --------------------------- Newsgroups: comp.os.vxworks Subject: Is anyone use C++ with T2? Date: Sun, 20 Feb 2000 19:24:06 +0200 From: "Leyf Gen" Organization: Motorola CIG Message-ID: <88p7q5$oc51@nntp.cig.mot.com> We have evaluation of vxWorks tools and we just want to know if anyone have experience with using C++ for development on vxWorks? Does vxWorks has good support for C++ development, if any special problems occurs? For the beginning we can't even download an C++ file to target, an error occurs. Thanx for HELP! --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Spurious interrupt with PPC860/850 Date: Sat, 19 Feb 2000 08:31:58 GMT From: john_94501@my-deja.com Organization: Deja.com - Before you buy. Message-ID: <88lkdu$ch1$1@nnrp1.deja.com> References: <24fq4.751$gi5.864@client> <88e21o$v9k$1@ndsukns01.ndsuk.com> The MVME260x boards that I used a while back had an interrupt controller on them that was derived from PC parts... it had a "feature" that it generated spurious interrupts on int 7 (the parallel port in most cases). I think that the VxWorks BSPs for those boards even had a dummy handler that just incremented a counter so you could tell how bad the device was! Maybe you're seeing the same thing! TTFN, John... In article <88e21o$v9k$1@ndsukns01.ndsuk.com>, "Will Fookes" wrote: > Jonathan, > > We had a similar problem with the PPC860SAR when we enabled the D-cache - > turned out it was the CPM Error interrupt. > > We fixed it by changing a couple of lines in init.s in the bootrom... > > /* Initialise instruction support control register (ICTRL) */ > lis r5,HIADJ(0x00000007) /* was 0x00000006 */ > addi r5,r5,LO(0x00000007) /* was 0x00000006 */ > mtspr ICTRL,r5 > > I understand that this disables the placing of certain debug information on > the bus. > > HTH, > Will > > Jonathan Cheng wrote in message <24fq4.751$gi5.864@client>... > > > >I am using PPC860 / 850 BSP in our custom board and I receive the following > >error message "uninitialized interrupt" from time to time. > > > >I check the offending interrupt and it is Level 7 interrupt. ( Interrupt > #15 > >in vxWorks. ) As Level 7 interrupt is mask off, so I assume it is not a > >hardware problem. It seems to be a software problem because Level 7 > >interrupt is the default encoding in SIVEC register. I check the code in > >ppc860Intr.c and it seems OK to me. > > > >Is anyone experience any similar problem? > > > >Any idea? > > > >Thanks > > > >Jonathan Cheng > > > > > > Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: VxSim on Solaris, with SIO & END Date: Wed, 23 Feb 2000 04:00:33 GMT From: tttt9999@my-deja.com Organization: Deja.com - Before you buy. Message-ID: <88vm0v$34k$1@nnrp1.deja.com> I am running into some problems when trying to migrate from SIMNT (vxsim on NT) to SIMSOLARIS (vxsim on solaris). I was trying to include the END and the SIO driver, both gave me errors when compiling and linking respectively. It turned out that I was missing $WIND_BASE/target/config/solaris/{sysSerial.c,configNet.h} It worked fine on the SIMNT env, and it does have those 2 files in the $WIND_BASE/target/config/simpc. Has anyone successfully running SIMSOLARIS with both SIO and END enabled ? Am I just missing those files during installations ? Thx. T9 Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Spurious interrupt with PPC860/850 Date: Mon, 21 Feb 2000 02:50:08 GMT From: pkockritz@home.com (Pete Kockritz) Organization: @Home Network Message-ID: References: <24fq4.751$gi5.864@client> <88dp3g$831$1@news.mch.sbs.de> In article <88dp3g$831$1@news.mch.sbs.de>, "Richard Marsh" wrote: > > I am using PPC860 / 850 BSP in our custom board and I receive the > following > > error message "uninitialized interrupt" from time to time. > > > > I check the offending interrupt and it is Level 7 interrupt. ( Interrupt > #15 > > in vxWorks. ) As Level 7 interrupt is mask off, so I assume it is not a > > hardware problem. It seems to be a software problem because Level 7 > > interrupt is the default encoding in SIVEC register. I check the code in > > ppc860Intr.c and it seems OK to me. > > Yes. we have exactly the same problem with our 860T's. We spent ages trying > to find out what was happening, but in the end we couldn't find anything > wrong in the code. We could never trap the problem using an ICE. Eventually > we tracked the problem down (using good old logMsg) to a CPM error interrupt > that occurs when the mask is removed. There is no mask availbale for this > interrupt. The problem was made worse by the fact that some boards had the > problem (perhaps happening once every couple of days), and some didn't > (never). In the end we just put a dummy handler in that ignored the > interrupt. We haven't noticed any side effects (yet!) and the fix has in for > about 6 months. I had this problem on an 860 using a custom BSP derived from the FADS860 BSP. If you don't have the most up to date ppc860Intr.c there is a problem if you have WindView enabled. With WindView enabled the ISR did not properly handle nested interrupts and you would see the 'uninitilized interrupt' message. The newer versions of the ISR handle that properly. You might check to see if your file is up to date. Regards, Pete - -- +-----------------------------------------------------------------+ | Pete Kockritz mailto:pkockritz@home.com | | This space for rent. | +-----------------------------------------------------------------+ --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Information on VxWorks Date: Wed, 23 Feb 2000 15:38:41 +0800 From: Zhao Changyi Organization: School of Computing, NUS Message-ID: <38B38E19.BC3F68F7@comp.nus.edu.sg> References: <38A3374A.87D0A8AF@drs.ca> <38ad5e78.0@seat.s3two.ie> Sender: Zhao Changyi simply go to the windriver web site. Rao wrote: > Can anyone advise me how to get a basic information, utility and use on > VxWorks...I am a novice to this environment and looking after the projects > related to Drivers...Thanx a million...Rapolu --------------------------- Newsgroups: comp.os.vxworks Subject: Re: etherInputHook with END driver Date: Mon, 21 Feb 2000 12:03:19 +0530 From: Yusuf Motiwala Organization: Lucent Technologies, Columbus, Ohio Message-ID: <38B0DC2F.377CE975@lucent.com> References: <38AEB10F.8CCB1BA9@Acm.Org> <38b02a3d.979759599@news.inter.net.il> etherInputHook works with END/NPT drivers but with added argument. etherInputHookAdd(FUNCPTR hookRoutine, char *device, int Unit) Regards, Yusuf Ran Shalgi wrote: > On Sat, 19 Feb 2000 07:04:47 -0800, "Randy Miyazaki (Go Dodgers > Blue!)" wrote: > > >will etherInputHook work with the END driver ? > > > >doesn't seem to. i have to compile the non-END version to get the Hook calls to > >work. any suggestions ? > > > >...randy > > With most (all ?) of the WRS-supplied END drivers, the answer is no. > Their new END drivers simply never invoke the hook function. > > Suggestions ? Either use non-END drivers, or use END drivers and get > your packets off the MUX by installing a protocol in "snarfing" mode. > Consult the VxWorks Network Protocol Toolkit for details, or search > www.deja.com for earlier postings along the same lines. Installing a > snarfing protocol is not much more complicated than using an > EtherHook. > > Ran Shalgi > > -- > Visit Bizarre++: web page dedicated to some eccentric C++ aspects > http://home.beseen.com/technology/bizarre_cpp/ > > Replace 127.0.0.1 with expand.com when replying by e-mail --------------------------- Newsgroups: comp.os.vxworks Subject: Re: VxSim on Solaris, with SIO & END Date: Wed, 23 Feb 2000 00:27:48 -0800 From: David Holloway Organization: Wind River Systems Message-ID: <38B39A04.5B359925@csua.berkeley.edu> References: <88vm0v$34k$1@nnrp1.deja.com> You are going to have a hard time using a serial driver with a simulator that doesn't emulate a serial port. tttt9999@my-deja.com wrote: > > I am running into some problems when trying to migrate from > SIMNT (vxsim on NT) to SIMSOLARIS (vxsim on solaris). I was > trying to include the END and the SIO driver, both gave me > errors when compiling and linking respectively. > > It turned out that I was missing > $WIND_BASE/target/config/solaris/{sysSerial.c,configNet.h} > > It worked fine on the SIMNT env, and it does have those 2 files > in the $WIND_BASE/target/config/simpc. > > Has anyone successfully running SIMSOLARIS with both SIO > and END enabled ? Am I just missing those files during installations ? > > Thx. > > T9 > > Sent via Deja.com http://www.deja.com/ > Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: problem whith downloading simple C++ program to simulator Date: Mon, 21 Feb 2000 09:02:00 +0200 From: "Stas" Organization: Motorola CIG Message-ID: <88qnst$8bb1@nntp.cig.mot.com> Dear Sirs, I encounter a problem when downloading a simple C++ program to simulator. the prog creates two tasks that exchange messages, I use wrapper classes for tasks and queues treatment. While downloading the program throws an error : - ----------------------------------------- Unresolved symbols encountered while downloading Errors while downloading D:/Tornado/target/proj/QueuePrj/SIMNTgnu/QueuePrj.out: __$_8VXWError _myValue__7VXWMsgQ _vxwThrow__FRC8VXWError ___8VXWErrori - ----------------------------------------- if I remove construction of objects (see below) the download succeed - ----------------------------------------- queue1 = new VXWMsgQ( MAX_MSGS, MAX_MSG_LEN, MSG_Q_PRIORITY ); if ( errno != OK ) return(ERROR); task1 = new VXWTask( "tTask1", 200, 0, STACK_SIZE, (FUNCPTR) taskFunc1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 ); - ----------------------------------------- thank you and best regards, - -- Stas Zinchik Motorola Israel Wireless Access Systems Division Tel: 972-3-568-4305 Fax: 972-3-565-8205 Email: mailto:stasz@milcse.cig.mot.com (mailto:BZS011@email.mot.com) --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Is anyone use C++ with T2? Date: 23 Feb 2000 08:59:06 GMT From: hat@se-46.wpa.wtb.tue.nl (Albert Hofkamp) Organization: Eindhoven University of Technology, The Netherlands Message-ID: References: <88p7q5$oc51@nntp.cig.mot.com> <88rtr2$rfg$1@newsmaster01.nextra.at> Reply-To: a.t.hofkamp@tue.nl On Mon, 21 Feb 2000 08:52:10 +0100, Mario Semo wrote: >i recompiled my already existing C++ code and it works fine. there were some >small problems like: Same here, no major problems. > automatic template resolution (doesnt work at vxSim at all (-frepo)) and >fails to compile some sources an all environments (which are compileable >with any other c++ compiler i have here) The c++ compiler they deliver is egcs, which has much better template and STL support than the old gcc 2.7.2.x of T1. Unfortunately, egcs is not finished. Especially in the area of templates, a lot of development is happening, so expect some problems with templates. Last but not least, egcs is a bit more strict wrt the latest c++ standard, and it flags some violations against the standard which other compilers let pass. Albert - --- Look ma, windows without Windows !! --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Question about romstart() Date: 21 Feb 2000 10:13:32 GMT From: johan.borkhuis@peektraffic.nl Organization: XS4ALL Internet BV Message-ID: <88r34c$fci$2@news1.xs4all.nl> References: <38AF1DD5.477AE807@tampabay.rr.com> Pranav Koushik wrote: : As I understand the bootup sequence is as follows : : rominit.s -> romStart() ->usrinit(). : Now romstart() copies the code from the flash to the RAM. After this I : am writing some code which writes and reads the the RAM from the : following addrs : (RAM_LOW_ADRS + ROM_COPY_SIZE) to (SYS_MEM_TOP). : The image in the flash is compressed. ROM_COPY_SIZE is 0x7FF00 i guess : and there is 32MB RAM on board. I think the bootimage is decompressing itself into the upper part of the memory. The VxWorks image itself is loaded into the lower part of the memory. So when you write to SYS_MEM_TOP you are overwriting the bootimage itself. Groeten, Johan - -- o o o o o o o . . . ___________________________________ o _____ || Johan Borkhuis | .][__n_n_|DD[ ====_____ | johan@borksoft.xs4all.nl | >(________|__|_[_________]_|________________________________| _/oo OOOOO oo` ooo ooo 'o!o!o o!o!o` === VxWorks page: http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html === --------------------------- Newsgroups: comp.os.vxworks Subject: Re: restart OS without reloading object Date: 21 Feb 2000 10:16:01 GMT From: johan.borkhuis@peektraffic.nl Organization: XS4ALL Internet BV Message-ID: <88r391$fci$3@news1.xs4all.nl> References: <3u0s4.6098$Qx5.74092@news.rdc1.tx.home.com> Bo Yu wrote: : Hello, : Suppose the system with vxworks is running, then if I do not : want to reload the object, I just jump to sysInit ( usrInit, usrRoot : afterwards ). Could the whole system be reinitialized as the first : time ? When you include your own SW with the VxWorks bootable image, yes it is possible. When you want to dynamically load modules and keep them over a restart, no that is not possible, because of the linking process. After you load an object VxWorks has to dynamically link this object to resolve all the missing symbols. Groeten, Johan - -- o o o o o o o . . . ___________________________________ o _____ || Johan Borkhuis | .][__n_n_|DD[ ====_____ | johan@borksoft.xs4all.nl | >(________|__|_[_________]_|________________________________| _/oo OOOOO oo` ooo ooo 'o!o!o o!o!o` === VxWorks page: http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html === --------------------------- Newsgroups: comp.os.vxworks Subject: Contractor required with VX works and C Date: Wed, 23 Feb 2000 10:50:42 GMT From: tpuffett@my-deja.com Organization: Deja.com - Before you buy. Message-ID: <890e22$ivi$1@nnrp1.deja.com> I am looking for a contractor with embedded C and VXworks experience to work in the South West of the UK for 6 months. The work will involve systems software and hardware design in an R&D environment. If you are interested, please email me at tpuffett@my-deja.com Thanks Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Multi-homing Multiple IP address per physical interface Date: Wed, 23 Feb 2000 12:46:15 +0200 From: Ilia Organization: Telegate Message-ID: <38B3BA77.DE0B6657@telegate.co.il> References: Tim Michals wrote: > > Does VxWorks TCP/IP stack support multi-IP addresses per interface? If not > this there a way to do this? Definitely DOES. VxWorks 5.3 with SENS stack supports that. I believe, that VxWorks 5.4 (Tornado II) too. Pre - SENS versions does not. Set the 1-st IP address with ifAddrSet(), as usual, and than add the additional ones with the ifAddrAdd(). We're do that and it works with 2 IP addresses per interface. AFAIK, the printed manual doesn't mention the ifAddrAdd(), but it is described in the online manual and "vxman". Best regards. Ilia. - -- +================================= | Bosis Ilia | Senior Software engineer | Terayon Information Systems inc. | E-mail: iliab@terayon.com | WWW: http://www.terayon.com +================================== --------------------------- Newsgroups: comp.os.vxworks Subject: Flash Booting Date: Wed, 23 Feb 2000 11:41:20 GMT From: pjs1@hotmail.com Organization: @Home Network Canada Message-ID: <38b3c6ee.32442254@news> Hi: I am in the process of trying to load the Vxworks kernel from the M-System Flash 2000. Everytime it tries to load it indicates that the file is 22 + 0 + 0 size then crashes. As anyone seen this before? Thanks, Patrick Shaw Computing Devices Canada pat.shaw@cdott.com --------------------------- End of New-News digest ********************** From vxwexplo-errs@csg.lbl.gov Wed Feb 23 06:37:01 2000 From: John_W_Cosgrove@res.raytheon.com Date: Wed Feb 23 06:37:03 PST 2000 Subject: Floating Point Underflows Dear vxWorks'ers, We are porting a unix application to vxWorks. On unix this application generates a floating point underflow, and continues executing using these very small numbers... I want the same thing to occur in the port. On our target, this underflow stops processing. (target is PPC 750, using the CPU=PPC604 switch. MVME-2700 using 5.3.1 BSP=1.1/4 for MVME-2604 patched with Moto's '2700 patch). Worse (or Better???) is that this is a FORTRAN application compiled by the Green Hills FORTRAN compiler. I seem to remember that exceptions can be disabled.... but how??? Thansk in advance, John Cosgrove Raytheon Electronic Systems (401)842-4167 From vxwexplo-errs@csg.lbl.gov Wed Feb 23 11:32:06 2000 From: Tom Antles Date: Wed Feb 23 11:32:08 PST 2000 Subject: why -mlongcall doesn't fix PPC32meg problem? We have 2 problems related to -mlongcall. Anyone know any answers? 1) We recompiled our application and our vxworks image with the -mlongcall switch, and enabled our bsp to use the memory above 32meg, but we still get the error: LOADER_RELOCATION_OFFSET_TOO_LARGE whenever we excecute 'ld < app' in the windshell. We are using C++ and do have static constructors. Is it true that this -mlongcall does not work for static constructors? 2) we used -mlongcall to create a new bootrom and bsp, and compiled with RAM_HIGH_ADRS at 62Meg in Makefile and config.h, instead of 32Meg. The target would not boot up and load the vxworks image. So it appears that -mlongcall doesn't work for moving the bootrom above the 32Meg limit either! Anyone know why? Since it doesn't do anything for us, what DOES -mlongcall work for?! Tom Antles ---------------------- tom_antles@agilent.com Phone: (509) 921-3590 Fax: (509) 921-3991 From vxwexplo-errs@csg.lbl.gov Wed Feb 23 17:27:53 2000 From: Prakash Dhavali Date: Wed Feb 23 17:27:55 PST 2000 Subject: Why VxSim compilers/linkers SO SLOW on WindowsNT I use Tornado2.0 on WindowsNT4.0 VxSim compilers and linkers(ccsimpc,ldsimpc) are VERY VERY slow as compared to other target/MIPS compilers/linkers(ccmips,ldmips). It's understood that MIPs compilers/linkers have reduced instruction sets to parse and interpret, BUT I was not expecting VxSim compiler/linkers to be SO slow. Has anyone experienced the same performance problems with VxSim? How does debug flag (-g) affect the performance of compilers and linkers? Thanks in advance Prakash From vxwexplo-errs@csg.lbl.gov Thu Feb 24 01:04:30 2000 From: Navin Kumar Agrawal Date: Thu Feb 24 01:04:32 PST 2000 Subject: Re: Why VxSim compilers/linkers SO SLOW on WindowsNT Tornado Hi Folks, Any help for the follwoing will be highly appreciated. I am porting an RSVP code originally developed for Sun Solaris to VxWorks. The solaris version of code uses MIB-II tables of IP device for getting matchinter interface inddx number. In case of Solaris implementation, it is possible to get complete IP route table in one shot without specifying any destination address. But I couldn't find any such similar function in VxWorks. The function m2IpRouteTblEntryGet() gets just one entry. Also to me it looked like that the result of this function is not appropriate. My target system is having just one entry in its routing table ie. correspoding to its own IP address. routing table on my target system: dest-192.168.100.0 nexthop-192.168.100.140 etc... Now when I tried using the above mentioned function for resolving the address 192.168.100.45, it returns failure, though this address would be located in same subnet. Can anybody help me in this regard? Thanks for your time, Regards, Navin PS: I call m2IpRouteTblEntryGet() function with just destination address filled. --- the vxWorks Users Group Exploder wrote: > Submitted-by vxwexplo-errs@csg.lbl.gov Wed Feb 23 > 17:27:53 2000 > Submitted-by: Prakash Dhavali > > > > I use Tornado2.0 on WindowsNT4.0 > > VxSim compilers and linkers(ccsimpc,ldsimpc) are > VERY VERY slow as > compared to other target/MIPS > compilers/linkers(ccmips,ldmips). It's > understood that MIPs compilers/linkers have reduced > instruction sets to > parse and interpret, BUT I was not expecting VxSim > compiler/linkers to > be SO slow. > > Has anyone experienced the same performance problems > with VxSim? > How does debug flag (-g) affect the performance of > compilers and > linkers? > > Thanks in advance > Prakash > > ********** > > This is a user group mailing list for vxWorks > related topics > see http://csg.lbl.gov/vxworks for posting and > other info > > __________________________________________________ Do You Yahoo!? Talk to your friends online with Yahoo! Messenger. http://im.yahoo.com From vxwexplo-errs@csg.lbl.gov Thu Feb 24 04:03:15 2000 From: Vxworks Exploder Date: Thu Feb 24 04:03:18 PST 2000 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Thu Feb 24 04:03:11 PST 2000 Subject: Re: Typical Thread Switching Time for MPC8260 Subject: taskIdVerify causing exception! Subject: RTL 8139 driver ? Subject: Re: UDP Broadcast/Multicast Subject: Re: Real Time Clock support? Subject: Reading DOS partitions on IDE Subject: Problem with sysProcNumSet for mv1604 Subject: Setting the TLB with in VxWorks (MPC860 MMU/Cache) Subject: Re: gcc 2.7.2 and C++ Template Classes - NFG! Subject: Re: Floating Point Underflows Subject: How to execute Startup script Subject: switching IP addresses on vxWorks Subject: Re: Alignment in Structures Subject: Re: Setting the TLB with in VxWorks (MPC860 MMU/Cache) Subject: Re: VxWorks Shell via the serial interface Subject: Stdio without console problem+solution Subject: Typical Thread Switching Time Subject: Is munchless C++ linking possible Subject: Re: Intel StrataFlash28F128AJ3 & TrueFFS Subject: Re: Problem with sysProcNumSet for mv1604 Subject: Re: Is anyone use C++ with T2? Subject: Re: VxSim on Solaris, with SIO & END Subject: Embedded Systems SW Engineer. Subject: Re: Gathering CPU utilization by process. Subject: Disk Defrag Using DosFSII Subject: [Q]: losing target connection Subject: Re: Gathering CPU utilization by process. Subject: Re: VxWorks vs. Linux Subject: Number of /tyCo/xx devices Subject: Re: taskIdVerify causing exception! Subject: Re: [Q]: losing target connection Subject: Re: Driver list Subject: Re: Purify or equivalent under vxSim ? Subject: Re: restart OS without reloading object Subject: Does anybody know for VxWorks on ARM710A (preferably CL-PS7111 CPU) based products Subject: memory leak deleting a pipe Subject: Re: RTL 8139 driver ? Subject: Re: Number of /tyCo/xx devices Subject: Re: Gathering CPU utilization by process. Subject: Re: taskIdVerify causing exception! Subject: Typical Thread Switching Time for MPC8260 Subject: Re: taskIdVerify causing exception! Subject: Re: Number of /tyCo/xx devices Subject: semDelete error Subject: Driver list Subject: Re: taskIdVerify causing exception! Subject: Making vxWorks_rom ? ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: Typical Thread Switching Time for MPC8260 Date: 23 Feb 2000 12:05:01 GMT From: johan.borkhuis@peektraffic.nl Organization: XS4ALL Internet BV Message-ID: <890idd$nmt$1@news1.xs4all.nl> References: <88u6d8$7i8$1@bcrkh13.ca.nortel.com> Serge Martineau wrote: : I'm interested in knowing the typical thread switching time in # of : clock cycles : or in # of intructions for the MPC8260 PowerQUICC II. The version of : VxWorks : is 5.4. I got a benchmark test from VxWorks for the MPC860, maybe they also have one for the MPC8260. For the 860 the time for a "Raw context switch time" is 11 uS on a 40Mhz MBX860 board. Groeten, Johan - -- o o o o o o o . . . ___________________________________ o _____ || Johan Borkhuis | .][__n_n_|DD[ ====_____ | johan@borksoft.xs4all.nl | >(________|__|_[_________]_|________________________________| _/oo OOOOO oo` ooo ooo 'o!o!o o!o!o` === VxWorks page: http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html === --------------------------- Newsgroups: comp.os.vxworks Subject: taskIdVerify causing exception! Date: Wed, 23 Feb 2000 12:21:09 -0000 From: "Will Fookes" Organization: NDS UK Message-ID: <890hno$eqd$1@ndsukns01.ndsuk.com> Hi, I am using Vxworks 5.3.1/Tornado 1.0.1 with a PPC860. When my system starts, I attempt to print out any old log messages remaining in RAM from the previous run (we know that the RAM stays up for about 20 seconds, so this has been a useful aid to finding out what caused the thing to reset!!) Of course, we expect the RAM to be liable to corruption, so we are very cautious how we process these old log messages. They each consist of a text string (we ensure it has a null termination) and a few numbers that we interpret (eg. error severity - we range-check it before looking up the text interpretation in a table). One of the numbers is the taskId of the task that generated the log message. We call taskName and check the return value - if null, the task doesn't exist, so we print the hex value of the taskId rather than the name. However, I am now seeing some slight memory corruption which causes taskName to crash - it looks as if the taskId is used as a pointer and de-referenced BEFORE it has been validated, causing the task to crash if the taskId value is outside the memory map. OK, so perhaps I can use taskIdValid to validate the taskId before attempting to call taskName? NOT SO: this seems to do the same thing... - -> taskIdVerify 0x2274dd90 Exception number 768: Task: 0x40ae38 (t4) data access program counter: 0x1ca210 access address: 0x2274ddc0 machine status: 0x9032 1cff48 vxTaskEntry +5c : funcCallWrapper ([]) 1bc74c funcCallWrapper+a4 : taskIdVerify ([]) value = 0 = 0x0 Does anyone know an easy way to validate a taskId ? Will Fookes --------------------------- Newsgroups: comp.os.vxworks Subject: RTL 8139 driver ? Date: Mon, 21 Feb 2000 15:11:15 +0100 From: =?iso-8859-1?Q?=D8ystein?