From mmuri@qualcomm.com Tue Jan 3 12:09:56 1995 From: mmuri@qualcomm.com (Mark Muri) Date: Tue Jan 3 12:10:03 PST 1995 Subject: Generating VMEBUS cycles from 167 Sys Con Anyone, I've been experimenting with using a VME167 to generate VMEBUS cycles to another 3rd party card. I have a 167 as the System controller and other 167 boards (which do a successful sm boot). By twiddling with the bits in the VMECHIP2, I have been very successful at generating VMEBUS addresses from the non-system-controller cards. I cannot, for the life of me, generate one from the System Controller! All I get are bus errors. The VMECHIP2 indicates MLOB, but I never see the cycle show up on the backplane (using a VMETRO). (As you know, a System Controller who exports his local memory as the VxWorks shared memory pool cannot even access that pool with the VMEBUS address that the non-system-contoller cards do - it must use it's local bus address.) I believe that it's some simple interaction between the dual role of the Master VMECHIP2 as one who's trying to generate the address PLUS the one in charge of arbitrating the bus. I've done the proper sysPhysMemDesc changes and, since vxMP is installed, the changes show up properly when I do a vmContextShow. The address space is writable and non-cached. Anyone run into this type of problem? I suppose there's a simple explanation and maybe I have been staring at it for tooooooooo long... Mark <=================o=====================================================> | Mark Muri | E-Mail: mmuri@qualcomm.com (619) Work: 658-3303 | | Senior Engineer | Office: Q-244A Fax: 658-2108 | | QUALCOMM, Inc <+> Car: 895-8315 | | 6455 Lusk Blvd., San Diego, CA 92121 Pager: 636-0446 | <=======================================================================> From stevem@syd.csa.com.au Thu Jan 5 00:50:17 1995 From: stevem@syd.csa.com.au (Steven McCoy) Date: Thu Jan 5 00:50:30 PST 1995 Subject: Bus Error with 32MByte MVME167 Our Ada application executes without a problem on a MVME167-B (25 MHz, 16 MBytes) and a MVME167-33A (33 MHz, 16 MBytes). A recent upgrade to the MVME167-34A (33 MHz, 32 MBytes) has suddenly caused the same code to generate a Bus error or at times an Illegal Instruction. This always occurs when executing i() or checkStack() commands from the shell. Note that we do not have a problem with the pi() command. e.g. Bus Error Program Counter: 0x00000008 Status Register: 0x3000 Access Address : 0x08eb7376 <-- Way out of range. Special Status : 0x0125 shell restarted. or Illegal Instruction Program Counter: 0x00000012 Status Register: 0x3008 shell restarted. The problem will still occur if the 34A 32 MByte memory module is used on the 33A. The 16Mbyte module works fine on the 34A. Hence, there appears to be a problem with memory initialisation or something with memory modules greater than 16 MBytes. Are there any VxWorks changes required when using a 167 with more than 16 MBytes of memory? We are using the following, VADSworks version 2.0.3(b). VxWorks (for Motorola MVME167) version 5.0.2b. Kernel: WIND version 2.0. Any ideas would be appreciated, Steven McCoy CSC Australia A Unit of Computer Sciences Corporation stevem@assip.csasyd.oz.au From daemon@vxw.ee.lbl.gov Thu Jan 5 04:00:24 1995 From: daemon@vxw.ee.lbl.gov Date: Thu Jan 5 04:00:32 PST 1995 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Thu Jan 5 04:00:13 PST 1995 Subject: Motorola MC68681 driver Subject: FAQ and ftp site info Subject: FAQ and ftp site info Subject: ld MEMORY command, multiple regions per section possible? Subject: Re: FAQ and ftp site info ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Motorola MC68681 driver Date: Wed, 4 Jan 1995 13:37:38 GMT From: bjm_vloedbeld@hgl.signaal.nl (B.J.M. Vloedbeld) Organization: Hollandse Signaal apparaten BV Message-ID: <1995Jan4.133738.17760@hgl.signaal.nl> Sender: usenet@hgl.signaal.nl (USENET News System) Hello, We are currently developing a Vxworks application with a 68020 processor and a MC68681 duart. question : We are looking now for a MC68681 driver that supports the serial IO and that uses the MC68681 timer for the system clock. Please email me if you have some information. Thanks, --Ben-- - -- ============================================================================== = B.J.M. Vloedbeld Dept:SDA-DID-DIM Org:Hollandse Signaalapparaten B.V. = = Adress : P.O.box 42 7550 GD Hengelo, The Netherlands = = Phone :+31-74-483994 FAX:+31-74-484009 email:bjm_vloedbeld@hgl.signaal.nl = =============================================================[Unclassified]=== --------------------------- Newsgroups: comp.os.vxworks Subject: FAQ and ftp site info Date: 04 Jan 1995 21:37:02 GMT From: wilson@ssdd475a.erim.org (Paul Wilson) Organization: Environmental Research Institute of Michigan, Ann Arbor, Michigan Message-ID: Sender: news@newsspool.erim.org Is there an FAQ available for this group? Where might I obtain it? Also, I have a requirement to produce a Sun readable TAR tape from a vxworks application. Is there a netware package to do that? Thanks in advance for assistance. regards, Paul Wilson wilson@erim.org --------------------------- Newsgroups: comp.os.vxworks Subject: FAQ and ftp site info Date: 04 Jan 1995 22:09:32 GMT From: wilson@ssdd475a.erim.org (Paul Wilson) Organization: Environmental Research Institute of Michigan, Ann Arbor, Michigan Message-ID: Sender: news@newsspool.erim.org Is there an FAQ available for this group? Where might I obtain it? Also, I have a requirement to produce a Sun readable TAR tape from a vxworks application. Is there a netware package to do that? Thanks in advance for assistance. regards, Paul Wilson wilson@erim.org --------------------------- Newsgroups: comp.os.vxworks,gnu.gcc.help Subject: ld MEMORY command, multiple regions per section possible? Date: 4 Jan 95 22:26:52 GMT From: kab@dtl.labs.tek.com (Kent Black) Organization: DTL, Tektronix, Inc. Keywords: a.out, MEMORY Message-ID: <6986@crl.labs.tek.com> Followup-To: comp.os.vxworks Sender: news@crl.labs.tek.com I would like to be able to put a globally accessible object (an OS specific symbol table for debugging) into some slow (or even optional) memory. I'm thinking a chunk of ROM for text, a smallish chunk of RAM for data and bss, and "somewhere else" for this other big chunk of initialized data (otherwise I'm short of usable RAM). Using a.out format (68k arch), is it possible to have a section, in particular .data, divided among memory regions? Consider the following ld script: OUTPUT_FORMAT("a.out-sunos-big") /* output .o file format */ TARGET("a.out-sunos-big") /* input .o file format */ ENTRY("_romSP") /* entry point */ MEMORY { rom : ORIGIN = 0xfB000000, LENGTH = 2M ram : ORIGIN = 0x06020000, LENGTH = 512K } SECTIONS { .text : { CREATE_OBJECT_SYMBOLS *(.text) _etext = .; } > rom .data : { *(.data) _builtForRom = .; SHORT(0); _edata = .; } > ram .bss SIZEOF(.data) + ADDR(.data) : { *(.bss) [ COMMON ] _end = .; __end = .; } > ram } Now, my first thought is that I could simply add an optional read of "symtable.o" into some arbitrary address from the .data segment, but that fails when .data > ram doesn't "fit". Clever alternatives welcome. - -- kab --------------------------- Newsgroups: comp.os.vxworks Subject: Re: FAQ and ftp site info Date: Thu, 5 Jan 1995 08:15:17 GMT From: bjm_vloedbeld@hgl.signaal.nl (B.J.M. Vloedbeld) Organization: Hollandse Signaal apparaten BV Message-ID: <1995Jan5.081517.13934@hgl.signaal.nl> References: Sender: usenet@hgl.signaal.nl (USENET News System) Paul Wilson (wilson@ssdd475a.erim.org) wrote: : Is there an FAQ available for this group? Where might I obtain it? : Also, I have a requirement to produce a Sun readable TAR tape from : a vxworks application. Is there a netware package to do that? : Thanks in advance for assistance. : regards, : Paul Wilson : wilson@erim.org here some lines from the original vxworks FAQ : The FAQ is currently posted to comp.os.vxworks, news.answers and comp.answers on the 15th of every month. This FAQ was compiled by hjb@netcom.com using comments by readers of comp.os.vxworks as well as his own limited knowledge of VxWorks. Credits appear at the end. Comments and indications of doubt are enclosed in []s in the text. Each section begins with dashes ("-") on a line of their own, then the section number. This should make searching for a specific section easy. This FAQ is also available via anonymous ftp in: rtfm.mit.edu:/pub/usenet/news.answers/vxworks-faq.Z ftp.uu.net:/usenet/news.answers/vxworks-faq.Z cs.toronto.edu:/pub/usenet/comp.answers/vxworks-faq.Z ftp.netcom.com:/pub/hjb/vxfaq.gz - -- ============================================================================== = B.J.M. Vloedbeld Dept:SDA-DID-DIM Org:Hollandse Signaalapparaten B.V. = = Adress : P.O.box 42 7550 GD Hengelo, The Netherlands = = Phone :+31-74-483994 FAX:+31-74-484009 email:bjm_vloedbeld@hgl.signaal.nl = =============================================================[Unclassified]=== --------------------------- End of New-News digest ********************** From crispen@eight-ball.hv.boeing.com Thu Jan 5 06:26:29 1995 From: crispen@eight-ball.hv.boeing.com (Bob Crispen) Date: Thu Jan 5 06:26:35 PST 1995 Subject: Re: comp.os.vxworks newsdigest kab@dtl.labs.tek.com (Kent Black) sez: >I would like to be able to put a globally accessible object (an OS >specific symbol table for debugging) into some slow (or even optional) >memory. I'm thinking a chunk of ROM for text, a smallish chunk of RAM >for data and bss, and "somewhere else" for this other big chunk of >initialized data (otherwise I'm short of usable RAM). > >Using a.out format (68k arch), is it possible to have a section, in >particular .data, divided among memory regions? [snip] >Clever alternatives welcome. This isn't particularly clever, and I'm not completely sure it addresses your problem, but you may find some use for it in your eventual solution. Change the line in vxworks/config/yourboard/sysLib.c (in sysMemTop()) that reads: static char *memTop = (char *)(LOCAL_MEM_LOCAL_ADRS + LOCAL_MEM_SIZE); to: static char *memTop = (char *)(LOCAL_MEM_LOCAL_ADRS + LOCAL_MEM_SIZE - 0x10000); and gen up another bootable version of VxWorks. If you want to keep this memory from being clobbered when you hit reset, do it for your VxWorks boot ROM, too. That gives you the top 64K (or whatever) of memory that's Real Manly Memory By God -- none of this girly-man dynamically allocated memory business. ;-) +-------------------------------+--------------------------------------+ | Bob Crispen | Who will babysit the babysitters? | | crispen@foxy.hv.boeing.com +--------------------------------------+ | (205) 461-3296 |Opinions expressed here are mine alone| +-------------------------------+--------------------------------------+ From mea@mclean.sparta.com Thu Jan 5 06:50:24 1995 From: mea@mclean.sparta.com (Mike Anderson) Date: Thu Jan 5 06:50:30 PST 1995 Subject: Re: FAQ Greetings! The FAQ is no longer being kept by Hwajin. It is now located at: teacher.mclean.sparta.com (157.185.32.10). With the holidays, bringing new machines on-line, running new LANs to support them and end of the year deliverables, etc., I'm afraid I've let the FAQ fall behind. I'm making every effort to catch it up to date and have the new FTP machine for the FAQ with Mosaic home page, etc. on-line by the end of the month. Sorry for the delay, =============================================================================== __ Real-Time System Development, Integration, Training and Services //\\ // \\ Mike Anderson // /\ \\ Director, Real-Time Systems // / \ \\ SPARTA, Inc. Voice : (703) 448-0210 ext. 235 // \ \\ 7926 Jones Branch Drive FAX : (703) 734-3323 \\ \ // Suite 900 EMAIL : mea@mclean.sparta.com \\ \ / // McLean, VA 22102 \\ \/ // "Software development is like making \\ // a baby... You can't make a baby in one \\// month by impregnating nine women. -- "Pride in Performance" Some things just take time." =============================================================================== From daemon@vxw.ee.lbl.gov Fri Jan 6 04:00:22 1995 From: daemon@vxw.ee.lbl.gov Date: Fri Jan 6 04:00:29 PST 1995 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Fri Jan 6 04:00:13 PST 1995 Subject: SCSI floppy disk drive Subject: Stability ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: SCSI floppy disk drive Date: Thu, 5 Jan 1995 19:44:28 GMT From: jch@netcom.com (Jack Hwang) Organization: NETCOM On-line Communication Services (408 261-4700 guest) Message-ID: This is an old news. Teac discontinue SCSI floppy disk drive FD235SJ. But due to some bureautic clumsiness, the company I am working with did not act on time to secure the supplies. Now it all comes to me to find alternatives. Does anyone know there is a replacement available which is compatiable with MVME147? (Even 147 itself is discontinued! DXXN!!!) --------------------------- Newsgroups: comp.os.vxworks Subject: Stability Date: 4 Jan 1995 19:50:52 GMT From: prabhala@penafiel.irvine.dg.com (Venkata Prabhala) Organization: Data General Corporation, Research Triangle Park, NC Keywords: VxWorks Message-ID: <3eeu6s$hcs@dg-irv.irvine.dg.com> I am a new to RTOS's. But I am certainly UNIX type! I have been wondering if someone on this group could comment on the stability of VxWorks as a development environment in a Unix host-based environment. How well does VxWorks hold up to its promise of remote host-based development/debugging? I would like to hear both positive and negative comments to help my decision. It would also help if someone could point out where I can obtain comparative studies about different RTOSs like VxWorks, VRTX, QNX, pSOS, Solaris etc. You may e-mail me if you think it would increase the S/N ratio on the news group. Thanks. - -- Venkat - -- Venkata Prabhala, Data General Corporation | (prabhala@irvine.dg.com) 2603 Main St., Ste. 360 | FAX: +1 714 724 3956 Irvine, CA 92714 | VOICE: +1 714 724 3569 --------------------------- End of New-News digest ********************** From djb@rtp.co.uk Fri Jan 6 05:24:49 1995 From: djb@rtp.co.uk (David Bullimore) Date: Fri Jan 6 05:24:56 PST 1995 Subject: X11- BDF bit map fonts Dear All, Does anyone know of a utility that converts Berkeley 'BDF' bit map fonts to 'PCF' format for use with the Visicom X server? David *********************************************** * David Bullimore * * Real Time Products Ltd. (RTP Ltd.) * * Unit 8 * * Holt Court * * Aston Science Park * * Birmingham * * B7 4EJ * * England * * * * Phone: +44.21.333 6955 * * Fax: +44.21.333 5433 * * * * Email djb@rtp.co.uk * *********************************************** From cook@positron.med.ge.com Fri Jan 6 06:59:41 1995 From: cook@positron.med.ge.com (Mike Cook) Date: Fri Jan 6 06:59:48 PST 1995 Subject: DOS & large drive Anybody out there using large drives (> 4 gigabyte) with VxWorks DOS? I've recently configured VxWorks 5.1.1 DOS with the Seagate Elite 9 scsi 9.8 gigabyte drive. With the exception that ioctl( fd, FIONFREE, &freeBytes ) doesn't work (you can't fit 9.8 gig in a 'long'!), it is functional. Performance is the issue. If I merely read/write multiples of clusters, being careful to align the writes on cluster boundaries, performance is about what I expect for our controller (2.5-3 meg per sec with xfers > 1 meg). Note that we strictly use contiguous files on the order of 3-100 megabytes. But when I read/write to file offsets not on cluster boundaries... performance goes right out the window (less than half). I believe this is due to the large cluster size of 0x25000 (74 sectors of 2K bytes), and DOS appears to read/write at a cluster granularity. For example, if I want to read 10K bytes from an arbritrary offset within my contiguous file, DOS reads an entire cluster of 150K bytes (or two clusters if the requested 10K crosses a cluster boundary). Even writes greater than a cluster can result in two additional cluster reads if you don't write even multiples of clusters at cluster boundaries. The DOS implementation forces me to a minimum cluster size of 0x25000 bytes. (9.8 gig / 64K clusters). So going to a smaller cluster size is not an option with this drive if I want to access the entire drive as one DOS volume. I contend, that if DOS would not do the 'buffered I/O' or only do it on a sector (or a few sectors) basis rather than on a cluster basis, my performance problems would go away. Question 1: Is there anyway I can turn off the buffered I/O or reduce the buffer size? Ideally, I would like to be able to be able to open a dos file, seek to an offset that is not cluster aligned and start reading/writing without having the system read/write more than two additional disk sectors. Question 2: If the answer to Question 1 is NO (unless you have source code)....I don't. I'm tempted to try to read/write to the DOS contiguous files using Raw I/O. Would this be advisable? Question 3: Is there a simple means to obtain the clustor number (or ideally the disk sector) for the start of a DOS contiguous file? (to aid my raw i/o if I choose to go that route) Question 4: Do you have any other suggestions as to how to improve the performance? Mike Cook GE Medical Systems cookm@med.ge.com (414) 548-4791 From leonid@rst.co.il Sat Jan 7 23:54:04 1995 From: leonid@rst.co.il (Leonid Rosenboim) Date: Sat Jan 7 23:54:10 PST 1995 Subject: Re: SCSI floppy disk drive > Date: Thu, 5 Jan 1995 19:44:28 GMT > From: jch@netcom.com (Jack Hwang) > This is an old news. Teac discontinue SCSI floppy disk drive FD235SJ. > But due to some bureautic clumsiness, the company I am working with > did not act on time to secure the supplies. Now it all comes to me > to find alternatives. Does anyone know there is a replacement available > which is compatiable with MVME147? (Even 147 itself is discontinued! > DXXN!!!) Jack, this has been expected to happen for some time. The SCSI floppies are too expensive anyway, and turns out Teac didn't find them profitable either. Here are a couple of ideas you could consider: 1. If replacing the 147, find a board with both SCSI and a CP-stype floppy controller on board. Take a look at Wind River's 3rd party catalog. 2. Floptical drives have been suggested more then once on this list as a viable alternative for floppies. I haven't seen them in use yet. 3. This is my presonal favorite: QIC tape drives are quite popular in SCSI-world, and not very expensive either. (mini-QICs are cheaper yet). It is just so happens that we sell a software solution for SCSI tapes on VxWorks. It lets you do backup / restore, software updates, and even boot from a tape when starting from scratch. You must have a hard disk on the system, or take a chunk of RAM and build a RAM-disk, cause the tape is not direct access. Please contact me directly for more details regarding the SCSI tape solution. ----------------------------------------------------------------------- Leonid Rosenboim Phone: +972-3-559-8144 R S T Software Industries Ltd. +972-50-307-142 3 Hashikma st. Industrial Zone, Fax: +972-3-559-8244 P.O.Box 11502, AZUR 58017, Israel E-Mail: leonid@rst.co.il From ntmtv.com!isaac@ntmtv.portal.com Mon Jan 9 14:23:38 1995 From: isaac@ntmtv.com Date: Mon Jan 9 14:23:45 PST 1995 Subject: Re: SCSI floppy disk drive > Submitted-by leonid@rst.co.il Sat Jan 7 23:54:04 1995 > Submitted-by: leonid@rst.co.il (Leonid Rosenboim) > > > Date: Thu, 5 Jan 1995 19:44:28 GMT > > From: jch@netcom.com (Jack Hwang) > > This is an old news. Teac discontinue SCSI floppy disk drive FD235SJ. > > But due to some bureautic clumsiness, the company I am working with > > did not act on time to secure the supplies. Now it all comes to me .... > > Jack, this has been expected to happen for some time. The SCSI floppies > are too expensive anyway, and turns out Teac didn't find them profitable > either. Here are a couple of ideas you could consider: > > 1. If replacing the 147, find a board with both SCSI and a CP-stype floppy > controller on board. Take a look at Wind River's 3rd party catalog. > > 2. Floptical drives have been suggested more then once on this list > as a viable alternative for floppies. I haven't seen them in use yet. > > 3. This is my presonal favorite: QIC tape drives are quite popular > in SCSI-world, and not very expensive either. (mini-QICs are cheaper yet). > It is just so happens that we sell a software solution for SCSI tapes on We are faced with a similiar problem as Jack with the finding a replacement to the 4MB Teac SCSI floppy drive. Are there many others with the same problem? I was wondering what others experience with the Floptical has been. I like the Floptical solution because it has a SCSI interface, has increased capacity and also will read 2MB floppies. But will it also get discontinued like the 4MB drives? Are there any major users of the Floptical, with sufficient demand to keep the company in business? Are there any companies that make equipment for Software Duplication on the Floptical or QIC tape drives? Any info would be appreciated. Thanks ------------------------------ Roy Isaac UUCP:{ames,mcdcup}!ntmtv!isaac Northern Telecom INTERNET: isaac@ntmtv.com Mountain View, CA PHONE: 415-940-2539 From del@rst.com Mon Jan 9 14:28:52 1995 From: del@rst.com (Mark Del Giorno) Date: Mon Jan 9 14:29:00 PST 1995 Subject: Programming FLASH on the MV162 I'm looking for some source code that can burn FLASH on a MVME162 (it uses the Intel 28F020). My ultimate goal is to be able to burn my application into FLASH and run on power up (embedded - with no ethernet attached). Wind River has a way that I can download a hex file via a serial link - this is o.k., but I can't believe I can't accomplish the same thing via the ethernet connection. When I asked them for code to burn the FLASH, I got the following response: TITLE: Programming Flash on the MV162 from within vxWorks KEYWORDS: MV162, Flash SYSTEM: All DESCRIPTION: Unsupported. SYMPTOM(S): Customer wishes to programmatically modify the flash eprom on a MV162. RESOLUTION: Programmatically writing flash from vxWorks is strongly discouraged. This is reportedly very messy and very slow. You have to erase all of flash prior to reprogramming. This is definitely unsupported. Just because it's "messy and slow" doesn't seem like a real good reason to me that Wind River should not support it!!!!! Thanks in advance for any help. _/_/_/_/ _/_/_/_/ _/_/_/_/_/ Mark Del Giorno (del@rst.com) _/ _/ _/ _/ (410) 876-9200 Fax: 876-9470 _/_/_/_/ _/_/_/_/ _/ Robotic Systems Technology _/ _/ _/ _/ 1110 Business Parkway South _/ _/ _/_/_/_/ _/ Westminster, MD 21157 From pf1122@tcville.es.hac.com Mon Jan 9 15:47:14 1995 From: pf1122@tcville.es.hac.com (Kim Chacon) Date: Mon Jan 9 15:47:22 PST 1995 Subject: Re: Programming FLASH on the MV162 Mark Del Giorno wrote: --------------------------------------------------------------------- > Submitted-by del@rst.com Mon Jan 9 14:28:52 1995 > Submitted-by: del@rst.com (Mark Del Giorno) > > I'm looking for some source code that can burn FLASH on a MVME162 > (it uses the Intel 28F020). My ultimate goal is to be able to burn > my application into FLASH and run on power up (embedded - with no > ethernet attached). Wind River has a way that I can download a hex > file via a serial link - this is o.k., but I can't believe I can't > accomplish the same thing via the ethernet connection. We have done this. We have installed a Flash EEPROM industry pack on our MVME162 card. We purchased software from a company named CompWare Corp. The contact person at Compware that we have dealt with is Paul Anderson (602-998-8650). Call him to get price information. I did not personally order the software, so I don't have that information. The software contains a flashLib, which provides C functions such as: flashShow(), flashWrite(), eraseFlash(), etc. We have successfully run our code from a Dos Filesystem relocated from a RAM disk device to the Flash EEPROM. The files were relocated via writeFlash(). I hope this helps, Kim M. Chacon Hughes Aircraft Company email: chacon@tcville.es.hac.com From daemon@vxw.ee.lbl.gov Tue Jan 10 04:00:49 1995 From: daemon@vxw.ee.lbl.gov Date: Tue Jan 10 04:01:14 PST 1995 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Tue Jan 10 04:00:24 PST 1995 Subject: Re: Getting printf() to work in "real-time" Subject: Re: NTDS boards Subject: CPU Idle time on Vxworks Subject: Re: RAMDISK/ROMDISK Subject: Re: VADSWorks (Ada on VxWorks) Subject: Re: Getting printf() to work in "real-time" Subject: CPU idle time on VxWorks -- Again Subject: Changing file times. Subject: re: RAMDISK/ROMDISK Subject: Re: CPU idle time on VxWorks -- Again Subject: Re: Changing file times. Subject: Re: CPU Idle time on Vxworks Subject: Has anyone ported scandir() ? Subject: Re: VADSWorks (Ada on VxWorks) Subject: Re: GreenHills Ada on VxW Subject: Re: NTDS boards Subject: Re: CPU Idle time on Vxworks Subject: ld error Subject: Re: CPU Idle time on Vxworks Subject: Re: Driver/Source for a Motorola MVME376 Ethernet pcb Subject: Re: TCP/IP & Mbuf configuration Subject: Re: Can VxWorks coexist with other OS on VMEbus? Subject: Embedded Systems Design Symposium Subject: Green Hills Subject: Timer interrupts in VxWorks Subject: Re: Timer interrupts in VxWorks Subject: Re: Timer interrupts in VxWorks Subject: FTP site for MVE166/167 Subject: Help! ISR causes NFS to die! Subject: dosFs VOLUME_NOT_AVAILABLE Subject: Motorola MC68681 driver Subject: FAQ and ftp site info Subject: FAQ and ftp site info Subject: ld MEMORY command, multiple regions per section possible? Subject: Re: FAQ and ftp site info Subject: SCSI floppy disk drive Subject: Stability Subject: Re: SCSI floppy disk drive Subject: Fortran compiler & tools/ Vxworks on Alpha Subject: Re: DOS & large drive Subject: Database on VxWorks Subject: Re: Changing file times. Subject: Looking For A File Transfer Protocol Subject: Looking for new VxWorks File System Subject: Can't ftp thor.atd.ucar.edu Subject: Re: Can't ftp thor.atd.ucar.edu Subject: Fortran Compiler for Alpha/VXworks Subject: Cartridge Tape Driver Subject: smObjLocalToGlobal() Question Subject: Email address Wind River Systems, Visicom ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: Getting printf() to work in "real-time" Date: Mon, 19 Dec 1994 11:44:07 GMT From: jh@ns.ph.liv.ac.uk (Jomar Honsi) Organization: University of Liverpool Nuclear Structure Group Message-ID: References: <9412162057.AA03904@lassen> Reply-To: jh@ns.ph.liv.ac.uk Sender: news@liverpool.ac.uk (News System) In article AA03904@lassen, stan@lassen.rti.com (Stan Schneider) writes: >>> Submitted-by del@rst.com Thu Dec 15 10:48:59 1994 > >>> for (i = 0 ; i < n ; i++) >>> { >>> if (some_signal() > TOO_HIGH) >>> printf("Error - cycle %d\n", i); >>> } > >>> I don't get anything printed to the screen until the go() routine exits >>> completely, then I get a complete dump of all the cycle numbers that the >>> errors occurred. > >Could this be simply that you're starving the netTask with an infinite loop at >tShell priority? Try setting the shell to a lower priority via >taskPrioritySet. > > -- Stan > Why not use logMsg ? (and watch the priorities) - --- ** Jomar Honsi Phone: + (44)(51)7943398 ** Department of Physics E-mail: jh@ns.ph.liv.ac.uk ** University of Liverpool, UK --------------------------- Newsgroups: comp.os.vxworks Subject: Re: NTDS boards Date: 19 Dec 1994 20:58:19 GMT From: Fran Horan Organization: JHU/APL Message-ID: <3d4s5b$4km@aplcomm.jhuapl.edu> References: <199412161938.OAA21574@piranha.ssd.ray.com> NTDS boards We use the Antares VMENIO board sold by Visicom in San Diego Ca. for several of our mission-critical embedded systems on Navy ships. Their phone number is 619-457-2111. We have several drivers developed in house but Visicom was planning to release a driver last time I talked to them. I would investigate theirs first and see if it is made to work with vxWorks. We are happy with their product and support. Fran Horan franfh@aplexus.jhuapl.edu FranHoran@aol.com --------------------------- Newsgroups: comp.os.vxworks Subject: CPU Idle time on Vxworks Date: 19 Dec 1994 17:00:08 -0600 From: sureshr@bnr.ca (Suresh Rao) Organization: Bell-Northern Research, Richardson, TX Message-ID: <3d539o$58d@crchha82.bnr.ca> Sender: sureshr@bnr.ca (Marco Marchetti) I need to find out the CPU Idle time (CPU load statistics) on Vxworks. Does anybody have any ideas about how to get it ? --------------------------- Newsgroups: comp.os.vxworks Subject: Re: RAMDISK/ROMDISK Date: Tue, 20 Dec 1994 01:15:03 GMT From: miked@wrs.com (Mike Deliman) Organization: Wind River Systems, Inc. Message-ID: References: Sender: news@wrs.com (News Manager) In article rsgawera@wg.icl.co.uk writes: >Dear VxWorkers, > Is there any easy way of building a file system in Flash memory ie >some sort of RAMdisk. We have a board running vxworks with a Flash card and we >want to store software on the Flash card. If a simple file system were created, >we could easily pull software off the card and also update the software. > > Thanks in advance, > Raj Gawera > >Raj Gawera >ICL CSD >email : rsgawera@wg.icl.co.uk Raj, check out the sources in /src/usr - look at the memDrv.c and ramDrv.c files. Build something like this for your flash.... "flashDrv.c", and use it with dosFs. Sincerely -miked - -- - ------------------------------------------------------------------------------ Mike Deliman <> miked@wrs.com <> phone: 510-748-4100 <> facs: 510-814-2164 Snail Mail: Wind River Systems, 1010 Atlantic Ave, Alameda CA 94501 USA - ------------------------------------------------------------------------------ --------------------------- Newsgroups: comp.os.vxworks Subject: Re: VADSWorks (Ada on VxWorks) Date: 20 Dec 1994 08:58:42 GMT From: corey@rational.com (Corey Ashford) Organization: Rational Software Corporation Message-ID: <3d66c2$snb@rational.rational.com> References: <00988FB6.3321E960.19@cub.kscorp.com> Bob Dowling (dowling@cub.kscorp.com) wrote: : Recently, I replied to a question from Matt Whiting (whiting_ms@corning.com) : about Ada compilers available for use with VxWorks. I wrote: : : > I've used the Rational/Verdix VADSworks cross-compiler on SUNs for the : > Motorola 68020/030 on the ubiquitous MVME-147. They can be reached at : > (800) 289-8237 or (703) 318-5800. : John R. Cobarruvias (cobarruvias@asd1.jsc.nasa.gov) saw this on : comp.os.vxworks, and sent me the following, which may interest you: : This is the only one I have heard of. VADSworks is pretty good. I have used : it before with an R3000 board. I would be VERY cautious with the Rational : APEX product coupled with VADSWorks. Try it before you buy it. : VADSWorks has been successfully used by a number of people. It's a great set : of tools. : A couple of things you should be aware of: : 1) VADSWorks is VxWorks coupled with Verdix Ada. It is sold by Verdix (now : Rational) as a single product. : 2) Because of 1, you get your tech support from Rational/Verdix NOT from : [Wind River]. : 3) It has caused us much grief if Verdix doesn't or can't answer our : questions. Then they had to go to WRS then get back to us. We couldn't ask : WRS directly. I think this MAY have changed recently. This hasn't changed much. : 4) Verdix gets VxWorks from WRS, then integrates it with VADS. Because of : this, we were usually behind on VxWorks newest releases and fixes since it : has to go through Verdix first. The problem was worse since we were using a : little-used board. 