= Straume Organization: NTNU , student Message-ID: <38B14783.EDF396BF@itk.ntnu.no> Hi, I have a ICP Nova-600 EBX Pentium Embedded Board with a built-in RTL8139 network device. I wonder if someone has used this chipset under VxWorks, and if there is any available drivers? Maybe someone has ported some BSD network drivers already. If so could you describe it if it was easy ? Thanks, Øystein --------------------------- Newsgroups: comp.os.vxworks Subject: Re: UDP Broadcast/Multicast Date: Wed, 23 Feb 2000 13:23:03 -0000 From: "Dave Korn" Organization: (Posted via) U-NET Internet Ltd. Message-ID: References: <88vmu0$ldl1@overload.lbl.gov> Bharath Channakeshava wrote in message <88vmu0$ldl1@overload.lbl.gov>... >Is it possible for "B" to find out if the packets he receives on port N,has come >to him due a multicast/broadcast/unicast i.e. I need to be able to say that >packet "a" came to me because it a broadcast,packet "b" came to me since I had >joined the multicast group,packet "c" came to me as it was unicast to me and so >on. Have you looked up recvfrom in the manual ? It fills in a struct sockaddr with the sender's address for you. You could always look in there.... DaveK - -- I wouldn't recommend sex, drugs or insanity for everyone, but they've always worked for me. -- Hunter S. Thompson --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Real Time Clock support? Date: 23 Feb 2000 14:45:07 +0100 From: hartmutp@bidro145.bbn.hp.com (Hartmut Pohl) Message-ID: <38b3e463_1@isoit370.bbn.hp.com> References: <385A7EC8.FB44AF4F@t-online.de> <385AAAE8.D8EDF244@terisa.com> <83j3a3$do5$1@nnrp1.deja.com> In article <83j3a3$do5$1@nnrp1.deja.com>, John Fusco writes: > In article <385AAAE8.D8EDF244@terisa.com>, > jeremey@terisa.com wrote: >> Heribert Vollkhausen wrote: >> > >> > I'm using VxWorks on a Intel 80486 based PC. Does VxWorks contain > any >> > routines to get (and set) the time from the real time clock? >> > Does 'clock_gettime( )' read the real time clock? >> > >> >> I'm also on a 486... yes, clock_gettime() and clock_settime() apply >> to the realtime clock. In fact, the only allowed clock id is >> CLOCK_REALTIME. At least that's the case on the version of things >> I'm using. >> >> Hope that helps, >> Jeremey. > > I don't know about the PC archicture, but in my experience these > routines do not actually set the realtime clock chip. They do set the > time in the software clock, but this is only good until the next reset. That's right, it sets only the OS clock. To set the RTC, you have to write your own stuff. At least in vxWorks 5.3 Best Regards, Hartmut - -- Hartmut_Pohl@agilent.com - -- Niemanden kann man so leicht taeuschen wie sich selbst. --------------------------- Newsgroups: comp.os.vxworks Subject: Reading DOS partitions on IDE Date: Wed, 23 Feb 2000 15:43:12 +0200 From: "Vova Lifliand" Organization: CompuLab Message-ID: <890nsi$65h$1@news.netvision.net.il> Reply-To: "Vova Lifliand" I use IdeDrv and DosFs library for IDE disks. After I mount hard disk, VxWorks seems to assume, that the (only) hard disk partition begins at sector 0, while DOS has first partition table and the partition itself. As a result I see one file system in DOS and completely another in VxWorks (the same disk space seems to have two FATs). Is it possible to see DOS formatted hard disks (fdisk + format) in VxWorks? Can IdeDrv perform hard disk autodetect? - -- Vova Lifliand voval@compulab.co.il CompuLab Tel.: 04-8222140#0108, 054-763655 --------------------------- Newsgroups: comp.os.vxworks Subject: Problem with sysProcNumSet for mv1604 Date: Wed, 23 Feb 2000 15:09:27 +0100 From: Andreas Johansson Organization: Ericsson Microwave Systems AB Message-ID: <38B3EA17.57C213B7@emw.ericsson.se> Hello, I am using VxWorks 5.3 together with a mv1604 board, and BSP version 1.1/4. The function sysProcNumSet (sysLib.c) sets the VME address of GCSR in the vmechip2 - vmebus slave GCSR group address register. When that is done, the variable 'sysProcNum' is used, should not the variable 'procNum' be used instead? As it is now gets all processors the same base address of their GCSR-register. Part of sysProcNumSet in sysLib.c - --- if (sysProcNum < 15) { /* VME address of GCSR is function of processor number */ temp = *VMECHIP2_LBTVCR; temp &= ~(0xfff00000); /* clear bits */ *VMECHIP2_LBTVCR = (temp | (sysProcNum << 20) | (GCSR_GROUP_ADDR << 24)); } - --- Thanks in advance, Andreas Johansson Ericsson Microwave Systems AB --------------------------- Newsgroups: comp.os.vxworks Subject: Setting the TLB with in VxWorks (MPC860 MMU/Cache) Date: Wed, 23 Feb 2000 10:30:23 -0600 From: Sandy Rivera Organization: ADC Telecommunications Message-ID: <38B40B1F.35CA05AA@adc.com> The MPC860 doesn't have BAT registers to setup the blocks of memory to write-through, write-back, or Inhibit. It uses the TLB to accomplish the different caching schemes on the different memory blocks. How do you setup these tables within vxWorks? I found were you enable the cache (cacheEnable function), but how is the TLB populated? What are the steps needed to enabling cache? --------------------------- Newsgroups: comp.os.vxworks Subject: Re: gcc 2.7.2 and C++ Template Classes - NFG! Date: 23 Feb 2000 18:21:25 +0100 From: hartmutp@bidro145.bbn.hp.com (Hartmut Pohl) Message-ID: <38b41715_1@isoit370.bbn.hp.com> References: <8676lu$a53$1@sulawesi.lerc.nasa.gov> [Posted and mailed] Dave, we've had the same problem with our old g++ compiler. It seems like you have the same version of '96 that comes with Tornado 1.0 We claimed the problem at Wind River and got a patch. With the updgrade to Tornado 1.01 we got a newer g++ (gcc version 2.7.2.3) also without the template problem. If you can afford it, switch to Tornado 1.01. Hope this helps, Hartmut In article <8676lu$a53$1@sulawesi.lerc.nasa.gov>, "Dave Hall" writes: > > I am trying to compile some c++ code for a PowerPC 603e target. > I am using the GNU compiler that comes bundled with Tornado 1.0.1: > > > Reading specs from /export/home0/wind/host/sun4-solaris2/lib/gcc-lib/ > powerpc-wrs-vxworks/cygnus-2.7.2-960126/specs > gcc version cygnus-2.7.2-960126 > > > The problem is that the compiler is screwing up the use of Template classes. > I am attaching some sample code that demonstrates the problem. > This code compiles and links just fine, including the munch, etc step. > When I download to the target, I get this: > > Undefined Symbols: > Stack::Stack(int) > > I need a fix for this problem! > > > Here's the sample code, which works fine in Microsoft VC++: > > ---------------------------------------------------------------------------- > ------------ > > #include > #include > > template > class Stack{ > public: > Stack(int = 10); > ~Stack() { delete [] stackPtr; } > int push (const T&); > int pop(T&); > int isEmpty() const { return top == -1; } > int isFull() const { return top == size -1 ;} > private: > int size; > int top; > T *stackPtr; > }; > > template > Stack::Stack(int s) > { > size = s; > top = -1; > stackPtr = new T[size]; > } > > template > int Stack::push(const T &item) > { > if(isFull() ) { > stackPtr[++top] = item; > return 1; > } > return 0; > } > > template > int Stack::pop(T &popValue) > { > if(isEmpty() ) { > popValue = stackPtr[top--]; > return 1; > } > return 0; > } > > > int dhmain() > { > Stack floatStack(5); > float f = 1.1; > cout << "Pushing elements onto floatStack\n"; > while (floatStack.push(f) ) { > cout << f << ' '; > f+= 1.1; > } > > cout << "\n Stack full. Cant push " << f; > cout << "\n Popping elements off \n" << endl; > > > > > while( floatStack.pop(f) ) > cout << f << " "; > > > cout << "\n Stack empty, cant pop" << endl; > > > return 0 ; > } > > // INSTANTIATION !!!! > template class Stack; > > > > > > > - -- Hartmut_Pohl@agilent.com - -- Freunde verstaendigen sich nicht, sie verstehen sich. Ernst Zacharias --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Floating Point Underflows Date: Wed, 23 Feb 2000 09:20:07 -0800 From: Ken Norlin Organization: NASA Dryden FRC Message-ID: <38B416C7.2CF4BD7F@dfrc.nasa.gov> References: <890tj9$ldl2@overload.lbl.gov> John_W_Cosgrove@res.raytheon.com wrote: > > Dear vxWorks'ers, > > We are porting a unix application to vxWorks. On unix this application > generates a floating point underflow, and continues executing using these > very small numbers... > > I want the same thing to occur in the port. On our target, this underflow > stops processing. > (target is PPC 750, using the CPU=PPC604 switch. MVME-2700 using 5.3.1 > BSP=1.1/4 for MVME-2604 patched with Moto's '2700 patch). > > Worse (or Better???) is that this is a FORTRAN application compiled by the > Green Hills FORTRAN compiler. > Hi, On my MVME-2605 board, I use the following: /* Disable floating point underflow/overflow exception handling */ fp_mask = vxFpscrGet(); fp_mask &= (~(_PPC_FPSCR_OE | _PPC_FPSCR_UE)); vxFpscrSet(fp_mask); I found this info in the back of the VxWorks 5.3.1 Programmer's Guide (pg 605). Ken N. - -- Ken Norlin --- Simulation Engineer NASA Dryden Flight Research Center ken.norlin@dfrc.nasa.gov (661) 258-2046 FAX (661) 258-2792 --------------------------- Newsgroups: comp.os.vxworks Subject: How to execute Startup script Date: Wed, 23 Feb 2000 15:00:11 GMT From: m.fischer@steinbichler.de (Manfred Fischer) Organization: roNet GmbH Rosenheim Message-ID: <38b3f5b1.25555657@news.ronet.de> --------------------------- Newsgroups: comp.os.vxworks Subject: switching IP addresses on vxWorks Date: Tue, 15 Feb 2000 03:12:23 GMT From: captain_underwear@my-deja.com Organization: Deja.com - Before you buy. Message-ID: <88ag6k$jkq$1@nnrp1.deja.com> Maybe this belongs to comp.unix.networking, but I was wondering what would happen in a vxworks environment under the following circumstances: I have 3 cards A, B , C. B & C are hot standby cards, and B is currently active, and is getting UDP and TCP messages from A. B crashes, and C wants to take over its role. Can C just send an arp message to say that B's IP address is now associated with its ethernet address ? Will the TCP connections actually stay up if the switchover is in a reasonable amount of time ? thanks in advance for any help Siddharth Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Alignment in Structures Date: Sat, 19 Feb 2000 18:39:26 GMT From: pwhicker@home.net (Paul Whicker) Organization: @Home Network Message-ID: References: <38A93BAA.F5307BFC@hotmail.com> In the end the alignment and pack pragmas are nothing more than advice to the compiler. Mostly they work though. Having said that you are aligning to 1? Well surely that's to any single byte boundary, or in other words anywhere. Perhaps I misunderstand, but wouldn't 2, 4 or 8 be more useful? Paul In article <38A93BAA.F5307BFC@hotmail.com>, "B. Lampi" wrote: >How can I set the alignment for a single structure? >Does "@pragma align 1" not work? > >Thanks for any advice. >Bernhard > >PS: >Host: PC, NT 4.0, Tornado 2 >target: PPC821, vxWorks >GNU C compiler for PPC 2.7.2-960126 > > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Setting the TLB with in VxWorks (MPC860 MMU/Cache) Date: 23 Feb 2000 13:10:18 -0500 From: Bill Pringlemeir Organization: Factory of the mind Message-ID: References: <38B40B1F.35CA05AA@adc.com> Sender: bpringle@DeadDuck Under VxWorks, you do not write the TLB's directly. This is accomplished by a 'MMU driver'. I beleive that the proper way to do this is to look at sysLib.c for the array sysPhysMemDesc[]. This contains the mechanism that ultimately constructs the TLB's. There are also allocation functions that give you cache safe memory for use in drivers. Things can be complicated if you wish to use features that aren't under the VxWorks model. Ie, the ARM architechures with MMUs have a cacheable, and bufferable attributes; This is like two cache types. hth, Bill >>>>> "Sandy" == Sandy Rivera writes: Sandy> The MPC860 doesn't have BAT registers to setup the blocks Sandy> of memory to write-through, write-back, or Inhibit. It Sandy> uses the TLB to accomplish the different caching schemes on Sandy> the different memory blocks. How do you setup these tables Sandy> within vxWorks? I found were you enable the cache Sandy> (cacheEnable function), but how is the TLB populated? What Sandy> are the steps needed to enabling cache? - -- Danger, end of message ahead! --------------------------- Newsgroups: comp.os.vxworks Subject: Re: VxWorks Shell via the serial interface Date: Sat, 19 Feb 2000 19:12:10 GMT From: pwhicker@home.net (Paul Whicker) Organization: @Home Network Message-ID: References: <38AC1B8D.8E272F8D@radguard.com> <88hbn8$su2$1@news1.xs4all.nl> Hmm, well I have a T2 installation on this laptop that I have never used and when I look I find the "target shell" components are all off by default. So I'd need to switch them on then I'd need to include either a standalone or downloadable symbol table (see development tools, symbol table components.) You'll need to tell it include the symbol table then under initialization components yuu'll need to select initialization on and pick a source (downloadable or built in). Next I'd look and check to see that INCLUDE_PC_CONSOLE is switched off (target, hardware, serial, PC console). While you are there you'll find that one of the parameters of SIO is CONSOLE_TTY and you can select which port is going to be used with that. Even if you aren't using a PC I think you'll still find that the default is no target symbol table and no target shell. It's a two minute job. HTH, Paul In article <88hbn8$su2$1@news1.xs4all.nl>, johan.borkhuis@peektraffic.nl wrote: >Rafi Hacham wrote: >: Hi All !!! > >: I want to be able to use the VxWorks 5.4 shell via the serial interface >: rather then >: the Ethernet interface (Tornado 2 environment) which is my default. >: Does anybody knows how to customize the BSP in order to use the shell >: via the >: serial interface? > >The Target shell is default on one of the serial lines. Check you >configuration to see if it is enabled and on wich serial line it is. The >host shell however (that is the one within Tornado) is using the network >connection. > >Groeten, > Johan > --------------------------- Newsgroups: comp.os.vxworks Subject: Stdio without console problem+solution Date: Thu, 17 Feb 2000 17:02:07 GMT From: trahn@my-deja.com Organization: Deja.com - Before you buy. Message-ID: <88h9ic$bqs$1@nnrp1.deja.com> Hello, maybe someone already run into this problem, at least it might help to know about... This is a somewhat academic question: What happens if I set stderr to (-1) using > ioGlobalStdSet (2, -1) and start some output, e.g. via > fprintf(stderr, "Nirvana") In fact, this is exactly what is done in the startup code from VxWorks 5.3.1, x86 BSP, if neither TTY device nor PC console is defined. Unfortunately, it seems to cause serious memory and/or timing problems on our target (however, without any ports on the target debugging is not easily done). I will fix the problem by creating a file descriptor on the null device and mapping the std out/err to this fd (all in usrConfig.c), but _not_ stdin (makes shell very active). With Tornado 2, WRS presents an improved version: the problem is scattered into several files, like sysComms.c, usrPcConsole.c and usrSerial.c I wonder about the response from WRS support :-) Yours Thomas Rahn ===/\\ C a s a b l a n c a ===// \\ ===/////\\ \\ // II //\\ ===// \\ \\// II \\// ================================================ MacroSystem Digital Video AG Thomas Rahn Forschung und tel: +49(0)6151-272579 Entwicklung fax: +49(0)6151-272577 email: trahn@macrosystem.de ================================================ Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Typical Thread Switching Time Date: Mon, 21 Feb 2000 15:02:59 -0500 From: "Serge Martineau" Organization: Bell Northern Research Message-ID: <88s5lj$o8f$1@bcrkh13.ca.nortel.com> Hello, I'm interested in knowing the typical thread switching time in # of clock cycles or in # of intructions. Thanks, Serge --------------------------- Newsgroups: comp.os.vxworks Subject: Is munchless C++ linking possible Date: 23 Feb 2000 18:50:51 GMT From: jnance@wrtph0qb.nortelnetworks.com (Jim Nance) Organization: Bell Northern Research Message-ID: <891a6b$hiq$1@bcrkh13.ca.nortel.com> Hello All, I am working on a VxWorks based project that is written in C++. Part of the build process involves partially linking the executable and running a tcl script called munch that takes the symbols table for this executable and generating global constructor calls. This is slow, and I was wondering if there was another way to link things that did not require running munch? Thanks, Jim --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Intel StrataFlash28F128AJ3 & TrueFFS Date: Wed, 23 Feb 2000 12:03:15 -0700 From: Doug Gibbons Organization: Lucent Technologies Message-ID: <38B42EF3.A02C1DEE@lucent.com> References: <88h1kv$427$1@pollux.ip-plus.net> Thomas Fox wrote: > > Hello, > has someone experience implementing Intel StrataFlash28F128AJ3 MTD for > TrueFFS. > We will use it with VxWorks 5.4 on a PowerPC 603 Board. > > Thanks, Thomas One of the interesting things about Strataflash is that each cell can hold 4 values instead of the usual 2. If a powerdown situation occurs while a cell is being programmed or erased, you could be left with an unexpected intermediate value in the cell. Some flash file systems maintain data integrity in the face of powerdown situations by way of checkpoint flags written into flash. I'd want some guarantee out of TrueFFS that any such mechanisms will still work correctly even when mapped onto Strataflash. One technique might be to implement binary checkpoint flags as pairs of bits - 00 or 11 - anything else is evidence of a power failure. - -- Lee Douglas Gibbons L u c e n t T e c h n o l o g i e s ldgibbons@lucent.com Bell Labs Innovations --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Problem with sysProcNumSet for mv1604 Date: Wed, 23 Feb 2000 18:43:23 GMT From: Minh.Lang@jpl.nasa.gov Organization: Jet Propulsion Laboratory - Pasadena CA Message-ID: <38b4287f.89779035@news.jpl.nasa.gov> References: <38B3EA17.57C213B7@emw.ericsson.se> On Wed, 23 Feb 2000 15:09:27 +0100, Andreas Johansson wrote: >Hello, > >I am using VxWorks 5.3 together with a mv1604 board, and BSP version >1.1/4. > >The function sysProcNumSet (sysLib.c) sets the VME address of GCSR in >the vmechip2 - vmebus slave GCSR group address register. When that is >done, the variable 'sysProcNum' is used, should not the variable >'procNum' be used instead? As it is now gets all processors the same >base address of their GCSR-register. sysProcNum was set to procNum, then the VMECHIP2's Slave GCSR Group Address Register was set to a **fixed** GCSR_GROUP_ADDR (bit 31-24) then the processor board ID (processor number) was set in bit 23-20. If you have more than 1 board, isn't this enough to differentiate one from another? = Minh = --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Is anyone use C++ with T2? Date: Wed, 23 Feb 2000 11:20:01 -0800 From: "Dave Craig" Organization: Verio Message-ID: References: <88p7q5$oc51@nntp.cig.mot.com> <88rtr2$rfg$1@newsmaster01.nextra.at> Hi, We have a very large C++ project which was developed with T1.0.1 that I'm gradually porting to TII. Here's the problems I've some accross...none are that bad so far, but the more C++ features you use the more problems you'll hit (templates, exceptions). - -frepo was broken but is now patched. However, there are still problems with linking across different drives on WIN32 hosts and linking with the BSP is a real trial if you don't want multiple definitions of some template symbols. c++filt386 doesn't work properly on some templated symbols. The symbols remain mangled. You may have problems with C++ exceptions especially when porting from T1.0.1. The compiler defaults to compiling in C++ exception handling. There's an SPR #29680 on the fact that this causes intermittent crashes at interrupt time...(the result of an interrupt time __get_eh_context call). The rule is unless you need C++ exceptions compile with -fno-exceptions. If you do need exceptions you'll have to take care to not call constructors at interrupt time. No basic_string<> types other than basic_string work (so no internationalization with wstring). I think that's it, the debugger is much better than in T1.0.1, as is the compiler (much faster, more errors). Dave Craig Palo Alto, CA. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: VxSim on Solaris, with SIO & END Date: Wed, 23 Feb 2000 19:25:33 GMT From: tttt9999@my-deja.com Organization: Deja.com - Before you buy. Message-ID: <891c79$9at$1@nnrp1.deja.com> References: <88vm0v$34k$1@nnrp1.deja.com> <38B39A04.5B359925@csua.berkeley.edu> But it works on SIMNT, why is it hard to support the same thing on SIMSOLARIS ? I would think that vxsim was implemented on the Sun/Solaris first ... then on the NT , no ? In article <38B39A04.5B359925@csua.berkeley.edu>, David Holloway wrote: > You are going to have a hard time using a serial > driver with a simulator that doesn't emulate a serial port. > > tttt9999@my-deja.com wrote: > > > > I am running into some problems when trying to migrate from > > SIMNT (vxsim on NT) to SIMSOLARIS (vxsim on solaris). I was > > trying to include the END and the SIO driver, both gave me > > errors when compiling and linking respectively. > > > > It turned out that I was missing > > $WIND_BASE/target/config/solaris/{sysSerial.c,configNet.h} > > > > It worked fine on the SIMNT env, and it does have those 2 files > > in the $WIND_BASE/target/config/simpc. > > > > Has anyone successfully running SIMSOLARIS with both SIO > > and END enabled ? Am I just missing those files during installations ? > > > > Thx. > > > > T9 > > > > Sent via Deja.com http://www.deja.com/ > > Before you buy. > Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Embedded Systems SW Engineer. Date: Mon, 21 Feb 2000 15:43:50 -0500 From: "Tom Tevendale" Organization: IGS - Information Gateway Services Message-ID: <88s7ns$b7a$1@news.igs.net> As a publicly traded company, CALIAN Technology Services (TSE: CTY) has been providing technical and project management support since 1982. With annual revenues in excess of $80 million dollars and more then 1,600 dedicated employees, CALIAN provides diversified technology and professional services throughout Canada and the United States. To learn more about Calian Technology Services please visit http://www.calian.com Our Ottawa Client is looking for an Embedded Systems Software Engineer. Overview: As a fast growing GPS commercial products company, our client has an immediate requirement for an Embedded Systems Software Engineer. As an ideal candidate you will feel comfortable with the use of laboratory equipment tools including logic analyzers and oscilloscopes to analyze and debug prototype systems. You are creative and enjoy the creation of new products. Responsibilities: Designing unit test and acceptance test plans Microcontroller selection Operating system selection Development tool specification Specification of hardware prototypes for software development Layering and partitioning of communications tasks Cellemetry Software interface Spread Spectrum/Circuit Switched modem interface Pager interface Code functional modules. As an Embedded Systems Software Engineer, you are experienced with C, assembler, working with windows based platforms and are familiar with industry tools such as VX Works, PSOS and QNX. Mandatory Qualifications: 2 + years or more experience in the design of embedded systems software. Desirable Qualifications: Experience with Real Time systems on microcontroller based multi-tasking operating systems. A communications background. Education Requirements: A Bachelors Degree in Electrical Engineering, or equivalent. If you are interested, please submit a resume (as a Word attachment) that highlights your experience to: t.tevendale@calian.com and quote reference number #2435 in your subject line. For more information, please contact Tom Tevendale Phone 613-599-8600 ext. 306 Toll Free 877-2254264 ext. 306 t.tevendale@calian.com --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Gathering CPU utilization by process. Date: Wed, 23 Feb 2000 20:02:06 GMT From: Ted Dennison Organization: Deja.com - Before you buy. Message-ID: <891ebq$aug$1@nnrp1.deja.com> References: <88rvh6$e7h$1@nnrp1.deja.com> <88uf4g$66o$1@nnrp1.deja.com> In article <88uf4g$66o$1@nnrp1.deja.com>, Ted Dennison wrote: > ScopeProfile also has a task-level CPU utilization profiling function. > I'm not too sure how much I trust it though. We're currently tracking > down a CPU hog which Windview and Spy both show clearly, but for some > reason ScopeProfile doesn't. Perhaps I just don't know how to read its > output properly or something. OK, I figured this one out. It turns out that the (rather old) version of ScopeProfile that we have processes its data from a low priority task. Thus incorrect information in the presence of a CPU hog is to be expected. It collects the data at iterrupt priority, but processes it at low priority, so the data just queued up without getting processed until I froze the CPU hogging task. At that point I did get good info that showed me the exact routine hogging the CPU. Damn nice. It would have taken *days* to figure that out without a profiler. RTI tells me that the most recent version of ScopeProfile processes the data on the host machine instead of the target. That should prevent these kinds of issues. - -- T.E.D. http://www.telepath.com/~dennison/Ted/TED.html Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Disk Defrag Using DosFSII Date: Wed, 23 Feb 2000 21:44:33 GMT From: pikorad@my-deja.com Organization: Deja.com - Before you buy. Message-ID: <891kbv$fn0$1@nnrp1.deja.com> Does anyone know of any vxworks utility or code that does a defrag of IDE hard disk. Basically, I have a hard drive mounted as one one my devices, and I want to be able to defrag it just like you can to any other drive using Windows Utilities. Any ideas? Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: [Q]: losing target connection Date: Thu, 24 Feb 2000 00:40:48 GMT From: ipmobilestephen@my-deja.com Organization: Deja.com - Before you buy. Message-ID: <891umg$n6t$1@nnrp1.deja.com> We are developing using Tornado NT version connected to Motorola boards running VxWorks. Something in the application we are writing is going crazy, and we'd randomly lose connection to the target. We lose both the Ethernet connection and serial terminal output and are forced to reboot the card. This gives us very little means of tracking down what is causing this. Does anybody have any suggestions or know of tools that might help out in the above situation? Thanks! Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Gathering CPU utilization by process. Date: 21 Feb 2000 19:38:11 GMT From: johan.borkhuis@peektraffic.nl Organization: XS4ALL Internet BV Message-ID: <88s473$nqc$1@news1.xs4all.nl> References: <88rvh6$e7h$1@nnrp1.deja.com> Willie Peloquin wrote: : Hello, : I typically work in QNX and I intend to transition over to VxWorks. : Could someone give me a quickstart on process profiling? Try spy. Look at spyLib for a description. Groeten, Johan - -- o o o o o o o . . . ___________________________________ o _____ || Johan Borkhuis | .][__n_n_|DD[ ====_____ | johan@borksoft.xs4all.nl | >(________|__|_[_________]_|________________________________| _/oo OOOOO oo` ooo ooo 'o!o!o o!o!o` === VxWorks page: http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html === --------------------------- Newsgroups: comp.os.vxworks Subject: Re: VxWorks vs. Linux Date: Sat, 19 Feb 2000 19:30:51 -0500 From: Thomas Davis Organization: Interpath Communications, Inc. Message-ID: <38AF35BB.805AAEF1@galois.net> References: <87t4tl$55u$1@news2.kornet.net> <38AE0C80.2A354436@earthlink.net> Hi, Interesting that this conversation took this turn. To think that you have to buy commercial to get quality is simply untrue. That is a Microsoft statement and we can see what it is going to get them. Linux is a shining example of how free can be better, faster and less resource intensive than any other OS. In fact, if anyone here wants to challenge Linux in speed and resource usage against any other OS in the world, I welcome the result (leaving RTOS out as I have not actually used the embedded one). My first experience with Linux was back when it didn't have a major revision number. I was in Clemson trying desperately to find something that doesn't crash all the time. I tried OS/2 (before Warp), and it was great but sooooo slow as it really needed memory and that was when that Memory factory burnt down causing supper high SIMM prices. I only had 4 Megs. I wanted Solaris but I just plain could not afford it. So I tried Linux. It was incredible, strong blazing speeds and multitasking too. I would compile while I was modified a dozen files and downloading stuff over my fast 2400 Baud modem. It took me two weeks to figure out how to get X running because I had to calculate and set up all those monitor settings. But when I got it up, the picture was absolutely perfect, 10 times better than all the other OS's. Now that I told you that, I was not (at that time) some computer geek, I was a Math major. I did know Windows and DOS well but that wasn't any help. The point is, that although it was technical, it didn't take a genius it is nothing compared to what a vxWorks developer deals with everyday. In this case free was better. Today I still use Linux and am happy to say old Gates can kiss my butt as I do not and will not ever use any of the garbage his company stole and called their own. I do my work on Solaris which I am happy to say is every bit as awesome as people say but in speed Linux kicks its butt; however, in reliability I give it to Solaris. Linux is so fast because it caches a lot risking some data while Solaris does not. Back at Linux, Linux is enjoying rapid application development right now in every field. In fact Linux is the fastest growing everything be it Desktop, Server, Embedded or even Processor port. WindRiver better watch their back because the Penguin is coming. About me, I love vxWorks, I am responsible for many kernels and have written countless drivers and applications. I also love Solaris and have within countless everything for it and Motif. I like Linux but I do not develop on it, I use it. Mostly this is because 1) it already has everything I need and 2) I do enough development at work. A note about GNU coding style, a have seen a lot and some of it, I have to admit, is just plain ugly (don't believe me, look at GNU make). But also, I have seen some good coding like GNU ftp. The GNU primary goal is portability and you can bet I support that. However, I don't agree that new Technology should be held back which GNU causes with their strict guidelines. But thinking about code quality, does anyone think including source in source good coding. That is how BSP's are setup by Wind River. That just bothers the hell out of me. Well anyway, I enjoy my safe world in vxWorks and Solaris, but I am not so blind to think that Linux isn't going to change that. In fact I know it will. Personally, I am betting on Java. That is where my attention and money is. Thomas lcb wrote: > > $2995 gets you a seat of VxWorks. $50K buys you 10 seats of Tornado Power > Professional. Good luck on every getting out a product with a free OS unless > you have a group of engineers working 80 to 100 hours a week for months. > > ChaeHo You wrote: > > > Hello, VxWorks users... > > > > I'm concerned about Linux development. And I'm interested about the price of > > VxWorks. > > Our company sales the Dy4 systems' products. That company makes the VME > > board for military/commercial version. And the main RTOS of the boards is > > VxWorks. > > That company determined a joining a Linux company,AMIRIX. If RTLinux would > > be ported to Dy4 boards, there will be many interesting changes in RTOS > > market. > > I think the price of the VxWorks is too high. The price of Tornado II per 1 > > user > > is about 50000 us dollars, isn't it? > > The problem is the capability of Linux. The developing speed of Linux is > > amazing. Maybe in a near future, RTLinux will compete VxWorks. > > > > Best Regards... - -- Galois, Inc. http://www.galois.net We color the world! --------------------------- Newsgroups: comp.os.vxworks Subject: Number of /tyCo/xx devices Date: Tue, 22 Feb 2000 11:31:35 +0100 From: Gilles Bannay Organization: GESPAC Message-ID: <38B26587.80FA2595@gespac.ch> Hello, I have to test some serial multiport board. on a x86 target (up to 16 ports). So I modified some BSP parameters: #define NUM_TTY 16 #define N_UART_CHANNELS 16 I initialized a 16 NS16550_CHAN entry table and only the devices from /tyCo/0 to /tyCo/9 are created. Can I increase this limit from 10 to 16 ? or is it a system limitation ? Thanks, Gilles. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: taskIdVerify causing exception! Date: Thu, 24 Feb 2000 05:45:22 GMT From: snaphook@southwesternbell.net (Bruce) Organization: SBC Internet Services Message-ID: <38b4c501.16956413@news.swbell.net> References: <890hno$eqd$1@ndsukns01.ndsuk.com> In comp.os.vxworks "Will Fookes" wrote: >However, I am now seeing some slight memory corruption which causes taskName >to crash - it looks as if the taskId is used as a pointer and de-referenced >BEFORE it has been validated, causing the task to crash if the taskId value >is outside the memory map. The task ID certainly is an address. >OK, so perhaps I can use taskIdValid to validate the taskId before >attempting to call taskName? >NOT SO: this seems to do the same thing... > >-> taskIdVerify 0x2274dd90 > >Exception number 768: Task: 0x40ae38 (t4) > >data access >program counter: 0x1ca210 >access address: 0x2274ddc0 >machine status: 0x9032 >1cff48 vxTaskEntry +5c : funcCallWrapper ([]) >1bc74c funcCallWrapper+a4 : taskIdVerify ([]) >value = 0 = 0x0 > >Does anyone know an easy way to validate a taskId ? That is interesting. You could certainly get a list of all current task IDs and search for your ID in that list? I don't think you can expect the task IDs to always be the same. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: [Q]: losing target connection Date: Thu, 24 Feb 2000 06:25:56 GMT From: pwhicker@home.net (Paul Whicker) Organization: @Home Network Message-ID: References: <891umg$n6t$1@nnrp1.deja.com> Use WDB in serial mode. Paul In article <891umg$n6t$1@nnrp1.deja.com>, ipmobilestephen@my-deja.com wrote: >We are developing using Tornado NT version >connected to Motorola boards running VxWorks. >Something in the application we are writing >is going crazy, and we'd randomly lose connection >to the target. We lose both the Ethernet connection >and serial terminal output and are forced to reboot >the card. This gives us very little means of >tracking down what is causing this. > >Does anybody have any suggestions or know of tools >that might help out in the above situation? > >Thanks! > > > >Sent via Deja.com http://www.deja.com/ >Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Driver list Date: Thu, 24 Feb 2000 06:27:06 GMT From: pwhicker@home.net (Paul Whicker) Organization: @Home Network Message-ID: <_a4t4.330$so2.5973@news1.frmt1.sfba.home.com> References: <88uhdd$na6$1@pegasus.csx.cam.ac.uk> Possibly on the web site, certainly from your local sales guy, he can get it from their marketing people. Paul In article <88uhdd$na6$1@pegasus.csx.cam.ac.uk>, hugh@mrao.cam.ac.uk wrote: >Hi, > I'm trying to locate a recent VxWorks driver list. >The long FAQ mantions vxdrivers.gz but I can't locate this with >WWW search engines. > Can anyone please point me to a recent list of VxWorks drivers >(looking for GPIB control) Many Thnks, Hugh Gibson --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Purify or equivalent under vxSim ? Date: Thu, 24 Feb 2000 06:30:13 GMT From: pwhicker@home.net (Paul Whicker) Organization: @Home Network Message-ID: References: <38B24E31.24F4BCC1@europem01.nt.com> If you use the Diab Data compilers and SDS Single Step you can achieve most of what purify does, I think it's called RTA or something like that. You said embedded systems but you don't specify an architecture, I think Diab/SDS are available for slightly less architectures than vxWorks but most are covered. Paul In article <38B24E31.24F4BCC1@europem01.nt.com>, Gabriel Lang wrote: >Does anyone know if it's possible to use >purify under vxWorks (or vxSim) and if not >is there a similar product for embedded >systems ? > >TTFN > >Gab. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: restart OS without reloading object Date: Thu, 24 Feb 2000 06:34:12 GMT From: pwhicker@home.net (Paul Whicker) Organization: @Home Network Message-ID: References: <3u0s4.6098$Qx5.74092@news.rdc1.tx.home.com> <88r391$fci$3@news1.xs4all.nl> Surely the locations would not have changed and the object would have been linked when it was initially loaded so I believe this answer is inadequate. I think you are more likely to fall on your face because of uninitialized data which on invocation of a newly loaded object is presumed to be zero? To call the object code you'd have to call via a pointer to a known location or add the symbol by hand. Alternatively just ROM the whole thing and it's no longer an issue, the code is always there regardless. HTH Paul In article <88r391$fci$3@news1.xs4all.nl>, johan.borkhuis@peektraffic.nl wrote: >Bo Yu wrote: >: Hello, > >: Suppose the system with vxworks is running, then if I do not >: want to reload the object, I just jump to sysInit ( usrInit, usrRoot >: afterwards ). Could the whole system be reinitialized as the first >: time ? > >When you include your own SW with the VxWorks bootable image, yes it is >possible. When you want to dynamically load modules and keep them over a >restart, no that is not possible, because of the linking process. After >you load an object VxWorks has to dynamically link this object to resolve >all the missing symbols. > >Groeten, > Johan > --------------------------- Newsgroups: comp.os.vxworks Subject: Does anybody know for VxWorks on ARM710A (preferably CL-PS7111 CPU) based products Date: Thu, 24 Feb 2000 06:35:51 GMT From: "nlazovic" Organization: Sympatico Message-ID: Does anybody know for VxWorks on ARM710A (preferably CL-PS7111 CPU) based development ? I would like to get info on general experience working in that environment. Sincerely, Nenad Lazovic NBS Transaction Systems Inc. nlazovic@idirect.com --------------------------- Newsgroups: comp.os.vxworks Subject: memory leak deleting a pipe Date: Thu, 24 Feb 2000 08:11:28 +0100 From: "Markus MItterer" Organization: KPNQwest customer news service Message-ID: Hi VxWorkers, In my code I create a pipe-device with pipeDevCreate(). When I don`t need the pipe any more I want to delete it. For this purpose I work with the functions iosDevFind() and iosDevDelete(). After calling this functions the device was deleted from system, but now I have a memory leak, which was created by pipeDevCreate(). How can I delete the pipe completly or what is the workaround? I can`t find a sence why there is no delete function for a pipe. Thank you in advance, Markus Mitterer, STK --------------------------- Newsgroups: comp.os.vxworks Subject: Re: RTL 8139 driver ? Date: Tue, 22 Feb 2000 13:46 +0000 (GMT Standard Time) From: rew@cp.cix.co.uk (Richard Wenham) Organization: CIX - Compulink Information eXchange Message-ID: References: <38B14783.EDF396BF@itk.ntnu.no> Reply-To: rew@cp.cix.co.uk WRS have a beta driver for the RTL 8139, contact your salesperson. Richard In article <38B14783.EDF396BF@itk.ntnu.no>, ostraume@itk.ntnu.no (=?iso-8859-1?Q?=D8ystein?= Straume) wrote: > Hi, > > I have a ICP Nova-600 EBX Pentium Embedded Board with a built-in RTL8139 > network device. I wonder if someone has used this chipset under VxWorks, > and if there is any available drivers? > > Maybe someone has ported some BSD network drivers already. If so could > you describe it if it was easy ? > > Thanks, Øystein --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Number of /tyCo/xx devices Date: 22 Feb 2000 14:17:53 GMT From: johan.borkhuis@peektraffic.nl Organization: XS4ALL Internet BV Message-ID: <88u5qh$poo$1@news1.xs4all.nl> References: <38B26587.80FA2595@gespac.ch> Gilles Bannay wrote: : Hello, : I have to test some serial multiport board. : on a x86 target (up to 16 ports). : So I modified some BSP parameters: : #define NUM_TTY 16 : #define N_UART_CHANNELS 16 : I initialized a 16 NS16550_CHAN entry table and : only the devices from /tyCo/0 to /tyCo/9 are created. : Can I increase this limit from 10 to 16 ? : or is it a system limitation ? The devices are created in usrSerial.c in config\comps\src (when you are using Tornado 2.0, otherwise look for the call ttyDevCreate). Here a call is placed to the function sysSerialChanGet. This function is in your sysSerial.c, and this function returns a pointer to a SIO device structure. Here also the support for more than 10 channels should be supported. Groeten, Johan - -- o o o o o o o . . . ___________________________________ o _____ || Johan Borkhuis | .][__n_n_|DD[ ====_____ | johan@borksoft.xs4all.nl | >(________|__|_[_________]_|________________________________| _/oo OOOOO oo` ooo ooo 'o!o!o o!o!o` === VxWorks page: http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html === --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Gathering CPU utilization by process. Date: Tue, 22 Feb 2000 14:19:52 GMT From: John or Jenn Organization: Who, us organized??? Message-ID: References: <88rvh6$e7h$1@nnrp1.deja.com> In article <38b24e2c$1@pull.gecm.com> Stephen Walsh, stephen.walsh@gecm.com writes: >There are two tools that are good for profiling and task monitoring. These >are Windview and Stethoscope. Actually, he asked for task level profiling, and that¹s provided by Spy (accessible from the browser, or the shell, IIRC). PerformancePak has a function level profiling tool (from the Stethoscope folks, RTI, but not part of Stethoscope). --------------------------- Newsgroups: comp.os.vxworks Subject: Re: taskIdVerify causing exception! Date: Thu, 24 Feb 2000 07:56:40 GMT From: john_94501@my-deja.com Organization: Deja.com - Before you buy. Message-ID: <892o7n$8br$1@nnrp1.deja.com> References: <890hno$eqd$1@ndsukns01.ndsuk.com> Hi, It seems that the case you are trying to cover is when the task ID is in fact an address outside of the mapped memory (task IDs, like all WIND object IDs at the moment, are simply the addresses of the structures). The simplest way to cover this eventuality is to assess whether the task ID is a valid address. There are a few ways to do this: 1) vxMemProbe() will determine whether the memory at the address is accessible (you only need read access). 2) vmTranslate() could also be used to determine this if you have it (I don't think that PPC does) 3) You could range check the value against the known valid RAM addresses in your system. Simple, but effective! Hope that's some help! John... In article <890hno$eqd$1@ndsukns01.ndsuk.com>, "Will Fookes" wrote: > Hi, > > I am using Vxworks 5.3.1/Tornado 1.0.1 with a PPC860. > > When my system starts, I attempt to print out any old log messages remaining > in RAM from the previous run (we know that the RAM stays up for about 20 > seconds, so this has been a useful aid to finding out what caused the thing > to reset!!) > > Of course, we expect the RAM to be liable to corruption, so we are very > cautious how we process these old log messages. They each consist of a text > string (we ensure it has a null termination) and a few numbers that we > interpret (eg. error severity - we range-check it before looking up the text > interpretation in a table). > > One of the numbers is the taskId of the task that generated the log message. > We call taskName and check the return value - if null, the task doesn't > exist, so we print the hex value of the taskId rather than the name. > > However, I am now seeing some slight memory corruption which causes taskName > to crash - it looks as if the taskId is used as a pointer and de-referenced > BEFORE it has been validated, causing the task to crash if the taskId value > is outside the memory map. > > OK, so perhaps I can use taskIdValid to validate the taskId before > attempting to call taskName? > NOT SO: this seems to do the same thing... > > -> taskIdVerify 0x2274dd90 > > Exception number 768: Task: 0x40ae38 (t4) > > data access > program counter: 0x1ca210 > access address: 0x2274ddc0 > machine status: 0x9032 > 1cff48 vxTaskEntry +5c : funcCallWrapper ([]) > 1bc74c funcCallWrapper+a4 : taskIdVerify ([]) > value = 0 = 0x0 > > Does anyone know an easy way to validate a taskId ? > > Will Fookes > > Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Typical Thread Switching Time for MPC8260 Date: Tue, 22 Feb 2000 09:27:54 -0500 From: "Serge Martineau" Organization: Bell Northern Research Message-ID: <88u6d8$7i8$1@bcrkh13.ca.nortel.com> I'm interested in knowing the typical thread switching time in # of clock cycles or in # of intructions for the MPC8260 PowerQUICC II. The version of VxWorks is 5.4. Thanks, Serge --------------------------- Newsgroups: comp.os.vxworks Subject: Re: taskIdVerify causing exception! Date: Thu, 24 Feb 2000 08:32:06 -0000 From: "Will Fookes" Organization: NDS UK Message-ID: <892omk$jlm$1@ndsukns01.ndsuk.com> References: <890hno$eqd$1@ndsukns01.ndsuk.com> <892o7n$8br$1@nnrp1.deja.com> Hi John, Some good ideas here, though I plumped for the simple and obvious option (3) - but my feeling is that a function like taskIdVerify should be capable of (safely) assessing for itself whether _any_ value passed in is a valid taskId. I mean, "Why buy a dog, and bark yourself?" Will john_94501@my-deja.com wrote in message <892o7n$8br$1@nnrp1.deja.com>... >Hi, > >It seems that the case you are trying to cover is when the task ID is in >fact an address outside of the mapped memory (task IDs, like all WIND >object IDs at the moment, are simply the addresses of the structures). > >The simplest way to cover this eventuality is to assess whether the task >ID is a valid address. There are a few ways to do this: > >1) vxMemProbe() will determine whether the memory at the address is >accessible (you only need read access). > >2) vmTranslate() could also be used to determine this if you have it (I >don't think that PPC does) > >3) You could range check the value against the known valid RAM addresses >in your system. Simple, but effective! > >Hope that's some help! > >John... > >In article <890hno$eqd$1@ndsukns01.ndsuk.com>, > "Will Fookes" wrote: >> Hi, >> >> I am using Vxworks 5.3.1/Tornado 1.0.1 with a PPC860. >> >> When my system starts, I attempt to print out any old log messages >remaining >> in RAM from the previous run (we know that the RAM stays up for about >20 >> seconds, so this has been a useful aid to finding out what caused the >thing >> to reset!!) >> >> Of course, we expect the RAM to be liable to corruption, so we are >very >> cautious how we process these old log messages. They each consist of >a text >> string (we ensure it has a null termination) and a few numbers that we >> interpret (eg. error severity - we range-check it before looking up >the text >> interpretation in a table). >> >> One of the numbers is the taskId of the task that generated the log >message. >> We call taskName and check the return value - if null, the task >doesn't >> exist, so we print the hex value of the taskId rather than the name. >> >> However, I am now seeing some slight memory corruption which causes >taskName >> to crash - it looks as if the taskId is used as a pointer and >de-referenced >> BEFORE it has been validated, causing the task to crash if the taskId >value >> is outside the memory map. >> >> OK, so perhaps I can use taskIdValid to validate the taskId before >> attempting to call taskName? >> NOT SO: this seems to do the same thing... >> >> -> taskIdVerify 0x2274dd90 >> >> Exception number 768: Task: 0x40ae38 (t4) >> >> data access >> program counter: 0x1ca210 >> access address: 0x2274ddc0 >> machine status: 0x9032 >> 1cff48 vxTaskEntry +5c : funcCallWrapper ([]) >> 1bc74c funcCallWrapper+a4 : taskIdVerify ([]) >> value = 0 = 0x0 >> >> Does anyone know an easy way to validate a taskId ? >> >> Will Fookes >> >> > > >Sent via Deja.com http://www.deja.com/ >Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Number of /tyCo/xx devices Date: Tue, 22 Feb 2000 18:02:44 +0100 From: Gilles Bannay Organization: GESPAC Message-ID: <38B2C134.6BAA3D97@gespac.ch> References: <38B26587.80FA2595@gespac.ch> <88u5qh$poo$1@news1.xs4all.nl> johan.borkhuis@peektraffic.nl wrote: > The devices are created in usrSerial.c in config\comps\src (when you are > using Tornado 2.0, otherwise look for the call ttyDevCreate). Here a > call is placed to the function sysSerialChanGet. This function is in your > sysSerial.c, and this function returns a pointer to a SIO device > structure. Here also the support for more than 10 channels should be > supported. In the usrSerial.c file, I replaced the ugly code of the itos() function (look at... and you'll understand...) with this one: /*************/ static char *itos(int val) { static char str[12]; int i = 11; str[i] = '\0'; do { str[--i] = '0' + (val % 10); val /= 10; } while (val!=0); return &str[i]; } /*************/ and now it's OK !! - -- ____ | | Gilles Bannay | | Software Engineer | | mailto:gilles.bannay@gespac.ch | | _____ ____________ _____ _____ | ___|_____||_____ | | _____|| |____||____________||_____||_____||_____ | Gespac S.A. | http://www.gespac.ch chemin des Aulx 18 http://www.gespac.com 1228 Geneva tel: +41 22 706 21 52 Switzerland Fax: +41 22 794 64 77 --------------------------- Newsgroups: comp.os.vxworks Subject: semDelete error Date: Thu, 24 Feb 2000 18:17:00 +0900 From: "Lee Sang Min" Organization: Electronics and Telecommunications Research Institute (ETRI) Message-ID: <892t02$d2s$1@news.etri.re.kr> func_a() { SEM_ID sem; sem = semBCreate( SEM_Q_FIFO, SEM_EMPTY ); func_b( sem ); } func_b( SEM_ID sem ) { func_c( (int) sem ); } func_c( int sem ) { SEM_ID mySem; mySem = (SEM_ID) sem; semDelete( mySem ); } Result: semDelete error - incorrect object ID memPartFree - invalid block address Is this wrong? --------------------------- Newsgroups: comp.os.vxworks Subject: Driver list Date: 22 Feb 2000 17:35:41 GMT From: hugh@mrao.cam.ac.uk Organization: University of Cambridge, England Message-ID: <88uhdd$na6$1@pegasus.csx.cam.ac.uk> Sender: Hi, I'm trying to locate a recent VxWorks driver list. The long FAQ mantions vxdrivers.gz but I can't locate this with WWW search engines. Can anyone please point me to a recent list of VxWorks drivers (looking for GPIB control) Many Thnks, Hugh Gibson --------------------------- Newsgroups: comp.os.vxworks Subject: Re: taskIdVerify causing exception! Date: Thu, 24 Feb 2000 09:22:06 GMT From: john_94501@my-deja.com Organization: Deja.com - Before you buy. Message-ID: <892t7s$cc3$1@nnrp1.deja.com> References: <890hno$eqd$1@ndsukns01.ndsuk.com> <892o7n$8br$1@nnrp1.deja.com> <892omk$jlm$1@ndsukns01.ndsuk.com> Hi Will, It's a trade off between performance and security I guess... since the test is used a lot in the kernel's task related functions to make sure that the ID you pass is a really a valid, living task and not something else, it could be used frequently. To add the more comprehensive tests would make all the task functions a little slower (and the range check solution doesn't work on all CPUs so it can get quite complex, and very much arch dependent). Since it is mainly there to check that a once valid task ID hasn't since become invalid (e.g. the task has exited), the extra checks shouldn't be necessary in many cases! What works for one person, doesn't for the next. I have heard that people have asked for all these tests to be removed from the kernel code so that they get pure, fast object manipulations (admittedly though, that is normally for the semaphore ops, which have similar validation, where it is a larger percentage of the time since the actual operation is so fast); others want more thorough checks... It's tough to please all! John... Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Making vxWorks_rom ? Date: Thu, 24 Feb 2000 12:11:19 +0200 From: "cyrille darmon" Organization: Motorola Message-ID: <893042$1bh$1@schbbs.mot.com> few question on how to build an vxWorks application ? I am working with a MVME2700 Motorola board. I have create a bootable project wither the mv2700 BSP. I add to my project a simple application ( Hello.c in the demo) . I want to build an .ELF application of vxWorks to be able to run it under the PPC1-Bug (Motorola firmware). without TORNADO . I have already success to build a Ram based VxWorks image.(vxWorks) and run it under the PPC1-Bug. now I try to build a RAM based image that starts in ROM.(vxWorks_rom) and when I try to make elftobin on the file I have an error : The error: C:\project\Project3\default>elftobin vxWorks_rom.bin seg1 : Expected load address 0xfff00100 but file address is 0x00111670 What do you suggest me to do ? Best Regards, Cyrille Darmon Email :< > --------------------------- End of New-News digest ********************** From vxwexplo-errs@csg.lbl.gov Thu Feb 24 07:16:30 2000 From: "Parkin, Robert" Date: Thu Feb 24 07:16:32 PST 2000 Subject: unsuscribe vxworks From vxwexplo-errs@csg.lbl.gov Thu Feb 24 07:18:18 2000 From: "Gaurav Kulshreshtha" Date: Thu Feb 24 07:18:20 PST 2000 Subject: Configuring END_DEV_TABLE I am trying to hook up my module to be used by IP in VxWorks environment. Can somebody provide me pointers on how to do it. regards Gaurav ______________________________________________________ Get Your Private, Free Email at http://www.hotmail.com From vxwexplo-errs@csg.lbl.gov Thu Feb 24 07:56:08 2000 From: Henrik Bergstrom Date: Thu Feb 24 07:56:10 PST 2000 Subject: muxSend() and 'chained' packets? Dear VxWorks users, I am currently evaluating VxWorks, with special attention to the networking code. One thing I want to do but have not yet succeeded with is sending multiple packets via the muxSend() call. My impression is that it should be possible to 'link' many packets by setting mBlkHdr.mNextPkt to an appropriate value and then call muxSend() to send two (or more packets) with just one call. However, the source code for the ln7990End driver (included in the standard x86 distribution) pays no attention to mBlkHdr.mNextPkt so I assume that at least that driver won't work with 'linked' packets. Have I missed something important or is this functionality only implemented 'in the manuals' so far? Please let me know if you can shed some light on this matter! Regards, Henrik - -- Henrik Bergstrom +46-8-628 28 28 (phone) Software Engineer +46-8-628 64 14 (fax) Intertex Data AB henrik.bergstrom@intertex.se From vxwexplo-errs@csg.lbl.gov Fri Feb 25 00:21:28 2000 From: teit.uldal@philips.com Date: Fri Feb 25 00:21:30 PST 2000 Subject: using vxVMI with a pc486 BSP Dear VxWorkers, We are trying to use the vxVMI option for an x86 target - but don't seem to be able to get it working under Tornado 1.0.1. Both when INCLUDE_MMU_BASIC and INCLUDE_MMU_FULL are defined, the downloaded image hangs when calling 'usrMmuInit'. We are using a 'pc486' BSP as a starting point and we do not have a BIOS on our target but our own boot code. Thanks in Advance, Teit Tornado vxWorks ---------------------------------------------------------------------------------------------------------- Teit Uldal Software Engineer Digital Networks Product Services Philips Industrial Activities nv Kempische Steenweg 293 B-3500 Hasselt, Belgium tlf / fax: +32 (0)11 305816 / 958 email: teit.uldal@philips.com From vxwexplo-errs@csg.lbl.gov Fri Feb 25 01:11:25 2000 From: "h17352" Date: Fri Feb 25 01:11:27 PST 2000 Subject: vxworks This is a multi-part message in MIME format. ------=_NextPart_000_0004_01BF7FB3.56480FE0 Content-Type: text/plain; charset="gb2312" Content-Transfer-Encoding: quoted-printable ------=_NextPart_000_0004_01BF7FB3.56480FE0 Content-Type: text/html; charset="gb2312" Content-Transfer-Encoding: quoted-printable
 