680x0 should have less of a problem. Starting Q1 1995, the VADSworks product will be layered on top of VxWorks in such a way that allows us to not need to alter any VxWorks source. This will enable us to keep up with VxWorks releases in a timely manner. In fact, the layer is so clean that the VADSworks product will be treated like an add-on product to VxWorks. This product has been in development for a couple of years and runs on 68k, MIPS, and soon DEC Alpha targets. : 5) If you go with APEX, then you have another layer to integrate. APEX over : VADS over VxWorks. I would be very cautious about this product. Try before : you buy. We did and we won't use it. APEX II (which is a combination of APEX's front-end tools with VADS back-end and debugger + plus some new features) will be integrated with VADSworks. This will be a much smoother product than the present combination of APEX and VADS. I don't know the projected date, but I'd guess it's at least Q3 1995. - - Corey == My comments do not represent any official position of Rational Software Corp. Please contact a sales rep for schedules == Internet: corey@rational.com Sneakernet: down_the_hall!right_at_the_mens_room!past_the_stain_on_the_wall!... Rational Software Corporation 1600 N.W. Compton Drive, Suite #357 Aloha, OR 97006 ph# (503) 690-1116 x6710 --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Getting printf() to work in "real-time" Date: Tue, 20 Dec 1994 14:15:32 GMT From: clegg@ssd.fsi.com Organization: Flightsafety International, Broken Arrow, OK Message-ID: <1994Dec20.141532.31676@dev1.ssd.fsi.com> References: <9412162057.AA03904@lassen> Reply-To: clegg@ssd.fsi.com Sender: news@dev1.ssd.fsi.com >In article AA03904@lassen, stan@lassen.rti.com (Stan Schneider) writes: > Submitted-by del@rst.com Thu Dec 15 10:48:59 1994 > > for (i = 0 ; i < n ; i++) > { > if (some_signal() > TOO_HIGH) > printf("Error - cycle %d\n", i); > } > > I don't get anything printed to the screen until the go() routine exits > completely, then I get a complete dump of all the cycle numbers that the > errors occurred. > I didn't see the orignal posting, however, if you are logging into the VxWorks system over ethernet, you may need to flush(stdout) the buffer once in a while to get real time output. R. Clegg clegg@ssd.fsi.com --------------------------- Newsgroups: comp.os.vxworks Subject: CPU idle time on VxWorks -- Again Date: 20 Dec 1994 08:53:23 -0600 From: sureshr@bnr.ca Organization: Bell-Northern Research, Richardson, TX Message-ID: <3d6r53$76q@crchha82.bnr.ca> Sender: sureshr@bnr.ca (Marco Marchetti) Spy command gives timing for individual tasks. I was looking for some command which gives the overall picture. --------------------------- Newsgroups: comp.os.vxworks Subject: Changing file times. Date: Tue, 20 Dec 1994 15:21:27 GMT From: heller@nm.picker.com (Mark Heller) Organization: Picker International, Nuclear Medicine Division Message-ID: <1994Dec20.152127.9206@picker.com> Sender: news@picker.com Hi All, I'm running VxWorks 5.1.1 and looking for a way to change the dates on files on a Dos filesystem. After going through the documentation I can't find an equivalent call to the Unix utime function. Does anyone know how to change the date on a VxWorks file. Thanks - ----- - -Mark - ---------------------------------------------------------------- Mark Heller Picker International Inc. heller@nm.picker.com Ohio Imaging Nuclear Medicine Division phone 216-473-7626 23060 Miles Road fax 216-473-7390 Bedford Heights, Ohio 44128 - ---------------------------------------------------------------- --------------------------- Newsgroups: comp.os.vxworks Subject: re: RAMDISK/ROMDISK Date: Tue, 20 Dec 1994 16:24:04 GMT From: dat85715@indirect.com (D'Anne Thompson) Organization: Internet Direct, indirect.com Message-ID: References: <9412200920.AA00735@rst.co.il> Sender: dat85715@bud.indirect.com In article <9412200920.AA00735@rst.co.il> leonid@rst.co.il (Leonid Rosenboim) writes: > > We recently completed a project which used a Radstone SME-1 4 MByte > > Flash EEPROM board as a WFRM (Write Few, Read Many) solid state > > "disk". We used the vxWorks-supplied ramDrv as a model to create a > > flashDrv driver. We only had to stick some Radstone-specific Flash > > EEPROM programming code into the ramBlkWrt() function. Next, we had > > vxWorks create a DOS filesystem on the board. ... > > This is not a very good idea: the DOS filesystem is updating certaing > sectors very frequently (FATs, root dir, dirs), and the Flash parts > all have a limitted erase-cycle life-time (means you can only erase > a sector up to 100,000 times, after that the sector is dead). > Depending on the type of parts you are using, and their respective > erase cycle limit, the board will die on you. > > The other drawback of this simplistic approach is that if a power > failure occurs in the midst of the DOS file system updating the FAT, > i.e. just after it erased the FAT sector(s) but before they have > been reprogrammed, then you loose some or all of your files. > > The only situation where such a solution is feasible, is when the files > are updated only during software version upgrades or system reconfiguration, > i.e. very infrequently and in a controlled manner. This is by no means > a general solution ! For a solution using the Microsoft Flash File System, I would contact RAMIX, Inc. The do pcmcia cards and software for VxWorks, and have software for this. There are two standards for file systems on Flash devices and the Microsoft FFS is one of them (can't recall the other). It implements a 'wear levelling' strategy to spread the usage among the flash blocks equally. I would suggest contacting RAMIX at 1-818-349-6772. Tell them D'Anne sent you (maybe they'll give me a discount next time). D'Anne Thompson dat85715@indirect.com dee@ioinc.tucson.az.us --------------------------- Newsgroups: comp.os.vxworks Subject: Re: CPU idle time on VxWorks -- Again Date: Tue, 20 Dec 1994 16:39:22 GMT From: dat85715@indirect.com (D'Anne Thompson) Organization: Internet Direct, indirect.com Message-ID: References: <3d6r53$76q@crchha82.bnr.ca> Sender: dat85715@bud.indirect.com In article <3d6r53$76q@crchha82.bnr.ca> sureshr@bnr.ca writes: > > Spy command gives timing for individual tasks. I was looking for some command > which gives the overall picture. If you are looking for a module that will give you a simple percentage of cpu busy (or idle) try the VxWorks archive 'loadMeter.shar'. The manual page is what you will find there. To get the source you will have to write to 'bob@noao.edu' and ask him for a copy (identify yourself and what you plan to do with the software - he's government funded and they want to know who is getting their software and what they are doing with it - 'technology transfer' you know). Anyway here's a synopsis. loadMeter(2) SYSTEM CALLS loadMeter(2) loadMeter.c - cpu load meter module SYNOPSIS loadMeterSysInit - initialize the load meter system loadMeterPercentSet - set a load meter value loadMeterPercentGet - get current average load value (in percent) loadMeterOnOffSet - turn load meter on/off loadMeterOnOffGet - return current on/off status of load meter system loadMeterAsString - string representation of current load STATUS loadMeterSysInit (int delay, int samples) STATUS loadMeterPercentSet (double percent) double loadMeterPercentGet (void) STATUS loadMeterOnOffSet (BOOL flag) BOOL loadMeterOnOffGet (void) char *loadMeterAsString (void) DESCRIPTION This module implements a cpu load meter for a vxWorks system. Current load is based upon the number of cycles through the idle task loop in a given period of time. The higher the number, the lower the cpu loading. A low number of cycles represents a high cpu loading. (Idle time is (100% - load%)). D'Anne Thompson dat85715@indirect.com dee@ioinc.tucson.az.us --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Changing file times. Date: 20 Dec 94 17:55:14 GMT From: froeber@bbn.com (Fred Roeber) Organization: Bolt, Beranek and Newman Inc. Message-ID: References: <1994Dec20.152127.9206@picker.com> heller@nm.picker.com (Mark Heller) writes: > I'm running VxWorks 5.1.1 and looking for a way to change > the dates on files on a Dos filesystem. See the function dosFsTimeSet in VxWorks. Fred - -- | Fred J Roeber, BBN Systems and Technologies | | 50 Enterprise Center, Suite 201 | | Middletown, RI 02842-1087 (401) 849-2543 | | froeber@bbn.com | --------------------------- Newsgroups: comp.os.vxworks Subject: Re: CPU Idle time on Vxworks Date: 20 Dec 1994 11:08:20 GMT From: gournay@news.cea.fr (J.F. Gournay) Organization: Resumix, Inc., Santa Clara, CA Message-ID: <3d6dv4$939@anemone.saclay.cea.fr> References: <3d539o$58d@crchha82.bnr.ca> Reply-To: gournay@news.cea.fr Use the spy comand. It will give you some outputs like the following: NAME ENTRY TID PRI total % (ticks) delta % (ticks) - -------- -------- ----- --- --------------- --------------- tExcTask _excTask 3dcc54 0 0% ( 23) 0% ( 0) tLogTask _logTask 3da354 0 0% ( 0) 0% ( 0) tShell _shell 3abc70 1 0% ( 1) 0% ( 1) tRlogind _rlogind 3bab28 2 0% ( 0) 0% ( 0) tTelnetd _telnetd 3b8c84 2 0% ( 0) 0% ( 0) tRlogOutTask _rlogOutTask 2cf300 2 1% ( 59) 2% ( 14) tRlogInTask _rlogInTask 2a7d0c 2 0% ( 0) 0% ( 0) tSpyTask _spyTask 2a67f4 5 2% ( 82) 2% ( 11) tRdbTask _rdbTask 3b6254 20 0% ( 0) 0% ( 0) timeStamp _gts_select 2d4ac0 32 0% ( 1) 0% ( 0) callback _callbackTask 2d6bb0 40 0% ( 0) 0% ( 0) scanEvent 2b6688 41 0% ( 0) 0% ( 0) tNetTask _netTask 3d5fc8 50 0% ( 14) 0% ( 0) scanPeriod 2b8f48 53 0% ( 1) 0% ( 1) scanPeriod 2bb808 54 0% ( 0) 0% ( 0) tFtpdTask _ftpdTask 3b3d9c 55 0% ( 0) 0% ( 0) scanPeriod 2be0c8 55 0% ( 0) 0% ( 0) scanPeriod 2c0988 56 0% ( 0) 0% ( 0) callback _callbackTask 2d9844 57 0% ( 0) 0% ( 0) scanPeriod 2c3248 57 0% ( 0) 0% ( 0) scanPeriod 2c5b08 58 0% ( 0) 0% ( 0) scanPeriod 2c83c8 59 0% ( 0) 0% ( 0) callback _callbackTask 2dc4d8 65 0% ( 0) 0% ( 0) tPortmapd _portmapd 3b776c 100 0% ( 0) 0% ( 0) CA TCP _req_server 2d2b78 181 0% ( 0) 0% ( 0) CA UDP _cast_server 2d1600 182 0% ( 0) 0% ( 0) CA online _rsrv_online_notify_task 2d0078 183 0% ( 0) 0% ( 0) taskwd 2ddde4 200 0% ( 0) 0% ( 0) KERNEL 0% ( 32) 0% ( 3) INTERRUPT 0% ( 11) 0% ( 0) IDLE 93% ( 3426) 94% ( 495) TOTAL 96% ( 3650) 98% ( 525) --------------------------- Newsgroups: comp.os.vxworks Subject: Has anyone ported scandir() ? Date: 20 Dec 1994 17:44:54 -0800 From: gnn@tx.abekas.com (George Neville-Neil) Organization: Abekas Video Systems, Inc. Message-ID: Hi Folks, I was wondering if anyone has ported Berkeley's scandir function to vxWorks? Thanks, George - -- George V. Neville-Neil work: gnn@abekas.com home:gnn@netcom.com Home Page: ftp://ftp.netcom.com/pub/gnn/gnn.html - -- George V. Neville-Neil work: gnn@abekas.com home:gnn@netcom.com Home Page: ftp://ftp.netcom.com/pub/gnn/gnn.html --------------------------- Newsgroups: comp.os.vxworks Subject: Re: VADSWorks (Ada on VxWorks) Date: 21 Dec 1994 05:40:35 GMT From: COBARRUVIAS@ASD1.JSC.NASA.GOV (COBARRUVIAS_JOHN) Organization: NASA/JSC Message-ID: <3d8f4j$att@pendragon.jsc.nasa.gov> References: <00988FB6.3321E960.19@cub.kscorp.com> <3d66c2$snb@rational.rational.com> In <3d66c2$snb@rational.rational.com> corey@rational.com writes: [points 1-4 deleted] > > : 5) If you go with APEX, then you have another layer to integrate. APEX over > : VADS over VxWorks. I would be very cautious about this product. Try before > : you buy. We did and we won't use it. > > APEX II (which is a combination of APEX's front-end tools with VADS > back-end and debugger + plus some new features) will be integrated with > VADSworks. This will be a much smoother product than the present > combination of APEX and VADS. I don't know the projected date, but I'd > guess it's at least Q3 1995. > Again, my point still stands. Try before you buy. I've heard about this integration of front end products (which are pretty good) and backend products (which are also pretty good) way back in '89 when it was called rcf. Now it is APEX and now APEX II and I still haven't SEEN the integration. Try before you buy. > > - Corey > > == My comments do not represent any official position of Rational Software > Corp. Please contact a sales rep for schedules == > > Internet: corey@rational.com > Sneakernet: down_the_hall!right_at_the_mens_room!past_the_stain_on_the_wall!... > Rational Software Corporation > 1600 N.W. Compton Drive, Suite #357 > Aloha, OR 97006 > ph# (503) 690-1116 x6710 --------------------------- Newsgroups: comp.os.vxworks Subject: Re: GreenHills Ada on VxW Date: 21 Dec 94 11:43:33 -0400 From: dyer@nrlvx1.nrl.navy.mil (Doug Dyer) Organization: NRL SPACE SYSTEMS DIVISION Message-ID: <1994Dec21.114333.1515@nrlvx1> References: <9412160956.AA01103@rtp.co.uk> In article <9412160956.AA01103@rtp.co.uk>, djb@rtp.co.uk (David Bullimore) writes: > Re: >> original question from Matt Whiting (whiting_ms@corning.com) >> about Ada compilers available for use with VxWorks. > > GreenHills are selling their Ada compiler for use with VxWorks. Has > anyone had experience of using this combination? > > I got a rather lengthy dose from a greenhills sales demo recently. Seems their one (_big_) advantage (which he didn't really understand) was that there was apparently no modification of the vxworks kernel. Therefore, unlike vadsworks, vxworks may be purchased and supported by WRS directly (and upgrades such as 5.2 should come quicker - no waiting on rational to rebuild their version of it). One problem for us is that they no longer support the SGI (but "for us" they'd do it... yeah). He shied away from providing references in a big way. Im still interested with their low cost, and since I sometimes have to modify the kernel in some way with the sometimes wacky hardware I deal with, an unmodified WRS kernel is a good plus, IMHO. Also, I feel that WRS is more supportive for kernel mods than I sense rational is (IMO). There was the usual dosage of compiler specs, and their code management stuff... Im so uninterested in that. I was/am looking for a reliable transparent ADA for vxworks and walked away with an unsure feeling on their package. Anyone care to rant and rave about MULTI? BTW, their MULTI GUI is for the *DOGS*. bye, Doug - -- Doug Dyer - dyer@alx.sticomet.com | Real-time systems . * //[]// Software Technology, Inc. (STI) | Space flight/ground software ^^ . DC office: (703) 329-9707 | vxWorks BSPs, etc. . --------------------------- Newsgroups: comp.os.vxworks Subject: Re: NTDS boards Date: Wed, 21 Dec 1994 21:21:16 GMT From: rrailey@relay.nswc.navy.mil (Rene H. Railey) Organization: NSWCDD, Dahlgren, VA 22448 Message-ID: References: <199412161938.OAA21574@piranha.ssd.ray.com> <3d4s5b$4km@aplcomm.jhuapl.edu> Sender: news@relay.nswc.navy.mil (oanews) In article <3d4s5b$4km@aplcomm.jhuapl.edu> Fran Horan writes: >Path: oanews!relay2!relay!night.primate.wisc.edu!caen!zip.eecs.umich.edu!newshost.marcam.com!news.mathworks.com!europa.eng.gtefsd.com!howland.reston.ans.net!news.sprintlink.net!uunet!netnews.jhuapl.edu!fran.jhuapl.edu!franfh >From: Fran Horan >Newsgroups: comp.os.vxworks >Subject: Re: NTDS boards >Date: 19 Dec 1994 20:58:19 GMT >Organization: JHU/APL >Lines: 11 >Distribution: world >Message-ID: <3d4s5b$4km@aplcomm.jhuapl.edu> >References: <199412161938.OAA21574@piranha.ssd.ray.com> >NNTP-Posting-Host: fran.jhuapl.edu >X-Newsreader: Nuntius Version 1.2 >X-XXMessage-ID: >X-XXDate: Mon, 19 Dec 1994 20:59:19 GMT >NTDS boards >We use the Antares VMENIO board sold by Visicom in San Diego Ca. for >several of our mission-critical embedded systems on Navy ships. Their >phone number is 619-457-2111. We have several drivers developed in house >but Visicom was planning to release a driver last time I talked to them. >I would investigate theirs first and see if it is made to work with >vxWorks. We are happy with their product and support. >Fran Horan >franfh@aplexus.jhuapl.edu >FranHoran@aol.com In addition we are in looking at a NTDS board from Sabtech Industries. The Hawk model has been out for sometime and a new Falcon model is up for release. You will find that the Antares cards are simple in their design with no onboard CPU for decision. The Sabtech Boards have a bp driver for vxWorks and as of now use a 68020 processor for more flexability. We have used Antares Cards in the past and have been very pleased. This move to Sabtech is to explore other options. A Sabtech rep can be reached at NewTech Associates at (301) 937-6822. They are Sabtech resellers. Rene H. Railey ! This is my opinion Naval Surface Warfare Center ! and does not necessarily Dahlgren Division Code F33 ! reflect the views of my Dahlgren, VA 22448-5000 ! employer. (703) 663-6276 rrailey@relay.nswc.navy.mil P.S Elvis is a deadhead --------------------------- Newsgroups: comp.os.vxworks Subject: Re: CPU Idle time on Vxworks Date: Thu, 22 Dec 1994 14:18:31 GMT From: clegg@ssd.fsi.com Organization: Flightsafety International, Broken Arrow, OK Message-ID: <1994Dec22.141831.25502@dev1.ssd.fsi.com> References: <3d539o$58d@crchha82.bnr.ca> <3d6dv4$939@anemone.saclay.cea.fr> Reply-To: clegg@ssd.fsi.com Sender: news@dev1.ssd.fsi.com In <3d6dv4$939@anemone.saclay.cea.fr>, gournay@news.cea.fr (J.F. Gournay) writes: >Use the spy comand. It will give you some outputs like the following: > Be careful when using the spy command. We have found that under certain circumstances it is unstable. During one particularly bad instance, it was saying that a task was using 89-100% of the CPU time when the task was actually using 10% to a max of 30%. The same task was also indicated by spy() to be using 0% of the CPU time later when it was running under the same state using the same amount of time. We eventually gave up on spy and wrote our own routines which used the VMEChip2 timers and allowed us to measure min/max and avg time inside a task. Granted, in a multitasking system, these times may include execution of higher priority tasks, but I am really only concerned with how much a frame a task takes up in ANY circumstance and not just the average, which is the only thing that spy can measure. Even though spy uses the PCCChip2 timers, I believe you could gather the same information easily by counting the number of timer interrupts (ticks) between times and adding the timer count to get the current "time" in microseconds. FYI Roger Clegg clegg@ssd.fsi.com --------------------------- Newsgroups: comp.os.vxworks Subject: ld error Date: 22 Dec 1994 21:07:10 GMT From: bqv@se19.wg2.waii.com (Bang Q. Vu) Organization: Western Geophysical Exploration Products Keywords: ld Message-ID: <3dcppu$3k7a@ep130.wg2.waii.com> Reply-To: vu@wg2.waii.com I used the "ld" command to load an object file from a SPARC station to a MVME147 target and got the following error: loadAoutLib error: insufficient memory for externals table (need 209872 bytes) ld error: error reading file. I used the ld option that add only global symbols to the symbol table. Am I getting a "too many symbols" error here? If so, how do I overcome this limitation? - -- Bang Vu Western Geophysical, a division of Western Atlas International Inc. ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ Email : vu@wg2.waii.com Voice : (713) 964-6192 Fax : (713) 964-6218 --------------------------- Newsgroups: comp.os.vxworks Subject: Re: CPU Idle time on Vxworks Date: 22 Dec 1994 22:56:19 GMT From: berg@fnal.fnal.gov (David Berg) Organization: FERMILAB, Batavia, IL Message-ID: <3dd06j$25m@fnnews.fnal.gov> References: <1994Dec22.141831.25502@dev1.ssd.fsi.com> Reply-To: berg@fnal.fnal.gov In article 25502@dev1.ssd.fsi.com, clegg@ssd.fsi.com () writes: >In <3d6dv4$939@anemone.saclay.cea.fr>, gournay@news.cea.fr (J.F. Gournay) writes: >>Use the spy comand. It will give you some outputs like the following: >> >Be careful when using the spy command. We have found that under certain >circumstances it is unstable. During one particularly bad instance, ... spy() uses the auxiliary clock, which for the mv167, is implemented by a PCCchip2 timer. The system clock for the mv167 is also implemented using a PCCchip2 timer. Although they are separate counters, they both derive from the same clock, and are therefore synchronous. The default tic rates are 100 Hz for the auxiliary clock and 60 Hz for the system clock, which are fairly commensurate (least common multiple is 300). Kernel scheduling events triggered by the system clock have a good chance of maintaining a constant phase with respect to the auxiliary clock; therefore, spy() is prone to systematic errors. The best alternative is to implement the auxiliary clock using a timer that is truly independent of the system clock timer, eg, one of the VMEchip2 timers. You might also vary the tic rates; try using prime numbers, eg, 101 and 61. If you really want to be sure that you're measuring what you want to, I would agree that the best approach is to code your own routines using a PCCchip2 timer. - --------------------------+--------------------------------------------- "A nanosecond here, | David M. Berg (berg@fndaut.fnal.gov) a nanosecond there, | Computing Division and pretty soon you're | Fermilab MS120 talking about Real Time." | PO Box 500 tel: 708-840-3021 (apologies to the late | Batavia IL 60510 fax: 708-840-2783 Sen. Everett Dirksen) | --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Driver/Source for a Motorola MVME376 Ethernet pcb Date: Fri, 23 Dec 1994 02:09:08 GMT From: hjb@netcom20.netcom.com (squeedy) Organization: Peaceful Star, Oakland, CA Message-ID: References: <199412201653.JAA15400@den.mcd.mot.com> Reply-To: hjb@netcom.com Sender: hjb@netcom.com (squeedy) >>>>> "Lenny" == Lenny Bates writes: In article <199412201653.JAA15400@den.mcd.mot.com> Lenny Bates writes: Lenny> Greetings, Does anyone know if a driver and source code for Lenny> the Motorola MVME376 (Hawk) Ethernet interface pcb is Lenny> avaliable. A driver for VXworks, HP/UX or Solaris would be Lenny> considered. the UNIX drivers for the MVME37X (at least MVME372 and MVME374) i have worked on previously are based on BPP (buffered-pipe protocol) and CE (common environment) standard over VME backplane, which were motorola-specific standards for talking to intelligent periperal boards. as far as i know the firmware running on the "stock" MVEM37X expects the BPP/CE for all packet exchanges to and from the host CPU. this mechanism was painfully slow, and it was very hard to get any good performance from these boards. i have never heard of anyone doing a driver that uses BPP/CE for vxworks. the drivers for MVME37X i have hacked on while at unisoft were part of a STREAMS TCP/IP package. the driver interface to the IP layer was different from any other modern STREAMS network device drivers. i believe motorola SVR3.2 for delta boxes with NSE option includes this TCP/IP along with the device driver. this TCP/IP was different from anyone else's; it was entirely written from scratch for STREAMS environment, before any real concrete standards for TPI or DLPI existed. some of the older TPI interface is present in the implementation, but the device driver interface is very different, and would pose serious rewriting for modern operating systems, whether or not the target TCP/IP suite is based on STREAMS. an older release of Vxworks had a optional VME backplane STREAMS driver for this UNIX TCP/IP. i believe there are several sites using this driver currently within SVR3.X on motorola MVME cpu based systems. if you compare this backplane device driver with the original BSD style backplane driver you can figure out how to port a NSE MVME37X driver to HP/UX and SunOS, both of which are based on BSD networking code. as for the BPP/CE motorola might still have some documentation on this. - -- Hwa-Jin Bae (hjb@netcom.com) Peaceful Star, 4001 San Leandro St. #3, Oakland, CA 94601 (510)436-2884 --------------------------- Newsgroups: comp.os.vxworks Subject: Re: TCP/IP & Mbuf configuration Date: Fri, 23 Dec 1994 02:38:14 GMT From: hjb@netcom20.netcom.com (squeedy) Organization: Peaceful Star, Oakland, CA Message-ID: References: <9412142227.AA04474@jupiter.mtvyp> <3cqj6i$s30@fnnews.fnal.gov> Reply-To: hjb@netcom.com Sender: hjb@netcom.com (squeedy) >>>>> "David" == David Berg writes: In article <3cqj6i$s30@fnnews.fnal.gov> berg@fnal.fnal.gov (David Berg) writes: Sanjay> I was wondering how to change the number of mbufs used by Sanjay> the network stack on vxWorks. From what I understand, the Sanjay> way to do this is change a constant in the mbuf.h header Sanjay> file and recompile. Is this correct? in Vxworks 5.1 and later, yes. in vxworks 5.0 and earlier, no. Sanjay> There is a concern that our system is having a Sanjay> producer/consumer problem with the mbufs. Can we check Sanjay> the validity of this scenario out by using mbufShow? I Sanjay> noticed a peg saying "number of times failed to find Sanjay> space". Does this tell you the number of times an mbuf Sanjay> was requested, but the request was blocked because there Sanjay> were not any mbufs available at the time. Are there any Sanjay> other ways? David> I believe this is the number of times it had to malloc more David> mbufs. What happens when you reach the max, or run out of David> memory? Your network hangs until something frees up. the mbufs are allocated one or more clusters at a time. the number you mention is an indication of the number of times this allocation was tried (i.e. memPartAlloc was attempted to mbuf memory partition) but failed. the network code tries its best to drain all the mbufs it can, but the problem may not go away due to incorrect use of mbufs in device drivers. David> I encountered a problem where one of the symptoms was that David> the number of allocated mbufs would shoot up, far above David> what should normally be used. (There was a problem in the David> network interface driver in the way it queued jobs to David> netTask, which runs them serially.) I mention this because David> you may also be seeing a symptom of an underlying problem, David> in which case simply allocating more mbufs won't fix it. a good network device driver should do minimal amount of mbuf management to prevent mbuf exhaustion. especially, smart network drivers that loan mbufs to protocol layers have to take care and not loan out too many (the user may not be reading), and alway fall back to copy mode if loaning doesn't look like it's the right thing to do. - -- Hwa-Jin Bae (hjb@netcom.com) Peaceful Star, 4001 San Leandro St. #3, Oakland, CA 94601 (510)436-2884 --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Can VxWorks coexist with other OS on VMEbus? Date: Fri, 23 Dec 1994 02:18:30 GMT From: hjb@netcom20.netcom.com (squeedy) Organization: Peaceful Star, Oakland, CA Message-ID: References: <3co9fa$gm8@usenet.INS.CWRU.Edu> Reply-To: hjb@netcom.com Sender: hjb@netcom.com (squeedy) >>>>> "Yoohwan" == Yoohwan Kim writes: In article <3co9fa$gm8@usenet.INS.CWRU.Edu> yxk14@po.CWRU.Edu (Yoohwan Kim) writes: Yoohwan> I'm working on a project with multiple robots and Yoohwan> controllers. We want to develop a C++ program on a Yoohwan> separate host computer to control the existing robot Yoohwan> controller. Our controller has motion processors, vision Yoohwan> processors, I/O boards, system processor, and a Motorola Yoohwan> 68030 board which may be used to run a downloaded Yoohwan> program. Rather than developing tedious shared memory and Yoohwan> message passing routines, we prefer to use an existing Yoohwan> product. The whole VMEbus is governed by the V+ Yoohwan> Operating system which runs on the system processor. Will Yoohwan> VxWorks cooperate with the the existing OS ? yes. Yoohwan> It seems that even in heterogeneous system, all the Yoohwan> processor boards should run VxWorks to properly Yoohwan> communicate, is that true ? If VxWorks cannot do it, is Yoohwan> there any other real time OS which can coexist with other Yoohwan> OS on the same VMEbus ? vxworks coexists fine with other OS'es on VMEbus. i have done work that involved SunOS, Solaris, and Motorola SVR3 running on the same backplane as multiple VxWorks based boards. - -- Hwa-Jin Bae (hjb@netcom.com) Peaceful Star, 4001 San Leandro St. #3, Oakland, CA 94601 (510)436-2884 --------------------------- Newsgroups: comp.os.vxworks Subject: Embedded Systems Design Symposium Date: Thu, 22 Dec 1994 12:54:49 GMT From: keesb@neth.hp.com (Kees Bos EMO) Organization: HP-Sales Office-The Netherlands Message-ID: <3dbsup$88o@hpuamsa.neth.hp.com> [ Article crossposted from comp.realtime ] [ Author was Kees Bos EMO (keesb@neth.hp.com) ] [ Posted on 22 Dec 1994 12:52:01 GMT ] Dear Designer, This is the final notice for the 1995 Hewlett-Packard Embedded Ssytems Design Symposium. The symposium is a one-day event consisting of the presentation of two technical papers. There is NO charge for attendance. Papers will be given by industry experts and consultants on a variety of important topics as detailed at the end of this notice. If you are interested in attending please reply to this message to: keesb@hpaichd.neth.hp.com Please include your postal mailing address so that an invitation can be sent to you immediately. Regards, Kees Bos Hewlett-Packard ************************************************************************* Venues and dates are as follows: Date Stop ----- ---- Jan 30 Madrid Feb 1 Toulouse Feb 2 Grenoble Feb 3 Les Ulis Feb 6 Bracknell Feb 7 Manchester Feb 8 Edinburgh Feb 10 Dublin Feb 13 Brussels Feb 14 Bad Homburg Feb 15 Munich Feb 17 Milano Feb 20 Zurich Feb 22 Vienna Feb 24 Berlin Feb 27 Helsinki Mar 1 Copenhagen Mar 3 Amsterdam --- ---- Papers Presented as follows: Methods for Successful Embedded Hardware Design. This session includes several examples of current methods and skills, as well as recommendations for streamlining product design. The module, which targets experienced engineers, covers a range of topics including complex system design, ASIC design, programmable logic design, PC board design, custom design considerations, hardware/firmware/software architecture trade-offs, and real-time system design. Using a Real-Time Kernel in Embedded Systems. This session examines various techniques for mapping out software to work in conjunction with real-time kernels. The presentation investigates performance reliability tradeoffs, shared data decisions, memory allocation methods, error-prone software strategies (interrupt routines, re-entrant code), task management, module partitioning, interrupt routines (data sharing, memory buffers, polling), asynchronous service routines, timers, and stack problems. --------------------------- Newsgroups: comp.os.vxworks Subject: Green Hills Date: Sun, 25 Dec 1994 23:36:36 From: terra@tiac.net (Ophira Rosolio Aharonson) Organization: Terra Computers Inc. Keywords: Fortran compilers Message-ID: Looking for info about Green Hills (addrees, phone, fax, e-mail) or any other supplier of FORTRAN compiler under VXworks, Thanks Ophira terra@tiac.net Ophira Rosolio-Aharonson Terra@Computers Inc. terra tiac.net --------------------------- Newsgroups: comp.os.vxworks,comp.os.realtime Subject: Timer interrupts in VxWorks Date: 26 Dec 1994 21:17:07 -0500 From: partho@cs.umd.edu (partho pratim mishra) Organization: U of Maryland, Dept. of Computer Science, Coll. Pk., MD 20742 Message-ID: <3dntf3$12p@seine.cs.umd.edu> I was thinking of driving a VxWorks task off the timer interrupt (i.e using the timerLib facilities). I would like to be able to do things at a fairly fine time granularity e.g. on the order of several hundred milliseconds. The VxWorks manuals suggest that running the clock at a frequency greater than 600 Hz will chew up most of the available cpu cycles. >tickAnnounce(2) VXWORKS REFERENCE MANUAL tickAnnounce(2) >NAME > tickAnnounce() - announce a clock tick to the kernel >SYNOPSIS > void tickAnnounce (void) >DESCRIPTION > This routine informs the kernel of the passing of time. It > should be called from an interrupt service routine that is > connected to the system clock. The most common frequencies > are 60Hz or 100Hz. Frequencies in excess of 600Hz are an > inefficient use of processor power because the system will > spend most of its time advancing the clock. By default, > this routine is called by usrClock() in usrConfig.c. Does anyone have any practical experience with driving the system at a higher clock rate eg 1000 Hz or 2000 Hz? Any idea why 600 Hz is the magic number given that interrupt processing times are supposed to be less than 100 microsecs. Thanks Partho Mishra --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Timer interrupts in VxWorks Date: 27 Dec 1994 08:41:06 GMT From: vwnemg@NEON2.iap.uni-bonn.de (Patrick Umlauf) Organization: Institut fuer Angewandte Physik, Universitaet Bonn Message-ID: <3dojv2$nlt@news.rhrz.uni-bonn.de> References: <3dntf3$12p@seine.cs.umd.edu> partho pratim mishra wrote in article <3dntf3$12p@seine.cs.umd.edu> : > > I was thinking of driving a VxWorks task off the timer interrupt >(i.e using the timerLib facilities). I would like to be able to do >things at a fairly fine time granularity e.g. on the order of several >hundred milliseconds. The VxWorks manuals suggest that running the >clock at a frequency greater than 600 Hz will chew up most of the >available cpu cycles. > >>tickAnnounce(2) VXWORKS REFERENCE MANUAL tickAnnounce(2) > >>NAME >> tickAnnounce() - announce a clock tick to the kernel > >>SYNOPSIS >> void tickAnnounce (void) > >>DESCRIPTION > >> This routine informs the kernel of the passing of time. It >> should be called from an interrupt service routine that is >> connected to the system clock. The most common frequencies >> are 60Hz or 100Hz. Frequencies in excess of 600Hz are an >> inefficient use of processor power because the system will >> spend most of its time advancing the clock. By default, >> this routine is called by usrClock() in usrConfig.c. > > Does anyone have any practical experience with driving the system >at a higher clock rate eg 1000 Hz or 2000 Hz? Any idea why 600 Hz >is the magic number given that interrupt processing times are >supposed to be less than 100 microsecs. > >Thanks > >Partho Mishra > My version of VxWorks only accepts rates of 50 or 100 Hz for the system clock. These numbers are predefined in the source of sysLib.c. Changes are of course possible but not useful. Since the usrClock controls task scheduling and system duties every tick uses lots of cpu-time for changing environment variables or performing delay and watchdog facilities. 100 ticks per second should therefore be the upper limit, otherwise the system will be occupied with itself. But why not using the sysAuxClk-Routines (sysLib.h)? The auxClk-timer's interrupts only call a user definable routine, nothing else. In my sources this routine is just an incremental counter to provide a clock of 1 msec resolotion. I use it with a dac-device as a function generator for any desirable wave forms to scan a laser. I think the rates for auxClk should not exceed 2000 Hz to preserve system efficiency. oooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooo /\ Patrick Umlauf / \ Institut fuer Angewandte Physik / \ Universitaet Bonn / \ Wegelerstrasse 8 / \|/ \ D-53115 Bonn, Germany / > *----- \ / /|\ \ FAX 0228 - 733474 / \ /________________\ E-Mail: vwnemg@NEON2.iap.uni-bonn.de oooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooo --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Timer interrupts in VxWorks Date: 27 Dec 1994 18:22:57 GMT From: berg@fnal.fnal.gov (David Berg) Organization: FERMILAB, Batavia, IL Message-ID: <3dpm21$44c@fnnews.fnal.gov> References: <3dntf3$12p@seine.cs.umd.edu> Reply-To: berg@fnal.fnal.gov In article 12p@seine.cs.umd.edu, partho@cs.umd.edu (partho pratim mishra) writes: > I was thinking of driving a VxWorks task off the timer interrupt >(i.e using the timerLib facilities). I would like to be able to do >things at a fairly fine time granularity e.g. on the order of several >hundred milliseconds. The VxWorks manuals suggest that running the >clock at a frequency greater than 600 Hz will chew up most of the >available cpu cycles. Even 60 Hz, which is 17 msec granularity, would meet a "several hundred milliseconds" requirement. What rate do you really need? > > Does anyone have any practical experience with driving the system >at a higher clock rate eg 1000 Hz or 2000 Hz? Any idea why 600 Hz >is the magic number given that interrupt processing times are >supposed to be less than 100 microsecs. We routinely run mv167's with a system clock rate of 1000 Hz to obtain millisecond scheduling resolution; for our application, this consumes only about 2% of the idle cycles, so the impact is minimal. I don't know where WRS obtained the 600 Hz figure. Actual impact will depend upon application, and of course on the specifics of the target board. We use 33 MHz MVME167s, for which the system clock is based on a timer that can be set to any rate within a wide range. Other boards may have timers with limited programmability, so they can only be set to a few specific rates (eg, 50 Hz, 100 Hz). - --------------------------+--------------------------------------------- "A nanosecond here, | David M. Berg (berg@fndaut.fnal.gov) a nanosecond there, | Computing Division and pretty soon you're | Fermilab MS120 talking about Real Time." | PO Box 500 tel: 708-840-3021 (apologies to the late | Batavia IL 60510 fax: 708-840-2783 Sen. Everett Dirksen) | --------------------------- Newsgroups: comp.os.vxworks Subject: FTP site for MVE166/167 Date: 28 Dec 1994 19:13:48 GMT From: Steve Rejto Organization: MIT Lincoln Laboratory Message-ID: <3dsddc$phr@testnews.llan> Is there an FTP site which has VxWorks drivers on it. In particular i'm looking for routines for source code to drivers for the MVE166/167 (serial ports.parrallel port,SCSI) etc. Thanks, srejto@ll.mit.edu --------------------------- Newsgroups: comp.os.vxworks Subject: Help! ISR causes NFS to die! Date: 30 Dec 1994 17:24:09 -0600 From: sabolcik@natinst.com (Ross Sabolcik) Organization: National Instruments, Austin, TX Message-ID: <3e24qp$4gt@eagle.natinst.com> Hello All: I am having the following problem which has been causing me great grief. I have a small test program running on my embedded computer which generates a series of bus errors, then attemtps to read and write to a NFS mounted remote file. The program is shown below: int toto(void) { for(;;) { A16Base = 0xffff0000; for(la = 0; la < 85; la++){ manId = (uint16)*((uint16 *)(A16Base + 0xC000 + (0x40 * 5) + 0)); manId = (uint16)*((uint16 *)(A16Base + 0xC000 + (0x40 * 6) + 0)); manId = (uint16)*((uint16 *)(A16Base + 0xC000 + (0x40 * 7) + 0)); } /*read the loop count */ if ( (fd = fopen(OUTFILE, "r")) == NULL) { printf("ERROR : fdFichier = 0x%x\n", *fd); return(-1); } fscanf(fd,"%ld", &count); fclose(fd); count++; /* write the loop count */ if ( (fd = fopen(OUTFILE, "w+")) == NULL) { printf("ERROR : fichier : fdFichier = 0x%x\n", *fd); return(-1); } fprintf(fd,"%ld", count); fclose(fd); } } The pointer dereferences shown above cause Bus Errors (I do not see this problem when the pointer dereference doesn't Bus Error). After running a variable number of loops, one of the fopen calls will fail causing the program to exit. When I get back to the shell prompt, I have lost all of my remote file systems. Any attempt to change into a remote directory or examine the files on a remote file system fails. I am able to do a stack dump from the shell prompt when I attempt to access a remote file and here is what I get: 44e70 _ls +1e : _opendir (449e6) 22716 _opendir +e : _open (449e6, 0, 0) 2a65c _open +80 : _iosOpen (321fa0, 31dee4, 0, 0) 2b314 _iosOpen +32 : 1ae54 ([321fa0, 31dee4, 0, 0, 449e7]) 1af1e _nfsUnmount +1ba: _nfsLookUpByName (321fae, 31dd90, 3220ee, 31dd28, 31dd08) 1bd5a _nfsLookUpByName+da : 1c920 (321fae, 186a3, 2, 4, &_xdr_diropargs, 31dcb4, &_xdr_diropres, 31dd28) 1c94c _nfsFileAttrGet+328: 1c78a (3fdb84, 321fae, 186a3, 2) 1c828 _nfsFileAttrGet+204: _clntudp_create (31d7b0, 186a3, 2, 5, 0, 3fdb88) 3eb5e _clntudp_create+24 : _clntudp_bufcreate ([31d7b0, 186a3, 2, 5, 0]) 3e9cc _clntudp_bufcreate+88 : _pmap_getport (31d7b0, 186a3, 2, 11) 3f0f8 _pmap_getport +94 : 3eb66 (3fdb5c, 3, &_xdr_pmap, 31d6bc, &_xdr_u_short, 31d6b6, 3c, 0) 3ecec _clntudp_create+1b2: _semQPut (100, 31d664, 0, 0, 3237d0) Now the semQPut looks purty suspicious to me. So what I am thinking is..... something (most likely my ISR) is corrupting the memory location used by the semQPut shown above. This corruption makes the semaphore appear to be taken/locked. This causes the file system call shown above to wait for the semaphore to free. However, no one has taken the semaphore (it was corrupted) and eventually the call times out. My problem is that semQPut is an undocumented kernel call (I can't find any documentation), so I am not sure what it really does. THE QUESTION IS HERE | | --- --- \ / \ / \ / \/ Does anyone with more vxWorks NFS experience have any idea what is really going on in the stack dump I've shown above? Any help would be greatly appreciated. - -------------------------------------------------------------------------------- Ross Sabolcik VXI Software Engineer National Instruments 6504 Bridge Point Parkway Austin, TX 78730-5039 - -------------------------------------------------------------------------------- - -------------------------------------------------------------------------------- Email: sabolcik@natinst.com - -------------------------------------------------------------------------------- --------------------------- Newsgroups: comp.os.vxworks Subject: dosFs VOLUME_NOT_AVAILABLE Date: 28 Dec 1994 18:59:56 -0800 From: alann@tx.abekas.com (Alan Nishioka) Organization: Abekas Video Systems. Inc. Message-ID: <3dt8nc$2lq@tx.abekas.com> THE SITUATION: We are using VxWorks 5.1 with a dosFs filesystem on an Iomega Io20s scsi floptical drive. THE SYMPTOMS: If you insert a disk and then immediately access it, dosFs returns: - -> ls error reading entry (errno=0x380001) value = -1 = 0xffffffff = scsiready.o_bss + 0xc7cb8463 The errno is for S_dosFsLib_VOLUME_NOT_AVAILABLE. Subsequent accesses return (forever): - -> ls Can't open ".". value = -1 = 0xffffffff = scsiready.o_bss + 0xc7cb8463 THE PROBLEM: DosFs gets a device not ready and refuses to try again. The problem is more severe with a 20MB floptical than it is with a 1.44MB floppy because the floptical drive is not ready for a longer period of time. Our current solution of inserting a disk and counting slowly to ten is not acceptable for our customers :-) THE QUESTION: Does anybody know how to get around this? I have looked in all the manuals and in some mail about using vxWorks with floppies. We don't have source to dosFsLib. _______________________________________________________________________________ Alan Nishioka alann@abekas.com 415 599-3024 Abekas Video Systems 101 Galveston Drive Redwood City CA 94063 --------------------------- Newsgroups: comp.os.vxworks Subject: Motorola MC68681 driver Date: Wed, 4 Jan 1995 13:37:38 GMT From: bjm_vloedbeld@hgl.signaal.nl (B.J.M. Vloedbeld) Organization: Hollandse Signaal apparaten BV Message-ID: <1995Jan4.133738.17760@hgl.signaal.nl> Sender: usenet@hgl.signaal.nl (USENET News System) Hello, We are currently developing a Vxworks application with a 68020 processor and a MC68681 duart. question : We are looking now for a MC68681 driver that supports the serial IO and that uses the MC68681 timer for the system clock. Please email me if you have some information. Thanks, --Ben-- - -- ============================================================================== = B.J.M. Vloedbeld Dept:SDA-DID-DIM Org:Hollandse Signaalapparaten B.V. = = Adress : P.O.box 42 7550 GD Hengelo, The Netherlands = = Phone :+31-74-483994 FAX:+31-74-484009 email:bjm_vloedbeld@hgl.signaal.nl = =============================================================[Unclassified]=== --------------------------- Newsgroups: comp.os.vxworks Subject: FAQ and ftp site info Date: 04 Jan 1995 21:37:02 GMT From: wilson@ssdd475a.erim.org (Paul Wilson) Organization: Environmental Research Institute of Michigan, Ann Arbor, Michigan Message-ID: Sender: news@newsspool.erim.org Is there an FAQ available for this group? Where might I obtain it? Also, I have a requirement to produce a Sun readable TAR tape from a vxworks application. Is there a netware package to do that? Thanks in advance for assistance. regards, Paul Wilson wilson@erim.org --------------------------- Newsgroups: comp.os.vxworks Subject: FAQ and ftp site info Date: 04 Jan 1995 22:09:32 GMT From: wilson@ssdd475a.erim.org (Paul Wilson) Organization: Environmental Research Institute of Michigan, Ann Arbor, Michigan Message-ID: Sender: news@newsspool.erim.org Is there an FAQ available for this group? Where might I obtain it? Also, I have a requirement to produce a Sun readable TAR tape from a vxworks application. Is there a netware package to do that? Thanks in advance for assistance. regards, Paul Wilson wilson@erim.org --------------------------- Newsgroups: comp.os.vxworks,gnu.gcc.help Subject: ld MEMORY command, multiple regions per section possible? Date: 4 Jan 95 22:26:52 GMT From: kab@dtl.labs.tek.com (Kent Black) Organization: DTL, Tektronix, Inc. Keywords: a.out, MEMORY Message-ID: <6986@crl.labs.tek.com> Followup-To: comp.os.vxworks Sender: news@crl.labs.tek.com I would like to be able to put a globally accessible object (an OS specific symbol table for debugging) into some slow (or even optional) memory. I'm thinking a chunk of ROM for text, a smallish chunk of RAM for data and bss, and "somewhere else" for this other big chunk of initialized data (otherwise I'm short of usable RAM). Using a.out format (68k arch), is it possible to have a section, in particular .data, divided among memory regions? Consider the following ld script: OUTPUT_FORMAT("a.out-sunos-big") /* output .o file format */ TARGET("a.out-sunos-big") /* input .o file format */ ENTRY("_romSP") /* entry point */ MEMORY { rom : ORIGIN = 0xfB000000, LENGTH = 2M ram : ORIGIN = 0x06020000, LENGTH = 512K } SECTIONS { .text : { CREATE_OBJECT_SYMBOLS *(.text) _etext = .; } > rom .data : { *(.data) _builtForRom = .; SHORT(0); _edata = .; } > ram .bss SIZEOF(.data) + ADDR(.data) : { *(.bss) [ COMMON ] _end = .; __end = .; } > ram } Now, my first thought is that I could simply add an optional read of "symtable.o" into some arbitrary address from the .data segment, but that fails when .data > ram doesn't "fit". Clever alternatives welcome. - -- kab --------------------------- Newsgroups: comp.os.vxworks Subject: Re: FAQ and ftp site info Date: Thu, 5 Jan 1995 08:15:17 GMT From: bjm_vloedbeld@hgl.signaal.nl (B.J.M. Vloedbeld) Organization: Hollandse Signaal apparaten BV Message-ID: <1995Jan5.081517.13934@hgl.signaal.nl> References: Sender: usenet@hgl.signaal.nl (USENET News System) Paul Wilson (wilson@ssdd475a.erim.org) wrote: : Is there an FAQ available for this group? Where might I obtain it? : Also, I have a requirement to produce a Sun readable TAR tape from : a vxworks application. Is there a netware package to do that? : Thanks in advance for assistance. : regards, : Paul Wilson : wilson@erim.org here some lines from the original vxworks FAQ : The FAQ is currently posted to comp.os.vxworks, news.answers and comp.answers on the 15th of every month. This FAQ was compiled by hjb@netcom.com using comments by readers of comp.os.vxworks as well as his own limited knowledge of VxWorks. Credits appear at the end. Comments and indications of doubt are enclosed in []s in the text. Each section begins with dashes ("-") on a line of their own, then the section number. This should make searching for a specific section easy. This FAQ is also available via anonymous ftp in: rtfm.mit.edu:/pub/usenet/news.answers/vxworks-faq.Z ftp.uu.net:/usenet/news.answers/vxworks-faq.Z cs.toronto.edu:/pub/usenet/comp.answers/vxworks-faq.Z ftp.netcom.com:/pub/hjb/vxfaq.gz - -- ============================================================================== = B.J.M. Vloedbeld Dept:SDA-DID-DIM Org:Hollandse Signaalapparaten B.V. = = Adress : P.O.box 42 7550 GD Hengelo, The Netherlands = = Phone :+31-74-483994 FAX:+31-74-484009 email:bjm_vloedbeld@hgl.signaal.nl = =============================================================[Unclassified]=== --------------------------- Newsgroups: comp.os.vxworks Subject: SCSI floppy disk drive Date: Thu, 5 Jan 1995 19:44:28 GMT From: jch@netcom.com (Jack Hwang) Organization: NETCOM On-line Communication Services (408 261-4700 guest) Message-ID: This is an old news. Teac discontinue SCSI floppy disk drive FD235SJ. But due to some bureautic clumsiness, the company I am working with did not act on time to secure the supplies. Now it all comes to me to find alternatives. Does anyone know there is a replacement available which is compatiable with MVME147? (Even 147 itself is discontinued! DXXN!!!) --------------------------- Newsgroups: comp.os.vxworks Subject: Stability Date: 4 Jan 1995 19:50:52 GMT From: prabhala@penafiel.irvine.dg.com (Venkata Prabhala) Organization: Data General Corporation, Research Triangle Park, NC Keywords: VxWorks Message-ID: <3eeu6s$hcs@dg-irv.irvine.dg.com> I am a new to RTOS's. But I am certainly UNIX type! I have been wondering if someone on this group could comment on the stability of VxWorks as a development environment in a Unix host-based environment. How well does VxWorks hold up to its promise of remote host-based development/debugging? I would like to hear both positive and negative comments to help my decision. It would also help if someone could point out where I can obtain comparative studies about different RTOSs like VxWorks, VRTX, QNX, pSOS, Solaris etc. You may e-mail me if you think it would increase the S/N ratio on the news group. Thanks. - -- Venkat - -- Venkata Prabhala, Data General Corporation | (prabhala@irvine.dg.com) 2603 Main St., Ste. 360 | FAX: +1 714 724 3956 Irvine, CA 92714 | VOICE: +1 714 724 3569 --------------------------- Newsgroups: comp.os.vxworks Subject: Re: SCSI floppy disk drive Date: Fri, 6 Jan 1995 14:04:38 GMT From: clegg@ssd.fsi.com Organization: Flightsafety International, Broken Arrow, OK Message-ID: <1995Jan6.140438.24584@dev1.ssd.fsi.com> References: Reply-To: clegg@ssd.fsi.com Sender: news@dev1.ssd.fsi.com In , jch@netcom.com (Jack Hwang) writes: >This is an old news. Teac discontinue SCSI floppy disk drive FD235SJ. >But due to some bureautic clumsiness, the company I am working with >did not act on time to secure the supplies. Now it all comes to me >to find alternatives. Does anyone know there is a replacement available >which is compatiable with MVME147? (Even 147 itself is discontinued! >DXXN!!!) > I would appreciate a copy of any information you have concerning a good alternative. If possible, could you tell me what the media code is for the suggested drive? Thanks in advance, Roger Clegg clegg@ssd.fsi.com --------------------------- Newsgroups: comp.os.vxworks Subject: Fortran compiler & tools/ Vxworks on Alpha Date: Fri, 6 Jan 1995 13:07:14 From: terra@tiac.net (Ophira Rosolio Aharonson) Organization: Terra Computers Inc. Keywords: Fortran, Programing tools, Message-ID: Looking for info on suppliers of Fortran compilers for Vxworks on Alpha. Thanks Ophira Rosolio-Aharonson Terra@Computers Inc. terra tiac.net --------------------------- Newsgroups: comp.os.vxworks Subject: Re: DOS & large drive Date: Fri, 6 Jan 1995 19:04:33 GMT From: jfinley@netcom.com (John Finley) Organization: NETCOM On-line Communication Services (408 261-4700 guest) Message-ID: References: <9501061458.AA04567@positron.med.ge.com> Mike Cook (cook@positron.med.ge.com) wrote: : Anybody out there using large drives (> 4 gigabyte) with VxWorks DOS? [snip] : Question 1: : Is there anyway I can turn off the buffered I/O or reduce the buffer size? : Ideally, I would like to be able to be able to open a dos file, seek to an : offset that is not cluster aligned and start reading/writing without having the : system read/write more than two additional disk sectors. : Question 2: : If the answer to Question 1 is NO (unless you have source code)....I don't. : I'm tempted to try to read/write to the DOS contiguous files using Raw I/O. : Would this be advisable? : Question 3: : Is there a simple means to obtain the clustor number (or ideally the disk : sector) for the start of a DOS contiguous file? (to aid my raw i/o if I choose : to go that route) : Question 4: : Do you have any other suggestions as to how to improve the performance? : Mike Cook : GE Medical Systems : cookm@med.ge.com : (414) 548-4791 I've never used a >4GB drive, but I do have two-year-old dosFsLib source. Q1: Cluster buffering looks pretty central to everything. The only time it's not used is when you open the entire disk as a 'raw volume'. I don't think you can turn it off or make it other than the cluster size. Q2: This seems reasonable, however... Q3: It's very simple in the source, but the necessary structure typedef (DOS_FILE_DESC), which contains the number of the first cluster in the file, is local to dosFsLib.c. Converting from cluster number to sector number is a scale and offset. I guess I don't know how you'd get the DOS_FILE_DESC* from the fd anyway, but that may be easy. Someone with current source (WRS) could write a function to get the sector offset of the start of a file pretty easily. If you can get them to look at it. I think it'd be a terrific ioctl to add to the library. Q4: If you're only writing, be sure to open the file write-only. It looks like this will avoid reading in the cluster where your write ends (if not aligned). If you do this, then for both writes and reads, the extra overhead of non-cluster-alignment looks to be (a) bcopy of data to/from the cluster buffer for partial cluster at beginning/end of read or write; and (b) extra overhead of more and smaller calls to the SCSI driver to do the transfer. Sorry I don't have a solution, John Finley jfinley@netcom.com --------------------------- Newsgroups: comp.os.vxworks Subject: Database on VxWorks Date: Sat, 7 Jan 1995 09:08:34 GMT From: aisic@actcom.co.il (Ilan aisic) Organization: ACTCOM - Internet Services in Israel Message-ID: Sender: news@wang.com Does anyone know if there's a commercial or non-commercial database that is available on VxWorks? - --ilan --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Changing file times. Date: Wed, 4 Jan 1995 00:14:57 GMT From: georg@sgl.ists.ca (Georg Feil) Organization: Institute for Space and Terrestrial Science Message-ID: References: <1994Dec20.152127.9206@picker.com> Sender: news@newshub.ists.ca froeber@bbn.com (Fred Roeber) writes: >heller@nm.picker.com (Mark Heller) writes: >> I'm running VxWorks 5.1.1 and looking for a way to change >> the dates on files on a Dos filesystem. >See the function dosFsTimeSet in VxWorks. Fred Just a comment on using dosFsTimeSet() ... this would have to be done in somewhat of a kludgy fashion, e.g. you would "back date" the dosFs time using dosFsTimeSet(), then do a dummy modify of the file, then return the dosFs time to normal. This requires that no other dosFs modify activity takes place, or other files may get erroneous times. Also it won't work if you've used dosFsDateTimeInstall(). An alternate way is to modify the in-memory DOS_DIR_ENT dosde_time and dosde_date fields, then set dosde_modified and flush the file. Ideally you would need to take certain VxWorks system semaphores while you do this to be safe (I could give more info if needed). Georg. - -- Georg Feil Space Geodynamics Laboratory | Internet: georg@sgl.ists.ca Institute for Space and Terrestrial Science | Phone: (416) 665-5458 4850 Keele St./North York/Ont/Canada/M3J 3K1 | Fax: (416) 665-1815 --------------------------- Newsgroups: comp.os.vxworks Subject: Looking For A File Transfer Protocol Date: 7 Jan 1995 17:32:21 -0800 From: randyr@treehouse.org (Randy Riggins) Organization: The Treehouse Message-ID: <3enfb5$86@treehouse.org> I don't know much about FTP but, Wind River's FTP documentation states that it copies the entire file into local memory before performing the transfer. Thus restricting the size of the largest file that can be transferred to the amount of the available local memory. I need a means of transferring 1M to 2G byte files between VxWorks targets. Using NFS to provide transparent file access is not an option. Does any know of a different VxWorks based file transfer protocol that streams the data between the disk and the network? Are there other versions of FTP available? Thanks, Randy Riggins --------------------------- Newsgroups: comp.os.vxworks Subject: Looking for new VxWorks File System Date: 9 Jan 95 10:35:48 EST From: tpd@icf.hrb.com (THOMAS P. DONNELLY) Organization: HRB Systems, Inc. Message-ID: <1995Jan9.103548.22647@hrbicf> Does anyone know of a third party that provides a VxWorks file system besides DOS and RT11? We have all sorts of problems with the buggy DOS F.S. and RT11 hasn't the flexibility of subdirectories. One that can handle large disks (greater than 1 gig) would be ideal. Thanks in advance Tom --------------------------- Newsgroups: comp.os.vxworks Subject: Can't ftp thor.atd.ucar.edu Date: 9 Jan 1995 18:09:33 GMT From: bonaker@amasd.anatcp.rockwell.com (Joe Bonaker) Organization: Rockwell Message-ID: <3eru4t$to@huron.eel.ufl.edu> Reply-To: bonaker@amasd.anatcp.rockwell.com I can no longer anonymously ftp(8) into the VxWorks archive site. Has this service been discontinued or superceded by the E-mail queries? % ftp thor.atd.ucar.edu Connected to thor.atd.ucar.edu. 421 Service not available, remote server has closed connection ftp> Thanks, Joe B. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Can't ftp thor.atd.ucar.edu Date: 9 Jan 1995 19:13:07 GMT From: vanandel@stout.atd.ucar.edu (Joe Van Andel) Organization: National Center for Atmospheric Research Message-ID: <3es1s3$766@ncar.ucar.edu> References: <3eru4t$to@huron.eel.ufl.edu> Sender: vanandel@canoe (Joe Van Andel) The VxWorks archive has been moved to ftp://ftp.atd.ucar.edu/pub/vxworks - -- Joe VanAndel Internet: vanandel@ncar.ucar.edu National Center for Web http://www.atd.ucar.edu/jva/home.html Atmospheric Research --------------------------- Newsgroups: comp.os.vxworks Subject: Fortran Compiler for Alpha/VXworks Date: Mon, 9 Jan 1995 17:24:01 From: terra@tiac.net (Ophira Rosolio Aharonson) Organization: Terra Computers Inc. Message-ID: Needed: info availability of Fortran compiler and tools for Alpha VXworks. Thanks Ophira Rosolio-Aharonson Terra@Computers Inc. terra tiac.net --------------------------- Newsgroups: comp.os.vxworks Subject: Cartridge Tape Driver Date: Thu, 5 Jan 95 09:24:29 GMT From: TaeKyung Chung Organization: SNU,KOREA Message-ID: <1995Jan5.092429.1479@news.snu.ac.kr> Sender: usenet@news.snu.ac.kr (NEWS POSTER) I want to receive about how to install Cartridge Tape Driver in VxWorks. If anyone know about that, mail or posting in this newsgroup. Help me.... And, I want to know that "Backup Device in VxWorks using any device" Thanks you...... ================= My Email Address : g1khahn@super.samsung.co.kr tkchung@jurasic.shi3.samsung.co.kr =================================== --------------------------- Newsgroups: comp.os.vxworks Subject: smObjLocalToGlobal() Question Date: Thu, 5 Jan 95 09:45:00 GMT From: TaeKyung Chung Organization: SNU,KOREA Message-ID: <1995Jan5.094500.1684@news.snu.ac.kr> Sender: usenet@news.snu.ac.kr (NEWS POSTER) Hello.... Happy New year !!!! I use 2 CPU Card (MVME167) and VxWorks O/S... For Example, One MVME167 use this routine int START smNameAdd(START,smObjLocalToGlobal( int A ),T_SM_BLOCK) Another MVME167 use routine int *StartFlag=0 int START_SEM if (smNameFind(START_SEM,&StartFlag,&type,NO_WAIT) == OK) StartFlag= (int *)smObjectGlobalToLocal(StartFlag); In this case, I want to set *StartFlag is 1, and then I want to pass StartFlag value using smObjLocalToGlobal Function. But, The value StartFlag is not changed when another MVME167 acess StartFlag value. Help me.... How to change StartFlag value......? and then another Function call without using smObjLocalToGlobal(). anyway.... I want to change Global parameter.... and then any MVM167 will read changed same value..... please.... help me...... ================ My Email Address : g1khahn@super.samsung.co.kr tkchung@jurasic.shi3.samsung.co.kr ======================================= --------------------------- Newsgroups: comp.os.vxworks Subject: Email address Wind River Systems, Visicom Date: 10 Jan 95 10:49:34 GMT From: wouters@ict.nl (Rick Wouters) Organization: ICT Automatisering Deventer B.V. Keywords: Motif, VxWorks, GUI, X Message-ID: <2088@ictser.UUCP> Reply-To: wouters@ict.nl Sender: news@ictser.UUCP Could someone send me email addresses of the above mentioned companies ? Because I'm looking for information about Motif GUI builders and/or toolkits for VxWorks, any suggestions and experience related to this topic would be welcome, too. Thanks in advance. Rick. - ------------------------------------------------------------------------ Rick Wouters ICT Automatisering Deventer B.V. e-mail: wouters@ict.nl P.O. Box 701 Telephone: +31-(0)5700-33888 7400 AS Deventer Fax: +31-(0)5700-21362 The Netherlands - ------------------------------------------------------------------------ --------------------------- End of New-News digest ********************** From gphillips@se01.wg2.waii.com Tue Jan 10 04:55:11 1995 From: "Gary L. Phillips" Date: Tue Jan 10 04:55:18 PST 1995 Subject: UNSUBSCRIBE UNSUBSCRIBE From randy@opid330.litton.com Tue Jan 10 08:54:54 1995 From: randy@opid330.litton.com (randy) Date: Tue Jan 10 08:55:01 PST 1995 Subject: Flash EEPROM Packs Anybody have a source of Flash EEPROM packs that might attach to a MVME147 and MVME167 card? Tanx ***************************************** * * * Randy Ryan * * Litton Data Systems * * * * Email: randy@opid330.dsd.litton.com * * (Note: Return Address on header is * * incorrect) * * * ***************************************** From ddavies@xmission.com Tue Jan 10 22:05:52 1995 From: Doug Davies Date: Tue Jan 10 22:06:00 PST 1995 Subject: Re: Programming FLASH on the MV162 > Submitted-by: del@rst.com (Mark Del Giorno) > > I'm looking for some source code that can burn FLASH on a MVME162 > (it uses the Intel 28F020). My ultimate goal is to be able to burn > my application into FLASH and run on power up (embedded - with no > ethernet attached). Wind River has a way that I can download a hex > file via a serial link - this is o.k., but I can't believe I can't > accomplish the same thing via the ethernet connection. When I asked > them for code to burn the FLASH, I got the following response: I have burned several boot ROMs over the ethernet on both the mv162 and the mv166 (both of which have on-board flash). This is done by booting the board with the Motorola bug ROM, and using its facilities to burn the flash. As I recall, the procedure for the mv162 is something like this: 1) Compile your application and convert the 'hex' output (Motorola S record) to a binary format. 2) Remove jumper GPI3 (pins 9 & 10 of J22 jumper header) on the mv162. (Caveat, I haven't done this for a while and in the back of my mind I remember something about this jumper location not being correct for either the mv162 or the mv166 . You may want to consult the mv162 documentation, what you want is the board to boot up using the Motorola bug ROM). 3) Power up the mv162, also insure that the Ethernet port on the mv162 is connected to the network. 4) After the '162-BBug>' prompt, type (or verify the following: 162-BBug>niot Controller LUN = 00? Device LUN = 00? Node Control Memory Address = FFE10000? Client IP Address = 0.0.0.0? Server IP Address = 0.0.0.0? Subnet IP Address Mask = 255.255.255.255? Broadcast IP Address = 0.0.0.0? BOOTP/RARP Request Retry = 00? TFTP/ARP Request Retry = 00? 162-BBug>niop Controller LUN = 00? Device LUN = 00? Get/Put = G? File Name = ? Memory Address = 0000E000? Length = 00000000? Byte Offset = 00000000? Note that the input address on the mv162 entered above is the location that the binary file will be loaded into. At this point you should see some status indicating that the file loaded successfully. If you get an error statups of 0400, this indicates that the RTC clock is not running. To fix this error, use the 'set' command to set the date & time and get the RTC clock running, ie: 162-BBug>set 0929941405 The format is MMDDYYHHMM. After running the 'set' command repeat the 'niop' command and then continue. 5) Now copy the binary file from RAM into the flash: 162-BBug>pflash :20000 ffa00000 Program FLASH Memory (Y/N)? y Note that at this point you should get status about the progress of the command. This will take several minutes to complete. The syntax of the pflash command is: pflash : 6) Re-install jumper GPI3 (pins 9 & 10 of J22 jumper header) on the mv162 and power up. Hope this helps (and that it works... ) -Douglas Davies (ddavies@xmission.com) From daemon@vxw.ee.lbl.gov Wed Jan 11 04:00:27 1995 From: daemon@vxw.ee.lbl.gov Date: Wed Jan 11 04:00:34 PST 1995 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Wed Jan 11 04:00:17 PST 1995 Subject: VxWorks drivers for Sbus FDDI cards Subject: VME to Sbus adaptors Subject: sysStartType ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: VxWorks drivers for Sbus FDDI cards Date: 10 Jan 1995 22:00:59 GMT From: mcdono@atcwest.com (Tim McDonough) Organization: Alliant Techsystems Message-ID: <3ev02r$3or@news.cerf.net> Reply-To: mcdono@atcwest.com I've found numerous Single Attached FDDI Single Slot SBUS cards for our SPARC host but no drivers. Anyone know where I can find a driver for VxWorks 5.1 and such a card? Thanks - --- Tim McDonough | The opinions voiced are my own and Alliant Techsystems | do not reflect those of my employer mcdono@atcwest.com | or the Susquehanna Hat Company. --------------------------- Newsgroups: comp.os.vxworks Subject: VME to Sbus adaptors Date: 10 Jan 1995 18:03:25 -0700 From: spitcher@oz.sim.es.com (Stanford Pitcher) Organization: A poorly-installed InterNetNews site Keywords: VMEbus SBus Message-ID: <3evaot$8nr@oz.sim.es.com> Hello, Our application requires a 'hard' real-time link between a SPARCstation running Solaris 2.3 and a MVME-162 card running vxWorks. We need to keep the two applications (one on each system - cycling at rates up to 50 Hz) in sync with other. We hoped to use TCP/IP sockets but are experiencing frequent glitches where the Sun's app does not respond for several frames (sometimes as many as 30). We are assuming that although we are running the Sun's app at real-time priority level, the net task (which runs at some level we can't change) is being bumped which causes the glitch. Another guess is that occasionally the net task hiccups for some reason. Anybody know??? Anyway, if we can't get sockets to work. We'll probably try a bus adaptor between the SBus and the VMEbus. The one we are considering is made by Bit 3. They have two models: with and without DMA. They supply some 'utility' code in FORTH or you can buy their 'support software' which includes a device driver, read/write/lseek, interrupt handlers, etc. They don't mention what its coded in. Finally, my questions: 1) Does anyone have any opinions/experiences on bus adapters in general and 3 Bits products (including their software) in particular? 2) Does anyone have any ideas on how we might improve our ethernet performance? Thanks alot, Stan spitcher@sim.es.com --------------------------- Newsgroups: comp.os.vxworks Subject: sysStartType Date: Wed, 11 Jan 1995 08:46:04 GMT From: rsgawera@wg.icl.co.uk (Raj Gawera) Organization: International Computers Limited Message-ID: Reply-To: rsgawera@wg.icl.co.uk Sender: news@oasis.icl.co.uk Dear VxWorkers, I am having trouble with interrogating sysStartType in usrInit(). It always seems to be 0 ie warm start, even if I power off-on the board. Anyone know any tricks that have to be done to use the sysStartType variable ? I am using a Radstone ENET 1 board with v5.1 vxWorks. cheers, Raj Raj Gawera rsgawera@wg.icl.co.uk --------------------------- End of New-News digest ********************** From @lazarus.nrtc.northrop.com:lam@csd.nad.northrop.com Wed Jan 11 09:49:32 1995 From: Christine Lam Date: Wed Jan 11 09:49:39 PST 1995 Subject: Re: Thanks Thanks to whoever replied to my email regarding socket recvfrom time out problem. Christine Lam From randy@opid330.litton.com Wed Jan 11 15:39:04 1995 From: randy@opid330.litton.com (randy) Date: Wed Jan 11 15:39:10 PST 1995 Subject: Compile options Does anybody know what the convention(s) are for levels of optimization with driver code. I was attempting to do level 2 optimization. The problem is, for the register locations that are strobed, multiple strobes to that address appear to be getting cut out of compiled code. Perhaps because the compiler sees that I'm writing the same value to an address? This is with the GNU C compiler supplied with VxWorks 5.1. Tanx ***************************************** * * * Randy Ryan * * Litton Data Systems * * * * Email: randy@opid330.dsd.litton.com * * (Note: Return Address on header is * * incorrect [not under my control * * either!) * * * ***************************************** From stevem@syd.csa.com.au Wed Jan 11 16:56:34 1995 From: stevem@syd.csa.com.au (Steven McCoy) Date: Wed Jan 11 16:56:41 PST 1995 Subject: Re: Bus Error with 32MByte MVME167 Re: > Our Ada application executes without a problem on a MVME167-B > (25 MHz, 16 MBytes) and a MVME167-33A (33 MHz, 16 MBytes). A recent > upgrade to the MVME167-34A (33 MHz, 32 MBytes) has suddenly > caused the same code to generate a Bus error or at times an Illegal > Instruction. This always occurs when executing i() or checkStack() > commands from the shell. Note that we do not have a problem with > the pi() command. > > e.g. > > Bus Error > Program Counter: 0x00000008 > Status Register: 0x3000 > Access Address : 0x08eb7376 <-- Way out of range. > Special Status : 0x0125 > > shell restarted. > > or > > Illegal Instruction > Program Counter: 0x00000012 > Status Register: 0x3008 > > shell restarted. Thanks for all the replies regarding this problem. It turns out that our problem was caused by an Ada task entry that was too long. The task IFF_SERVICE_RECEIVED_POINT_TO_POINT_MESSAGES_TASK_TYPE in the Ada package IFF_SERVICE_RECEIVED_POINT_TO_POINT_MESSAGES will have the following name and entry in the VxWorks symbol table, adaT28 _A_iff_service_received_point_to_point_messages_task_type.iff_service_received_point_to_point_messages The entry length here is 100 characters excluding the _A prefix. This task works fine on a 16 MByte board but causes problems on a 32 MByte board. The limit appears to be 99 characters when using 32 MBytes. I'm not sure why this is the case. An extra byte appears to be required for something else. Someone may want to shed some light here. The i(), and checkStack() commands impose name limits, "char name[MAX_SYS_SYM_LEN + 1]", where MAX_SYS_SYM_LEN = 100 from sysSymTbl.h. The Ada tasks are automatically assigned the names adaT1, adaT2, ... The MAX_SYS_SYM_LEN also applies for the entry point in the symbol table where our problem lay. Steven McCoy CSC Australia A Unit of Computer Sciences Corporation stevem@assip.csasyd.oz.au From elliotts@stanilite.com.au Wed Jan 11 17:51:36 1995 From: elliotts@stanilite.com.au (Stephen Elliott - ENG - TEMP) Date: Wed Jan 11 17:51:43 PST 1995 Subject: Re: Compile options randy@opid330.litton.com (randy) wrote: > > ... > I was attempting to do level 2 optimization. The problem is, for the > register locations that are strobed, multiple strobes to that address > appear to be getting cut out of compiled code. Perhaps because the > compiler sees that I'm writing the same value to an address? > ... You should be using "volatile" on the strobed regiser variables. -- Steve Elliott - Stanilite Electronics, 424, Lane Cove Rd, NSW 2113, Australia Voice : +61 2 325 1667 Fax : +61 2 325 1799 eMail : elliotts@stanilite.com.au From stan@lassen.rti.com Wed Jan 11 22:25:53 1995 From: stan@lassen.rti.com (Stan Schneider) Date: Wed Jan 11 22:26:00 PST 1995 Subject: Re: Compile options >> Submitted-by randy@opid330.litton.com Wed Jan 11 15:39:04 1995 >> Does anybody know what the convention(s) are for levels of optimization >> with driver code. >> >> I was attempting to do level 2 optimization. The problem is, for the >> register locations that are strobed, multiple strobes to that address >> appear to be getting cut out of compiled code. Perhaps because the >> compiler sees that I'm writing the same value to an address? >> >> This is with the GNU C compiler supplied with VxWorks 5.1. >> gcc will (almost?) always optimize "unnecessary" memory accesses, unless the variable is declared "volatile", regardless of optimization level. This is in compliance with the ANSI spec. So, adding "volatile" to your hardware register definitions is the right way to fix this (IMHO). Gcc also has a HACK called the "-volatile" flag that makes everything act as if it were declared volatile. It's good for testing to see if the optimizer has bitten you, but should otherwise be avoided, as it really messes up the compiler's performance. HTH, -- Stan =============================================================================== = = = = Stan Schneider = email: stan@rti.com = = Real-Time Innovations, Inc. = Phone: (408) 720-8312 = = 954 Aster, Sunnyvale, CA 94086 = Fax: (408) 720-8419 = = = = =============================================================================== From sergi@hpbpq11.bpo.hp.com Thu Jan 12 00:36:01 1995 From: Sergi Casas(QA Software testing) Date: Thu Jan 12 00:37:47 PST 1995 Subject: Re: Compile options > > Submitted-by randy@opid330.litton.com Wed Jan 11 15:39:04 1995 > Submitted-by: randy@opid330.litton.com (randy) > > Does anybody know what the convention(s) are for levels of optimization > with driver code. > > I was attempting to do level 2 optimization. The problem is, for the > register locations that are strobed, multiple strobes to that address > appear to be getting cut out of compiled code. Perhaps because the > compiler sees that I'm writing the same value to an address? > > This is with the GNU C compiler supplied with VxWorks 5.1. > > Tanx > ***************************************** > * * > * Randy Ryan * I think your problem is actually caused by the optimization process, but you can solve it easily by using the keyword "volatile" when declaring the pointers to those hardware registers that you'd like to make sure are accessed as many times as your C program refers to them. Example: volatile char *ParallelControlReg; ... { ParallelControlReg |= STROBE; ParallelControlReg |= STROBE; ParallelControlReg &= ~STROBE; ... } Hope this helps. -- |============================================================================| | Sergi Casas ##### / ##### Firmware Engineer | | Hewlett-Packard Company ### /_ _ ### | | Barcelona Division (BCD) ## / / / / ## e-mail: sergi@hp-bpo.bpo.hp.com | | Avda. Graells, 501 ## / / /_/ ## Phone : +34 3 582 1475 | | 08190 S. Cugat (Barcelona) ### / ### Fax : +34 3 582 2515 | | SPAIN ##### / ##### HP-Telnet: 712 1475 | |============================================================================| From daemon@vxw.ee.lbl.gov Thu Jan 12 04:00:27 1995 From: daemon@vxw.ee.lbl.gov Date: Thu Jan 12 04:00:36 PST 1995 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Thu Jan 12 04:00:17 PST 1995 Subject: sysStartType and sysInit() Subject: Final Call For Papers: SIGPLAN Workshop on LCTRTS Subject: problem with time() Subject: Re: problem with time() ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: sysStartType and sysInit() Date: Wed, 11 Jan 1995 13:24:17 GMT From: rsgawera@ss11.wg.icl.co.uk (Raj Gawera) Organization: International Computers Limited Message-ID: Reply-To: rsgawera@ss11.wg.icl.co.uk Sender: news@oasis.icl.co.uk Dear VxWorkers, After a bit of digging around and following the vxWorks boot sequence, I came across the sysInit routine in sysALib.s. This routine gets called after decompression from romStart() and is the entry point for the decompressed vxWorks. the call to sysInit has the startType as a parameter but the sysinit routine does not look at this parameter, it simply calls usrInit() with a parameter of BOOT_WARM_AUTOBOOT (which is 0). It is this value which is then stored in sysStartType so every boot looks like a warm boot. I altered sysInit() to look at the parameter it was passed on the stack and forward it to usrInit() so sysInit now looks like : _sysInit: movew #0x3700,sr /* temporarily, for the debug stage */ movel a7@(0x4),d7 /* put startType in d7 */ movel #_sysInit,a7 /* set stack to grow down from code */ movel d7,a7@- /* push start type from d7 */ jsr _usrInit /* never returns - starts up kernel */ This works and sysStartType now reflects the reboot type. The question is, is the original sysInit routine bugged, or was it designed to make every boot look like a warm boot ? I know sysALib.s is part of the BSP but I've checked two 680x0 BSPs and their both identical. Can anyone with SPARC/other BSPs shed any light on their sysInit() routines. This is all on a 680x0 based ENET 1 with v5.1 standalone vxWorks. cheers, Raj Raj Gawera rsgawera@wg.icl.co.uk --------------------------- Newsgroups: comp.os.vxworks Subject: Final Call For Papers: SIGPLAN Workshop on LCTRTS Date: 11 Jan 1995 14:24:59 -0500 From: sshong@cs.umd.edu (Seongsoo Hong) Organization: U of Maryland, Dept. of Computer Science, Coll. Pk., MD 20742 Message-ID: <3f1bab$keb@bird.cs.umd.edu> CALL FOR PAPERS ACM SIGPLAN Workshop on Languages, Compilers and Tools for Real-Time Systems (In Conjunction with ACM SIGPLAN PLDI/PEPM) La Jolla, California June 21-22, 1995 ACM SIGPLAN LCT-RTS '95 is an interface between two dynamic fields of computer science and engineering: programming languages and real-time systems. The time is right for this workshop: top researchers in these areas are addressing many similar problems, but with slightly different perspectives and technologies. LCT-RTS provides a forum where these researchers can share their results and directions, and where they can potentially form new collaborations based on common interests. MOTIVATION: Until recently real-time systems development was the province of experienced specialists, who were faced with a variety of custom kernels, non-standard languages and vendor-specific device interfaces. System integration inevitably involved a complicated process of taking timing measurements, hand-tuning the code, and then re-measuring. These ad-hoc techniques have not scaled to support modern systems. There is a growing desire to adopt advanced design strategies, standard kernels, reusable modules, generic languages and the like. Also, the majority of real-time developers is longer drawn from the ranks of embedded controls experts; rather, it is composed of animators, physicists, video producers, musicians, medical technicians, automotive engineers, manufacturing engineers, etc. New software approaches are needed to support these new systems, and this new generation of real-time programmers. THE WORKSHOP: ACM SIGPLAN LCT-RTS '95 is devoted to investigating software technologies for contemporary real-time systems. Original submissions are invited in all areas relevant to this theme, including (but not restricted to) the following topics: * Programming Languages for Real-Time: Industrial and Research * Design: Requirements, System Specification, Analysis * Exception Handling: Semantics, Policies, Mechanisms * Prototyping Languages * Timing Analysis: Static and Dynamic Approaches * Scheduling Analysis * Realtime on RISC: Caches, Register Windows, Pipelines * Realtime