------=_NextPart_000_0004_01BF7FB3.56480FE0-- From vxwexplo-errs@csg.lbl.gov Fri Feb 25 05:10:40 2000 From: "Fitch, Chris" Date: Fri Feb 25 05:10:43 PST 2000 Subject: GUIs for Embedded Systems using VxWorks Fellow VxWorkers, We are trying to find a replacement for our current method of building GUIs: RTX Windows and an X-Windows based GUI Builder. What other products are available? For additional background: our system employs a Power PC Single Board Computer with an attached Video Card. Thank you in advance, Chris Fitch LM NE&SS From vxwexplo-errs@csg.lbl.gov Fri Feb 25 05:24:04 2000 From: "Straub, Robert" Date: Fri Feb 25 05:24:06 PST 2000 Subject: NE-2000 Ethernet card VxWorks Driver Hello All, Our department is using Tornado 1.0.1 on PC development hosts and PC-compatible targets with STD-BUS (8-bit) architecture. We have a 10Base-T network. I am looking for an alternate source for a VxWorks driver for an NE-2000-compatible Ethernet controller (WinSystems LPM/MCM-NE2000) that is known to work with an 8-bit-wide data path. WindRiver distributes an NE-2000-compatible driver, if_ene.c, but our previous experience suggests that it might not work in our environment, or with our card. Operation in 8-bit versus 16-bit mode appears to be an issue. We and WindRiver have modified if_ene.c extensively, and we still do not have a robust Ethernet interface. WinSystems claims that the WRS-supplied driver should work. I would also like to hear responses from anyone who can support that claim, or disprove it. Thank you for any help or information you can provide. Robert J. Straub Telephonics Corporation straub@telephonics.com From vxwexplo-errs@csg.lbl.gov Fri Feb 25 08:08:13 2000 From: Jim Way Date: Fri Feb 25 08:08:15 PST 2000 Subject: VxWorks Ethernet END Failure This message is in MIME format. Since your mail reader does not understand this format, some or all of this message may not be legible. ------_=_NextPart_001_01BF7FA9.CEF68048 Content-Type: text/plain; charset="iso-8859-1" VxWorkers, This is a reposting of an earlier problem, but with more information. I'm using VxWorks 5.4 and Tornado II with a Motorola PPC860 processor (with on-chip Ethernet support). I have configured END drivers, FNTP support, and SNTP client and SNTP server. For what it's worth, my code is written in C++. On occasion, the unit ceases talking out the Ethernet ports. I monitor this in two ways: 1. connecting a telnet session and activating a once-a-second report. 2. once a minute, testing the return code from a ping of a known good address. Typically I first notice that the telnet session has stopped updating. The telnet session no longer accepts data from me either. I can go to one of my serial connections and observe that the unit is still operating normally, except for Ethernet. I can issue commands and get data. However, Ethernet related commands such as ping will fail. To recover from this failure, I must reset the unit. This creates an interruption in the Ethernet based outputs of the unit. Being an SNTP time server, this is unacceptable. When I had several boxes available for testing, I noticed that often two boxes would fail in this way at the same time. I'm down to two boxes now (sales and marketing demos gobbled the rest). The simultaneous failures suggested to me that there was something external that was triggering the problem. The only common connection between the units was the corporate Ethernet connection that connects the units to our desktop workstations (Wintel). This also sees all corporate network traffic: email, Internet, ERP, etc. I've placed a Network Associates Sniffer on the network to monitor for failures. I've seen a few CRC and fragment errors that did not appear to affect the unit. After eight days of monitoring with the Sniffer, I finally monitored during an Ethernet failure. I saw no unusual Ethernet activity and no new error types that had not already been successfully handled by my unit. Using the debug port, I've confirmed that my cluster utilizations are good. The unit apparently has never not been able to allocate clusters when it needed to. Memory is available for malloc(4096). Maximum stack utilizations have never been more than 25% of the available task stack. This leads me to a number of questions: 1. What would cause this Ethernet failure? 2. Is there a status bit or interrupt available through either VxWorks or the PPC that I can examine the health of the Ethernet subsystem? 3. Is there a software-only mechanism to clear or to reset only the Ethernet subsystem without having to wait for an entire VxWorks reboot? 4. Is there another question I should be asking? Any and all advice, comments, and suggestions are welcomed. TIA, Jim -------------------------------------------- Jim Way, Software Engineer Datum Austin (Austron Inc.) voice: 512.721.4170 fax : 512.721.4087 email: jwayATaustronDOTcom (no spam please) -------------------------------------------- ------_=_NextPart_001_01BF7FA9.CEF68048 Content-Type: text/html; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable VxWorks Ethernet END Failure