Memory Management and Garbage Collection * Support for Parallelism and Data Placement * Program Transformation and Optimization for Real-Time Performance * Profiling, Measurement and Debugging * System Integration and Testing Of particular interest are case studies, or experimental results based on application-building experiences; for example in interactive graphics, imaging, manufacturing, etc. Papers should report new research, and should not exceed 5000 words (approximately 10 pages typeset 10-point on 16-point spacing, or 15 typewritten double-spaced pages). Short papers are also welcomed, which describe existing implementations, work-in-progress, or new problems and important issues. Short papers should not exceed 3000 words (6 pages). All accepted papers will be presented at the workshop and published in the proceedings, which will be distributed at the workshop. SUBMISSION: Please submit seven (7) copies of papers, to: ACM SIGPLAN LCT-RTS Attn: Richard Gerber Department of Computer Science University of Maryland College Park, MD 20742 USA Papers will be reviewed the program committee for appropriateness of content and presentation. Proceedings will be distributed at the workshop. IMPORTANT DATES Submission of draft paper: 23 January 1995 Notification of acceptance: 22 March 1995 Final version due: 15 May 1995 PROGRAM COMMITTEE Alan Burns (University of York) Richard Gerber, Co-Chair (University of Maryland) Rajiv Gupta (Univ of Pittsburgh) Mary Hall (Caltech) Connie Heitmeyer (Naval Research Lab) Insup Lee (University of Pennsylvania) Al Mok (University of Texas at Austin) Thomas Marlowe, Co-Chair (Seton Hall University, NJIT RTCL) Steve Tjiang (Synopsys Inc.) PROGRAM CO-CHAIRS Richard Gerber Thomas Marlowe Department of Computer Science Department of Mathematics University of Maryland Seton Hall University College Park, MD 20742, USA South Orange NJ 07079, USA telephone: +1 301 405 2710 telephone: +1 201 761 9784 fax: +1 301 405 6707 fax: +1 201 761 9596 rich@cs.umd.edu marlowe@cs.rutgers.edu --------------------------- Newsgroups: comp.os.vxworks Subject: problem with time() Date: 11 Jan 1995 20:20:20 GMT From: mwette@mr-ed.jpl.nasa.gov (Matt Wette) Organization: Jet Propulsion Laboratory Message-ID: <3f1ei4$17n@grover.jpl.nasa.gov> On any unix machine, time() provides the number of seconds since 1970. On VxWorks time() provides the number of seconds since the first time() call after reboot (at least on 5.1.1). Anyone know of a way to way to set the # of seconds so I can get time() to work properly? Matt - -- matthew.r.wette@jpl.nasa.gov --------------------------- Newsgroups: comp.os.vxworks Subject: Re: problem with time() Date: 11 Jan 1995 21:14:23 GMT From: hmp@frc2.frc.ri.cmu.edu (Henning Pangels) Organization: Field Robotics Center, CMU Message-ID: References: <3f1ei4$17n@grover.jpl.nasa.gov> Reply-To: hmp@frc2.frc.ri.cmu.edu (Henning Pangels) In article <3f1ei4$17n@grover.jpl.nasa.gov> mwette@mr-ed.jpl.nasa.gov (Matt Wette) writes: > On any unix machine, time() provides the number of seconds since 1970. > On VxWorks time() provides the number of seconds since the first > time() call after reboot (at least on 5.1.1). Anyone know of a way to > way to set the # of seconds so I can get time() to work properly? > > Matt > If your target board has a time-of-day/calendar chip (such as the MK48T02 or similar), you can write your own version of time() that accesses this chip. I've done this successfully before by including the code in usrConfig.c. I'm not certain right now, but you may have to remove the original time() from the vxWorks library. Without such a chip, I'm guessing the only way to get actual time is via the net, e.g. NTP or something like that, or through external hardware (like a digitizer and camera looking at a wristwatch :-) - -Henning - -- Henning Pangels | hmp@cs.cmu.edu | Field Robotics Center Senior Research Programmer| (412) 268-8911 |Carnegie-Mellon University - ---------------------------------------------------------------------------- I am a .signature virus. Copy me into your .signature today! --------------------------- End of New-News digest ********************** From leonid@rst.co.il Thu Jan 12 06:41:50 1995 From: leonid@rst.co.il (Leonid Rosenboim) Date: Thu Jan 12 06:41:56 PST 1995 Subject: Re: problem with time() The basic VxWorks package lacks the knowledge of absolute time, i.e. it only can start from the time it booted last. What you can do, is get the absolute time from somewhere else, and use the function clock_settime() to set the time once, then you can call time() or clock_gettime() to get the current time. This is practically sufficient for most purposes. In some rare cases you may want to fiddle with clcok_setres() to change the presumed sysClk frequency in order for your target time() to be in sync with an external time reference. There are basically three ways to get the absolute time at initialization time, and optionaly synchronize with later on: 1. The chapest is the battery-backed Reat Time Clock on your board (Mot use MK48T0x, others the 7170) 2. Use the "rdate" protocol to get the time from your closest Unix station on the network. Take the Unix "rdate" program source and port it to VxWorks (really easy stuff). 3. There is a wide range of time reference devices that can be hooked up via a serial link, their price ranges widely, so is their accuracy. If your target LAN is somehow hooked up into the Internet, that's your cheapest source for accurate time. My experience with NTP is not so good, these protocols average clocks of network stations, in my experience all SUNs are inaccurate. And NTP wastes an unnecesarily large amount of BW and CPU. That is unless you need a millisecond time synchronization ... I hope this clears things a bit. ----------------------------------------------------------------------- Leonid Rosenboim Phone: +972-3-559-8144 R S T Software Industries Ltd. Fax: +972-3-559-8244 3 Hashikma st. Industrial Zone, E-Mail: leonid@rst.co.il P.O.Box 11502, AZUR 58017, Israel WindRiver Distributor From leif@tsl.uu.se Thu Jan 12 07:32:37 1995 From: leif@tsl.uu.se Date: Thu Jan 12 07:32:43 PST 1995 Subject: Driver for GreenSpring IP-Stepper Does anyone have a VxWorks driver or interface routines for the GreenSpring IP-Stepper that they would like to share... leif ------------------------------------------------------------------------------- Leif Thuresson, The Svedberg Laboratory, Uppsala University, Sweden leif@tsl.uu.se From hebo@mbari.org Thu Jan 12 08:34:24 1995 From: Bob Herlien Date: Thu Jan 12 08:34:32 PST 1995 Subject: Re: problems with time() > On any unix machine, time() provides the number of seconds since 1970. > On VxWorks time() provides the number of seconds since the first > time() call after reboot (at least on 5.1.1). Anyone know of a way to > way to set the # of seconds so I can get time() to work properly? > > Matt Get the usrTime library from the VxWorks archive. It implements time(), gettimeofday(), settimeofday(), etc, in a Unix-compatible fashion. It includes code for the MK48T08 real-time clock chip, as implemented on the MVME-162 CPU board. If you use some other clock chip or CPU, you'll need to modify this code, but it should be very straightforward. -------------------------------------------------------------- Bob Herlien MBARI (Monterey Bay Aquarium Research Institute) hebo@mbari.org From hebo@mbari.org Thu Jan 12 08:47:58 1995 From: Bob Herlien Date: Thu Jan 12 08:48:04 PST 1995 Subject: Re: problem with time() I sent my earlier response to this question before reading Leonid's response below. Although I usually agree with Leonid's advice, I couldn't let this pass without comment. > My experience with NTP is not so good, these protocols average > clocks of network stations, in my experience all SUNs are inaccurate. > And NTP wastes an unnecesarily large amount of BW and CPU. That is > unless you need a millisecond time synchronization ... Granted, the original requestor just wanted a working time() function, so NTP is overkill. In fact, the only part of the NTP port for VxWorks that he needs is usrTime, which is much simpler to configure than NTP itself. But I disagree with your comments. Suns have had, indeed, a checkered history with regard to accurate time; mostly, I understand (I don't use them myself) due to dropping clock interrupts. I believe that these problems have been solved. In the early days of NTP, Suns required particular tweaking. But I'm constantly amazed that NTP is able to keep such accurate time while using such miniscule system resources, particularly net and CPU bandwidth. In steady state, it sends messages to your peers once every 16 minutes. -------------------------------------------------------------- Bob Herlien MBARI (Monterey Bay Aquarium Research Institute) hebo@mbari.org From mea@mclean.sparta.com Thu Jan 12 10:57:15 1995 From: mea@mclean.sparta.com (Mike Anderson) Date: Thu Jan 12 10:57:21 PST 1995 Subject: IP Multicast Greetings! This has been asked before, but we'll try again since I haven't seen any responses. Has anyone implemented (or are in the process of implementing) IP Multicast under VxWorks? It doen't have to be free (although that would be nice ;-). Regards, =============================================================================== __ Real-Time System Development, Integration, Training and Services //\\ // \\ Mike Anderson // /\ \\ Director, Real-Time Systems // / \ \\ SPARTA, Inc. Voice : (703) 448-0210 ext. 235 // \ \\ 7926 Jones Branch Drive FAX : (703) 734-3323 \\ \ // Suite 900 EMAIL : mea@mclean.sparta.com \\ \ / // McLean, VA 22102 \\ \/ // "Software development is like making \\ // a baby... You can't make a baby in one \\// month by impregnating nine women. -- "Pride in Performance" Some things just take time." =============================================================================== From guthrie@power.amasd.anatcp.rockwell.com Thu Jan 12 12:06:47 1995 From: guthrie@power.amasd.anatcp.rockwell.com (Bob Guthrie) Date: Thu Jan 12 12:06:54 PST 1995 Subject: IP Multicast To all: Mike Anderson writes the following: > Submitted-by mea@mclean.sparta.com Thu Jan 12 10:57:15 1995 > Submitted-by: mea@mclean.sparta.com (Mike Anderson) > > Greetings! > > This has been asked before, but we'll try again since I haven't seen any > responses. Has anyone implemented (or are in the process of implementing) > IP Multicast under VxWorks? It doen't have to be free (although that would > be nice ;-). > > Regards, I would also be interested in responses to this question. Robert Guthrie email: guthrie@power.amasd.anatcp.rockwell.com From wgsu12!rmarplcd@kns.com Thu Jan 12 13:58:35 1995 From: wgsu12!rmarplcd@kns.com (Russell Marple) Date: Thu Jan 12 13:58:42 PST 1995 Subject: 3rd party C++ class libraries Greetings, I'm looking into using the WindC++ Gateway. I'd like to hear from anyone which has used any 3rd party class libraries with VxWorks (ex. Rogue Wave's libraries tools.h++ or views.h++). Any comments on what worked (or didn't) would be appreciated. I'm using: VxWorks 5.1.1 MVME147 target Regards, Russ +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ + Russ Marple e-mail: marple@kns.com + Kulicke and Soffa Industries, Inc. voice: 215-784-6232 + 2101 Blair Mill Road fax: 215-659-7588 + Willow Grove, PA 19090 +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ From sidney@alfred.ca.boeing.com Thu Jan 12 15:28:25 1995 From: sidney@alfred.ca.boeing.com Date: Thu Jan 12 15:28:31 PST 1995 Subject: GE FANUC DNC2 Communication Protocol Implementation Has anyone implemented the FANUC DNC2 communication protocol on VxWorks or on any Unix machine ? Sidney *************************************************** * email: sidney@alfred.ca.boeing.com * * voice: (206)931-2140 * * usSnail: P.O. Box 3707 M/S 5K-14 * * Seattle, Wa 98124 * * * *Disclaimer: The contents of this message in no * * way reflects the Opinions either real* * or imaginary of the Boeing Company. * * All Opinions/rantings are my own and * * only I am responsible for them. * *************************************************** From vogel@fz.telekom.de Fri Jan 13 05:08:37 1995 From: vogel@fz.telekom.de (Veit Vogel) Date: Fri Jan 13 05:08:44 PST 1995 Subject: 3rd party C++ class libraries Hi, > I'm looking into using the WindC++ Gateway. I'd like to hear from anyone > which has used any 3rd party class libraries with VxWorks (ex. Rogue Wave's > libraries tools.h++ or views.h++). Any comments on what worked (or didn't) > would be appreciated. We have just the same question especially for tools.h++ and we appreciate any comments. Does someone has any experiences???? We are using VxWorks version 5.1.1 on Force boards 30 and 40 together with WindC++ Gateway. Thanks for comments and greetings to the outside world Veit Vogel ================================================= | Veit Vogel, FZ 222 c | | Telekom Forschungszentrum | | | | Am Kavalleriesand 3 | | D-64295 Darmstadt | | Germany | | | | Phone: +49 6151 83 2573 | | Fax: +49 6151 83 4590 | | email: vogel@fz.telekom.de | ================================================= From mkrokosz@astro.ge.com Fri Jan 13 06:17:59 1995 From: Matthew Krokosz Date: Fri Jan 13 06:18:05 PST 1995 Subject: Non-Blocking Socket "send/write" I have a vxWorks application that connects to a remote application via TCP/IP sockets. Once connected to the remote application, the vxWorks application sends data to the remote application. I need to prevent the "write" call to the socket from blocking. I have tried using the ioctl with the FIONBIO option but that did not work. The condition I am seeing this write block is when the remote application does not service its socket. I have produced a test case where the remote application goes into an infinite loop once it starts to receive data via the socket. This causes the remote application not to read any more data from its TCP receive buffer. Once the remote application goes into the infinite loop, the vxWorks application which is writing to the socket will block after writing X bytes to the sockets. I think that X bytes is equal to the size of the TCP receive buffer on the remote side plus the size of the TCP send buffer on the vxWorks side. By observing the send buffer with "inetstatShow" it is apparent that the write does block after the send buffer fills to its max, which probably occurs after the remote receiver buffer fills to its max. Ideally I would like to make this "write/send" system call non-blocking, but using the ioctl did not work. I could prevent the vxWorks application from getting into this blocking situation if I could obtain the number of bytes in the send buffer before writing to the socket. I could then prevent the application from writing to the socket if it would max out the send buffer and create a blocking situation. I can not seem to find out how to obtain the number of bytes in the send buffer. Any help would be appreciated! ************************************************************************* | Matthew Krokosz email: mkrokosz@astro.ge.com | | Martin Marietta Astro Space phone: (609) 490-7194 | | Princeton, NJ | | | ************************************************************************* From eugenel@condorsys.com Fri Jan 13 09:45:58 1995 From: eugenel@condorsys.com (Eugene Leung X714) Date: Fri Jan 13 09:46:04 PST 1995 Subject: MVME162 IndustryPacks Does anyone know if someone makes an IndustryPack mezzanine for a 1553B interface? And what about ARINC 429? Eugene ================================================================================ - - - Eugene Leung + email: eugenel@condorsys.com - - Sr. S/W Engineer + phone: (408)371-9580 ext 714 - - Condor Systems Inc. + fax: (408)371-9589 - - 2133 Samaritan Dr. - - San Jose, CA 95124 - - - ================================================================================ From jtosey@glenayre.com Fri Jan 13 10:07:34 1995 From: jtosey@glenayre.com (Joseph Tosey [4572]) Date: Fri Jan 13 10:07:42 PST 1995 Subject: Maximizing Ethernet Throughput We are evaluating a board using the 82596 ethernet chip. According to the specs, it can handle full duplex transfers. Question: Has you tried connecting two ethernet devices, point-to-point, using a flavor of NULL-modem cable? One application for this is connecting a high end server to an ethernet switch so that the switch acts as a concentrator. Doing this should allow transfers of 10 Mbps in each direction for an aggregate throughput of 20 Mbps. How did you do it? What kind of cabling did you use? Any special drivers? What results did you get? ------------------------------------------------------------------------------- Joseph Tosey Fax: +1 604 293 4317 Manager, OS & Networking Reception: +1 604 293 1611 Glenayre Electronics Voice Mail: +1 604 293 4399 x4572 1570 Kootenay St, Vancouver, BC V5K 2B8 Timezone: PST/PDT (same as US/CA) Canada From mea@mclean.sparta.com Fri Jan 13 15:40:56 1995 From: mea@mclean.sparta.com (Mike Anderson) Date: Fri Jan 13 15:41:03 PST 1995 Subject: Re: Maximizing Ethernet Throughput Greetings! > Submitted-by: jtosey@glenayre.com (Joseph Tosey [4572]) > > We are evaluating a board using the 82596 ethernet chip. According > to the specs, it can handle full duplex transfers. > > Question: Has you tried connecting two ethernet devices, > point-to-point, using a flavor of NULL-modem cable? > > One application for this is connecting a high end server to > an ethernet switch so that the switch acts as a concentrator. > Doing this should allow transfers of 10 Mbps in each > direction for an aggregate throughput of 20 Mbps. > > How did you do it? What kind of cabling did you use? > Any special drivers? What results did you get? I have used a special Null modem-type cable to hook two twisted pair XCVRs back to back as a point-to-point Ethernet. You only got 10 MBPs, but it did work and only required 2 $40 XCVRs and a piece of twisted-pair wiring with RJ-45 jacks on them. I presume that, if your SBC had a TP interface on it, that you could hook two of them back to back with just a $5 cable. I've never used the Ethernet switches, so I can't comment on that aspect. Regards, =============================================================================== __ Real-Time System Development, Integration, Training and Services //\\ // \\ Mike Anderson // /\ \\ Director, Real-Time Systems // / \ \\ SPARTA, Inc. Voice : (703) 448-0210 ext. 235 // \ \\ 7926 Jones Branch Drive FAX : (703) 734-3323 \\ \ // Suite 900 EMAIL : mea@mclean.sparta.com \\ \ / // McLean, VA 22102 \\ \/ // "Software development is like making \\ // a baby... You can't make a baby in one \\// month by impregnating nine women. -- "Pride in Performance" Some things just take time." =============================================================================== From Greg.Gordon@dlep1.itg.ti.com Sat Jan 14 19:27:46 1995 From: Greg.Gordon@dlep1.itg.ti.com Date: Sat Jan 14 19:27:52 PST 1995 help From daemon@vxw.ee.lbl.gov Sun Jan 15 04:00:26 1995 From: daemon@vxw.ee.lbl.gov Date: Sun Jan 15 04:00:34 PST 1995 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Sun Jan 15 04:00:17 PST 1995 Subject: Writing GUIs with an embedded OS Subject: Re: Writing GUIs with an embedded OS Subject: Any RIP-2, OSPF, etc. VxWorks ports? Subject: Re: VME to Sbus adaptors Subject: Re: MVME162 IndustryPacks Subject: VxWorks development environment? ------------------------------------------------------- Newsgroups: comp.os.vxworks,comp.os.os9 Subject: Writing GUIs with an embedded OS Date: 13 Jan 1995 20:13:12 GMT From: ffdixon@bgtys3.bnr.ca (Frederick Dixon) Organization: Bell-Northern Research Keywords: GUI Message-ID: <3f6mso$q9l@bcarh8ab.bnr.ca> Hi everyone, I'm looking for advice from people who have experience writing GUI's with embedded operating systems -- specifically VX-Works or OS/9. I work at Bell-Northern Research, the research arm of Northern Telecom. My current project involves building a prototype hand-held device that has a touch-screen LCD interface. We're searching for an embedded OS for the device to drive its 160x240 LCD display. A lot of design work is going into the user interface -- ink notes, buttons, windows, scrolling lists, and animation, and sound - -- requiring the OS (and its libraries) to handle complex graphics. We don't want to write the low-level graphics code ourselves! Instead, we would like to design the interface with a GUI builder. RAM and ROM space are at a premium. X-Windows is just way to big. I would very interested in hearing from anyone who has experience in these areas. Cheers,... Fred Dixon, ffdixon@bnr.ca Software Developer, Bell-Northern Research --------------------------- Newsgroups: comp.os.vxworks,comp.os.os9 Subject: Re: Writing GUIs with an embedded OS Date: Fri, 13 Jan 95 21:17:18 GMT From: arb@rtsi.com (Allan R. Batteiger) Organization: Real-Time Services Inc Keywords: GUI Message-ID: <3f6ubu$j57@news.onramp.net> References: <3f6mso$q9l@bcarh8ab.bnr.ca> ffdixon@bgtys3.bnr.ca (Frederick Dixon) wrote: >I'm looking for advice from people who have experience writing >GUI's with embedded operating systems -- specifically VX-Works or OS/9. > >I work at Bell-Northern Research, the research arm of Northern >Telecom. My current project involves building a prototype hand-held >device that has a touch-screen LCD interface. > >We're searching for an embedded OS for the device to drive its 160x240 >LCD display. A lot of design work is going into the user interface -- >ink notes, buttons, windows, scrolling lists, and animation, and sound >-- requiring the OS (and its libraries) to handle complex graphics. We >don't want to write the low-level graphics code ourselves! Instead, >we would like to design the interface with a GUI builder. > >RAM and ROM space are at a premium. I can give you @ options for OS9 Option 1: Gwindows Produced by Gespac. Very nice user I/F very easy to work with and build user apps. I can provide users references if you would like to e-mail or talk to them. Option 2: RAVE Produced by Microware Also has a nice user I/F Advantage is built in sound support. Gwindows does not directly support sound. Possible Disadvantage: You need to check with Microware But I am not sure if RAVE is supported with OS9 3.0 (latest version). I may be able to provide user references but it has been a while. - ------------------------------------- Allan R. Batteiger - President Real-Time Services Inc. E-mail: arb@rtsi.com PH. (214) 245-4239 --------------------------- Newsgroups: comp.os.vxworks,comp.protocols.tcp-ip Subject: Any RIP-2, OSPF, etc. VxWorks ports? Date: 13 Jan 1995 22:45:45 GMT From: bonaker@amasd.anatcp.rockwell.com (Joe Bonaker) Organization: Rockwell Keywords: VxWorks RIP OSPF BGP GateD Message-ID: <3f6vqp$43p@huron.eel.ufl.edu> Followup-To: poster Reply-To: bonaker@amasd.anatcp.rockwell.com Does anyone have VxWorks versions of the following Internet protocols for share/sale? (I'm only aware of the routed/RIP implementation on the VxWorks ftp archive.) Cornell's GateD -- kitchen sink: RIP-[12], OSPF2, EGP2, BGP[234] ... RIP-2 -- RFC 1388 OSPF -- Version 2, RFC 1247 BGP -- Version 3, RFC 1268 or Version 4, RFC 1467 Path MTU Discovery -- RFC 1191 Router Discovery -- RFC 1256 (in.rdisc via ICMP raw sockets) IEEE 802 netif -- Example driver supporting both 802 and Ether encap Please E-mail any relevant info; I'll followup in comp.os.vxworks. Thanks, Joe B. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: VME to Sbus adaptors Date: Sat, 14 Jan 1995 17:42:07 +0000 From: david@signus.demon.co.uk (David Firth) Organization: Home Message-ID: <1995Jan14.174207.585@signus.demon.co.uk> References: <3evaot$8nr@oz.sim.es.com> Sender: usenet@demon.co.uk Stanford Pitcher (spitcher@oz.sim.es.com) wrote: : Hello, : 50 Hz) in sync with other. We hoped to use TCP/IP sockets but are : experiencing frequent glitches where the Sun's app does not respond for : several frames (sometimes as many as 30). We are assuming that although : we are running the Sun's app at real-time priority level, the net task If the delay occurs roughly once every 30 seconds it might be the update process. We have suffered a similar problem using SunOS 4.1.3 which is why we are considering vxworks! - -- David Firth --------------------------- Newsgroups: comp.os.vxworks Subject: Re: MVME162 IndustryPacks Date: 15 Jan 1995 05:28:51 GMT From: pjg@tesla.svl.trw.com (Paul Gyugyi) Organization: TRW Avionics & Surveillance Group Message-ID: References: <9501131745.AA10930@csibtfr> In article <9501131745.AA10930@csibtfr> eugenel@condorsys.com (Eugene Leung X714) writes: Does anyone know if someone makes an IndustryPack mezzanine for a 1553B interface? And what about ARINC 429? Ballard Technology makes an ARINC 429 IndustryPack. They also will package it as a Motorola 162 with up to four ARINC IP's, with firmware to store the individual data in pre-assigned VME adresses (in case you are looking for a low software development solution). Each IP module has two HI-8282 chips, providing a total of four inputs and two outputs. Ballard's number is (206) 339-0281. Disclaimer: I have not used these; I just have some literature from them. - -- =--=--=--=--=--=--=--=--=--=--=--=--=--=--=--=--=--=--=--=--=--= | Paul Gyugyi Paul_Gyugyi@smtp.svl.trw.com | | paul@gyugyi.com gyugyi@earthsea.stanford.edu | =--=--=--=--=--=--=--=--=--=--=--=--=--=--=--=--=--=--=--=--=--= --------------------------- Newsgroups: comp.os.vxworks Subject: VxWorks development environment? Date: Sat, 14 Jan 1995 10:14:27 +0100 From: henryg@abc.se (Henry Gessau) Organization: None whatsoever Message-ID: Hi all, We are in the process of changing our embedded hardware from 80186-based to 68360. At the same time we will go from a home-made monitor to VxWorks. We are a bit worried about the iron required for a decent development environment. It seems that the best (best supported, most utilities) is some sort of Unix workstation (Sun?). The problem is that we (about 50 of us) would then have to aquire Unix workstations since we are currently running PC's. What would we be missing out on if we stuck to our PC's? Surely there is some support for MS-Windows? How about Windows NT? - -- Henry Gessau henryg@abc.se / sighg@eb.se / 100073.2031@compuserve.com --------------------------- End of New-News digest ********************** From jtosey@glenayre.com Sun Jan 15 19:57:48 1995 From: jtosey@glenayre.com (Joseph Tosey [4572]) Date: Sun Jan 15 19:57:55 PST 1995 Subject: Re: Maximizing Ethernet Throughput > > We are evaluating a board using the 82596 ethernet chip. According > > to the specs, it can handle full duplex transfers. > > > > Question: Have you tried connecting two ethernet devices, > > point-to-point, using a flavor of NULL-modem cable? > > > > One application for this is connecting a high end server to > > an ethernet switch so that the switch acts as a concentrator. > > Doing this should allow transfers of 10 Mbps in each > > direction for an aggregate throughput of 20 Mbps. > > > > How did you do it? What kind of cabling did you use? > > Any special drivers? What results did you get? > > I have used a special Null modem-type cable to hook two twisted pair > XCVRs back to back as a point-to-point Ethernet. You only got 10 MBPs, but > it did work and only required 2 $40 XCVRs and a piece of twisted-pair > wiring with RJ-45 jacks on them. I presume that, if your SBC had a > TP interface on it, that you could hook two of them back to back with > just a $5 cable. I've never used the Ethernet switches, so I can't comment > on that aspect. I'm not sure that we're on the same wavelength. From what I understand, you are still running half-duplex, while I wish to run full-duplex. Did I understand you correctly? I.e., both ends simultaneously transmitted and received? In the 1992 Intel "32-Bit Local Area Network (LAN) Component User's Manual" (82596 manual), section 1.26.2 "Full-Duplex Operation" paragraph, it says: -- "The 82596 does not use the CSMA/CD protocol when it is configured for full-duplex operation. During full-duplex operation the 82596 can transmit regardless of the state of its receiver--with one exception. The 825896 will not transmit a new frame if the receiver is continuously active between the previous frame transmission and the new transmit-pending frame. During the full duplex operation, the IFS timer is inactive and collisions cannot occur." -- This is the capability I wish to deploy. I first came across this idea a few years ago when I read an article about someone who connected to a drone sub through a three-mile cable using a similar technique. Thanks in advance. ------------------------------------------------------------------------------- Joseph Tosey Fax: +1 604 293 4317 Manager, OS & Networking Reception: +1 604 293 1611 Glenayre Electronics Voice Mail: +1 604 293 4399 x4572 1570 Kootenay St, Vancouver, BC V5K 2B8 Timezone: PST/PDT (same as US/CA) Canada From leonid@rst.co.il Sun Jan 15 23:41:36 1995 From: leonid@rst.co.il (Leonid Rosenboim) Date: Sun Jan 15 23:41:43 PST 1995 Subject: Re: Maximizing Ethernet Throughput If you got two Ethernet stations with 10baseT sockets, it's quite easy to connect them point to point (see below). This however does not necesarily mean it' going to work full duplex just because of that. Some chip/driver combinations can do full-duplex, some don't. These that do, usually are like this not intentionally, cause Full Duplex Ethernet is a new thing. In 10BaseT there is also a DTE/DCE-like relation: the work station, PC or 10baseT MAU are like DTEs, the HUB or LANswitch is kinda DCEish. Usually you connect a station to a HUB, which requires a one-to-one cable, connecting ins {1,2}; {3,6}. The parens determine the pairs. If for some reason you want to connect two DTEs point-to-point, just make a cable like this: 1 -> 3 2 -> 6 3 -> 1 6 -> 2 You can get by without a HUB in some situations by this. You can also connect two HUBs to one another with the same type of cable. Special note to Joseph Tosey: your Ethernet switch is probably a DCE so a twisted cable wont do it for you. ----------------------------------------------------------------------- Leonid Rosenboim Phone: +972-3-559-8144 R S T Software Industries Ltd. or: +972-50-307-142 3 Hashikma st. Industrial Zone, Fax: +972-3-559-8244 P.O.Box 11502, AZUR 58017, Israel E-Mail: leonid@rst.co.il From daemon@vxw.ee.lbl.gov Mon Jan 16 04:00:36 1995 From: daemon@vxw.ee.lbl.gov Date: Mon Jan 16 04:00:43 PST 1995 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Mon Jan 16 04:00:16 PST 1995 Subject: Re: VxWorks development environment? ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: VxWorks development environment? Date: 15 Jan 1995 15:53:22 GMT From: btadlock@metronet.com Organization: Texas Metronet Message-ID: <3fbgdj$ib3@feenix.metronet.com> References: In article henryg@abc.se (Henry Gessau) wrote: > Hi all, > > We are in the process of changing our embedded hardware from > 80186-based to 68360. At the same time we will go from a home-made > monitor to VxWorks. We are a bit worried about the iron required for > a decent development environment. > > It seems that the best (best supported, most utilities) is some sort > of Unix workstation (Sun?). The problem is that we (about 50 of us) > would then have to aquire Unix workstations since we are currently > running PC's. > > What would we be missing out on if we stuck to our PC's? Surely there > is some support for MS-Windows? How about Windows NT? You do not want to use ms-windows/nt based development toolsfor vxWorks development, you will be much happier with the Unix based development system. I just completed a vxworks project using a 68360 and we used a SUN Host and PC's running a ms-windows based Xserver called Xvision that connected to the sun and allowed us to run the SUN based development system from our PC's. email me direct if you want more details. BTW, I wrote almost all of the low level 68360 code/drivers for the project. Hope This Helps! btadlock@metronet.com --------------------------- End of New-News digest ********************** From djb@rtp.co.uk Mon Jan 16 06:30:49 1995 From: djb@rtp.co.uk (David Bullimore) Date: Mon Jan 16 06:30:55 PST 1995 Subject: Target Mode SCSI Dear All, Has anyone produced target mode SCSI support for VxWorks? We want to use a VxWorks sytem as a slave on the SCSI bus. Thanks David *********************************************** * David Bullimore * * Real Time Products Ltd. (RTP Ltd.) * * Unit 8 * * Holt Court * * Aston Science Park * * Birmingham * * B7 4EJ * * England * * * * Phone: +44.21.333 6955 * * Fax: +44.21.333 5433 * * * * Email djb@rtp.co.uk * *********************************************** From mea@mclean.sparta.com Mon Jan 16 06:33:54 1995 From: mea@mclean.sparta.com (Mike Anderson) Date: Mon Jan 16 06:34:01 PST 1995 Subject: Re: VxWorks Development Environment Greetings! > From: henryg@abc.se (Henry Gessau) > Hi all, > > We are in the process of changing our embedded hardware from > 80186-based to 68360. At the same time we will go from a home-made > monitor to VxWorks. We are a bit worried about the iron required for > a decent development environment. > > It seems that the best (best supported, most utilities) is some sort > of Unix workstation (Sun?). The problem is that we (about 50 of us) > would then have to aquire Unix workstations since we are currently > running PC's. > > What would we be missing out on if we stuck to our PC's? Surely there > is some support for MS-Windows? How about Windows NT? Well Henry, there are several alternatives. First, if your PCs are networked together, you can always purchase a single Sun (or one of the ISA-bus based Opus SPARCards that plug into your PC) and then use one of the many fine X-Windows emulators that run under MS-DOG. I've used Hummingbird's Exceed and VisonWare's Xvision products with great success. That seems to fit with WRS's licensing policy of 10 users on a single development host. You may want to buy a couple of hosts (and don't forget the 2nd/3rd development (fortunately at half price) license :-), to support all 50 of your developers. I can't immediately recall if 10 users means 10 *simultaneous* users or not. Anyone from WRS care to clarify? Alternatively, you could use VxWorks for Windows and run the Green Hills Multi compiler/debugger on the PC. I'm not sure if CPU32+ architectures are supported yet on that product, but I'm sure that they soon will be if they aren't already. However, for the number of people you're talking about, this approach would probably be more expensive than the Sun approach owing to the per seat cost of the Multi development environment. The advantage here is using your current PCs and the possible use of a floating license manager (Green Hills has this for the Sun, I'm not sure about the PC) such that you only buy, say 10 licenses, and only 10 people at a crack can be compiling/debugging simultaneously. I don't know about the status of support for Windows NT, but since Microsoft keeps slipping shipment (and I've heard many complaints about the current NT development environment), I don't know that I'd count on NT in 1995. Since programmers tend to spend far more time in the editor than actually compiling or using the source code debuggers (and you can always use printf statements ;-) this may work for you. Also, don't forget those target licenses and the ability to use the symbolic debugger provided with VxWorks. I used that one by itself for the first 4 years I worked with VxWorks and it's really quite powerful. HTH, =============================================================================== __ Real-Time System Development, Integration, Training and Services //\\ // \\ Mike Anderson // /\ \\ Director, Real-Time Systems // / \ \\ SPARTA, Inc. Voice : (703) 448-0210 ext. 235 // \ \\ 7926 Jones Branch Drive FAX : (703) 734-3323 \\ \ // Suite 900 EMAIL : mea@mclean.sparta.com \\ \ / // McLean, VA 22102 \\ \/ // "Software development is like making \\ // a baby... You can't make a baby in one \\// month by impregnating nine women. -- "Pride in Performance" Some things just take time." =============================================================================== From leonid@rst.co.il Tue Jan 17 00:46:52 1995 From: leonid@rst.co.il (Leonid Rosenboim) Date: Tue Jan 17 00:47:00 PST 1995 Subject: Milspec SPARC Could anyone give me a pointer to where I can buy an off-the-shelf Millitary board (VME) based on one of the SPARC processors ? ----------------------------------------------------------------------- Leonid Rosenboim Phone: +972-3-559-8144 R S T Software Industries Ltd. Fax: +972-3-559-8244 3 Hashikma st. Industrial Zone, E-Mail: leonid@rst.co.il P.O.Box 11502, AZUR 58017, Israel WindRiver Distributor From daemon@vxw.ee.lbl.gov Tue Jan 17 04:00:37 1995 From: daemon@vxw.ee.lbl.gov Date: Tue Jan 17 04:00:45 PST 1995 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Tue Jan 17 04:00:17 PST 1995 Subject: Micro Tec compiler? Subject: VxWorks phone # ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Micro Tec compiler? Date: 16 Jan 1995 21:27:34 GMT From: wenchao@po.EECS.Berkeley.EDU (Wenchao Yang) Organization: University of California, at Berkeley Message-ID: <3feoc6$s48@agate.berkeley.edu> Hi, We are planing to migrate VxWorks from MIPS R3000 to 68EC040, and the cross compiler we have for the 48EC040 is the Micro Tec cross compiler. So is there anyone who has successfully used the Micro Tec compiler for the VxWorks? Any informaiton would be greatly appreciated. - --wenchao-- wenchao@storage.tandem.com --------------------------- Newsgroups: comp.os.vxworks Subject: VxWorks phone # Date: 16 Jan 1995 18:03:31 GMT From: matt@lachman.com (Matt Harper) Organization: Lachman Technology, Inc., Naperville, IL Message-ID: <3fecdj$bff@genesis.npr.legent.com> Does anyone have the phone # / email address for the company that writes VxWorks? I would like to get some marketing liturature & product info. Thanks in advance, Matthew Harper Telephone: 708-505-9555 X317 Legent Corp., Lachman Networking Div Fax: 708-505-9574 1901 N. Naper Blvd. Internet: matt@npr.legent.com Naperville IL 60563 USA --------------------------- End of New-News digest ********************** From mcolavit@huey.jpl.nasa.gov Tue Jan 17 10:50:39 1995 From: mcolavit@huey.jpl.nasa.gov (Mark Colavita) Date: Tue Jan 17 10:50:46 PST 1995 Subject: IP module for f/o TAXI Hi, Does anyone know of an IP (IndustryPack) module providing a fiber optic TAXI interface using the HP DLT/DLR 6000 transmitter/receiver or similar? or even a copper implementation? Thanks. Mark Colavita mcolavit@huey.jpl.nasa.gov Jet Propulsion Laboratory, MS: 306-388 tel: 818-354-7835 4800 Oak Grove Dr. Pasadena, CA 91109 fax: 818-393-9471 From eugenel@condorsys.com Tue Jan 17 18:02:59 1995 From: eugenel@condorsys.com (Eugene Leung X714) Date: Tue Jan 17 18:03:05 PST 1995 Subject: psos exploder ? Does anyone know if there's an exploder for psos? ================================================================================ - - - Eugene Leung + email: eugenel@condorsys.com - - Sr. S/W Engineer + phone: (408)371-9580 ext 714 - - Condor Systems Inc. + fax: (408)371-9589 - - 2133 Samaritan Dr. - - San Jose, CA 95124 - - - ================================================================================ From johncal@sunshine.labs.tek.com Tue Jan 17 19:17:59 1995 From: johncal@sunshine.labs.tek.com (John Calvin) Date: Tue Jan 17 19:18:05 PST 1995 Subject: Power PC compilers Greetings: Were embarking on a project using the Beta version of the Wind River Power PC port. In talking to their sales people, so are a lot of other people. If this is truly the case then there are plenty of people with the same question we have. What compiler are people using for Power PC's? Currently our only semi supported option is the Green Hill's compiler and this needs to be purchased separately. Will anybody share their comments on how well this has worked out? Diab Data I hear is working on one, but it will be a while. GNU I've been advised, is 8 to 10 months away from a Power-PC compiler. Can anybody confirm this? Motorola apparently has a Power PC compiler, which according to some sources has the highest level of optimization. This is hardly surprising, as they know more about the Power PC than anyone. To use this would require its "ecoff" output files be converted to an "elf" format or a re- write of the targets symbolic linker to accept ecoff. Either way is equally unpleasant. Has anyone done this already? Any thoughts from others considering the same questions would be appreciated. Thanks in advance. Cheers John Calvin, Design Technology Research Lab, Tektronix Inc. 14150 S.W. Karl Braun Drive P.O. Box 500, M.S. 50-662 Beaverton OR 97077 U.S.A. John.C.Calvin@tek.com, 503-627-4840, Fax 503-627-7875 From bolz@mars.egt.fr Tue Jan 17 23:39:09 1995 From: Bernard Bolz Date: Tue Jan 17 23:39:15 PST 1995 Subject: GNU compiler on PC for 68k targets Hi everybody, We are using gcc on HP 9000 for 68k targets, and we would like to generate the same code from PC's. So we are looking for a port of gcc for PC's, which can produce code for 68k. Does anyone know of such a compiler? Any help would be greatly appreciated. B. BOLZ. -- +------------------------------------+------------------------------------+ | Bernard BOLZ | European Gas Turbines | | EGT PEDL | 3, avenue des Trois Chenes | | tel: (33) 84.55.63.88 | 90018 BELFORT Cedex | | fax: (33) 84.55.35.21 | FRANCE | | E-Mail: bbolz@egt.fr | | +------------------------------------+------------------------------------+ From mea@mclean.sparta.com Wed Jan 18 06:08:46 1995 From: mea@mclean.sparta.com (Mike Anderson) Date: Wed Jan 18 06:08:52 PST 1995 Subject: Re: GNU compiler on PC for 68k targets Greetings! > Submitted-by bolz@mars.egt.fr Tue Jan 17 23:39:09 1995 > Submitted-by: Bernard Bolz > > Hi everybody, > > We are using gcc on HP 9000 for 68k targets, and we would like to generate > the same code from PC's. > So we are looking for a port of gcc for PC's, which can produce code for 68k. > Does anyone know of such a compiler? > According to their ad in Embedded Systems Mag, Hundred Acre Consulting in Reno, Nevada (+1-702-829-9700) has such a beast with one year of support in source code for $495 ($US). I've not used it myself, but to get the sources and binaries ready to run with 1 year of support for <$500 sounds like a deal to me. If nothing else, it would get you close without having to start from scratch. Regards, =============================================================================== __ Real-Time System Development, Integration, Training and Services //\\ // \\ Mike Anderson // /\ \\ Director, Real-Time Systems // / \ \\ SPARTA, Inc. Voice : (703) 448-0210 ext. 235 // \ \\ 7926 Jones Branch Drive FAX : (703) 734-3323 \\ \ // Suite 900 EMAIL : mea@mclean.sparta.com \\ \ / // McLean, VA 22102 \\ \/ // "Software development is like making \\ // a baby... You can't make a baby in one \\// month by impregnating nine women. -- "Pride in Performance" Some things just take time." =============================================================================== From daemon@vxw.ee.lbl.gov Thu Jan 19 04:00:33 1995 From: daemon@vxw.ee.lbl.gov Date: Thu Jan 19 04:00:47 PST 1995 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Thu Jan 19 04:00:21 PST 1995 Subject: Re: problem with time() Subject: VME ATM with VxWorks drivers... Subject: How can I test for existence of a file? Subject: Re: VxWorks Development Environment Subject: Problems connecting to target via serial line Subject: FAQs, FTP & Web sites, etc. Subject: Opening Local DOS FS files Subject: Re: Non-Blocking Socket "send/write" Subject: Re: Problems connecting to target via serial line Subject: Re: Problems connecting to target via serial line Subject: execute() Subject: Re: Power PC compilers Subject: svc_run Subject: Re: VME ATM with VxWorks drivers... Subject: Re: Opening Local DOS FS files Subject: Bus Errors Subject: XDI SMT port to VxWorks Subject: IPOctal Subject: Re: VxWorks Development Environment ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: problem with time() Date: Fri, 13 Jan 1995 20:38:43 GMT From: georg@sgl.ists.ca (Georg Feil) Organization: Institute for Space and Terrestrial Science Message-ID: References: <3f1ei4$17n@grover.jpl.nasa.gov> Sender: news@newshub.ists.ca hmp@frc2.frc.ri.cmu.edu (Henning Pangels) writes: >Without such a chip, I'm guessing the only way to get actual time is >via the net, e.g. NTP or something like that, or through external >hardware (like a digitizer and camera looking at a wristwatch :-) One of the easiest ways to get time over the net is RDATE. Here is some sample code. Georg. - --------------------------- rdate.c -------------------------- #include #include #include #include #include #include #include /* well-known internet port number for rdate, from Unix /usr/include/netinet/in.h (missing in VxWorks) */ #define IPPORT_TIMESERVER 37 STATUS rdate_get(const char* hostname, unsigned long* time) /* * Get the time from remote host using RDATE protocol. * 'hostname' is the host name or inet address to use. * 'time' returns the time as an integer number of seconds since midnight, * January first, 1900 (subtract 2208988800 to convert to Unix time, * which is in seconds since midnight, January first, 1970). * Note that the rdate time format will overflow in about 2036. * Return value is VxWorks status code, errno==S_hostLib_UNKNOWN_HOST * means host name not found. */ { struct sockaddr_in socka; int sockfd; /* socket descriptor */ struct timeval timeout; /* timeout value to pass to connectWithTimeout() */ unsigned long tmptime; /* time value read */ bzero((char *)&socka, sizeof(socka)); socka.sin_family = AF_INET; /* Decode a.b.c.d numeric IP address, if illegal must be a host name */ socka.sin_addr.s_addr = inet_addr(hostname); if (socka.sin_addr.s_addr == -1L) { /* -1 means invalid address */ /* look up host name */ socka.sin_addr.s_addr = hostGetByName((char*)hostname); if (socka.sin_addr.s_addr == -1L) { /* -1 means error */ /* errno will be S_hostLib_UNKNOWN_HOST here if host name not found */ return(ERROR); } } socka.sin_port = IPPORT_TIMESERVER; if ((sockfd = socket(AF_INET, SOCK_STREAM, 0)) < 0) { return(ERROR); } /* call connect() with a timeout of 8 seconds */ timeout.tv_sec=8; timeout.tv_usec=0; if (connectWithTimeout(sockfd, (struct sockaddr *)&socka, sizeof(socka), &timeout) < 0) { close(sockfd); /* ignore errors, not crucial */ return(ERROR); } /* read 4 bytes from socket, the time */ if (nread(sockfd, (char*)&tmptime, sizeof(unsigned long)) < sizeof(unsigned long)) { close(sockfd); /* ignore errors, not crucial */ return(ERROR); } close(sockfd); /* ignore errors, not crucial */ /* Note that the ntohl macro is a no-op on a 68000 series processor, but technically should be there as the rdate protocol is defined to return bytes in network order. */ *time = ntohl(tmptime); return(OK); } - -- Georg Feil Space Geodynamics Laboratory | Internet: georg@sgl.ists.ca Institute for Space and Terrestrial Science | Phone: (416) 665-5458 4850 Keele St./North York/Ont/Canada/M3J 3K1 | Fax: (416) 665-1815 --------------------------- Newsgroups: comp.os.vxworks Subject: VME ATM with VxWorks drivers... Date: Tue, 17 Jan 1995 13:17:56 GMT From: heyes@codahp.cebaf.gov (Graham Heyes) Organization: Continuous Electron Beam Accelerator Facility Message-ID: Followup-To: comp.os.vxworks Sender: usenet@murdoch.acc.Virginia.EDU We are putting together an ATM network with several VME based VxWorks systems. I had a look three months ago and couldn't find a VME card with VxWorks drivers for ATM. Does anyone have any better information. If I don't find anything soon I'll be buying a FORE card and writing my own driver... - -- Graham - -<>-<>-<>-<>-<>-<>-<>-<>-<>-<>-<>-<>-<>-<>-<>-<>-<>-<>-<>-<>-<>-<>-<>-<>-<>-<>- Graham Heyes,CEBAF,12000 Jefferson Ave,Newport News,VA 23606,Tel:(804) 249-7030 --------------------------- Newsgroups: comp.os.vxworks Subject: How can I test for existence of a file? Date: Tue, 17 Jan 1995 13:23:28 GMT From: gamin@ireq-robot.hydro.qc.ca (Martin Boyer) Organization: La division Robotique de l'Institut de recherche d'Hydro-Quebec Message-ID: Followup-To: comp.os.vxworks Sender: news@ireq.hydro.qc.ca (Netnews Admin) I am trying to test for the existence of a file (in an NFS-mounted partition) under VxWorks 5.1.1. Testing the return value of the following functions gives me enough information, but it also prints an error message on the console if the file doesn't exist (which I don't want). stat(filename, &st) => OK if the file exists => Error if not (but also prints an error message) open(hostconfig, READ, 0) => a file descriptor if the file exists => Error if not (but also prints an error message) I was looking for something like the UNIX "access" function, but couldn't find it. I'd be willing ro "redirect standard error to /dev/null" if that would get rid of the error message and if I knew how to do that in VxWorks. Thanks for any help. - -- Martin Boyer mboyer@ireq-robot.hydro.qc.ca Division Robotique mboyer@ireq-robot.uucp Institut de recherche d'Hydro-Quebec +1 514 652-8412 Varennes, QC, Canada J3X 1S1 --------------------------- Newsgroups: comp.os.vxworks Subject: Re: VxWorks Development Environment Date: 17 Jan 1995 15:02:17 GMT From: btadlock@metronet.com Organization: Texas Metronet Message-ID: <3fgm5p$2v2@feenix.metronet.com> References: <9501161439.AA01068@locutus> In article <9501161439.AA01068@locutus> mea@mclean.sparta.com (Mike Anderson) wrote: > Greetings! > >> From: henryg@abc.se (Henry Gessau) >> Hi all, >> >> We are in the process of changing our embedded hardware from >> 80186-based to 68360. At the same time we will go from a home-made >> monitor to VxWorks. We are a bit worried about the iron required for >> a decent development environment. >> >> It seems that the best (best supported, most utilities) is some sort >> of Unix workstation (Sun?). The problem is that we (about 50 of us) >> would then have to aquire Unix workstations since we are currently >> running PC's. >> >> What would we be missing out on if we stuck to our PC's? Surely there >> is some support for MS-Windows? How about Windows NT? > > Well Henry, there are several alternatives. First, if your PCs are > networked together, you can always purchase a single Sun (or one of the > ISA-bus based Opus SPARCards that plug into your PC) and then use one > of the many fine X-Windows emulators that run under MS-DOG. I've used > Hummingbird's Exceed and VisonWare's Xvision products with great success. > That seems to fit with WRS's licensing policy of 10 users on a single > development host. You may want to buy a couple of hosts (and don't > forget the 2nd/3rd development (fortunately at half price) license :-), > to support all 50 of your developers. I can't immediately recall if 10 > users means 10 *simultaneous* users or not. Anyone from WRS care to > clarify? > > Alternatively, you could use VxWorks for Windows and run the Green > Hills Multi compiler/debugger on the PC. I'm not sure if CPU32+ > architectures are supported yet on that product, but I'm sure that they > soon will be if they aren't already. However, for the number of people > you're talking about, this approach would probably be more expensive > than the Sun approach owing to the per seat cost of the Multi > development environment. The advantage here is using your current PCs > and the possible use of a floating license manager (Green Hills has > this for the Sun, I'm not sure about the PC) such that you only buy, > say 10 licenses, and only 10 people at a crack can be > compiling/debugging simultaneously. I don't know about the status of > support for Windows NT, but since Microsoft keeps slipping shipment > (and I've heard many complaints about the current NT development > environment), I don't know that I'd count on NT in 1995. > > Since programmers tend to spend far more time in the editor than actually > compiling or using the source code debuggers (and you can always use printf > statements ;-) this may work for you. Also, don't forget those target > licenses and the ability to use the symbolic debugger provided with > VxWorks. I used that one by itself for the first 4 years I worked with > VxWorks and it's really quite powerful. Have you ever tried the vxWorks for ms-windows ? I here it leaves alot to be desired. Also, the greenhills development system is prob. the worst Ive ever seen and/or used. I tried for several months to get their system to work and never got anywhere. Their support was awful at best. The Wind Rivers platform may not be the best BUT at least it WORKS ! Do yourself a favor and purchase 2 or 3 SUN hosts and get Xvisions Xserver for the PC, You WILL BE GLAD YOU DID ! --------------------------- Newsgroups: comp.os.vxworks,jpl.comp.vxworks Subject: Problems connecting to target via serial line Date: Tue, 17 Jan 1995 17:55:34 GMT From: jhops@gumby.NoSubdomain.NoDomain (Jonathan M. Hops) Organization: Jet Propulsion Laboratory Message-ID: <1995Jan17.175534.6416@llyene.jpl.nasa.gov> Sender: jhops@gumby (Jonathan M. Hops) This question probably belongs on a Solaris bulletin board, post here first: I'm trying to setup my system so that I can get my hkv4f target console to come up in a shell on my Solaris 2.3 Sparc LX. I was able to get it to work once, but since then I keep getting a CAN'T ACCESS DEVICE when I issue a "cu". This occurs even after I cycle power and reboot both systems. Maybe I'm doing something wrong with the /etc/uucp/Devices file or not setting up my pmadm command correctly. Any ideas or references for me? _/ _/_/_/_/_/ _/ Jonathan M. Hops (818)354-1970 ============_/=_/======_/=_/========> Jet Propulsion Laboratory _/ _/_/_/_/_/ _/ M/S 238-528 _/ _/ _/ _/ 4800 Oak Grove Drive _/ _/ _/ _/ Pasadena, California _/_/_/_/_/ _/ _/_/_/_/_/_/ 91109 USA - -------------------------------------------------------------- "Integrity without knowledge is weak and useless, and knowledge without integrity is dangerous and dreadful." - -- SAMUEL JOHNSON 1709P1784 --------------------------- Newsgroups: comp.os.vxworks Subject: FAQs, FTP & Web sites, etc. Date: 17 Jan 1995 21:24:26 GMT From: joemct@hurricane.cds.tek.com ( CDS) Organization: Tektronix Colorado Data Systems, Englewood, Co. Message-ID: <3fhcia$lk0@tekgen.bv.tek.com> Net surfers, Please post pointers to vxWorks FAQs, web sites, FTP sites, and any other interesting repositories of "real-time" info that you know of. I would like to expand my knowledge and need a place to start. Thanks in advance. *************************************************** * McJoe can be found lurking in the following * * shadowy places: * * * * josephm@banyan.bv.tek.com * * TheMcJoe@aol.com * * * * "He knows that changes aren't permanent... * * But change is!!!" * * - Rush * *************************************************** --------------------------- Newsgroups: comp.os.vxworks Subject: Opening Local DOS FS files Date: Tue, 17 Jan 1995 23:32:10 GMT From: geger@phantom.den.mmc.com (George Eger) Organization: Martin Marietta Astronautics Group Keywords: DOS fopen Message-ID: <1995Jan17.233210.18052@den.mmc.com> Sender: news@den.mmc.com (News Admin) OK, I give. What's the trick to opening a file on a local DOS FS SCSI disk? I can ls() or ll() and see it. I can copy() it to stdout. But my program, which successfully opens files on my remote Sun and reads them, can't open the same file when copied to a local SCSI disk. In both cases, I'm