VxWorkers,

This is a reposting of an earlier problem, but with = more information.

I'm using VxWorks 5.4 and Tornado II with a Motorola = PPC860 processor (with on-chip Ethernet support).
I have configured END drivers, FNTP support, and = SNTP client and SNTP server.
For what it's worth, my code is written in = C++.

On occasion, the unit ceases talking out the Ethernet = ports. I monitor this in two ways:
1.  connecting a telnet session and activating = a once-a-second report.
2.  once a minute, testing the return code from = a ping of a known good address.

Typically I first notice that the telnet session has = stopped updating. The telnet session no longer accepts data from me = either. I can go to one of my serial connections and observe that the = unit is still operating normally, except for Ethernet. I can issue = commands and get data. However, Ethernet related commands such as ping = will fail.

To recover from this failure, I must reset the unit. = This creates an interruption in the Ethernet based outputs of the unit. = Being an SNTP time server, this is unacceptable.

When I had several boxes available for testing, I = noticed that often two boxes would fail in this way at the same time. = I'm down to two boxes now (sales and marketing demos gobbled the rest). = The simultaneous failures suggested to me that there was something = external that was triggering the problem. The only common connection = between the units was the corporate Ethernet connection that connects = the units to our desktop workstations (Wintel). This also sees all = corporate network traffic: email, Internet, ERP, etc.

I've placed a Network Associates Sniffer on the = network to monitor for failures. I've seen a few CRC and fragment = errors that did not appear to affect the unit. After eight days of = monitoring with the Sniffer, I finally monitored during an Ethernet = failure. I saw no unusual Ethernet activity and no new error types that = had not already been successfully handled by my unit.

Using the debug port, I've confirmed that my cluster = utilizations are good. The unit apparently has never not been able to = allocate clusters when it needed to. Memory is available for = malloc(4096). Maximum stack utilizations have never been more than 25% = of the available task stack.

This leads me to a number of questions:
1.  What would cause this Ethernet = failure?
2.  Is there a status bit or interrupt = available through either VxWorks or the PPC that I can examine the = health of the Ethernet subsystem?

3.  Is there a software-only mechanism to clear = or to reset only the Ethernet subsystem without having to wait for an = entire VxWorks reboot?

4.  Is there another question I should be = asking?

Any and all advice, comments, and suggestions are = welcomed.

TIA,
Jim

--------------------------------------------
Jim Way, Software Engineer
Datum Austin (Austron Inc.)
voice: 512.721.4170
fax  : 512.721.4087
email: jwayATaustronDOTcom (no spam please)
--------------------------------------------

------_=_NextPart_001_01BF7FA9.CEF68048-- From vxwexplo-errs@csg.lbl.gov Fri Feb 25 08:39:32 2000 From: Henrik Bergstrom Date: Fri Feb 25 08:39:35 PST 2000 Subject: Calling ipReceiveRtn() explicitly? Dear VxWorks users, I am trying to write a layer "between the IP-module and the END-driver" and in my current approach I do this by attaching a SNARF-protocol to the END-device in question. To me, this seems to be the only way of doing what I want to do, but if anyone has any better solutions, please don't hesitate to contact me. Now, over to my question. I would like to call ipReceiveRtn() explicitly to "simulate" incoming packets from a specific END-driver. I tried calling ipReceiveRtn() with all parameters except pSpare set to valid values but that generated an "ipProto: unknown device" error message so I guess the pSpare parameter is used by ipReceiveRtn(). Since pSpare is only available to the MUX through the muxBind() call it seems impossible for me to find out the correct value for pSpare. It of course has to be stored somewhere, but where? In target/h/end.h a structure called NET_PROTOCOL is declared which includes the pSpare element. Is this structure publically available? (The comments in end.h says it is part of "a singly linked list whose head is stored in a netDevice".) If anyone has any relevant information to share, please don't hesitate to send it to me. Regards, Henrik - -- Henrik Bergstrom +46-8-628 28 28 (phone) Software Engineer +46-8-628 64 14 (fax) Intertex Data AB henrik.bergstrom@intertex.se From vxwexplo-errs@csg.lbl.gov Fri Feb 25 10:36:05 2000 From: Denis BEURIVE Date: Fri Feb 25 10:36:08 PST 2000 Subject: Request for document about vxWorks driver Hi folks, I am searching (for a while now) documents that explain how to write drivers for vxWorks. I have contacted Wind River but I did not receive any answer. Does anybody know a book or a web site that contains information about vxWorks drivers ? Best regards, Denis BEURIVE _________________________________________________________ Do You Yahoo!? Get your free @yahoo.com address at http://mail.yahoo.com From vxwexplo-errs@csg.lbl.gov Fri Feb 25 23:00:38 2000 From: "A Supreeth Reddy" Date: Fri Feb 25 23:00:40 PST 2000 Subject: Tornado: WindView Tcl API library file Hi Folks, Any help for the following will be highly appreciated. I used the WindView application to collect data for calculating the CPU Usage and Memory Usage. But the "CPU Usage" option doesn't show any graph, rather it allows us to export the data to an "Excel Sheet". Anyway the excel sheet also wouldn't provide much information. So I found a solution to this problem in the WRS Frequently Asked Questions (FAQs) viz, TSR #133548 TITLE : Windview 2.0 CPU Usage Data Exported to Excel Does not Inc.... and TSR #130864 TITLE : How to use Event-Base TCL API to investigate an event log. Both the solutions provide a TCL script to be used to extract the CPU Usage from the .wvr file obtained from WindView. I encountered the following problem while using the tcl script. The 4th line of the tcl script contains load libwvapi.so When I do %source cpuUsage.tcl It immediately gives an error saying "libwvapi.so" is not found. I have extensively searched the whole of the Tornado II and WindView installation directories but couldn't find the library file. I use Tornado II on a WinNT machine. Can anybody help me in this regard? Thanks in advance, Supreeth. From vxwexplo-errs@csg.lbl.gov Sat Feb 26 04:03:13 2000 From: Vxworks Exploder Date: Sat Feb 26 04:03:16 PST 2000 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Sat Feb 26 04:03:09 PST 2000 Subject: Free source code for any RTOS ? Subject: Re: Purify or equivalent under vxSim ? Subject: Re: VxWorks vs. Linux Subject: Re: Free source code for any RTOS ? Subject: vmBaseGlobalMapInit() lengthy function? Subject: Re: semDelete error Subject: Finding file size. Subject: T1 to T2 Subject: 16C754 support Subject: C++ RTTI Overhead Subject: T2 Unix Man Pages, etc Subject: Re: semDelete error Subject: Re: semDelete error Subject: File corruption using DosFsLib Subject: Re: File corruption using DosFsLib Subject: Re: Using a standard off the shelf GDB for debugging Subject: Re: Finding file size. Subject: Re: File corruption using DosFsLib Subject: Re: 16C754 support Subject: TSFS Booting question Subject: Re: Free source code for any RTOS ? Subject: porting LAN(sonic-t: DP83936AV) to SA-1100 Subject: Job announcement for Germany Subject: Re: Is anyone use C++ with T2? Subject: How to put a telnet server into half-duplex mode? Subject: Re: why -mlongcall doesn't fix PPC32meg problem? Subject: Re: Making vxWorks_rom ? Subject: Re: semDelete error Subject: Re: Embedded Java OS by Esmertec ?? Subject: Re: Purify or equivalent under vxSim ? Subject: ostrstream performance problems Subject: Job Announcements - NASA/Cleveland Subject: How do I rebuild the VxWorks module library? Subject: Vxworks - fcntl() - tornado - GCC Subject: Re: using vxVMI with a pc486 BSP Subject: ifShow gives "interfacesMasksTbl" error for added End Driver Subject: Re: vmBaseGlobalMapInit() lengthy function? Subject: Re: VME interrupts interrupting VME interrupts Subject: What object of vxWorks can replacing the event of M$ Windows? Subject: IP on Vxworks 5.4 Subject: Re: GUIs for Embedded Systems using VxWorks Subject: Re: How do I rebuild the VxWorks module library? Subject: Re: IP on Vxworks 5.4 Subject: Re: Request for document about vxWorks driver Subject: Re: What object of vxWorks can replacing the event of M$ Windows? Subject: Re: VxWorks vs. Linux ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Free source code for any RTOS ? Date: Thu, 24 Feb 2000 17:40:27 -0800 From: vikram khandpur Organization: eur.cis.philips.com Message-ID: <38B5DD8B.6E89B2F9@blr.pin.philips.com> Hi, I have to write a basic RTOs for 8051 platform. Is there any free source code for a similar RTOs available for reference on the web ? thanks vikram --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Purify or equivalent under vxSim ? Date: Thu, 24 Feb 2000 13:32:59 +0100 From: Gabriel Lang Organization: Nortel Networks Message-ID: <38B524FB.E77CA546@europem01.nt.com> References: <38B24E31.24F4BCC1@europem01.nt.com> Paul Whicker wrote: > > If you use the Diab Data compilers and SDS Single Step you can achieve most of > what purify does, I think it's called RTA or something like that. > > You said embedded systems but you don't specify an architecture, I think > Diab/SDS are available for slightly less architectures than vxWorks but most > are covered. > > Paul > > In article <38B24E31.24F4BCC1@europem01.nt.com>, Gabriel Lang > wrote: > >Does anyone know if it's possible to use > >purify under vxWorks (or vxSim) and if not > >is there a similar product for embedded > >systems ? > > > >TTFN > > > >Gab. The host is on SunOS 5.6 and the Target on MPC860 Gab. - -- --------------------------- Newsgroups: comp.os.vxworks Subject: Re: VxWorks vs. Linux Date: Thu, 24 Feb 2000 09:54:51 -0800 From: John McClenny Organization: Posted via Supernews, http://www.supernews.com Message-ID: References: <87t4tl$55u$1@news2.kornet.net> <38AE0C80.2A354436@earthlink.net> In article <38AE0C80.2A354436@earthlink.net>, dblbrown@earthlink.net says... > $2995 gets you a seat of VxWorks. $50K buys you 10 seats of Tornado Power > Professional. Good luck on every getting out a product with a free OS unless > you have a group of engineers working 80 to 100 hours a week for months. > And that would be different from VxWorks how? --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Free source code for any RTOS ? Date: Thu, 24 Feb 2000 15:51:13 GMT From: Ted Dennison Organization: Deja.com - Before you buy. Message-ID: <893k1f$s29$1@nnrp1.deja.com> References: <38B5DD8B.6E89B2F9@blr.pin.philips.com> In article <38B5DD8B.6E89B2F9@blr.pin.philips.com>, vikram khandpur wrote: > I have to write a basic RTOs for 8051 platform. > Is there any free source code for a similar RTOs available for reference > on the web ? Sources for RTEMS is available in Ada and C at http://www.rtems.com/ and perhaps http://www.rtems.army.mil/rtems.html 11 supported platforms are listed, but 8051 isn't one of them. - -- T.E.D. http://www.telepath.com/~dennison/Ted/TED.html Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: vmBaseGlobalMapInit() lengthy function? Date: 24 Feb 2000 11:02:33 -0500 From: Bill Pringlemeir Organization: Factory of the mind Message-ID: Sender: bpringle@DeadDuck I am attempting to get an SA1110 board running with VxWorks. I noticed that the vmBaseGlobalMapInit(&sysPhysMemDesc[0], sysPhysMemDescNumEnt, TRUE) function in usrMMUinit.c takes a long time to execute (>30secs @150MHz). I have 32MB of SDRAM and 32MB of StrataFlash. Is the code allocating 1k pages for this memory or something like that? Is it possible to add 1-2 MB of flash and then add some more later? Or is this an unusually long. tia, Bill - -- Things should be described as simply as possible, but no simpler. - A. Einstein --------------------------- Newsgroups: comp.os.vxworks Subject: Re: semDelete error Date: Thu, 24 Feb 2000 16:56:02 -0000 From: "Stephen Walsh" Message-ID: <38b55fd4$1@pull.gecm.com> References: <892t02$d2s$1@news.etri.re.kr> Lee Sang Min wrote in message <892t02$d2s$1@news.etri.re.kr>... > >func_a() >{ > SEM_ID sem; > > sem = semBCreate( SEM_Q_FIFO, SEM_EMPTY ); > func_b( sem ); >} > >func_b( SEM_ID sem ) >{ > func_c( (int) sem ); The SEM_ID structure has just been cast to an int. This will mean you only pass the first int's worth of the SEM_ID struct. >} > >func_c( int sem ) >{ > SEM_ID mySem; > > mySem = (SEM_ID) sem; Here sem is only an int passed by value into the function. The copy info will be undefined as the information you want to copy is not available and you will get rubbish in the mySem structure. > semDelete( mySem ); >} > >Result: semDelete error - incorrect object ID > memPartFree - invalid block address > >Is this wrong? > To do it properly pass all instances of SEM_ID by reference and don't try to cast stuctures to int it will only cause you grief. Hope The Above Helps Steve ******************************************* Stephen A Walsh Senior Software Engineer BAE Systems ******************************************* Opinions are usually mine and mine alone ******************************************* --------------------------- Newsgroups: comp.os.vxworks Subject: Finding file size. Date: Thu, 24 Feb 2000 16:43:33 +0000 From: nsi5510000-Prout Organization: Lucent Technologies Message-ID: <38B55FB5.E0212666@lucent.com> Hi, Is there a method of finding the size in bytes of a file located on an nfs mounted file system? Thanks, Mark --------------------------- Newsgroups: comp.os.vxworks Subject: T1 to T2 Date: Thu, 24 Feb 2000 16:35:15 GMT From: fatihozcan@my-deja.com Organization: Deja.com - Before you buy. Message-ID: <893mk2$tvd$1@nnrp1.deja.com> Is there a manual such that it explains the steps moving from T1 to T2 in terms of changes in our source codes especially for IP related? thanks. Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: 16C754 support Date: 24 Feb 2000 17:35:28 GMT From: johan.borkhuis@peektraffic.nl Organization: Peek Traffic B.V. Keywords: 16C754,UART,VxWorks Message-ID: <893q50$kev$1@news1.xs4all.nl> Hello, We are looking for a driver for the 16C754. Can someone point me in the direction where I can find such a sio-driver? Groeten, Johan - -- o o o o o o o . . . ___________________________________ o _____ || Johan Borkhuis | .][__n_n_|DD[ ====_____ | johan@borksoft.xs4all.nl | >(________|__|_[_________]_|________________________________| _/oo OOOOO oo` ooo ooo 'o!o!o o!o!o` === VxWorks page: http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html === --------------------------- Newsgroups: comp.os.vxworks Subject: C++ RTTI Overhead Date: Thu, 24 Feb 2000 18:50:22 GMT From: normpgh@home.com Organization: Deja.com - Before you buy. Message-ID: <893uhb$4k8$1@nnrp1.deja.com> Hi, I was wondering if anyone out there has been using Runtime Type Information (RTTI) in a VxWorks C++ application. My question relates to the amount of overhead that is incurred by turning this feature on. I beleive that just enabling RTTI causes additional overhead within the construction of each object in the system. And that there is additional overhead if/when this information is actually used. But I have no way to quantify any performance degradation. If anyone has experience using RTTI under VxWorks, I would be very interested in hearing your opinion. Thanks in advance for any info you might provide. - NormPgh Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: T2 Unix Man Pages, etc Date: Thu, 24 Feb 2000 13:03:44 -0600 From: "R. L. Bob Irwin" Organization: Systems Interface Software Inc Message-ID: <38B58090.AA099BBB@systemsinterface.com> We've just installed T2.01 on a Solaris 2.7 host. Lost several days coz we weren't informed that it only installs under CDE interface, not OpenWindows. (a caution to others). Question 1: Cannot find any Unix Man pages for vxWorks library. Do they exist or can they be generated and installed? I've been spoiled in the past 6 years by being able to do a simple 'man -k keyword' and find all the matching manpages thru a small xterm window. Not thrilled with the html ref pages, you gotta know the function's name before getting the page. Searching is tedius. Question 2: Since we work on many different computers and OSes, Would prefer to use the DDD debugger gui with gdb talking to our remote PPC vxWorks targets. Anyone doing this currently? If so, would appreciate some basic directions in implementing it. Thanks, bob - -- R L Bob Irwin * bobi@systemsinterface.com * 281-484-6661 x26 * Houston TX --------------------------- Newsgroups: comp.os.vxworks Subject: Re: semDelete error Date: Thu, 24 Feb 2000 20:02:29 +0100 From: "Xavier.Montagne" Organization: Wanadoo, l'internet avec France Telecom Message-ID: <893vdm$svp$1@wanadoo.fr> References: <892t02$d2s$1@news.etri.re.kr> Yes, Stephen is right .... it's a problem of C ( or C++ because of the delete ) and not a problem about VxWorks ! The semDelete instruction allows the system to delete the SEMAPHORE object, but with your exemple, you don't pass the object but just the ID of an object.... and the compiler is waiting for the memory allocation of the object you want to delete !! Of course you haven't anyone .... because semBCreate was done for the variable "sem", and the job of semBCreate is to prepare the memory space for the variable ... so what is possible follows: func_a() { SEM_ID sem; sem = semBCreate( SEM_Q_FIFO, SEM_EMPTY ); func_delete( sem ); } func_delete( SEM_ID& mySem ) { semDelete( &mySem ); } In C++ you have to put the "&" before the variable you want to delete by reference.... so I think it must be the same with semDelete....! Bye for now Xavier "Lee Sang Min" a écrit dans le message news: 892t02$d2s$1@news.etri.re.kr... > > func_a() > { > SEM_ID sem; > > sem = semBCreate( SEM_Q_FIFO, SEM_EMPTY ); > func_b( sem ); > } > > func_b( SEM_ID sem ) > { > func_c( (int) sem ); > } > > func_c( int sem ) > { > SEM_ID mySem; > > mySem = (SEM_ID) sem; > semDelete( mySem ); > } > > Result: semDelete error - incorrect object ID > memPartFree - invalid block address > > Is this wrong? > > > > > > > > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: semDelete error Date: Thu, 24 Feb 2000 13:01:12 -0600 From: Andrew Johnson Organization: APS, Argonne National Laboratory Message-ID: <38B57FF8.7270C09E@aps.anl.gov> References: <892t02$d2s$1@news.etri.re.kr> Lee Sang Min wrote: > > func_a() > { > SEM_ID sem; > > sem = semBCreate( SEM_Q_FIFO, SEM_EMPTY ); > func_b( sem ); > } > > func_b( SEM_ID sem ) > { > func_c( (int) sem ); > } > > func_c( int sem ) > { > SEM_ID mySem; > > mySem = (SEM_ID) sem; > semDelete( mySem ); > } > > Result: semDelete error - incorrect object ID > memPartFree - invalid block address if (sem == NULL) ? - - Andrew - -- Complexity comes for free, Simplicity you have to work for. --------------------------- Newsgroups: comp.os.vxworks Subject: File corruption using DosFsLib Date: Thu, 24 Feb 2000 14:20:13 -0500 From: "John R." Message-ID: I'm using dosFs and the out of the box ATA driver. I'm getting intermittent file corruption. The corruption seems to be related to broken cluster chains or clusters out of range. Has anyone experienced this? Any possible explanations? Thanks - -John --------------------------- Newsgroups: comp.os.vxworks Subject: Re: File corruption using DosFsLib Date: 24 Feb 2000 19:40:34 GMT From: johan.borkhuis@peektraffic.nl Organization: XS4ALL Internet BV Message-ID: <8941fi$rrt$1@news1.xs4all.nl> References: John R. wrote: : I'm using dosFs and the out of the box ATA driver. I'm getting intermittent : file corruption. The corruption seems to be related to broken cluster chains : or clusters out of range. Have you tried DosFS 2.0? Groeten, Johan - -- o o o o o o o . . . ___________________________________ o _____ || Johan Borkhuis | .][__n_n_|DD[ ====_____ | johan@borksoft.xs4all.nl | >(________|__|_[_________]_|________________________________| _/oo OOOOO oo` ooo ooo 'o!o!o o!o!o` === VxWorks page: http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html === --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Using a standard off the shelf GDB for debugging Date: Thu, 24 Feb 2000 19:55:45 GMT From: Robert Klinkhammer Organization: Road Runner Message-ID: <38B58FE2.F85388ED@mediaone.net> References: <88v71l$oov$1@nnrp1.deja.com> Reply-To: klink@rmediaone.net yes. I believe that "target vxworks host" is supported. hjhall@my-deja.com wrote: > > Has anyone used an off the shelf GDB to debug with? I am trying to > debug a VxSim NT target and it always SIGSEGV's (seg fault), it doesn't > do it if I use the T2 gdb. For many reasons I want to use an off the > shelf GDB and not the one supplied by T2. Does anyone know why this is > happening? > > Thanks > > -Howard > > Sent via Deja.com http://www.deja.com/ > Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Finding file size. Date: Thu, 24 Feb 2000 11:56:53 -0800 From: Ken Norlin Organization: NASA Dryden FRC Message-ID: <38B58D05.FB86A1E1@dfrc.nasa.gov> References: <38B55FB5.E0212666@lucent.com> nsi5510000-Prout wrote: > > Hi, > > Is there a method of finding the size in bytes of a file located on an nfs > mounted file system? > > Thanks, > > Mark Hi Mark, Have you already tried 'stat()' or 'fstat()'? Ken N. - -- Ken Norlin --- Simulation Engineer NASA Dryden Flight Research Center ken.norlin@dfrc.nasa.gov (661) 258-2046 FAX (661) 258-2792 --------------------------- Newsgroups: comp.os.vxworks Subject: Re: File corruption using DosFsLib Date: Thu, 24 Feb 2000 15:28:37 -0500 From: "John R." Message-ID: <5rgt4.1867$zn4.1593@client> References: <8941fi$rrt$1@news1.xs4all.nl> I can not migrate to DosFS 2.0 at this time. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: 16C754 support Date: Thu, 24 Feb 2000 12:11:49 -0800 From: Ken Norlin Organization: NASA Dryden FRC Message-ID: <38B59085.ACED516A@dfrc.nasa.gov> References: <893q50$kev$1@news1.xs4all.nl> johan.borkhuis@peektraffic.nl wrote: > > Hello, > > We are looking for a driver for the 16C754. Can someone point me in the > direction where I can find such a sio-driver? > Hi, SBS Greensprings uses the 16C754 on their IP-OctalPlus-422 -- they sell a VxWorks driver for their IP. And Radstone uses the UART on their PMC654. So you could buy their drivers & adapt them to your needs. I got the 16C754 manual from the Exar web site & coded my own 'driver' for a simple asynchronous interface using the Greenspring IP. Hope that helps, Ken N. - -- Ken Norlin --- Simulation Engineer NASA Dryden Flight Research Center ken.norlin@dfrc.nasa.gov (661) 258-2046 FAX (661) 258-2792 --------------------------- Newsgroups: comp.os.vxworks Subject: TSFS Booting question Date: Fri, 25 Feb 2000 08:20:18 +0800 From: "Zhang Zhiqun" Organization: The news server of Changzhou TeleCom Message-ID: <894hou$2cub$1@news.cz.js.cn> Hi I do some development based on the PC486 target and Tornado II Windows Host. I want to use the target's COM2 to connect to the target-server. So, I put the following code into the PC486 BSP's config.h: #undef WDB_COMM_TYPE #undef WDB_TTY_CHANNEL #undef WDB_TTY_BAUD #define WDB_COMM_TYPE WDB_COMM_SERIAL /* default path is UDP sockets */ #define WDB_TTY_CHANNEL 1 /* default UDPL_SLIP channel */ #define WDB_TTY_BAUD 115200 #define INCLUDE_TSFS_BOOT In the command line, we make the vxWorks. It can connect to the target-server successfully. I create one bootable image based on the above BSP in the Tornado IDE and build the vxWorks. It can execute but can not connect to the target-server. The case is the target-server reloading automaticly after the vxWorks loading in the target. Why? Thanks and best regards. Zhang Zhiqun (Bula Cheung) bula@guoguang.com.cn Guoguang Electronic Group Co. http://www.guoguang.com.cn 93 Qingtan Road, Changzhou, Jiangsu, 213015, China Tel: +86 519 696 7935 ext. 2372 Fax: +86 519 696 9104 --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Free source code for any RTOS ? Date: Thu, 24 Feb 2000 21:02:37 -0800 From: "DuckE" Message-ID: <38b60c47.0@news.pacifier.com> References: <38B5DD8B.6E89B2F9@blr.pin.philips.com> Reply-To: "DuckE" Try posting this same question on comp.realtime. I recall seeing reference to an RTOS for the 8051 there, but didn't bookmark the location. SteveD vikram khandpur wrote in message news:38B5DD8B.6E89B2F9@blr.pin.philips.com... > Hi, > I have to write a basic RTOs for 8051 platform. > Is there any free source code for a similar RTOs available for reference > on the web ? > > thanks > vikram > > > --------------------------- Newsgroups: comp.os.vxworks Subject: porting LAN(sonic-t: DP83936AV) to SA-1100 Date: Fri, 25 Feb 2000 15:04:10 +0900 From: "sj lee" Organization: Electronics and Telecommunications Research Institute (ETRI) Message-ID: <89566a$34t$1@news.etri.re.kr> Hi experts! My vxWorks at strongARM board will use LAN. But my LAN chips(DP83936AV or DP83932) needs additional coding at BSP and wiring with strongARM is not simple for control signal. Do you have successful experience? Thank you. --------------------------- Newsgroups: comp.os.vxworks Subject: Job announcement for Germany Date: Fri, 25 Feb 2000 07:50:43 +0100 From: "Achim Zimmer" Organization: Colt Telecom GmbH Message-ID: <8958l9$s68$1@crusher.de.colt.net> Wir sind eine stark wachsende internationale Unternehmensgruppe im Bereich Telekommunikation. Für dieses Wachstum bauen wir auch unseren Entwicklungsbereich aus und suchen ab sofort Software-Entwicklungsingeneure für unseren Standort Dresden Beschreibung: Unsere Entwickler arbeiten an innovativen Lösungen für die Sprach-Daten-Integration sowie Sicherheit der Internet-Nutzung im Business-Umfeld. Ihre Aufgaben umfassen - - Mitarbeit bei der Erstellung von Anwendungskonzepten und Programmablaufplänen sowie deren hardwarenahe Umsetzung in ISDN- und IP-Netzwerken - - Softwareentwicklung am PC mit simulierter Hardware-Umgebung und/oder Zielhardware - - Entwicklung geeigneter Testmethoden unter Einsatz modernster Methoden und Tools - - Pflege und Weiterentwicklung bestehender Systeme Anforderungen: - - Kenntnisse in den wesentlichen Internet/Intranet-Technologien - - UNIX/LINUX Erfahrung - - Erfahrung in führenden Multitasking/Echtzeitbetriebssystemen z.B. in VxWorks und HW-Plattformen (Embedded Prozessoren) - - Programmiererfahrung in C, C++ - - Grundlegende Kenntnisse moderner HW-Plattformen - - Konzeptionelle und anwenderorientierte Denkweise - - Gute Englischkenntnisse in Wort und Schrift - - Teamfähigkeit, Spaß an der Arbeit und Sense of Humor Eine außergewöhnlich gute Arbeitsatmosphäre durch flache Hierarchien und große Freiheitsgrade stellen die Basis unseres jungen und sich dynamisch entwickelnden Unternehmens dar. Sowohl unsere Internationalität als auch unser Wachstum bieten weitere Entwicklungmöglichkeiten. Wenn Sie sich gewappnet fühlen, an der Entwicklung modernster Kommunikationslösungen für eine weltweite Anwendung mitzuarbeiten und gegebenenfalls auch Verantwortung zu übernehmen, dann senden Sie bitte Ihre aussagekräftigen Bewerbungsunterlagen unter Angabe Ihrer Gehaltsvorstellung und des frühestmöglichen Eintrittstermins an DICA Technologies AG z.H. Roland Geissler St. Petersburger Str. 15 Germany - 01069 Dresden oder per email an: rgeissler@dica.de - -- Thanks Achim Zimmer R&D DICA Technologies Have a nice day ! --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Is anyone use C++ with T2? Date: Thu, 24 Feb 2000 08:18:42 +0100 From: "Mario Semo" Organization: Nextra Telekom GmbH Message-ID: <8958v8$5md$1@newsmaster01.nextra.at> References: <88p7q5$oc51@nntp.cig.mot.com> <88rtr2$rfg$1@newsmaster01.nextra.at> Hello, >-frepo was broken but is now patched. However, there are still problems with >linking across different drives on WIN32 hosts and linking with the BSP is a >real trial if you don't want multiple definitions of some template symbols. a) i assume you mean SPR 29880 + patch for it. this is absolutly required. rya. b) there are still samples which compiles fine with other compilers. sample: sample: template class bar { public: void foobar() { FOO::Counter i; } }; class X { public: typedef int Counter; }; int main(int argc,char *argv[]) { bar bx; return 0; } fails to compile. c) -frepo is not supported on simpc (vxsim) d) what are the propblems in cross-drive template resolution? >You may have problems with C++ exceptions especially when porting from >T1.0.1. The compiler defaults to compiling in C++ exception handling. >There's an SPR #29680 on the fact that this causes intermittent crashes at >interrupt time...(the result of an interrupt time __get_eh_context call). >The rule is unless you need C++ exceptions compile with -fno-exceptions. If >you do need exceptions you'll have to take care to not call constructors at >interrupt time. uh. in which situations do i call a ctor at interrupt time? e.g. in a watchdog callback? anywhere else? when i just have "ordinary" code which waits on sema4, ... ? there is another problem with exceptions which makes me very unhappy: (pseudocode) public.h: extern void foo(); class ExceptionType {}; file1.cpp main() [ try { foo(); } catch (ExceptionType &exc) { printf 1 } catch(...) { printf 2 } } file2.cpp void foo() { ExceptionType aExc; throw aExc; } this works ok when file1.cpp and files2.cpp are compiled and then bound together into one binary image. but now assume they are compiled and then loaded each. first load file2.out and then file1.out. then a) on VxSim : still everything is ok. i see "1" printed. on other HW (real VxWorks) i see "2" printed. > >I think that's it, the debugger is much better than in T1.0.1, as is the >compiler (much faster, more errors). I see -O1 and -O2 as absolutly unuseable. i have about 10 TRAPS of the compiler (creating coredumps) with -O1 and -O2. compilers are SimPC, PPC, X86. --------------------------- Newsgroups: alt.winsock.programming,comp.os.vxworks,comp.protocols.tcp-ip Subject: How to put a telnet server into half-duplex mode? Date: Fri, 25 Feb 2000 09:04:56 GMT From: nedren@yahoo.com Organization: Deja.com - Before you buy. Message-ID: <895gjm$art$1@nnrp1.deja.com> Reply-To: nedren@yahoo.com Hi, I'm trying to write a simple interface to a telnet server. Basically what I want is to get lines from the server and then process the lines and then send something back. For each of the negotiation parameters the server sent me, I'm responding WON'T for DO, and DON'T for WILL. I would like the server to send me a "Go Ahead" when it finishes sending, so I can then start sending my stuff. How would I be able to put the server into this half-duplex mode? After I send all these WON'T and DON'Ts, the server responded something like but then it's still not sending out go ahead at the end! Also, another question, am I suppose to send a Go ahead at the end of each negotiation string I send out, and send a LF + CR at end of each ASCII Data stream? Thank you!! - Ned - Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: why -mlongcall doesn't fix PPC32meg problem? Date: Fri, 25 Feb 2000 09:26:49 GMT From: Computer Engineer Organization: Somewhere in the Earth Message-ID: <38B5DA24.D84207A7@netscape.net> References: <200002231931.LAA20321@altosax.spk.hp.com> Hello Tom, The PPC relocation problem has been fixed in Tornado 2.0. CE --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Making vxWorks_rom ? Date: Fri, 25 Feb 2000 09:25:38 GMT From: Computer Engineer Organization: Somewhere in the Earth Message-ID: <38B5D9DD.2247815@netscape.net> References: <893042$1bh$1@schbbs.mot.com> Hi Cyrille, Known problem. SPR#8845. Contact WindRiver support for updated elfToBin tool. CE --------------------------- Newsgroups: comp.os.vxworks Subject: Re: semDelete error Date: Fri, 25 Feb 2000 09:13:55 GMT From: john_94501@my-deja.com Organization: Deja.com - Before you buy. Message-ID: <895h4h$b82$1@nnrp1.deja.com> References: <892t02$d2s$1@news.etri.re.kr> There is nothing wrong with your code! It runs just fine for me under the simulator... I added #include "semLib.h", compiled it, "downloaded" it to the simulator under T2 and called func_a from the shell. It worked just fine! To comment on some of the other replies here... 1) A SEM_ID is only a 32 bit value (in fact, a pointer to the data structure representing the semaphore). Although not perhaps a sensible thing to do, you may cast it to an int without any fear of losing data. 2) C++ is not relevant here... semDelete() and the C++ delete operator are not related! Now, why do you get an error? Well, one possibility is that the semBCreate() failed and you actually do not have a semaphore to delete. In that case the ID you pass will be NULL (as another posting stated). Another possibility is that something else on your target is corrupting the memory for the task executing the code (esp. its stack). Finally, if this code is running as part of a task doing more things, you could have overflowed the stack of that task... TTFN, John... Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Embedded Java OS by Esmertec ?? Date: Fri, 25 Feb 2000 10:17:59 GMT From: pwhicker@home.net (Paul Whicker) Organization: @Home Network Message-ID: References: <88v41k$3vc@news-central.tiac.net> Good question. With TurboJ and fully compiled Java already available I don't suppose it would just be for speed. How complete is the underlying support? Just running Java won't replace all the other things an o/s does. I guess I'd use compiled Java for non-graphical and TurboJ for graphical apps. Then again by based on Java do you mean written in Java? With compiled Java you could as soon code the o/s in that as anything else. Too little information. Paul In article <88v41k$3vc@news-central.tiac.net>, "Eric Riseman" wrote: >How does the Esmertec Java RTOS differ from HP Chi or Sun's EJava ? > >How much sense does a hard, real time Java OS make ? > >Why would someone buy an RTOS based on Java instead of VxWorks or OSE ? > >What kind of company would be an ideal candidate for a >JAVA RTOS ? > >Please resond to: Eric@RapidLogic.com > >Thanks in Advance !! >Eric > > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Purify or equivalent under vxSim ? Date: Fri, 25 Feb 2000 10:25:08 GMT From: pwhicker@home.net (Paul Whicker) Organization: @Home Network Message-ID: <8Mst4.1577$so2.31988@news1.frmt1.sfba.home.com> References: <38B24E31.24F4BCC1@europem01.nt.com> <38B524FB.E77CA546@europem01.nt.com> I know the 860 is supported but I'm not certain which versions of SunOS/Solaris are supported. Try www.diab.com or ww.sdsi.com (I think). Paul In article <38B524FB.E77CA546@europem01.nt.com>, Gabriel Lang wrote: >Paul Whicker wrote: >> >> If you use the Diab Data compilers and SDS Single Step you can achieve most > of >> what purify does, I think it's called RTA or something like that. >> >> You said embedded systems but you don't specify an architecture, I think >> Diab/SDS are available for slightly less architectures than vxWorks but most >> are covered. >> >> Paul >> >> In article <38B24E31.24F4BCC1@europem01.nt.com>, Gabriel Lang >> wrote: >> >Does anyone know if it's possible to use >> >purify under vxWorks (or vxSim) and if not >> >is there a similar product for embedded >> >systems ? >> > >> >TTFN >> > >> >Gab. > > >The host is on SunOS 5.6 and the Target on MPC860 > >Gab. --------------------------- Newsgroups: comp.os.vxworks Subject: ostrstream performance problems Date: Fri, 25 Feb 2000 11:21:56 GMT From: dixon@crd.ge.com Organization: Deja.com - Before you buy. Message-ID: <895oki$ft2$1@nnrp1.deja.com> I'm working with a colleague at a remote location to troubleshoot a performance problem. The real software is quite complicated, but he has come up with a simple example that also exhibits the same behavior. Here's our basic loop: ostrstream* strId[5000]; for(i=0; i<5000; i++) { strId = new ostrstream; *(strId[i]) << "string" << "string2" << ... << "stringn"; } We time the new and the insertion. The code as written causes quite a bit of heap fragmentation, and I understand the impact of the fragmentation. We've tried to eliminate the fragementation effects using an alternate constructor: ostrstream* strId[5000]; char* buffer[5000]; for(i=0; i<5000; i++) { buffer[i] = new char[256]; strId[i] = new ostrstream(buffer[i], 256); *(strId[i]) << "string" << "string2" << ... << "stringn"; } Again we time the individual news and the insertion. Very repeatedly the time to do the insertion doubles at iteration 1000 and remains at the larger value for the rest of the test. I have not been able to reproduce this behavior locally, but I do not have an identical configuration. Does anyone have any suggestions as to what might be going on? TIA standard disclaimers apply Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Job Announcements - NASA/Cleveland Date: Fri, 25 Feb 2000 08:45:59 -0500 From: "Dave Hall" Organization: NASA Lewis Research Center Message-ID: <89616f$adh$1@sulawesi-fi.lerc.nasa.gov> NEEDED: TWO SENIOR LEVEL ENGINEERS FOR AN EXCITING NASA PROJECT IN CLEVELAND OHIO. Dynacs Engineering is a World Class aerospace contractor. We are developing a unique Microgravity research facility under contract with the NASA Glenn Research center. The Fluids and Combustion Facility is scheduled to fly on the International Space Station (US Lab) in the near future. We have the following openings for talented engineers who would like to put their mark on this historic project: POSITION 1: Real Time SW/System Architect. Able to analyze top level requirements, perform tradeoff studies and synthesize a real time, embedded system architecture with minimum supervision. Familiar with image processing, data acquisition and object oriented design tools. Preferred skills include VxWorks, Artisan, Solaris, C++, CORBA. BSCE or BSEE required. MS preferred. Five or more years of directly related experience. Background in formal software engineering methods (CMM Level 2/3, ISO9001, Mil-Std-2167, etc). POSITION 2: Real Time Software Engineer. Able to design, code and test VxWorks applications in C++, including device drivers. Expert in the use of Object Oriented design tools/methods to build real-world computer interfaces. Familiar with embedded data acquisition/control hardware technology. BSCS, BSEE or BSCE. Advanced degrees favored. 4-10 Years of directly related experience. Please forward resumes to: David Hall Dynacs Engineering Inc. 2001 Aerospace Parkway Brookpark, OH 44142 Email resumes to: David.Hall@grc.nasa.gov (Plain Text Only!) Visit us on the World Wide Web at http://www.dynacs.com. NASA maintains a web site devoted to our project at http://fcf.grc.nasa.gov. --------------------------- Newsgroups: comp.os.vxworks Subject: How do I rebuild the VxWorks module library? Date: Fri, 25 Feb 2000 14:57 +0000 (GMT Standard Time) From: rew@cp.cix.co.uk (Richard Wenham) Organization: CIX - Compulink Information eXchange Message-ID: Reply-To: rew@cp.cix.co.uk I need to modify the serial driver, i8250Sio.c. Having done this, how do I get it compiled, and replaced in target\lib\libI80486gnuvx.a? Making the BSP VxWorks image in the normal way does not spot the change to i8250Sio.c, and I cannot find specific instructions on rebuilding the library. Issuing a make from target\src\user results in a library which is about 600Kb instead of 2 Mb, with many undefined symbols when building the BSP. No response from WRS, so please can anybody point me in the right direction? Tornado 1.0.1, Windows NT host, PC486 target. Thanks, Richard Wenham Cranfield Precision --------------------------- Newsgroups: comp.os.vxworks Subject: Vxworks - fcntl() - tornado - GCC Date: Fri, 25 Feb 2000 17:31:15 +0100 From: "Gaël Trinquart" Organization: UUNET Message-ID: <896anv$c1m$3@reader1.fr.uu.net> It seems that the fcntl() function is not implemented on GCC. Does anyone have the source code from fcntl() ? --------------------------- Newsgroups: comp.os.vxworks Subject: Re: using vxVMI with a pc486 BSP Date: Fri, 25 Feb 2000 12:44:47 -0800 From: Harvey Taylor Organization: Organization? What organization?! Message-ID: <38B6E9BF.5669@despam.pangea.ca> References: <896dc7$q2o2@overload.lbl.gov> teit.uldal@philips.com wrote: > We are trying to use the vxVMI option for an x86 target - but don't seem to > be able to get it working under Tornado 1.0.1. > Both when INCLUDE_MMU_BASIC and INCLUDE_MMU_FULL are defined, > the downloaded image hangs when calling 'usrMmuInit'. > Two questions. What CPU are you using? [If it is a P2, usrMmuInit might be _very_ slow.] What is in your sysPhysMemDesc array [in sysLib.c]? - -het - --- "Assembly of Japanese bicycle require great peace of mind." - Instruction manual mentioned by R. Pirsig Harvey Taylor mailto:het@despam.portal.ca http://www.pangea.ca/~het --------------------------- Newsgroups: comp.os.vxworks Subject: ifShow gives "interfacesMasksTbl" error for added End Driver Date: Fri, 25 Feb 2000 18:57:16 GMT From: flyguy99@my-deja.com Organization: Deja.com - Before you buy. Message-ID: <896jab$3fh$1@nnrp1.deja.com> T2,target PPC860 custom board with motCpmEnd driver installed and working. I need to add a second End driver, so I started with motCpmEnd and stripped out the SCC stuff (this will not be a SCC driver). I get endLoad(), endStart, endIoctl() calls just like the Cpm driver. At the host shell I do "ifShow" and see the Cpm and lo drivers. muxShow displays: Device: cpm Unit: 0 Description: MPC860 Power-QUICC Enhanced Network Driver Protocol: IP 4.4 ARP Type: 2054 Recv 0xacf78 Shutdown 0xad23c Protocol: IP 4.4 TCP/IP Type: 2048 Recv 0xacf78 Shutdown 0xad174 Device: mpe Unit: 0 Description: PCI End Protocol: IP 4.4 ARP Type: 2054 Recv 0xacf78 Shutdown 0xad23c Protocol: IP 4.4 TCP/IP Type: 2048 Recv 0xacf78 Shutdown 0xad174 value = 0 = 0x0 ipAttach 0,"mpe" (my device name) succeeds, but then ifShow displays the cpm and lo devices followed by: can't read "interfacesMasksTbl(0x103cf4)": no such element in array ifAddrSet also fails. I have looked at the endObject and all seems to be OK as far as I can tell. Any ideas what is happening? Tim Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: vmBaseGlobalMapInit() lengthy function? Date: Fri, 25 Feb 2000 13:49:28 -0600 From: Andrew Johnson Organization: APS, Argonne National Laboratory Message-ID: <38B6DCC8.9E3D687B@aps.anl.gov> References: Bill Pringlemeir wrote: > > I am attempting to get an SA1110 board running with VxWorks. > I noticed that the vmBaseGlobalMapInit(&sysPhysMemDesc[0], > sysPhysMemDescNumEnt, TRUE) function in usrMMUinit.c takes a > long time to execute (>30secs @150MHz). I have 32MB of SDRAM > and 32MB of StrataFlash. Is the code allocating 1k pages for > this memory or something like that? It's creating page table entries (or whatever they're called on the SA arch) for all the memory you've got defined in sysPhysMemDesc. Your 64MB regions shouldn't take anywhere near that long to initialize though, have a look at all the entries you've defined and see if you can reduce the size of any, and in particular don't just define any large regions "in case you'll need it". On VME boards a big mistake is to declare most of the A32 address space this way, I think the memory overhead is something like 1K per MB on the 68K family and it adds long delays to startup. - - Andrew - -- Complexity comes for free, Simplicity you have to work for. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: VME interrupts interrupting VME interrupts Date: Fri, 25 Feb 2000 20:02:50 GMT From: John Fusco Organization: Deja.com - Before you buy. Message-ID: <896n59$64l$1@nnrp1.deja.com> References: <38ABFAA6.F13EA759@mail.ford.com> I would recommend getting a copy of the Universe II user's manual from www.tundra.com if you haven't already. There is a good discussion of interrupts and variable latency. http://www.tundra.com/fixed_docs/UniIIMan.pdf From what I have read, the UII will complete the IACK before the PCI interrupt is generated. The manual says that it is possible for a higher priority interrupt to miss preempting a lower priority one. There is a SW IACK on the UII, but I have never used it. It's not clear to me how that would work in this situation. In article <38ABFAA6.F13EA759@mail.ford.com>, "Connolly, F. T. (Francis)" wrote: > > --------------C1C695F96F856CA3E32F7A26 > Content-Type: text/plain; charset=us-ascii > Content-Transfer-Encoding: 7bit > > This question has to do with a higher priority VME interrupt interrupting a > lower priority one. Preempting here means that if a lower priority ISR > (interrupt service routine) is running, the higher priority one will "interrupt" > it, finish its ISR, and then the lower priority interrupt's ISR will continue. > > We are VxWorks "users", not "developers". Or, our application is a laboratory > engine control and instrumentation system that is VME and VXI with all hand > written C; we are not writing BSPs. However, we are not totally ignorant; we do > know a bit about VxWorks and its inner workings. Also, the main motivation > behind all this is that we have one interrupt for which we would like to keep > the latency "constant", which is the case on a 68k board. We are trying to get > this to work on a PPC board, on which the interrupt latency happens to not be > "constant". > > We are using VxWorks 5.3.1 (and 5.2 and even 5.1), but not really using Tornado > 1.?, since we use "5.2 mode" with the shell on the target and a startup script > or whatever it's called; we do it the old fashioned way. > > The VME specification prioritizes the IRQ levels (and then the daisy chain > prioritizes interrupters on the same level), and says that if two interrupts > occur simultaneously, the higher priority one will be serviced first. I haven't > found any spec. for what should happen if a higher priority interrupt is > asserted while a lower priority interrupt's ISR is executing. > > The context of the question is the situation on a 68k board versus a PPC (604) > board that uses a Universe II VME-PCI interface. On the 68K board a higher > priority interrupt does preempt a lower priority one (as observed using some > hardware signals), while on the particular PPC board, if the lower priority > interrupt is running while the higher priority one is asserted, the higher > priority one does not execute until the lower priority ISR has completed. > > We understand the concept of why this is so; the 68k interrupts _are_ > prioritized, while on the PPC BSP the Universe II uses one PCI interrupt for all > VME interrupts. Some people have said that one might be able to remap one or > more of the VME interrupts to another PCI interrupt. We can see how this might > happen (we have the Universe II user manual, among other info). > > Without enough detailed knowledge, it is hard to tell how difficult remapping an > interrupt would be, and if one would even want to do it (i.e., modify the BSP > for one particular situation). > > So, the question more or less is if anyone has any similar experience with, or > comments on this situation. What I would really like to know is if anyone knows > if we have any hope of readily doing this, or if it would get really ugly. I > don't wish to spend a lot of (or, rather, more) time on this. The vendor is > already (apparently) pondering the question as well. Thanks for any assistance > or comments. > > -- > Francis Connolly Technical Specialist > FMC-FAO-PD-R&VT-P&AE-CAPE-APPTE-CEFET-P/T ET > 313-594-4081 office (1115 GTL) fconnoll@ford.com > 313-317-7404 fax CDS ID: fconnoll > > --------------C1C695F96F856CA3E32F7A26 > Content-Type: text/html; charset=us-ascii > Content-Transfer-Encoding: 7bit > > > > This question has to do with a higher priority VME interrupt interrupting > a lower priority one. Preempting here means that if a lower priority ISR > (interrupt service routine) is running, the higher priority one will "interrupt" > it, finish its ISR, and then the lower priority interrupt's ISR will continue. >