using inFD = fopen(filenameStr, "r"); The across-the-net works great. The local gives back inFD = NULL, but errno is also 0. Any ideas will be welcome. I'm trying to read local script files, and it is kind of a drag when I can read them from across the net, but not off the local disk. TIA, GWE ||========================================================================== ||George Eger / geger@den.mmc.com || Voice - (303) - 971 - 6974 || ||Integ. Fault Tolerant Avionics || Fax - (303) - 977 - 1145 || ||Space Launch Systems || MS T320 || ||Martin Marietta Astronautics || P.O. Box 179, Denver CO 80201 || ||========================================================================== ||We are at a cusp - between the past when humans were more reliable than || ||computers and the future when computers are more reliable than humans. || ||========================================================================== ||All opinions (however truthful or misinformed) are my own. || ||========================================================================== --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Non-Blocking Socket "send/write" Date: Tue, 17 Jan 1995 22:48:13 GMT From: hjb@netcom.com (Hwa-Jin Bae) Organization: Peaceful Star, Oakland, CA Message-ID: References: <199501131418.JAA26364@a9926.ASTRO.GE.COM> In article <199501131418.JAA26364@a9926.ASTRO.GE.COM>, Matthew Krokosz wrote: > >I have a vxWorks application that connects to a remote application >via TCP/IP sockets. Once connected to the remote application, the >vxWorks application sends data to the remote application. I need >to prevent the "write" call to the socket from blocking. I have >tried using the ioctl with the FIONBIO option but that did not >work. FIONBIO does work. are you sure you are turning FIONBIO *on*? > >The condition I am seeing this write block is when the remote >application does not service its socket. I have produced a test >case where the remote application goes into an infinite loop once >it starts to receive data via the socket. This causes the remote >application not to read any more data from its TCP receive buffer. >Once the remote application goes into the infinite loop, the >vxWorks application which is writing to the socket will block >after writing X bytes to the sockets. I think that X bytes is >equal to the size of the TCP receive buffer on the remote >side plus the size of the TCP send buffer on the vxWorks side. >By observing the send buffer with "inetstatShow" it is apparent >that the write does block after the send buffer fills to its >max, which probably occurs after the remote receiver buffer fills >to its max. > if your socket level buffer on the send side is full, and you have FIONBIO turned on, your write() will return EWOULDBLOCK. >Ideally I would like to make this "write/send" system call >non-blocking, but using the ioctl did not work. I could >prevent the vxWorks application from getting into this blocking >situation if I could obtain the number of bytes in the send >buffer before writing to the socket. I could then prevent >the application from writing to the socket if it would max out >the send buffer and create a blocking situation. I can not >seem to find out how to obtain the number of bytes in the send >buffer. Any help would be appreciated! > > you can use getsockopt() with SOL_SOCKET, SO_SNDBUF, to retrieve the size of your socket's send buffer. - -- Hwa-Jin Bae (hjb@netcom.com) Peaceful Star, Oakland, CA --------------------------- Newsgroups: comp.os.vxworks,jpl.comp.vxworks Subject: Re: Problems connecting to target via serial line Date: Tue, 17 Jan 1995 23:00:26 GMT From: miked@wrs.com (Mike Deliman) Organization: Wind River Systems, Inc. Message-ID: References: <1995Jan17.175534.6416@llyene.jpl.nasa.gov> Sender: news@wrs.com (News Manager) In article <1995Jan17.175534.6416@llyene.jpl.nasa.gov> jhops@gumby.NoSubdomain.NoDomain (Jonathan M. Hops) writes: > >I'm trying to setup my system so that I can get my hkv4f target >console to come up in a shell on my Solaris 2.3 Sparc LX. > >I was able to get it to work once, but since then I keep >getting a CAN'T ACCESS DEVICE when I issue a "cu". Hello Mr. Hops, Check your /etc/remote file for the serial port your hkv4f is connected to, and be sure it's entry is correct. Then tip e.g. tip ttya and that should do it. Regards, -mike - -- - ------------------------------------------------------------------------------ Mike Deliman <> miked@wrs.com <> phone: 510-748-4100 <> facs: 510-814-2164 Snail Mail: Wind River Systems, 1010 Atlantic Ave, Alameda CA 94501 USA - ------------------------------------------------------------------------------ --------------------------- Newsgroups: comp.os.vxworks,jpl.comp.vxworks Subject: Re: Problems connecting to target via serial line Date: 17 Jan 1995 23:44:13 GMT From: mwette@mr-ed.jpl.nasa.gov (Matt Wette) Organization: Jet Propulsion Laboratory Message-ID: <3fhkod$2lj@grover.jpl.nasa.gov> References: <1995Jan17.175534.6416@llyene.jpl.nasa.gov> In article <1995Jan17.175534.6416@llyene.jpl.nasa.gov>, jhops@gumby.NoSubdomain.NoDomain (Jonathan M. Hops) writes: |> |> This question probably belongs on a Solaris bulletin board, |> post here first: |> |> I'm trying to setup my system so that I can get my hkv4f target |> console to come up in a shell on my Solaris 2.3 Sparc LX. |> |> I was able to get it to work once, but since then I keep |> getting a CAN'T ACCESS DEVICE when I issue a "cu". |> |> This occurs even after I cycle power and reboot both systems. |> |> Maybe I'm doing something wrong with the /etc/uucp/Devices file |> or not setting up my pmadm command correctly. |> |> Any ideas or references for me? I was able to talk to my HKV4F from Solaris 2 ("out of the box") using kermit ($ kermit -l /dev/term/b -b 9600). I don't know about cu, but I can toss a kermit binary your way if it will help. Matt - -- matthew.r.wette@jpl.nasa.gov --------------------------- Newsgroups: comp.os.vxworks Subject: execute() Date: Wed, 18 Jan 1995 00:08:07 GMT From: geger@phantom.den.mmc.com (George Eger) Organization: Martin Marietta Astronautics Group Keywords: shell yacc Message-ID: <1995Jan18.000807.21794@den.mmc.com> Sender: news@den.mmc.com (News Admin) I am running a user interface program from the VxWorks shell and would like to allow the user to 'escape' to the shell to execute shell commands. I found execute() in the call stack for the normal shell, and if I hand it a pointer to the character string of the command requested, it seems to do that command okay. The problem is, on exit from my program, I get error messages like: 0x77c468 (tShell): memPartFree: invalid block 0x719eb4 in partition 0x84176 due to yyparse() not being re-entrant. Is there a better way to do this? I really don't want to have to build my own shell command parser. Why can't execute() or yyparse() be made re-entrant? Is this fixed/improved/considered for VxWorks 5.2/5.X/6.0? TIA, GWE ||========================================================================== ||George Eger / geger@den.mmc.com || Voice - (303) - 971 - 6974 || ||Integ. Fault Tolerant Avionics || Fax - (303) - 977 - 1145 || ||Space Launch Systems || MS T320 || ||Martin Marietta Astronautics || P.O. Box 179, Denver CO 80201 || ||========================================================================== ||We are at a cusp - between the past when humans were more reliable than || ||computers and the future when computers are more reliable than humans. || ||========================================================================== ||All opinions (however truthful or misinformed) are my own. || ||========================================================================== --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Power PC compilers Date: Wed, 18 Jan 1995 18:04:40 GMT From: moore@ced.utah.edu (Tim Moore) Organization: University of Utah Center for Engineering Design Message-ID: References: <9501180316.AA16552@sunshine.LABS.TEK.COM> Sender: news@ced.utah.edu In article <9501180316.AA16552@sunshine.LABS.TEK.COM> johncal@sunshine.labs.tek.com (John Calvin) writes: GNU I've been advised, is 8 to 10 months away from a Power-PC compiler. Can anybody confirm this? gcc already generates code for the PowerPC. The sticking point might be gdb, gas, and the other binutils. However, they already generate / read rs6000 binaries, so it seems likely that PowerPC ports are pretty close. - -- Tim Moore Sarcos Research Corp. moore@ced.utah.edu "Wind in my hair - Shifting and drifting - Mechanical music - Adrenaline surge" - Rush --------------------------- Newsgroups: comp.os.vxworks Subject: svc_run Date: Wed, 18 Jan 1995 17:53:32 GMT From: shurstik@ceco.ceco.com (Susan L. Hurstik) Organization: Commonwealth Edison Company Message-ID: Reply-To: shurstik@ceco.ceco.com Sender: root@ceco.ceco.com (Operator) Hello, I need to support call-back functionality for an rpc app I wrote. In order to do this, I need to write my own svc_run routine. I have done this under solaris with no problem. Under vxWorks however, I'm not sure what parametes to pass to the select call. Has anyone done this before or has access to the source vxWorks uses for svc_run? Here is what the routine looks like on a Sun app: int asyncSvcRun() { fd_set readfds; struct timeval timeout; int dtbsize; int retVal; int i; extern int errno; struct rlimit rlp; getrlimit(RLIMIT_NOFILE, &rlp); dtbsize = (int)rlp.rlim_cur; readfds = svc_fdset; /* svc_fdset is a global var */ timeout.tv_sec = 0; timeout.tv_usec = 0; switch(i = select(dtbsize, &readfds, NULL, NULL, &timeout)){ case -1: case 0: retVal = 0; break; default: retVal = 1; svc_getreqset(&readfds); break; } return(retVal); } Thanks in advance - -Sue --------------------------- Newsgroups: comp.os.vxworks Subject: Re: VME ATM with VxWorks drivers... Date: Wed, 18 Jan 1995 04:50:19 GMT From: afraser@hookup.net (Andrew Fraser) Organization: HookUp Communication Corporation, Oakville, Ontario, CANADA Message-ID: References: In article heyes@codahp.cebaf.gov (Graham Heyes) writes: >We are putting together an ATM network with several VME based VxWorks >systems. I had a look three months ago and couldn't find a VME card with >VxWorks drivers for ATM. Does anyone have any better information. If I >don't find anything soon I'll be buying a FORE card and writing my own >driver... >-- Call Newbridge Microsystems (613) 592-0714. I know they have a VME ATM card, but I don't know what software or drivers are available. - -Andy ---------------------------------------------------------------- | Andrew Fraser ****** ***** | | Senior Hardware Engineer ***** * ***** | | **** *** ***** ***** ***** * | | Tel: (613) 596-9922 ext 251 *** **** **** ***** ***** ** | | Fax: (613) 596-0574 ** ***** *** ***** ***** *** | | Email: andrewf@dy4.isis.org * ***** ** ***** ***** **** | | afraser@hookup.net ****** * ***** ***** | | DY 4 Systems Inc. ***** | | 21 Fitzgerald Rd., Nepean ***** | | Ontario, Canada K2H 9J4 CUSTOMER FIRST, QUALITY ALWAYS | ---------------------------------------------------------------- --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Opening Local DOS FS files Date: Wed, 18 Jan 1995 05:14:16 GMT From: jfinley@netcom.com (John Finley) Organization: NETCOM On-line Communication Services (408 261-4700 guest) Message-ID: References: <1995Jan17.233210.18052@den.mmc.com> George Eger (geger@phantom.den.mmc.com) wrote: : OK, I give. What's the trick to opening a file on a local DOS FS SCSI [edit] : inFD = fopen(filenameStr, "r"); [edit] : ||========================================================================== : ||George Eger / geger@den.mmc.com || Voice - (303) - 971 - 6974 || You aren't perchance using unescaped backslashes, are you? The C string "\sd0\newfile.dat" has an embedded newline and something else; you'd probably want "\\sd0\\newfile.dat". (Or better, "/sd0/newfile.dat") - -John Finley jfinley@netcom.com --------------------------- Newsgroups: comp.os.vxworks Subject: Bus Errors Date: 18 Jan 1995 23:01:36 GMT From: jroberts@.jpl.nasa.gov (James A. Roberts) Organization: Jet Propulsion Laboratory, Pasadena, CA Message-ID: <3fk6kg$lap@grover.jpl.nasa.gov> Reply-To: jroberts@.jpl.nasa.gov I am having problems with random bus errors in bcopy and bfill when using msg queues and taskSpawn in 5.02-B on a hkv3f. I am using the patch for the fp exceptions from the FAQ (item 31) and it helped a lot, but I still have the problem when under a heavy load. any suggestions? - --- James A. Roberts --------------------------- Newsgroups: comp.os.vxworks Subject: XDI SMT port to VxWorks Date: 17 Jan 1995 23:55:15 GMT From: sri@buggy.tdd.sj.nec.com (Sri Chaganty) Organization: DCSD, NEC America, San Jose California Message-ID: <3fhld3$duv@elvis.syl.sj.nec.com> I am interested to know the expereinces of any one who might have ported XDI's SMT software to VxWorks environment. Please send your replies to my e-mail address at sri@dcsd.sj.nec.com Thanks in advance. Sri. --------------------------- Newsgroups: comp.os.vxworks Subject: IPOctal Date: 18 Jan 1995 19:06:52 GMT From: maskrey@pooh.sdd.trw.com (Ken Maskrey) Organization: Data Technologies Division, TRW Inc. Keywords: IP Serial Message-ID: <3fjosc$bik@yakko.sdd.TRW.COM> I remember someone a while back had source for one of the Octal Industry Packs(IPs). Could someone either send me the source, if available, or point me to where I can get it? Thanks, Ken Maskrey TRW --------------------------- Newsgroups: comp.os.vxworks Subject: Re: VxWorks Development Environment Date: 19 Jan 1995 07:50:45 GMT From: PJDJ44A@prodigy.com (Stan Hilinski) Organization: Prodigy Services Company 1-800-PRODIGY Message-ID: <3fl5kl$7che@usenetp1.news.prodigy.com> btadlock@metronet.com wrote: >Have you ever tried the vxWorks for ms-windows ? I here it leaves alot >to be desired. Also, the greenhills development system is prob. the worst >Ive ever seen and/or used. I tried for several months to get their system to >work and never got anywhere. Their support was awful at best. >The Wind Rivers platform may not be the best BUT at least it WORKS ! >Do yourself a favor and purchase 2 or 3 SUN hosts and get Xvisions >Xserver for the PC, You WILL BE GLAD YOU DID ! I won't dispute the gentleman's difficulties with Greenhill's Multi, but I think it's important to say they we've had a completely different experience with it. I personally think it's a terrific development tool for both vxworks and sun platforms. We have used it in at least one major project and plan to use it in others. As for support, it's been great. I e-mail problems and suggestions to customer support, and I usually get a response within a couple hours. And on the occasions I needed a patch, they've mailed them the same day. Naturally, I've resorted to the phone for real emergencies for instantaneous help. All in all, it's been a wonderful tool. Stan Hilinski, Goddard Spaceflight Center, Greenbelt, Md --------------------------- End of New-News digest ********************** From guthrie@power.amasd.anatcp.rockwell.com Thu Jan 19 08:17:07 1995 From: guthrie@power.amasd.anatcp.rockwell.com (Bob Guthrie) Date: Thu Jan 19 08:17:16 PST 1995 Subject: XDI SMT port to VxWorks (Sri Chaganty writes) > Newsgroups: comp.os.vxworks > Subject: XDI SMT port to VxWorks > Date: 17 Jan 1995 23:55:15 GMT > From: sri@buggy.tdd.sj.nec.com (Sri Chaganty) > Organization: DCSD, NEC America, San Jose California > Message-ID: <3fhld3$duv@elvis.syl.sj.nec.com> > > I am interested to know the expereinces of any one who might have > ported XDI's SMT software to VxWorks environment. > > Please send your replies to my e-mail address at > sri@dcsd.sj.nec.com > > > Thanks in advance. > > > Sri. > > --------------------------- I would also be interested in responses to this question. Robert Guthrie e-mail : guthrie@power.amasd.anatcp.rockwell.com From dunk@ftms.com Thu Jan 19 09:09:07 1995 From: Michael Dunk Date: Thu Jan 19 09:09:14 PST 1995 Subject: Re: svc_run > I need to support call-back functionality for an rpc app I wrote. > In order to do this, I need to write my own svc_run routine. > I have done this under solaris with no problem. Under vxWorks however, > I'm not sure what parameters to pass to the select call. Has anyone > done this before or has access to the source vxWorks uses for svc_run? The calling syntax to the select call is the same under vxWorks. The only tricky part to writing your own svc_run is accessing svc_fdset. Under vxWorks 5.0.3, svc_fdset is unique to each task. We accessed it using the following routine (this may have changed in 5.1): #include extern WIND_TCB *taskIdCurrent; void VCSGetfdset( fdPtr ) fd_set *fdPtr; { struct svc *ms = &taskRpcStatics->svc; *fdPtr = ms->svc_fdset; } We do not use this function anymore, as we found it more efficient to break up our application into multiple tasks, one of which uses the standard svc_run call. ---- ----------------------------------------- Michael F. Dunk Extrel FTMS Inc. 6416 Schroeder Road Madison, Wisconsin, 53711-2424 (608)-273-8262 dunk@ftms.com ----------------------------------------- From mea@mclean.sparta.com Thu Jan 19 15:30:21 1995 From: mea@mclean.sparta.com (Mike Anderson) Date: Thu Jan 19 15:30:28 PST 1995 Subject: Re: FAQs, FTP & Web Sites, etc. Greetings! Well, some progress is finally being made! We have a new address for the VxWorks FAQ. That address is: ftp.mclean.sparta.com (157.185.32.10). (The old address still works as well). Anonymous FTP is up and seems to be working. I've been trying to access the names I was given as FAQ masters on the net to post a new copy of the FAQ, but to no avail. Does anyone have the addresses of the folks I need to talk to about becoming the *official* keeper of the VxWorks FAQ? Any pointers would be appreciated. I'm also working on updating the FAQ to bring in the running threads about MV167 flash programming, etc. I am also soliciting any info from the group relating to available device drivers, consulting, etc. that would be appropriate for the FAQ and its related files. I hope to have a revised FAQ and associated files on-line by Feb 1. Please send any info you have and I'll do my best to include it in this new revision. We also have a *rudementary* web server up at: http://www.mclean.sparta.com I'm currently working on getting entries added for the User's Group archives mirror here as well as the FAQ and related files. I would like to have entries for the User's Group Meetings as well. Can the west coast stuckee please contact me so I can get that info into the hopper? For those groups that are adding their own web servers, if you would like a pointer from SPARTA's home page to yours for things like consulting or driver work, please include that info in your submissions to me and we'll add a link. Again, I apologize for the delays in getting all of this up to date and functioning. I always takes longer and costs more than you expect ;-). Regards, =============================================================================== __ Real-Time System Development, Integration, Training and Services //\\ // \\ Mike Anderson // /\ \\ Director, Real-Time Systems // / \ \\ SPARTA, Inc. Voice : (703) 448-0210 ext. 235 // \ \\ 7926 Jones Branch Drive FAX : (703) 734-3323 \\ \ // Suite 900 EMAIL : mea@mclean.sparta.com \\ \ / // McLean, VA 22102 \\ \/ // "Software development is like making \\ // a baby... You can't make a baby in one \\// month by impregnating nine women. -- "Pride in Performance" Some things just take time." =============================================================================== From stan@lassen.rti.com Thu Jan 19 18:13:10 1995 From: stan@lassen.rti.com (Stan Schneider) Date: Thu Jan 19 18:13:16 PST 1995 Subject: Nitro 60 Does anyone have any experience with the Nitro 60 board from Heurikon? -- Stan =============================================================================== = = = = Stan Schneider = email: stan@rti.com = = Real-Time Innovations, Inc. = Phone: (408) 720-8312 = = 954 Aster, Sunnyvale, CA 94086 = Fax: (408) 720-8419 = = = = =============================================================================== From daemon@vxw.ee.lbl.gov Fri Jan 20 04:00:31 1995 From: daemon@vxw.ee.lbl.gov Date: Fri Jan 20 04:00:41 PST 1995 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Fri Jan 20 04:00:20 PST 1995 Subject: Re: Opening Local DOS FS files Subject: help with ls() Subject: EtherXpress Pro Driver VxWorks Subject: Re: svc_run Subject: Restarting a task at a specified location? Subject: Re: help with ls() Subject: Re: GNU compiler on PC for 68k targets Subject: Re: Programming FLASH on the MV162 Subject: Re: Power PC compilers Subject: Re: VME ATM with VxWorks drivers... Subject: Info wanted on off-the-shelf MIL-SPEC embedded controller boards Subject: Re: Writing GUIs with an embedded OS ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: Opening Local DOS FS files Date: Thu, 19 Jan 1995 15:23:25 GMT From: geger@phantom.den.mmc.com (George Eger) Organization: Martin Marietta Astronautics Group Keywords: DOS fopen Message-ID: <1995Jan19.152325.20200@den.mmc.com> References: <1995Jan17.233210.18052@den.mmc.com> Sender: news@den.mmc.com (News Admin) In article <1995Jan17.233210.18052@den.mmc.com>, geger@phantom.den.mmc.com (George Eger) writes: |> OK, I give. What's the trick to opening a file on a local DOS FS SCSI |> disk? I can ls() or ll() and see it. I can copy() it to stdout. But |> my program, which successfully opens files on my remote Sun and reads them, |> can't open the same file when copied to a local SCSI disk. In both |> cases, I'm using |> |> inFD = fopen(filenameStr, "r"); |> |> The across-the-net works great. The local gives back inFD = NULL, but |> errno is also 0. |> |> Any ideas will be welcome. I'm trying to read local script files, and |> it is kind of a drag when I can read them from across the net, but |> not off the local disk. |> Actually, the way I get a filename is to read in a line of user ascii from stdin using fgets(), look for a '<', and call fopen with the rest of the string. Here's a condensed fragment of what I'm doing: extern int errno; FILE *inFD, *LastInFD; int get_token(char *inputLine, FILE *inFD, FILE *newFD) { if (fgets(&(inputLine[0]), MAX_CHAR_PER_LINE, inFD) != NULL) { if (input[0] == '<') { newFD = fopen(&(inputLine[1]), "r"); if (newFD == NULL) { logMsg("Failed to open script file %s (errno %d)\n", &(inputLine[1]), errno); return(FAILED_OPENING_SCRIPT); } else return(READING_NEW_SCRIPT); } else return(GOT_TOKEN); } else return(END_OF_INPUT); } I've tried using a file name including the device name of the local disk ( cd "/sd0/" ) just the filename ( I need help with the 'openDir' command and the 'stat', 'fstat' commands. I can open() a file and read from, but I can't get the fstat or stat of the file. Infact I can't even do a ls() or ll() command from the shell. I'm doing this accross the net. Any ideas? Thanks Steve --------------------------- Newsgroups: comp.os.vxworks Subject: EtherXpress Pro Driver VxWorks Date: 19 Jan 1995 18:56:48 GMT From: RRCM76A@prodigy.com (W Chenevert) Organization: Prodigy Services Company 1-800-PRODIGY Message-ID: <3fmclg$24j4@usenetp1.news.prodigy.com> We are looking for an EtherXpress Pro driver for VxWorks. Wind R doesn't seem intrested in developing one. They have driver for EtherXpress but not EtherXpree Pro and there is quite a difference in the hardware. If anyone knows where we can get this driver I would appreciate it. W.J. Chenevert (wjchenevert@halnet.com) - - W CHENEVERT RRCM76A@prodigy.com --------------------------- Newsgroups: comp.os.vxworks Subject: Re: svc_run Date: Thu, 19 Jan 1995 16:22:45 GMT From: mtvgae01@ntmtv.com (Geoffrey A Espin) Organization: Northern Telecom Inc, Mountain View, CA Message-ID: References: Sender: news@ntmtv.com In article , shurstik@ceco.ceco.com (Susan L. Hurstik) writes: |> I need to support call-back functionality for an rpc app I wrote. |>In order to do this, I need to write my own svc_run routine. |>I have done this under solaris with no problem. Under vxWorks however, |>I'm not sure what parametes to pass to the select call. |>Has anyone done this before or has access to the source vxWorks |>uses for svc_run? Here is what the routine looks like on a Sun app: Looks good to me. If you look in unsupported/demo/sprites there's an example. The svc_run() routine is called svc_doit(). |> struct rlimit rlp; |> getrlimit(RLIMIT_NOFILE, &rlp); |> dtbsize = (int)rlp.rlim_cur; AKA getdtablesize() which VxWorks never had. The undocumented variable contains this value which is defined in config/all/configAll.h by the macro NUM_FILES(50). Actually, the set width that you care about is the highest 'fd' bit set -- you could use ffsMsb() (finds most significant bit set) to specify the value (this might be more efficient for select()). Geoff - -- __/ _/ _/_/_/ Geoffrey Espin mtvgae01@ntmtv.com _/_|_/ _/ (415)940-2992 Fax (415)966-1067 _/ __/orthern _/elecom 685A E. Middlefield, PO Box 7277, Mtn View, CA 94039 --------------------------- Newsgroups: comp.os.vxworks Subject: Restarting a task at a specified location? Date: Thu, 19 Jan 1995 15:41:09 -0700 From: cobarruvias@asd1.jsc.nasa.gov (John R. Cobarruvias) Organization: NASA/JSC Message-ID: Followup-To: comp.os.vxworks Bear with me folks, I am trying to understand someone elses design and code from it.........(and worst of all, the designer is no longer in the area......) I have a task (TASK A) that does the following: 1) spawned at initialization time 2) performs task specific one time only initialization (such as ties a Signal Handler to a RESET_SIGNAL) 3) Then waits on a semaphore 4) Upon getting the semaphore, performs its task 5) Gives up the semaphore 6) Goes to 3 (wait mode again) But.....If the application receives a RESET interrupt, the ISR sends a signal to TASK A, if it is active (not waiting for the semaphore). The signal handler must then do the following: 1) reset various parameters 2) then instruct TASK A to give up the semaphore, and then restart at 3 above, which is to wait for the semaphore again. So the question is: How can the signal handler perform its processing, then instruct TASK A to restart at the point of waiting for the semaphore regardless of where in the code the task was "signaled" to stop? Hey folks, any help would be appreciated! John R. Cobarruvias --------------------------- Newsgroups: comp.os.vxworks Subject: Re: help with ls() Date: Thu, 19 Jan 1995 21:46:18 GMT From: hjb@netcom.com (Hwa-Jin Bae) Organization: Peaceful Star, Oakland, CA Message-ID: References: <3fm7si$h7t@agate.berkeley.edu> In article <3fm7si$h7t@agate.berkeley.edu>, Steven Schofield wrote: > >I need help with the 'openDir' command and the >'stat', 'fstat' commands. > >I can open() a file and read from, but I can't get >the fstat or stat of the file. Infact I can't even >do a ls() or ll() command from the shell. I'm doing >this accross the net. > vxworks has no consistency in this area because it does not have a consistent file system model; it does not have a real file system -- their filesystems are bunch of drivers that act like file systems. some of these drivers do a better job, some don't. not all of them do the same set of operations consistently. it all depends on which file system device driver you're using. - -- Hwa-Jin Bae (hjb@netcom.com) Peaceful Star, Oakland, CA --------------------------- Newsgroups: comp.os.vxworks Subject: Re: GNU compiler on PC for 68k targets Date: 19 Jan 1995 22:08:15 GMT From: billr@saab.CNA.TEK.COM (Bill Randle) Organization: Tektronix, Inc., Redmond, OR Message-ID: <3fmnsf$r58@marlyn.cna.tek.com> References: <9501180739.AA22729@mars.egt.fr> Reply-To: billr@saab.CNA.TEK.COM In article <9501180739.AA22729@mars.egt.fr>, Bernard Bolz writes: |> We are using gcc on HP 9000 for 68k targets, and we would like to generate |> the same code from PC's. |> So we are looking for a port of gcc for PC's, which can produce code for 68k. |> Does anyone know of such a compiler? You don't mention what OS on the PCs, but something to think about would be to run Linux (or similar OS) on the PCs [you can have split Linux/DOS partitions on the hard drive]. I've successfully ported gcc 2.6.2 to run under Linux/486 as a 68k cross compiler. Works just fine. [I did have to tweak the config file a little.] I also compiled the binutils to get as68k, ld68k, nm68k, et.al. - -- -Bill Randle Tektronix, Inc. billr@saab.CNA.TEK.COM --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Programming FLASH on the MV162 Date: 19 Jan 1995 18:57:05 GMT From: switt@triceratops.hac.com (Steve Witt) Organization: Hughes Aircraft Company Message-ID: <3fmcm1$1r9@hacgate2.hac.com> References: <199501101734.KAA03798@xmission.xmission.com> The manual page that comes with the MVME-162 BSP describes how to program the on-board FLASH over the Ethernet interface pretty well. 