We are VxWorks "users", not "developers". Or, our application is a laboratory > engine control and instrumentation system that is VME and VXI with all > hand written C; we are not writing BSPs. However, we are not totally ignorant; > we do know a bit about VxWorks and its inner workings. Also, the main motivation > behind all this is that we have one interrupt for which we would like to > keep the latency "constant", which is the case on a 68k board. We are trying > to get this to work on a PPC board, on which the interrupt latency happens > to not be "constant". >

We are using VxWorks 5.3.1 (and 5.2 and even 5.1), but not really using > Tornado 1.?, since we use "5.2 mode" with the shell on the target and a > startup script or whatever it's called; we do it the old fashioned way. >

The VME specification prioritizes the IRQ levels (and then the daisy > chain prioritizes interrupters on the same level), and says that if two > interrupts occur simultaneously, the higher priority one will be serviced > first. I haven't found any spec. for what should happen if a higher > priority interrupt is asserted while a lower priority interrupt's ISR is > executing. >

The context of the question is the situation on a 68k board versus a > PPC (604) board that uses a Universe II VME-PCI interface. On the 68K board > a higher priority interrupt does preempt a lower priority one (as observed > using some hardware signals), while on the particular PPC board, if the > lower priority interrupt is running while the higher priority one is asserted, > the higher priority one does not execute until the lower priority ISR has > completed. >

We understand the concept of why this is so; the 68k interrupts _are_ > prioritized, while on the PPC BSP the Universe II uses one PCI interrupt > for all VME interrupts. Some people have said that one might be able to > remap one or more of the VME interrupts to another PCI interrupt. > We can see how this might happen (we have the Universe II user manual, > among other info). >

Without enough detailed knowledge, it is hard to tell how difficult > remapping an interrupt would be, and if one would even want to do > it (i.e., modify the BSP for one particular situation). >

So, the question more or less is if anyone has any similar experience > with, or comments on this situation. What I would really like to know > is if anyone knows if we have any hope of readily doing this, or if it > would get really ugly. I don't wish to spend a lot of (or, rather, more)  > time on this. The vendor is already (apparently) pondering the question > as well. Thanks for any assistance or comments. > 

-- 
> Francis
Connolly          &nbs
p; Technical Specialist
> FMC-FAO-PD-R&VT-P&AE-CAPE-APPTE-CEFET-P/T ET   
> 313-594-4081 office (1115 GTL) fconnoll@ford.com
> 313-317-7404
fax           &nb
sp;    CDS ID: fconnoll
>   > > --------------C1C695F96F856CA3E32F7A26-- > > Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: What object of vxWorks can replacing the event of M$ Windows? Date: Fri, 25 Feb 2000 09:33:13 +0800 From: "joke" Organization: The news server of Changzhou TeleCom Message-ID: <894m3r$2hq8$1@news.cz.js.cn> Hello , Can you tell me what object of vxWorks can replacing the event object of M$ Windows or how to implement its in vxWorks? Regard, Haoyu Tang --------------------------- Newsgroups: comp.os.vxworks Subject: IP on Vxworks 5.4 Date: Sat, 26 Feb 2000 00:39:22 GMT From: gardhab@my-deja.com Organization: Deja.com - Before you buy. Message-ID: <8977bq$hu3$1@nnrp1.deja.com> hi, i have a few questions on VxWorks' IP capability - a. Does VxWorks support IpV6 (ie 64 bit IP addressing) ? b. Does VxWorks support IP aliasing? tks, karthik Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: GUIs for Embedded Systems using VxWorks Date: Sat, 26 Feb 2000 01:15:57 GMT From: snaphook@southwesternbell.net (Bruce) Organization: SBC Internet Services Message-ID: <38b72938.10118672@news.swbell.net> References: <896dc8$q2o3@overload.lbl.gov> In comp.os.vxworks "Fitch, Chris" wrote: >We are trying to find a replacement for our current method of building GUIs: >RTX Windows and an X-Windows based GUI Builder. What other products are >available? For additional background: our system employs a Power PC Single >Board Computer with an attached Video Card. Look at www.ease.com for their VxWorks port of Metawindows --------------------------- Newsgroups: comp.os.vxworks Subject: Re: How do I rebuild the VxWorks module library? Date: Sat, 26 Feb 2000 09:05:46 GMT From: pwhicker@home.net (Paul Whicker) Organization: @Home Network Message-ID: References: Cranfield Bedfordshire? I used to live in Bedford and cycle through there. Try "make CPU=80486 TOOL=gnu", I think that should build the object and add it to the archive... however that requires a certain amount of bravery. I normally don't do this, I just copy the source file to my BSP directory and then add the object to MACH_EXTRA in the Makefile. That should do it. Once you are confident you can use ar386 to replace the module in the library, check the manual but I think the archive command will be something like ar386 -rv lib80486gnuvx.a i8250Sio.o I think that should do it. Most of the WR filed engineers seem to be missing this week, I think they have some sort of meeting, I guess that's dumping everything on support. Paul In article , rew@cp.cix.co.uk wrote: >I need to modify the serial driver, i8250Sio.c. Having done this, how do >I get it compiled, and replaced in target\lib\libI80486gnuvx.a? Making >the BSP VxWorks image in the normal way does not spot the change to >i8250Sio.c, and I cannot find specific instructions on rebuilding the >library. Issuing a make from target\src\user results in a library which >is about 600Kb instead of 2 Mb, with many undefined symbols when building >the BSP. No response from WRS, so please can anybody point me in the >right direction? > >Tornado 1.0.1, Windows NT host, PC486 target. > >Thanks, > >Richard Wenham >Cranfield Precision > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: IP on Vxworks 5.4 Date: Sat, 26 Feb 2000 09:08:32 GMT From: pwhicker@home.net (Paul Whicker) Organization: @Home Network Message-ID: References: <8977bq$hu3$1@nnrp1.deja.com> Not as standard in both cases, however there is a NAT implementation and I think there is an IPv6 implementation too. I'm not sure but I think they got them with Routerware. HTH, Paul In article <8977bq$hu3$1@nnrp1.deja.com>, gardhab@my-deja.com wrote: >hi, >i have a few questions on VxWorks' IP capability - > >a. Does VxWorks support IpV6 (ie 64 bit IP >addressing) ? >b. Does VxWorks support IP aliasing? > >tks, >karthik > > >Sent via Deja.com http://www.deja.com/ >Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Request for document about vxWorks driver Date: Sat, 26 Feb 2000 09:11:18 GMT From: pwhicker@home.net (Paul Whicker) Organization: @Home Network Message-ID: References: <896kd7$q2o5@overload.lbl.gov> Try the drivers class documentation, their training people will probably sell you that. Then again you could always take the class, I found it useful and I got the manual and a week in SF too :-) Paul In article <896kd7$q2o5@overload.lbl.gov>, Denis BEURIVE wrote: >Hi folks, > >I am searching (for a while now) documents that >explain how to write drivers for vxWorks. I have >contacted Wind River but I did not receive any answer. > >Does anybody know a book or a web site that contains >information about vxWorks drivers ? > >Best regards, > >Denis BEURIVE >_________________________________________________________ >Do You Yahoo!? >Get your free @yahoo.com address at http://mail.yahoo.com > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: What object of vxWorks can replacing the event of M$ Windows? Date: Sat, 26 Feb 2000 09:15:36 GMT From: pwhicker@home.net (Paul Whicker) Organization: @Home Network Message-ID: References: <894m3r$2hq8$1@news.cz.js.cn> Perhaps you could describe what the event of M$ Windows is? That's probably not the area of expertise of people reading this. Tell us what you want to do and we'll tell you how to do it. :-) Paul In article <894m3r$2hq8$1@news.cz.js.cn>, "joke" wrote: >Hello , > > Can you tell me what object of vxWorks can replacing the event object of >M$ Windows or how to implement its in vxWorks? > >Regard, > >Haoyu Tang > > > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: VxWorks vs. Linux Date: 26 Feb 2000 12:01:55 GMT From: hugh@mrao.cam.ac.uk Organization: University of Cambridge, England Message-ID: <898fbj$rbe$3@pegasus.csx.cam.ac.uk> References: <87t4tl$55u$1@news2.kornet.net> <38AE0C80.2A354436@earthlink.net> Sender: :> $2995 gets you a seat of VxWorks. $50K buys you 10 seats of Tornado Power :> Professional. Good luck on every getting out a product with a free OS unless :> you have a group of engineers working 80 to 100 hours a week for months. I'm surprised you say this... My experience is far from that... in certain areas (especially complex device drivers) so-called 'profession' software is often hastily thrown together, and often full of bugs. As the source-code is secret, your only solution is to pester the driver makers in the hope that they will fix the problem. Having played with both, I would say that that Linux software is usually of exceptional quality, having been crafted by people who take time to get the bugs removed and optimised for speed, rather than being driven by the hasty need to stamp the glossy brochures with 'driver available' Hugh Gibson. --------------------------- End of New-News digest ********************** From vxwexplo-errs@csg.lbl.gov Sat Feb 26 08:03:14 2000 From: "Keith Buchanan" Date: Sat Feb 26 08:03:16 PST 2000 Subject: Re: VxWorks vs. Linux :> $2995 gets you a seat of VxWorks. $50K buys you 10 seats of Tornado Power :> Professional. Good luck on every getting out a product with a free OS unless :> you have a group of engineers working 80 to 100 hours a week for months. ...and to switch architectures or targets, you only pay the $2995 again. The investment in Tornado tools carries forward. The price is a drop in the bucket compared to the labor costs of development. >I'm surprised you say this... > >My experience is far from that... in certain areas (especially complex >device drivers) so-called 'profession' software is often hastily >thrown together, and often full of bugs. As the source-code is >secret, your only solution is to pester the driver makers in the hope >that they will fix the problem. Speaking for a group that writes complex device drivers... I must conclude that your observations are skewed by one or more poor experiences. My peers typically write very throughly tested professional drivers. It has been my experience that (for VxWorks anyway) this is the rule rather than the exception. The source code is 'secret' because we often have serious monies tied up in developing the code. Rarely does a single customer pay the entire costs of development. > Having played with both, I would say that that Linux software is >usually of exceptional quality, having been crafted by people who take time >to get the bugs removed and optimised for speed, rather than being driven >by the hasty need to stamp the glossy brochures with 'driver >available' Having developed complex drivers and applications for both, I say that the quality depends on the dedication and practices of the person or organization that wrote the software. Software is no different than any other engineering discipline in this regard and Linux enjoys no advantage over VxWorks. In fact, the Linux driver model is more complex. Linux and VxWorks are great tools but are still intended for very different applications. Using a free-OS (currently serving my mail!) for any purpose has unique risks. Choosing VxWorks for deadline driven applications is a no-risk decision. Call me for a glossy brochure, ------------------------------------ /\ Thomas Keith Buchanan /**\ Principal Engineer /****\ SPARTA, Inc. /****/.. 205 East IH 30 /****/.... Rockwall, TX 75087 .\****\__... ...\**\ /.... Voice: 972 772 4487 ....\**\/\.... FAX: 214 824 3670 ....\****\.. PCS: 972 672 9657 ....\****/ WWW: www.sparta.com .../***/ ./***/ SPARTA \**/ ====== \/ Pride In Performance From vxwexplo-errs@csg.lbl.gov Sun Feb 27 04:03:07 2000 From: Vxworks Exploder Date: Sun Feb 27 04:03:09 PST 2000 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Sun Feb 27 04:03:04 PST 2000 Subject: Re: Request for document about vxWorks driver Subject: Re: semDelete error Subject: Re: semDelete error Subject: Re: What object of vxWorks can replacing the event of M$ Windows? Subject: tempnam ? Subject: Re: VxWorks vs. Linux Subject: VxWorks for MIPS and PowerPC Subject: US-CA: Network SW Engr: RIP / BGP / OSPF, VxWorks + Subject: dosFs ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: Request for document about vxWorks driver Date: 26 Feb 2000 18:42:46 GMT From: johan.borkhuis@peektraffic.nl Organization: XS4ALL Internet BV Message-ID: <8996r6$c1o$1@news1.xs4all.nl> References: <896kd7$q2o5@overload.lbl.gov> Denis BEURIVE wrote: : Hi folks, : I am searching (for a while now) documents that : explain how to write drivers for vxWorks. I have : contacted Wind River but I did not receive any answer. : Does anybody know a book or a web site that contains : information about vxWorks drivers ? Take a look at the BSP development kit. Groeten, Johan - -- o o o o o o o . . . ___________________________________ o _____ || Johan Borkhuis | .][__n_n_|DD[ ====_____ | johan@borksoft.xs4all.nl | >(________|__|_[_________]_|________________________________| _/oo OOOOO oo` ooo ooo 'o!o!o o!o!o` === VxWorks page: http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html === --------------------------- Newsgroups: comp.os.vxworks Subject: Re: semDelete error Date: Fri, 25 Feb 2000 08:27:37 +0100 From: "Mario Semo" Organization: Nextra Telekom GmbH Message-ID: <899bgb$s4s$1@newsmaster01.nextra.at> References: <892t02$d2s$1@news.etri.re.kr> i assume you have included the required libs and have prototyped all functions well, since you didnt mention any environment informations we either have to guess or to ignore your mail. i just tried it with a complete sample code (with the correct return values given (neither of your functions is returning anything, but all are defined with default int return value) and i added some printfs (have you tried some printfs to add? have you tried the debugger?) anyway. cant reproduce. envir: T2 Fixpack3, with some patches additional installed (prinmary for open and for templates) target: vxSim as shipped with T2 Host is NT 4.05 US - -> ld func_a 4b4e9e0 4b4e9e0 value = 0 = 0x0 - -> and here is the source code it tried: #include #include void func_a(); void func_b( SEM_ID sem ); void func_c( int sem ); void func_a() { SEM_ID sem; sem = semBCreate( SEM_Q_FIFO, SEM_EMPTY ); printf("%x\n",sem); func_b( sem ); } void func_b( SEM_ID sem ) { func_c( (int) sem ); } void func_c( int sem ) { SEM_ID mySem; mySem = (SEM_ID) sem; printf("%x\n",mySem); semDelete( mySem ); } - -- regards, Mario Semo. http://www.kirchnersoft.com Lee Sang Min wrote in message <892t02$d2s$1@news.etri.re.kr>... > >Result: semDelete error - incorrect object ID > memPartFree - invalid block address > >Is this wrong? > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: semDelete error Date: Fri, 25 Feb 2000 08:30:53 +0100 From: "Mario Semo" Organization: Nextra Telekom GmbH Message-ID: <899bgf$s4s$2@newsmaster01.nextra.at> References: <892t02$d2s$1@news.etri.re.kr> <38b55fd4$1@pull.gecm.com> but semId isn't a structure. from semlib.h: typedef struct semaphore *SEM_ID; so he is casting a struct* to an int and reverse. PS: i cant rep his problem. - -- regards, Mario Semo. http://www.kirchnersoft.com Stephen Walsh wrote in message <38b55fd4$1@pull.gecm.com>... > >> >> mySem = (SEM_ID) sem; >Here sem is only an int passed by value into the function. The copy info >will be undefined as the information you want to copy is not available and >you will get rubbish in the mySem structure. >> semDelete( mySem ); >>} >> > >To do it properly pass all instances of SEM_ID by reference and don't try to >cast stuctures to int it will only cause you grief. > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: What object of vxWorks can replacing the event of M$ Windows? Date: Sat, 26 Feb 2000 21:19:23 +0100 From: "Mario Semo" Organization: Nextra Telekom GmbH Message-ID: <899cv5$n3$1@newsmaster01.nextra.at> References: <894m3r$2hq8$1@news.cz.js.cn> i assume you mean EventCreate/... look into semlib.h documentation. - -- regards, Mario Semo. http://www.kirchnersoft.com joke wrote in message <894m3r$2hq8$1@news.cz.js.cn>... >Hello , > > Can you tell me what object of vxWorks can replacing the event object of >M$ Windows or how to implement its in vxWorks? > --------------------------- Newsgroups: comp.os.vxworks Subject: tempnam ? Date: Sat, 26 Feb 2000 22:48:40 +0100 From: "Mario Semo" Organization: Nextra Telekom GmbH Message-ID: <899hmg$6da$1@newsmaster01.nextra.at> hello, i need a way to create a tempory filename / temporary file. i always use tempnam() for this, but tempnam is just in one include i dont knoe in vxworks, and it is not documented K:\LC32\L\RT\OOS\c\LCoosPersistenceFileSystem.cpp(304): warning: implicit declaration of function `int tempnam(...)' F:\Tornado>find32 tempnam *.h "F:\Tornado\target\h\private\stdioP.h" "F:\Tornado\host\x86-win32\i386-pc-mingw32\sys-include\stdio.h" F:\Tornado>find32 stdiop.h *.h "F:\Tornado\target\h\private\stdioP.h" what is this include ditectory. any other idea how to create a temporary filename? - -- regards, Mario Semo. http://www.kirchnersoft.com --------------------------- Newsgroups: comp.os.vxworks Subject: Re: VxWorks vs. Linux Date: Sat, 26 Feb 2000 19:03:38 -0500 From: "Jeff Creem" Organization: MV Communications, Inc. Message-ID: <899pkv$o4e$1@pyrite.mv.net> References: <898u7v$q2p6@overload.lbl.gov> Stuff Deleted.. > Linux and VxWorks are great tools but are still intended for very different > applications. Using a free-OS (currently serving my mail!) for any purpose > has unique risks. Choosing VxWorks for deadline driven applications is a > no-risk decision. > No risk??? Low risk perhaps... but please... I have had to beg for vxWorks source components several times to fix bugs that Wind River was too slow to fix (or to admit they even existed).. On the up side the local sales rep has always been good about giving us one or two source files if we were confident we knew where the problem was. --------------------------- Newsgroups: comp.os.vxworks Subject: VxWorks for MIPS and PowerPC Date: Sun, 27 Feb 2000 00:49:03 GMT From: bjd2000@my-deja.com Organization: Deja.com - Before you buy. Message-ID: <899s9u$98g$1@nnrp1.deja.com> I doing a little research on real-time kernels for the 32-bit MIPS processors and Motorola's Power PC. Has anyone reading this forum used VxWorks on either of these targets? If so, I would like to hear your comments on how satisfied you were ( or weren't ) with this choice of real-time OS and hardware platform. Thanks in advance for sharing your experiences. Sincerely, Brian D. Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: US-CA: Network SW Engr: RIP / BGP / OSPF, VxWorks + Date: Sat, 26 Feb 2000 21:27:38 -0800 From: N Square Corp Organization: N Square Corp. Message-ID: <38B8B5CA.4A283D8D@nsquare.com> Reply-To: arnab@nsquare.com Currently we are looking for a Title: Network SW Engr (Contractor) Good networking background ! Must have any / as many one of the following - BGP (Border Gateway Protocol) or RIP (Routing Information Protocol) or OSPF (Open Shortest Path First ) VxWorks a plus Job Loc: south Bay Area, CA Dur: 3 -6 mths with potential follow-on If you are interested in exploring this opp., please respond to this e-mail ID with a copy of your latest resume, - alongwith info on i> availability to join a new position ii> resident status in USA iii> day / even contact nos (best time to contact) iv> expected all incl. hourly bill rate (corp-to-corp or W2) Please do not change the subject line while replying ! - -- Sincerely Arnab DESIGN & CONSULTANCY SERVICES DEPT. N SQUARE CORPORATION 2100 Walsh Avenue, Ste.# B-1 Santa Clara, CA 95050 Tel # 408-654-0500 Fax # 408-654-0520 http://www.nsquare.com --------------------------- Newsgroups: comp.os.vxworks Subject: dosFs Date: Sun, 27 Feb 2000 08:39:28 GMT From: vladimir_l2439@my-deja.com Organization: Deja.com - Before you buy. Message-ID: <89ans0$rnh$1@nnrp1.deja.com> Is it possible to make dosFs driver tell disk driver about all sectors it doesn't need. For example when deleting file, dosFs only marks File Allocation Table, but actually doesn't mark the unnecessary sectors. I want dosFs to notify me about all sectors, it it no longer needs. Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- End of New-News digest ********************** From vxwexplo-errs@csg.lbl.gov Mon Feb 28 04:03:06 2000 From: Vxworks Exploder Date: Mon Feb 28 04:03:09 PST 2000 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Mon Feb 28 04:03:04 PST 2000 Subject: STD_IN failing on target shell Subject: Re: VxWorks for MIPS and PowerPC Subject: Re: GUIs for Embedded Systems using VxWorks Subject: PPC740 and MMU problem Subject: Intel 810 chipset ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: STD_IN failing on target shell Date: Sun, 27 Feb 2000 17:26:31 +0100 From: "Øyvind Holmeide" Organization: Telenor Online Public Access Message-ID: I have an application that is based on input/output through the target shell (INCLUDE_SHELL is defined). Output to the shell (STD_OUT) is OK, but input (STD_IN) is failing. The target shell gives the response "undefined symbol" for all inputs. Everything is fine if the application is started via the Tornado 2 shell. How can this be solved? regards Øyvind Holmeide --------------------------- Newsgroups: comp.os.vxworks Subject: Re: VxWorks for MIPS and PowerPC Date: Sun, 27 Feb 2000 17:45:52 GMT From: "Steve Jones" Organization: @Home Network Message-ID: References: <899s9u$98g$1@nnrp1.deja.com> I have used VxWorks on both a mpc860 and mpc850. I have had really good luck. I suggest purchasing a reference BSP like the EST860 or the VxWorks ADS860. The process of getting the BSP up and running is not as difficult as it may first seem. The hard part is to wade through all the possible options. Get with your FAE and understand all your requirements, they can help. Good Luck Steve Jones codejones@hotmail.com wrote in message news:899s9u$98g$1@nnrp1.deja.com... > I doing a little research on real-time kernels for the 32-bit MIPS > processors and Motorola's Power PC. Has anyone reading this forum used > VxWorks on either of these targets? If so, I would like to hear your > comments on how satisfied you were ( or weren't ) with this choice of > real-time OS and hardware platform. Thanks in advance for sharing your > experiences. > > Sincerely, > > Brian D. > > > Sent via Deja.com http://www.deja.com/ > Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: GUIs for Embedded Systems using VxWorks Date: Mon, 28 Feb 2000 08:47:07 +0100 From: Ole Asbjorn Fadum Organization: Telenor Online Public Access Message-ID: <38BA27FB.D3BA65A9@scanmar.no> References: <896dc8$q2o3@overload.lbl.gov> <38B6D4B7.7B2777BF@galois.net> If you are considering a web-server there is a free one to, named GoAhead Thomas Davis wrote: > "Fitch, Chris" wrote: > > > > Fellow VxWorkers, > > > > We are trying to find a replacement for our current method of building GUIs: > > RTX Windows and an X-Windows based GUI Builder. What other products are > > available? For additional background: our system employs a Power PC Single > > Board Computer with an attached Video Card. > > > > Thank you in advance, > > Chris Fitch > > LM NE&SS > > If you are networked, consider, ditching the video and go with an > embedded web server like WindWeb. The benefits are incredible. That is > the best tip I have for you. > > Thomas Davis > > -- > Galois, Inc. > http://www.galois.net > We color the world! --------------------------- Newsgroups: comp.os.vxworks Subject: PPC740 and MMU problem Date: Mon, 28 Feb 2000 12:26:08 +0200 From: Ofer Goren Organization: nortelnetworks Israel Message-ID: <38BA4D40.B15BBDF8@nortelnetworks.com> Hi. I have the following configuration: A PPC740 based board with up to 384MB of memory. I can reach this amount of memory using three memory slots of 128MB each, but each slot can have also 64MB simm, and here comes my problem. In the BAT table, I defined 2 IBATs and 2 DBATs. I initialize each for 256MB in sysLib.c. Then in sysSetDramBat(), I'm fixing the BATs, according to the real amount of memory I have. Please pay attention, that I can have memeory simms of 64MB also. Well, all went well for 128MB and 384MB, but for 256MB the board refused to come up. So I have 2 problems here: 1. If I have only 128MB, I deed to fix my First IBAT/DBAT to 128MB. What do I do with my second IBAT/DBAT? For now, I put there 128K, but this is probably a bug which will result unpredictable behaviour. So what do I do? 2. For 256MB, I put 128MB in my first IBAT/DBAT and 128MB in my second IBAT/DBAT. With that scenarion, my board did not come up. So I fixed it so my first IBAT/DBAT will be 256MB, and my second DBAT/IBAT will be 128K. again, bug, but at least the board comes up. Here is my initialized BATS: /* I BAT 0 - code resides within CPU SDRAM */ ((LOCAL_MEM_LOCAL_ADRS & _MMU_UBAT_BEPI_MASK) | _MMU_UBAT_BL_256M | _MMU_UBAT_VS | _MMU_UBAT_VP), ((LOCAL_MEM_LOCAL_ADRS & _MMU_LBAT_BRPN_MASK) | _MMU_LBAT_PP_RW), /* I BAT 1 - additional CPU bank */ ((LOCAL_MEM_LOCAL_ADRS2 & _MMU_UBAT_BEPI_MASK) | _MMU_UBAT_BL_256M | _MMU_UBAT_VS | _MMU_UBAT_VP), ((LOCAL_MEM_LOCAL_ADRS2 & _MMU_LBAT_BRPN_MASK) | _MMU_LBAT_PP_RW), .... ... /* D BAT 0 - data resides within CPU SDRAM */ ((LOCAL_MEM_LOCAL_ADRS & _MMU_UBAT_BEPI_MASK) | _MMU_UBAT_BL_256M | _MMU_UBAT_VS | _MMU_UBAT_VP), ((LOCAL_MEM_LOCAL_ADRS & _MMU_LBAT_BRPN_MASK) | _MMU_LBAT_PP_RW), /* D BAT 1 - additional CPU bank */ ((LOCAL_MEM_LOCAL_ADRS2 & _MMU_UBAT_BEPI_MASK) | _MMU_UBAT_BL_256M | _MMU_UBAT_VS | _MMU_UBAT_VP), ((LOCAL_MEM_LOCAL_ADRS2 & _MMU_LBAT_BRPN_MASK) | _MMU_LBAT_PP_RW), ... } And here is the fix done later in sysSetDramBat(): ULONG sysSetDramBat() { ULONG bs1, bs2; ULONG offset; switch (sysDramSize) { case (512*1024*1024): /* 512 MB */ /* I guess third bank is empty, and first two banks are 256 each */ bs1 = _MMU_UBAT_BL_256M; bs2 = _MMU_UBAT_BL_256M; offset = (256 * 1024 * 1024); break; case (384*1024*1024): /* 384 MB */ /* 128 in each bank */ bs1 = _MMU_UBAT_BL_256M; /* bank 1 and 2 */ bs2 = _MMU_UBAT_BL_128M; /* bank 3 */ offset = (256 * 1024 * 1024); break; case (256*1024*1024): /* 256 MB */ bs1 = _MMU_UBAT_BL_256M; /* bank 1 and 2 */ bs2 = _MMU_UBAT_BL_128K; /* bank 3 */ offset = (256 * 1024 * 1024); /* here, I know I have a bug. bs2 should be 0 or the same as bs1. But 0 is not possible, and when I tried to put the same value of bs1 in bs2, the board wont come up. So I put there the minimal memory avaliable, hoping later I'll have the time to research the bug, and figure it out. */ break; case (128*1024*1024): /* 128 MB */ bs1 = _MMU_UBAT_BL_128M; /* bank 1 and 2 */ bs2 = _MMU_UBAT_BL_128K; /* bank 3 */ offset = (128 * 1024 * 1024); /* here, I know I have a bug. bs2 should be 0 or the same as bs1. But 0 is not possible, and when I tried to put the same value of bs1 in bs2, the board wond come up. So I put there the minimal memory avaliable, hoping later I'll have the time to research the bug, and figure it out. */ break; default: /* 128 MB */ bs1 = _MMU_UBAT_BL_128M; /* bank 1 and 2 */ bs2 = _MMU_UBAT_BL_128K; /* bank 3 */ offset = (128 * 1024 * 1024); break; } /* I BAT 0 */ sysBatDesc[0] = ( (LOCAL_MEM_LOCAL_ADRS & _MMU_UBAT_BEPI_MASK) | bs1 | _MMU_UBAT_VS | _MMU_UBAT_VP); /* I BAT 1 */ sysBatDesc[2] = ( ((LOCAL_MEM_LOCAL_ADRS + offset) & _MMU_UBAT_BEPI_MASK) | bs2 | _MMU_UBAT_VS | _MMU_UBAT_VP); /* D BAT 0 */ sysBatDesc[8] = ( (LOCAL_MEM_LOCAL_ADRS & _MMU_UBAT_BEPI_MASK) | bs1 | _MMU_UBAT_VS | _MMU_UBAT_VP); /* D BAT 1 */ sysBatDesc[10] = ( ((LOCAL_MEM_LOCAL_ADRS + offset) & _MMU_UBAT_BEPI_MASK) | bs2 | _MMU_UBAT_VS | _MMU_UBAT_VP); return(bs1); } Can any one help? - -- Embedded SW Team Leader Email : ogoren@nortelnetworks.com Phone number: +9723-6456023 Fax number : +9723-6479579 ESN : 826-6023 http://www.nortelnetworks.com - ---------------------------------------------------- If you lend someone $20, and never see that person again, it was probably worth the investment. --------------------------- Newsgroups: comp.os.vxworks Subject: Intel 810 chipset Date: Mon, 28 Feb 2000 13:46:53 +0200 From: Leonid Rosenboim Organization: BitBand Technologies Ltd. http://www.bitband.com Message-ID: <38BA602D.8F468D14@bitband.com> Reply-To: leonid@bitband.com We are considering using the Intel 810 chipset, has anyone used it's advanced features ? Any other experience is welcome. Leonid --------------------------- End of New-News digest ********************** From vxwexplo-errs@csg.lbl.gov Mon Feb 28 09:25:58 2000 From: Paul Bade Date: Mon Feb 28 09:26:00 PST 2000 Subject: Vital Product Data and MV2400 BSP (vxWorks) Hi, I am having a problem with the 2400 BSP when it attempts to read the VPD (Vital Product Data). The bootroms work fine and the kernel loads across the network. But, it hangs in sysVpdInit(). If I comment this call out, the kernel boots and runs, but it doesn't enable level 2 cache and doesn't understand what the CPU is. Has anyone seen this problem before? -- +====================================================================+ | __ ____ __ __ | | /\ \ / \ \ /\_\__ /\ \ Johns Hopkins University | | / \_\ | /\ \ \ / / /\_\\ \ \ Applied Physics Lab. | | / /\ | |\ \/ \ \ / / / / / \ \ \ | | / \/ |_| \ /\ \_\ / / / / / \ \ \ Paul R. Bade | | / /\__/_/ \ \ \/_// / / / / / / / (443)-778-8681 | | / / / \ \_\ \ \/ / / / / / bade@jhuapl.edu | | \/_/ \/_/ \__/_/ \/_/ | | __ ____ __ __ | | /\ \ / \ \ /\ \ /\ \ | | / \_\ | /\ \ \ / \_\ / \ \ | | / /\ | |\ \/ \ \ / /\ |_| / /\ \ \ | | / \/ |_| \ /\ \_\ / / / | | \/ / \ \ | | / /\ / / \ \ \/_// / / / / / /\ \_\ | | \ \/ / / \ \_\ \ \/ / / \/ / / / | | \__/_/ \/_/ \__/_/ /_/_/ | | | +====================================================================+ From vxwexplo-errs@csg.lbl.gov Mon Feb 28 09:42:49 2000 From: sabarron@west.raytheon.com Date: Mon Feb 28 09:42:51 PST 2000 Subject: Routing with VxWorks 5.4.x All, Having trouble with the ifaddradd command in VxWorks. We use the ifaddradd command to allow our VxWorks production system to interact with our Unix development network without changing the "real" ip address of the production system. The problem is when we are trying to do static routes with a VxWorks PowerPC board. We have VME enclosures A, B and C. All have back plane address, but I really don't care about them, as I am only helping out getting the cards to talk over the network. B has a "real" IP address of 129.217.200.2 and a ifaddradd address of 134.206.204.1 A has a Ethernet card and a FDDI card. The Ethernet address has a real IP address of 134.6.200.1 a ifaddradd address of 134.206.200.1 and a ifaddradd address of 129.217.200.1. The FDDI has a real IP address of 134.2.200.1. C has a FDDI card with a real IP address of 134.2.203.1. We want C to send all 129.217.x.x traffic to A. Which should see the ifaddradd and send it out it's Ethernet card. Then VME b should hear it and respond. VME enclosures A and C can talk no problem. We can even ping from C to the ifaddradd address of 129.217.200.1. We can not get VME C and B to talk (no ping or any thing). The problem appears that the packets get lost. We are fairly sure that this is a problem of too many ifaddradd and how we have it setup. We have the routing correct, and we have seen this work, with fewer ifaddradd. Wondering if any one out there has more experience with ifaddradd? Alexander Barron Raytheon DCS NAMS San Diego From vxwexplo-errs@csg.lbl.gov Mon Feb 28 15:28:37 2000 From: John Moore Date: Mon Feb 28 15:28:39 PST 2000 Subject: MV2400, Force P750 Ethernet performance and the wind shell Hello, I'm loading files on both the Force PCORE750 and the Motorola MV2433. The Force board is loading the same files in about half the time as the Motorola. I'm running Tornado 2/5.4 on both and loading from the wind shell (same server, using NFS for each). I've looked through the archives and haven't found any recent problem reports on the MV2400 BSP ( I believe I'm using the latest drivers). Also the Force board takes about 22 seconds to load a 2 Megabyte file using the wind shell. Using the target shell, with the same kernel, the load time is under a second. Similar results on the Motorola board but the times are about twice as long. I am using 100 Base T and am at a loss to explain the order of magnitude difference in the shell load times. Comments ? Thanks, John Moore From vxwexplo-errs@csg.lbl.gov Tue Feb 29 00:57:18 2000 From: "Ganesh.N" Date: Tue Feb 29 00:57:21 PST 2000 Subject: Does VxWorks support usage of Tcl/Tk ? Hello All, Does VxWorks support usage of Tcl/Tk ? thanks in advance, Ganesh. From vxwexplo-errs@csg.lbl.gov Tue Feb 29 04:04:02 2000 From: Vxworks Exploder Date: Tue Feb 29 04:04:04 PST 2000 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Tue Feb 29 04:03:06 PST 2000 Subject: Re: How do I rebuild the VxWorks module library? Subject: Re: How do I rebuild the VxWorks module library? Subject: Re: How do I rebuild the VxWorks module library? Subject: dosFs Subject: Embedded int 3 handing with vxWorks. Subject: Re: builtin_new/builtin_delete Subject: Simple database code? Subject: Re: VxWorks for MIPS and PowerPC Subject: using tftpXfer Subject: Re: MibWay for VxWorks Subject: Re: GUIs for Embedded Systems using VxWorks Subject: a problem when using "ClearCase" and "Tornado2.0" together Subject: Stethoscope 5.0 Subject: memory autosizing using sysPhysMemTop Subject: dosFs 2.0 and TFFS question Subject: Windows Managers for VxWorks Subject: sysBusIntGen Subject: Re: How to put a telnet server into half-duplex mode? Subject: Re: Windows Managers for VxWorks Subject: How to get current time & date for timestamps for DOS-FS files ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: How do I rebuild the VxWorks module library? Date: Mon, 28 Feb 2000 14:21:54 +0200 From: "Shuky Persky" Organization: Bezeq International Ltd. Message-ID: <38ba67ee@news.bezeqint.net> References: - -- Hi, Copy it to the BSP directory. Don't forget to delete the dependxxxx file to enforce regeneration U don't need to touch the library. Enjoy, Shuky - ------------------------------ voice : 972-(0)50-432266 email : shukyp@bigfoot.com Richard Wenham wrote in message news:memo.20000225145726.259A@cpecran.compulink.co.uk... > I need to modify the serial driver, i8250Sio.c. Having done this, how do > I get it compiled, and replaced in target\lib\libI80486gnuvx.a? Making > the BSP VxWorks image in the normal way does not spot the change to > i8250Sio.c, and I cannot find specific instructions on rebuilding the > library. Issuing a make from target\src\user results in a library which > is about 600Kb instead of 2 Mb, with many undefined symbols when building > the BSP. No response from WRS, so please can anybody point me in the > right direction? > > Tornado 1.0.1, Windows NT host, PC486 target. > > Thanks, > > Richard Wenham > Cranfield Precision > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: How do I rebuild the VxWorks module library? Date: Mon, 28 Feb 2000 14:24 +0000 (GMT Standard Time) From: rew@cp.cix.co.uk (Richard Wenham) Organization: CIX - Compulink Information eXchange Message-ID: References: Reply-To: rew@cp.cix.co.uk > Cranfield Bedfordshire? I used to live in Bedford and cycle through > there. Yes, on the University campus, building 90. Pop in if you happen to be passing. > ...I just copy the source file to my BSP directory and That's what I decided to do, after engaging the brain. Much more sensible, and easier to explain to the config management system. Thanks. Richard In article , pwhicker@home.net (Paul Whicker) wrote: > Cranfield Bedfordshire? I used to live in Bedford and cycle through > there. > > Try "make CPU=80486 TOOL=gnu", I think that should build the object and > add it to the archive... however that requires a certain amount of > bravery. > > I normally don't do this, I just copy the source file to my BSP > directory and then add the object to MACH_EXTRA in the Makefile. That > should do it. > > Once you are confident you can use ar386 to replace the module in the > library, check the manual but I think the archive command will be > something like > > ar386 -rv lib80486gnuvx.a i8250Sio.o > > > I think that should do it. Most of the WR filed engineers seem to be > missing this week, I think they have some sort of meeting, I guess > that's dumping everything on support. > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: How do I rebuild the VxWorks module library? Date: Mon, 28 Feb 2000 14:24 +0000 (GMT Standard Time) From: rew@cp.cix.co.uk (Richard Wenham) Organization: CIX - Compulink Information eXchange Message-ID: References: <38ba67ee@news.bezeqint.net> Reply-To: rew@cp.cix.co.uk > Copy it to the BSP directory. That's what I did. Thanks. Richard In article <38ba67ee@news.bezeqint.net>, shuky@infogear.co.il (Shuky Persky) wrote: > > > -- > Hi, > > > Copy it to the BSP directory. > > Don't forget to delete the dependxxxx file to enforce regeneration > > U don't need to touch the library. > > Enjoy, > Shuky > ------------------------------ > voice : 972-(0)50-432266 > email : shukyp@bigfoot.com > Richard Wenham wrote in message > news:memo.20000225145726.259A@cpecran.compulink.co.uk... > > I need to modify the serial driver, i8250Sio.c. Having done this, > > how do > > I get it compiled, and replaced in target\lib\libI80486gnuvx.a? > > Making > > the BSP VxWorks image in the normal way does not spot the change to > > i8250Sio.c, and I cannot find specific instructions on rebuilding the > > library. Issuing a make from target\src\user results in a library > > which > > is about 600Kb instead of 2 Mb, with many undefined symbols when > > building > > the BSP. No response from WRS, so please can anybody point me in the > > right direction? > > > > Tornado 1.0.1, Windows NT host, PC486 target. > > > > Thanks, > > > > Richard Wenham > > Cranfield Precision --------------------------- Newsgroups: comp.os.vxworks Subject: dosFs Date: Mon, 28 Feb 2000 17:35:28 +0200 From: "Vladimir" Organization: NetVision Israel Message-ID: <89e4aa$j6k$1@news.netvision.net.il> Is it possible to make dosFs driver tell disk driver about all sectors it doesn't need. For example when deleting file, dosFs only marks File Allocation Table, but actually doesn't mark the unnecessary sectors. I want dosFs to notify me about all sectors, it it no longer needs. --------------------------- Newsgroups: comp.os.vxworks Subject: Embedded int 3 handing with vxWorks. Date: Mon, 28 Feb 2000 12:47:25 -0500 From: Burt Bicksler Organization: BBB Message-ID: Hi, I've been tasked with finding a way to invoke / break into the debugger conditionally from a running vxWorks application (vxWorks target). WRS has an int 3 instruction defined in one of the headers, but no examples of using this. The straight forward approach just locks things up (yes the debbuger is running when I try this). I then thought about using the Signal approach but WRS strikes again, and their GDB doesn't support signals with a vxWorks target. I have a kludge that works, but wondered if anyone had ever gotten an embedded int 3 to break into the debugger with vxWorks. Running on the Intel platform.. Thanks, Burt Bicksler Remove the spam bait from the return address if replying by e-mail. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: builtin_new/builtin_delete Date: Mon, 28 Feb 2000 19:03:36 +0100 From: "Xavier.Montagne" Organization: Wanadoo, l'internet avec France Telecom Message-ID: <89edcr$atu$1@wanadoo.fr> References: <38BAA71C.3C96@fzi.de> Hi, "Torsten Rupp" wrote in the message : 38BAA71C.3C96@fzi.de... > Hi, > > currently I try to load some program code including C++ new()/delete() > calls on a MV2400/PPC 750 board. I have a problem with the functions > __builtin_new and __builtin_delete, which seems not to be included > in the PPC target lib/BSP anymore (it was included in the lib for > MC68040 targets). Are these functions now included in some other > (additional) lib? Thanx for some hint. > > Torsten > > -- > Computer Science Research Center (FZI) phone : +49-721-93372-13 > Mobility Management and Robotics fax : +49-721-93372-19 > Haid-und-Neu-Str. 10-14 e-mail: rupp@fzi.de > D-76131 Karlsruhe (Germany) www : www.fzi.de/robot I've got exactly the same configuration, and haven't noted any problem with new and delete... Are you sure to put the correct lib (PPC604xxxx.a or something else) in the compiler rules ?? Do you work with Tornado II ....? If it is, give all the rules you put for your project.... Xavier --------------------------- Newsgroups: comp.os.vxworks Subject: Simple database code? Date: Mon, 28 Feb 2000 18:17:51 GMT From: Marjorie Krueger Organization: Vixel Message-ID: <38BABBB1.F4823986@vixel.com> Anyone know of any simple database code (2 or 3 keys at the most) that has a small enough footprint to be run in VxWorks? Thanks Marjorie Krueger Vixel Corporation --------------------------- Newsgroups: comp.os.vxworks Subject: Re: VxWorks for MIPS and PowerPC Date: 28 Feb 2000 14:12:28 -0500 From: Bill Pringlemeir Organization: Factory of the mind Message-ID: References: <899s9u$98g$1@nnrp1.deja.com> Sender: bpringle@DeadDuck I have used the MPC860. I didn't like the code size of the PowerPC. Does the MIPS instruction set have three register opcode with 32 registers? I think that 32 registers is excessive. It takes up opcode space and makes the PowerPC take lots of code space. Unless the particular PPC fits your peripheral requirements, I would choose something else. I think that the some of the ARM variants are as good as the MIPS processors. You didn't mention what kind of application you were targeting. If you are doing custom hardware, then the peripherals are important. The MPC860 has really nice capabilities for ethernet (but the processor sucks...). If you just need pure number crunching, I would choose MIPS or ARM for power consumption and lower cost (less RAM/ROM). fwiw, Bill Brian> I doing a little research on real-time kernels for the Brian> 32-bit MIPS processors and Motorola's Power PC. Has anyone Brian> reading this forum used VxWorks on either of these targets? Brian> If so, I would like to hear your comments on how satisfied Brian> you were ( or weren't ) with this choice of real-time OS Brian> and hardware platform. Thanks in advance for sharing your Brian> experiences. - -- We're off to see the wizard. Because because because.... of all the wonderful things he does. --------------------------- Newsgroups: comp.os.vxworks Subject: using tftpXfer Date: Mon, 28 Feb 2000 15:50:17 -0800 From: "Bernard Littau" Organization: Alternate Access Inc. Affordable, Reliable Access to the InterNet Message-ID: <38bb09d4_2@huge.aa.net> Hi All, Has anyone out there used tftpXfer? I am getting a return of ERROR, with no errno being set. I am chasing through the assembler code, and it seems to be failing because it is hard coded to get an ethernet address, via ifAddrGet with an interface of lo0. I am using ppp (or wish to :-) for this Xfer. Thanks, Bernard --------------------------- Newsgroups: comp.os.vxworks Subject: Re: MibWay for VxWorks Date: Mon, 28 Feb 2000 18:04:11 -0800 From: Vinh Du Organization: Wind River Message-ID: <38BB291B.ED4F875F@windriver.com> References: <38912550.4F448385@lucent.com> You can use MIBWay for WindWeb with changing the snmp. Mibway is design for the WindNET SNMP product. You should ask for a demo of this product if you're interested. Rapid Logic, Inc. Headquarters Phone: 510-749-0900 vinh Yusuf Motiwala wrote: > Hi, > > Is anybody has used the RapidLogic product, "MIBWay for Wind Web Server" > > Product URL: http://www.rapidlogic.com/products/mibway_wws.html > > We want to use this product to implement the WEB based management > (and possibly CLI also). We are using WindNET SNMP and would not > like to change our SNMP implementation. > > I would like to get any feedback on this product, about technical > issues, > performance, stability, documentation etc. I will very much appreciate > your > feedback. > > Thanks in Advance, > Yusuf - -- Thanks, vinh *************************** Vinh Du Senior Field Application Engineer Wind River vinh.du@WindRiver.com 408-487-9922 *************************** --------------------------- Newsgroups: comp.os.vxworks Subject: Re: GUIs for Embedded Systems using VxWorks Date: Mon, 28 Feb 2000 18:00:08 -0800 From: Vinh Du Organization: Wind River Message-ID: <38BB2828.1281B8B7@windriver.com> References: <896dc8$q2o3@overload.lbl.gov> You should consider using Zinc for vxWorks or Java for vxWorks as your GUI. http://www.wrs.com/products/html/zinc_ds.html http://www.windriver.com/internet/html/pjps_ds.html# vinh WR "Fitch, Chris" wrote: > Fellow VxWorkers, > > We are trying to find a replacement for our current method of building GUIs: > RTX Windows and an X-Windows based GUI Builder. What other products are > available? For additional background: our system employs a Power PC Single > Board Computer with an attached Video Card. > > Thank you in advance, > Chris Fitch > LM NE&SS --------------------------- Newsgroups: comp.os.vxworks Subject: a problem when using "ClearCase" and "Tornado2.0" together Date: Mon, 28 Feb 2000 20:52:43 -0500 From: rwang Organization: Hughes Network Systems Message-ID: <38BB266B.98845913@hns.com> References: <89f33m$4t02@overload.lbl.gov> Hello, We met a problem when we used "ClearCase" as version control tool in "Tornado" environment. The problem is we can not share the "Makefile" without checkout. For example, two users have their own views, j:\user1 and j:\user2, which are mapped from a unix server to a NT client. When user1 setup a new project in the sharing workspace, the Makefile is created automatically based on user1's mapping directory. The first two lines in the core information part of the makefile looks like: ## core information PRJ_DIR = J:/user1/projectname PRJ_FILE = projectname.wpj When the user2 only wants to make some changes in an existing source file of the same project. User2 should not need to check out the Makefile because those changes do not effect Makefile. But when user2 does so (does not check out the Makefile), he/she sees the exactly same Makefile as the user1. Then when user2 wants to compile and build the project, he/she will fail because the PRJ_DIR in his/her Makefile should be "J:\user2\projectname", not "J:/user1/projectname". I tried to set some environment variables, like "$(WIND_BASE)", in the PRJ_DIR in the Makefile. Such as: PRJ_DIR = $(MAPPED_USER_PRJDIR) But I did not know how to let Tornado know this environment variable or which Tornado environment variable can be used to solve this problem. Thanks a lot. Ran --------------------------- Newsgroups: comp.os.vxworks Subject: Stethoscope 5.0 Date: Tue, 29 Feb 2000 02:45:23 GMT From: MJM Organization: mjm et al Message-ID: <38BB3124.2A45485F@sympatico.ca> Reply-To: mike.mcnaughton@sympatico.ca Hello you lucky vxworks users: I have purchased Steth 5.0 and wish to install it tornado 1.01, vxworks 5.3.1. So far it has done nothing but frustrate the Hell out of any attempts to install it.. What is required in WDB land. What about libraries. What is the target server-server. WRS fly up their butts in the manual when they get into that and the sym_synch_lib. Is there someone among you who wouln't mind taking a few minutes to outline a detailed procedure of what you had to do to get it going. I suspect sym_synch_lib trouble too. I've tried in earnest to follow the wrs procedure to the finest detail but it just won't fall into place. There are some a priori details missiing. Help would be greatly appreciated. If you lived down the street from me and helped me out with this in a manner which led to my getting it up and running, I would buy you a 24-pack of beer. WRS should test their written explanation on normal people who just want to set up the @#$@%#@$ application without having to know every last bit of computer jragon that was ever conjured up. A step-by-step procure or a wizard would be really good. Thanks MJM --------------------------- Newsgroups: comp.os.vxworks Subject: memory autosizing using sysPhysMemTop Date: 29 Feb 2000 05:35:31 GMT From: tmiller652@aol.com (TMiller652) Organization: AOL http://www.aol.com Message-ID: <20000229003531.15846.00002347@ng-cd1.aol.com> Hello, I'm trying to have my vxworks system detect the amount of ram available to the system at startup and configure accordingly. I have a method that I can use to test for the size of physical memory, and i've included it in the function sysPhysMemTop. I thought this would override the "#defined" memory size as described in "LOCAL_MEM_SIZE" - but it doesn't seem to work. Does anyone know a reason why determining memory using sysPhysMemTop WOULDN"T override the previous definition of LOCAL_MEM_SIZE? Any assistance would be greatly appreciated. Tom Miller, Motion Engineering, Santa Barbara CA. --------------------------- Newsgroups: comp.os.vxworks Subject: dosFs 2.0 and TFFS question Date: Tue, 29 Feb 2000 05:49:30 GMT From: "Steve Jones" Organization: @Home Network Message-ID: Greetings to all, I have installed the new dosFs 2.0, and am having trouble getting backwards compatibility with TFFS. Has anyone successively made the switch to the new dosFs? I could use a few pointers. Thank in advance. Steve Jones codejones@hotmail.com --------------------------- Newsgroups: comp.os.vxworks Subject: Windows Managers for VxWorks Date: Tue, 29 Feb 2000 06:33:44 GMT From: jfwinter@easynet.co.uk (Philippe) Organization: Easynet Group plc Message-ID: <38bb67c0.135161586@news.easynet.co.uk> Is anyone aware of any Windows Managers (X-Windows) supported by the VxWorks Platform ? Thanks Phillipe --------------------------- Newsgroups: comp.os.vxworks Subject: sysBusIntGen Date: Tue, 29 Feb 2000 08:23:09 +0100 From: Tomas Larsson Organization: National Defence Research Establishment Message-ID: <38BB73DC.54D2FE30@lin.foa.se> Hi all! We have a configuration with two MVME2700-boards + VIPC618 and we generating interrups through a DUAL/PIT IP module sitting on the VIP- carrier. When trying to - -> sysBusIntGen(2,0x16) > value = -1 = 0xffffffff We have nothing ,that we know of, which could cause an VME-bus interrrupt Any suggestion is appriciated Tomas --------------------------- Newsgroups: alt.winsock.programming,comp.os.vxworks,comp.protocols.tcp-ip Subject: Re: How to put a telnet server into half-duplex mode? Date: 25 Feb 2000 07:26:50 -0500 From: James Carlson Message-ID: <86ya89zcvp.fsf@ironbridgenetworks.com> References: <895gjm$art$1@nnrp1.deja.com> Sender: carlson@helios nedren@yahoo.com writes: > I would like the server to send me a "Go Ahead" when it finishes > sending, so I can then start sending my stuff. How would I be able to > put the server into this half-duplex mode? You really do need to accept and request SGA. Many (probably most) servers don't support the go-ahead function at all. It's an anachronism at best. > Also, another question, am I suppose to send a Go ahead at the end of > each negotiation string I send out, and send a LF + CR at end of each > ASCII Data stream? Actually, CR followed by LF, not the other way around. NVT basically requires a fill character after CR. Block-oriented communicates on TELNET usually uses either NVT ASCII with end-of-line as the turn-around indication *or* uses IAC EOR to signal block end (as with tn3270). There's really no need to use go-ahead. - -- James Carlson, System Architect IronBridge Networks / 55 Hayden Avenue 71.246W Vox: +1 781 372 8132 Lexington MA 02421-7996 / USA 42.423N Fax: +1 781 372 8090 "PPP Design and Debugging" --- http://people.ne.mediaone.net/carlson/ppp --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Windows Managers for VxWorks Date: Tue, 29 Feb 2000 10:00:28 GMT From: Computer Engineer Organization: Somewhere in the Earth Message-ID: <38BB2807.CB8BF73E@netscape.net> References: <38bb67c0.135161586@news.easynet.co.uk> RtX-Windows is supported under Tornado 1.0.1 ONLY. --------------------------- Newsgroups: comp.os.vxworks Subject: How to get current time & date for timestamps for DOS-FS files Date: Tue, 29 Feb 2000 10:24:42 GMT From: m.fischer@steinbichler.de (Manfred Fischer) Organization: roNet GmbH Rosenheim Message-ID: <38bb9cf1.9562099@news.ronet.de> I'm working with a MV2700 PowerPC board with a DOS FS on a harddisk. Is there a possibilty to get current time and date as timestamp for new created or changed files. Is there a special mechanism to apply a timestamp automatically on files. --------------------------- End of New-News digest ********************** From vxwexplo-errs@csg.lbl.gov Tue Feb 29 04:22:15 2000 From: "Ganesh.N" Date: Tue Feb 29 04:22:17 PST 2000 Subject: Does VxWorks support usage of TCL/TK? hello, Does VxWorks support usage of TCL/TK? or do we need to port TCL/TK to be used in VxWorks ? thanks & regards, Ganesh. From vxwexplo-errs@csg.lbl.gov Tue Feb 29 06:50:43 2000 From: manish.nair@wipro.com Date: Tue Feb 29 06:50:45 PST 2000 Subject: Jumbo frames, priority queue, checksum Hi VxWorkers, Could you please tell me whether VxWorks network stacks can handle Jumbo frames, if so how do we enable the support for jumbo frames on the send side. Can it be enabled just by increasing the mtu size to around 9000 bytes? Is there any priority queue implementation in the VxWorks network stacks? If so, how can we obtain the priority of a packet ? Lets say if I have a card which supports the priority queue implementation and when I receive packets on a priority basis, what should I be doing when I pass the same packet to the upper layers using the receiveRtn in the END_OBJ, to indicate the priority level in which it was received? Is it possible to tell the network stacks, not to do the checksum computation for TCP/UDP headers, on both the send & receive side? I can get the job done in hardware, as my card supports it, so offloading of this job to the hardware could be a good proposition for me. Thanks & Regds Manish From vxwexplo-errs@csg.lbl.gov Tue Feb 29 07:10:57 2000 From: bob schulman Date: Tue Feb 29 07:10:59 PST 2000 Subject: Re: Windows Managers for VxWorks jfwinter@easynet.co.uk asks: Is anyone aware of any Windows Managers (X-Windows) supported by the VxWorks Platform ? Seaweed Systems has X client-side packages which support twm and, optionally, mwm. These packages are supported under both Tornado1 and Tornado2. Please see www.seaweed.com for more info. bob From vxwexplo-errs@csg.lbl.gov Tue Feb 29 14:41:54 2000 From: "Gerald Przybylski, Lawrence Berkeley National Lab, GTP@LBL.GOV" Date: Tue Feb 29 14:41:56 PST 2000 Subject: Need Generic (non-vxworks) boot prom or binary file for MVME-167 Can you please contact me if you can send me a binary file for a generic boot loader for an Motorola MVME-167? or if you have a boot prom you don't need... Thank you Gerald Przybylski gtp@lbl.gov From vxwexplo-errs@csg.lbl.gov Tue Feb 29 20:30:20 2000 From: "Ganesh.N" Date: Tue Feb 29 20:30:22 PST 2000 Subject: FW: Does VxWorks support usage of TCL/TK? hello, Does VxWorks support usage of TCL/TK? or do we need to port TCL/TK to be used in VxWorks ? thanks & regards, Ganesh.