'man mv162'. The Motorola BUG ROM is used to do this and there is a jumper on the board to tell the board to boot out of ROM or FLASH. The trickiest part of the whole operation for me was to figure out how to configure TFTP on the Sun to allow the file transfer. This isn't explained in the VxWorks documentation (and I'm not saying it should be) but you need to modify the entry in /etc/inetd.conf for TFTP. There is a directory on this line that is the directory from which your bootable image is TFTP'd. - -------------------------------------------------------------------------- Steve Witt switt@triceratops.hac.com Hughes Aircraft Company --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Power PC compilers Date: Thu, 19 Jan 1995 23:55:09 GMT From: chris@wrs.com (Chris Ford) Organization: Wind River Systems, Inc. Message-ID: References: <9501180316.AA16552@sunshine.LABS.TEK.COM> Sender: news@wrs.com (News Manager) Hello, johncal@sunshine.labs.tek.com (John Calvin) asks: > What compiler are people using for Power PC's? All the vendors you mentioned are fine companies. We work with several of them at Wind River, as each has strengths that appeal to certain customers. We advise you to consider standards compliance now so you aren't locked into a proprietary solution later. The leaders in the embedded community are rallying around the Embedded Application Binary Interface (EABI), a derivative of SVR4 for PowerPC. For PowerPC EABI information, please contact: Stephen Sobek | Kevin Burke Microcontroller Technologies Group| Technology Products Motorola | International Business Machines Corporation 6501 William Cannon Drive West | 3039 Cornwallis Road Mail Stop OE45 | Mail Stop H83/061 Austin, TX 78735 | Research Triangle Park, NC 27709 steve@avar.sps.mot.com | kevin_burke@vnet.ibm.com > John Calvin, Design Technology Research Lab, Tektronix Inc. > 14150 S.W. Karl Braun Drive P.O. Box 500, M.S. 50-662 > Beaverton OR 97077 U.S.A. > John.C.Calvin@tek.com, 503-627-4840, Fax 503-627-7875 - -- Chris Ford Wind River Systems Engineering - -- Chris Ford Wind River Systems Engineering --------------------------- Newsgroups: comp.os.vxworks Subject: Re: VME ATM with VxWorks drivers... Date: 19 Jan 1995 22:15:13 -0700 From: bgeer@xmission.com (bgeer) Organization: XMission Public Access Internet (801 539 0900) Message-ID: <3fngt1$b2v@xmission.xmission.com> References: afraser@hookup.net (Andrew Fraser) writes: >In article heyes@codahp.cebaf.gov (Graham Heyes) writes: >>We are putting together an ATM network with several VME based VxWorks >>systems. I had a look three months ago and couldn't find a VME card with >>VxWorks drivers for ATM. Does anyone have any better information. If I >>don't find anything soon I'll be buying a FORE card and writing my own >>driver... >>-- >Call Newbridge Microsystems (613) 592-0714. I know they have a VME ATM card, >but I don't know what software or drivers are available. Interphase also has ATM cards. Their first ATM was one of their 5211 FDDI cards depopulated with an ATM daughter card. I'm told they are nearing release of a new layout VME w/ ATM with a media (fiber, copper, ?) daughter card. Not sure about drivers yet, tho their 5211 has a readily available driver which might be re-cooked for ATM. Bob - -- <> Bob `Bear' Geer <> * / <> <> bgeer@xmission.com <> _o * o * ,_____/o <> <> Salt Lake City, <> -\<, * <\ ~~,__/ > ~~ <> <> Ootah, USA <> O/ O __ /__, / <> --------------------------- Newsgroups: comp.os.vxworks,comp.sys.m68k,alt.industrial.computing,sci.electronics,rec.radio.amateur.space Subject: Info wanted on off-the-shelf MIL-SPEC embedded controller boards Date: 18 Jan 1995 03:03:02 GMT From: grusin@benji.Colorado.EDU (GRUSIN MICHAEL) Organization: University of Colorado, Boulder Message-ID: <3fi0d6$hbr@CUBoulder.Colorado.EDU> We are looking for an off-the-shelf, MIL SPEC (883B), 68020/30/40, embedded system board for flight on a research satellite. If anyone has pointers to companies which supply such boards, I'd appreciate hearing about them! Thanks, -Mike Grusin (grusin@rintintin.colorado.edu) --------------------------- Newsgroups: comp.os.vxworks,comp.os.os9 Subject: Re: Writing GUIs with an embedded OS Date: Fri, 20 Jan 1995 11:25:57 GMT From: werner@sunscipw.cern.ch (Per Werner) Organization: CERN Keywords: GUI Message-ID: References: <3f6mso$q9l@bcarh8ab.bnr.ca> Reply-To: werner@sunscipw.cern.ch (Per Werner) Sender: news@news.cern.ch (USENET News System) I have used Motif 1.1 from Microware for GUI application which run on OS9, SunOS4.1.3 and ULTRIX 4.3. Microware Motif/X11R4 is not bug free but it works OK. Maybe Microware or someone else has a later port. I used G++ 1.40 and wrote a small class library with objects like Messsage, SelectionBox, FileSelectionBox, Question, ScrolledText etc. Hope this can help Per E. Werner **************************************************************************** * * * Mail address : Per.Werner@cern.ch * * Telephone : ++41 22 767 4932 * * Organization : CERN - European Laboratory for Particle Physics * * RD24 - ECP Division - Data Acquisition Systems Group * * RD24 WWW URL : http://www1.cern.ch/RD24 * * SCI WWW URL : http://www1.cern.ch/RD24 (is there another one ?) * * High Speed Interconnect (SCI, HIPPI, FiberChannel and ATM) - * * WWW URL : http://www1.cern.ch/HSI * * * **************************************************************************** --------------------------- End of New-News digest ********************** From sergi@hpbpq11.bpo.hp.com Fri Jan 20 05:24:27 1995 From: Sergi Casas(QA Software testing) Date: Fri Jan 20 05:24:37 PST 1995 Subject: Re: Restarting a task > > So the question is: How can the signal handler perform its processing, then > instruct TASK A to restart at the point of waiting for the semaphore > regardless of where in the code the task was "signaled" to stop? > > Hey folks, any help would be appreciated! > > > John R. Cobarruvias Look at "setjmp()", "longjmp()" mechanism in the vxWorks manuals. The explanation is not very good there, but you can find these calls in any standard C book. If you just don't return from the signal handler but rather jump into step 3 with longjmp() ... it should work; I've never tried this, though. Hope this helps. -- |============================================================================| | Sergi Casas ##### / ##### Firmware Engineer | | Hewlett-Packard Company ### /_ _ ### | | Barcelona Division (BCD) ## / / / / ## e-mail: sergi@hp-bpo.bpo.hp.com | | Avda. Graells, 501 ## / / /_/ ## Phone : +34 3 582 1475 | | 08190 S. Cugat (Barcelona) ### / ### Fax : +34 3 582 2515 | | SPAIN ##### / ##### HP-Telnet: 712 1475 | |============================================================================| From zot!Schilling.com!stevec@netcom.com Fri Jan 20 08:11:33 1995 From: stevec@schilling.com (Steve Cohan) Date: Fri Jan 20 08:11:40 PST 1995 Subject: Re: Info wanted on off-the-shelf MIL-SPEC embedded controller boards > We are looking for an off-the-shelf, MIL SPEC (883B), 68020/30/40, > embedded system board for flight on a research satellite. Radstone Technology has a line of MIL-STD-883 products, including an '020/'881 processor board. I don't know if there are any other processor boards, but there are other peripheral boards (memory, scsi, 1553, pwr supplies,...). Contact: Radstone Technology Corp. 20 Craig Rd. Montvale, NJ 07645 800-368-2738 ...................................................................... . . . Steve Cohan Telerobotics for nuclear, . . Control Systems Engineer subsea, and other hazardous . . Schilling Development, Inc. environments . . . . . . 1632 Da Vinci Court tel : 916 753 6718 . . Davis, CA 95616 fax : 916 753 8092 . . email : stevec@schilling.com . . . ...................................................................... From brwenholz@smtpgate.read.tasc.com Fri Jan 20 08:19:46 1995 From: Bruce Wenholz Date: Fri Jan 20 08:19:53 PST 1995 Subject: subscribe From vilardebo@nrlvax.nrl.navy.mil Fri Jan 20 09:02:04 1995 From: vilardebo@nrlvax.nrl.navy.mil Date: Fri Jan 20 09:02:11 PST 1995 Subject: dual port (VME) memory on FORCE CPU-30 and CPU-40 I have been trying to figure out how to set up dual port memory on the FORCE CPU-30 and CPU-40. I want to have 512K bytes of memory set for VME access (dual ported). I think this probably will require programming the FGA and moving sysMemTop(). But I don't have much detail. If anyone has an example in either C or Ada it would be greately appreciated. I am using the GNU C/C++, VxWorks 5.1, Compilier and the TARTAN v4.3.1 Ada compilier for VxWorks. Thanks Ken Vilardebo Mnemonics, Inc. From mea@mclean.sparta.com Fri Jan 20 09:14:11 1995 From: mea@mclean.sparta.com (Mike Anderson) Date: Fri Jan 20 09:14:35 PST 1995 Subject: Re: Restarting a task at a specified location? Greetings! John R. Cobarruvias writes: > > I have a task (TASK A) that does the following: > > 1) spawned at initialization time > 2) performs task specific one time only initialization (such as ties a > Signal Handler to a RESET_SIGNAL) > 3) Then waits on a semaphore > 4) Upon getting the semaphore, performs its task > 5) Gives up the semaphore > 6) Goes to 3 (wait mode again) > > But.....If the application receives a RESET interrupt, the ISR sends a > signal to TASK A, if it is active (not waiting for the semaphore). The > signal handler must then do the following: > > 1) reset various parameters > 2) then instruct TASK A to give up the semaphore, and then restart at 3 > above, which is to wait for the semaphore again. > > > So the question is: How can the signal handler perform its processing, then > instruct TASK A to restart at the point of waiting for the semaphore > regardless of where in the code the task was "signaled" to stop? > > Hey folks, any help would be appreciated! > Well, signal handlers work differently depending on which version of VxWorks you're using. First, a signal handler, when fired off, causes an asynchronous change in the thread of control of the process the signal handler is associated with. In plain English, this means that a signal handler works a lot like an ISR in the sense that a task can be running along, get a signal, jump to the signal handler and run the handler and then return back to the original point in the task and continue execution. Typically, signal handlers were used for error recovery but they can be used for both interprocess communication and for "event" notification. VxWorks only supports a few of the standard BSD signals, but you can write your own signal handling support for things such as the asynchronous I/O signal SIGIO as many of us in the VxWorks community already have. Now, as to the differences in behavior between versions. In 5.0.X, if your task was blocked at a semTake and you got a signal, the signal handler would only run when the task it was associated got resheduled. That is, after the semTake completed/timed out. Under 5.1.X, the signal handler will run even if the task was blocked at the semTake. Therefore, if you're running under 5.1.X, then your problem is simple. Just set your signal handler up to reset your variables, do the semGive if necessary (or do a taskRestart and let your original code do it for you) and use setjmp/longjmp to "goto" a particular location in your original task (this is not necessary if you used taskRestart). If you're using 5.0.X, the problem is a little more complicated. Your ISR would have to send the signal and then do a semGive (make sure the semaphore is not a mutex variety 'cause those can't be given from interrupt level) so the signal handler would run. Then have the signal handler do the cleanups as outlined in the 5.1.X discussion. Hope this helps, =============================================================================== __ Real-Time System Development, Integration, Training and Services //\\ // \\ Mike Anderson // /\ \\ Director, Real-Time Systems // / \ \\ SPARTA, Inc. Voice : (703) 448-0210 ext. 235 // \ \\ 7926 Jones Branch Drive FAX : (703) 734-3323 \\ \ // Suite 900 EMAIL : mea@mclean.sparta.com \\ \ / // McLean, VA 22102 \\ \/ // "Software development is like making \\ // a baby... You can't make a baby in one \\// month by impregnating nine women. -- "Pride in Performance" Some things just take time." =============================================================================== From vilardebo@nrlvax.nrl.navy.mil Fri Jan 20 09:28:30 1995 From: vilardebo@nrlvax.nrl.navy.mil Date: Fri Jan 20 09:28:36 PST 1995 Subject: Ada on VxWorks Re: Original question from Matt Whiting whiting_ms@corning.com about Ada on VxWorks. I have been using TARTAN Works V4.3.1 Ada compiler, it seems to be a pretty well behaved product. It runs on an unmodified VxWorks kernel (unlike VADS works). TARTAN is only up to 5.1 VxWorks currently. You can get information from Glen Lawson (813)724-3555 Fax:(813)726-8144 (813)724-1081 TARTAN, INC. McCormic Dr. Suite 103 Bldg 2651 Clearwater FL 34619 So far I have gotten good support from both TARTAN and Wind River. Ken Vilardebo Mnemonics, Inc. From mea@mclean.sparta.com Fri Jan 20 09:31:16 1995 From: mea@mclean.sparta.com (Mike Anderson) Date: Fri Jan 20 09:31:26 PST 1995 Subject: Re: Heurikon Nitro 60 Greetings! > > Submitted-by stan@lassen.rti.com Thu Jan 19 18:13:10 1995 > Submitted-by: stan@lassen.rti.com (Stan Schneider) > > > Does anyone have any experience with the Nitro 60 board from Heurikon? > > -- Stan We were one of Heurikon's beta-test sites for the Nitro 60 (50 MHz 68060) and we are now using the Nitro 60 in a couple of our production systems. I have to say that this thing is F-A-S-T! It toasts our 33 MHz 68040s by a factor of 3-4. Our Nitro 60 also sports dual SCSI and dual Ethernets via Heurikon's Corebus mezzanine bus. Very nice :-). We use them in some custom 3-slot VME bus crates that we build into Sampsonite briefcases and heat also does not seem to be a problem with these boards. They also support VME64 via the VIC64 interface. The VxWorks port is solid except for one minor problem. That is that the tRdbTask for remote source debugging hasn't been ported yet for the '060. Since my code doesn't have bugs, I haven't missed the debugger yet ;-). I'm told that the 5.2 release will support the VxGDB debugger. In the mean time, you can still use the symbolic debugger and printf/logMsg calls. HTH, =============================================================================== __ Real-Time System Development, Integration, Training and Services //\\ // \\ Mike Anderson // /\ \\ Director, Real-Time Systems // / \ \\ SPARTA, Inc. Voice : (703) 448-0210 ext. 235 // \ \\ 7926 Jones Branch Drive FAX : (703) 734-3323 \\ \ // Suite 900 EMAIL : mea@mclean.sparta.com \\ \ / // McLean, VA 22102 \\ \/ // "Software development is like making \\ // a baby... You can't make a baby in one \\// month by impregnating nine women. -- "Pride in Performance" Some things just take time." =============================================================================== From Doug.Owens@quickmail.llnl.gov Fri Jan 20 09:51:32 1995 From: "Doug Owens" Date: Fri Jan 20 09:51:39 PST 1995 Subject: Unsubscribe Subject: Time:9:49 AM OFFICE MEMO Unsubscribe Date:1/20/95 I was subscribed to the exploder for owens@lodtm.llnl.gov. Please unsubscribe - the system is gone and the mail is backing up. Thanks Doug Owens From vilardebo@nrlvax.nrl.navy.mil Fri Jan 20 10:21:00 1995 From: vilardebo@nrlvax.nrl.navy.mil Date: Fri Jan 20 10:21:07 PST 1995 Subject: Re: Info wanted on off-the-shelf MIL-SPEC embedded controller boards General A-Tronics has a 68040 based board with a built in 1553, contact keith@gac.usa.com Ken Vilardebo Mnemonics, Inc. From stan@lassen.rti.com Fri Jan 20 14:31:55 1995 From: stan@lassen.rti.com (Stan Schneider) Date: Fri Jan 20 14:32:01 PST 1995 Subject: Rom emulators Does anyone out there have experience with ROM emulators (good or bad)? Anyone using XLNT system's products? -- Stan =============================================================================== = = = = Stan Schneider = email: stan@rti.com = = Real-Time Innovations, Inc. = Phone: (408) 720-8312 = = 954 Aster, Sunnyvale, CA 94086 = Fax: (408) 720-8419 = = = = =============================================================================== From satish@ltis.loral.com Fri Jan 20 16:43:15 1995 From: Satish Mali Date: Fri Jan 20 16:43:21 PST 1995 Subject: Re: Rom emulators Excellent. I highly recommend them. Even though the price may look a little steep at the beginning, the time it saves is worth it. Thanks -satish ******************************************************************************** * Satish Mali email : satish@ltis.loral.com * * I speak only for myself and not for my employer * ******************************************************************************** From daemon@vxw.ee.lbl.gov Sat Jan 21 04:00:27 1995 From: daemon@vxw.ee.lbl.gov Date: Sat Jan 21 04:00:35 PST 1995 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Sat Jan 21 04:00:17 PST 1995 Subject: Fast SCSI disk write/read Subject: Re: Programming FLASH on the MV162 Subject: re: Info wanted on off-the-shelf MIL-SPEC embedded controller boards Subject: What happened to Telesystems SLW ? Subject: VxSim Subject: Re: VxSim ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Fast SCSI disk write/read Date: Fri, 20 Jan 1995 10:12:41 GMT From: pbiereic@eso.org (Peter Biereichel) Organization: ESO - European Southern Observatory, Garching by Munich Message-ID: <1995Jan20.101241.25243@eso.org> Sender: news@eso.org Does anyone know about a SCSI library/driver for fastest possible data transfer to a SCSI disk? A file system is not needed. We just want to dump data in "burst mode". - -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-= Peter Biereichel pbiereic@te1.hq.eso.org ph: ++49 89-32006-266 ESO, European Southern Observatory fax:++49 89-3202362 Karl-Schwarzschild-Str.2, D-85748 Garching, GERMANY - -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-= --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Programming FLASH on the MV162 Date: Fri, 20 Jan 1995 14:10:54 +0000 From: anj@mail.ast.cam.ac.uk (Andrew Johnson) Organization: Royal Greenwich Observatory Message-ID: References: <199501101734.KAA03798@xmission.xmission.com> <3fmcm1$1r9@hacgate2.hac.com> In article <3fmcm1$1r9@hacgate2.hac.com> switt@triceratops.hac.com (Steve Witt) writes: >The trickiest part of the whole operation for me was to figure >out how to configure TFTP on the Sun to allow the file transfer. >This isn't explained in the VxWorks documentation (and I'm not >saying it should be) but you need to modify the entry in /etc/inetd.conf >for TFTP. There is a directory on this line that is the directory >from which your bootable image is TFTP'd. You should be careful about moving tftpd away from /tftpboot, and make sure that the directory you use or any of its subdirectories does not contain any confidential data. TFTP is not secure (it can't be by definition), so anyone with network access to your system will be able to read these files. - - Andrew ,_}\_. Andrew Johnson, Technology Division / \ Royal Greenwich Observatory ~~~~~~~~| ,--/;. )~~~~ Madingley Road, Cambridge, UK ., |/ (/ U Phone: +44 (0)1223 374823 .`., /_\ . . Email: anj@mail.ast.cam.ac.uk wwWWWww' WWW: http://cast0.ast.cam.ac.uk/~anj/ --------------------------- Newsgroups: comp.os.vxworks Subject: re: Info wanted on off-the-shelf MIL-SPEC embedded controller boards Date: 20 Jan 1995 16:37:08 GMT From: msjl@ix.netcom.com (Mark Lambert) Organization: Netcom Message-ID: <3foork$jen@ixnews2.ix.netcom.com> References: <9501201600.AA08954@Schilling.com> > Radstone Technology has a line of MIL-STD-883 products, including an > '020/'881 processor board. I don't know if there are any other > processor boards, but there are other peripheral boards (memory, > scsi, 1553, pwr supplies,...). Can I just clarify the processor question: `020, `030, `040 and i860 boards are currently available. PowerPC coming soon. - -- Mark Lambert Radstone Technology Corp., Field Applications Engineer 3333 Bowers Avenue, Suite 130, Santa Clara, CA 95054 ++++++++++++++++++++++++++++++++++++++++++++++++++++ Tel: (408) 727-4585 +E-mail: lambert@radstone.co.uk + Fax: (408) 727-4797 + (NB. Please use this in preference to the + + address in the header of this posting)+ ++++++++++++++++++++++++++++++++++++++++++++++++++++ --------------------------- Newsgroups: comp.os.vxworks,comp.dcom.lans.ethernet Subject: What happened to Telesystems SLW ? Date: 20 Jan 1995 19:04:19 GMT From: hmp@frc2.frc.ri.cmu.edu (Henning Pangels) Organization: Field Robotics Center, CMU Message-ID: Reply-To: hmp@frc2.frc.ri.cmu.edu (Henning Pangels) A few years back, we purchased a wireless ethernet bridge (ARLAN 620) from Telesystems SLW, located at the time in Toronto, Canada. We've been pretty happy with the equipment, and were thinking of getting another unit for a new project - alas, they seem to have disappeared from the face of the earth! Would anyone happen to know their fate? In looking for an alternative, I'd be interested in hearing what people are using these days in the area of wireless networking. Any recommendations? Thanks, - -Henning - -- Henning Pangels | hmp@cs.cmu.edu | Field Robotics Center Senior Research Programmer| (412) 268-8911 |Carnegie-Mellon University - ---------------------------------------------------------------------------- Fast, Cheap, Good: Choose Any Two --------------------------- Newsgroups: comp.os.vxworks Subject: VxSim Date: Fri, 20 Jan 1995 20:04:54 GMT From: heiple@ct.picker.com (HEIPLE, HOLLY) Organization: Picker Keywords: VxSim Message-ID: <1995Jan20.200454.3862@picker.com> Reply-To: heiple@ct.picker.com Sender: news@picker.com Does anybody have any experience using VxSim? We are considering buying it based on the literature but I am wondering how much a developer would really use it. Why not just use the target? Any comments or words of wisdom? Thanks, Holly Heiple --------------------------- Newsgroups: comp.os.vxworks Subject: Re: VxSim Date: Sat, 21 Jan 1995 02:48:26 GMT From: marc@wrs.com (Marc Shepard) Organization: Wind River Systems, Inc. Keywords: VxSim Message-ID: References: <1995Jan20.200454.3862@picker.com> Sender: news@wrs.com (News Manager) In article <1995Jan20.200454.3862@picker.com> heiple@ct.picker.com writes: >Does anybody have any experience using VxSim? We are considering >buying it based on the literature but I am wondering how much a developer >would really use it. Why not just use the target? Any comments or words >of wisdom? I use it all the time here at WRS to prototype new code. You don't have to scrounge for hardware in order to test your code. You can call "PRINTF" (which calls the underlying UNIX printf) anywhere in your code, even exception handlers, ISRs (which are implemented as UNIX signals), or context switch code. Also, it's much faster to reboot than real hardware. - -- ``` (o o) - -----------------oOO--(_)--OOo-------------------- marc@wrs.com (510)814-2142 --------------------------- End of New-News digest ********************** From daemon@vxw.ee.lbl.gov Sun Jan 22 04:00:50 1995 From: daemon@vxw.ee.lbl.gov Date: Sun Jan 22 04:00:57 PST 1995 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Sun Jan 22 04:00:29 PST 1995 Subject: Position available Subject: Re: Info wanted on off-the-shelf MIL-SPEC embedded controller boards ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Position available Date: Fri, 20 Jan 1995 12:38:28 GMT From: dicicco@peb.com (Darrell S. DiCicco) Organization: Princeton Electronic Billboard, Inc. Keywords: image processing, real-time, video insertion Message-ID: <9501200738.AA28526@Performa> Reply-To: dicicco@peb.com (Darrell S. DiCicco) Sender: news@nntpxfer.psi.com January 20, 1995 Intended for posting on the real time users bulletin board. PEB is a fast growing Princeton based company applying computer vision and image processing to real-time advertising. We have an immediate opening for a software engineer/ programmer to join our team in developing the next generation of our propriety live video insertion system (L-vis). We are looking for a tenacious, goal-oriented, pro-active individual who works well within a team and delivers results on time. Extensive knowledge of C/C++ required. Familiarity with image processing techniques and/or real time systems necessary. Knowledge of Solaris OS, vxWorks and UNIX a plus. Competitive salary and comprehensive benefits offered. Send resume to rjr@peb.com or write/fax to: Roy Rosser PEB Inc. 47 Hulfish Street, Suite 500 Princeton, NJ 08542 (609) 924 9399 (609) 924 0634 --------------------------- Newsgroups: comp.os.vxworks,comp.sys.m68k,alt.industrial.computing,sci.electronics,rec.radio.amateur.space Subject: Re: Info wanted on off-the-shelf MIL-SPEC embedded controller boards Date: Sat, 21 Jan 1995 15:52:49 UNDEFINED From: lacovara@ollie.jsc.nasa.gov (R. C. Lacovara) Organization: GeoControl Systems, Incorporated Message-ID: References: <3fi0d6$hbr@CUBoulder.Colorado.EDU> In article <3fi0d6$hbr@CUBoulder.Colorado.EDU> grusin@benji.Colorado.EDU (GRUSIN MICHAEL) writes: >From: grusin@benji.Colorado.EDU (GRUSIN MICHAEL) >Subject: Info wanted on off-the-shelf MIL-SPEC embedded controller boards >Date: 18 Jan 1995 03:03:02 GMT >We are looking for an off-the-shelf, MIL SPEC (883B), 68020/30/40, >embedded system board for flight on a research satellite. If anyone >has pointers to companies which supply such boards, I'd appreciate >hearing about them! >Thanks, -Mike Grusin (grusin@rintintin.colorado.edu) There are some possibilities. Try Radstone. Let me know if you find a good one. R. C. Lacovara, Ph. D. Engineer GeoControl Systems voice: 713 333 2561 1720 NASA Road 1 fax: 713 333 2966 Houston, Texas 77058 Texas is the largest and grandest State of the United States. It retains the right to fly its flag at the same height as that of the United States. Texas must not be abbreviated, like NJ or RI: NJ fits between Houston and San Antonio, with RI to spare. Al --------------------------- End of New-News digest ********************** From micf@lannet.com Sun Jan 22 05:24:32 1995 From: micf@lannet.com (Michael Feldman) Date: Sun Jan 22 05:24:38 PST 1995 Subject: vxWorks ports of ATM stuff Hello, I was wondering if anyone has information on the commercial ports of an ATM stuff (signalling, ILMI, etc.) to vxWorks. If anyone has any information on this board, could you respond by e-mail? Thank you. Michael Feldman, Lannet Data Communications, Israel micf@lannet.com From daemon@vxw.ee.lbl.gov Mon Jan 23 04:00:25 1995 From: daemon@vxw.ee.lbl.gov Date: Mon Jan 23 04:00:33 PST 1995 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Mon Jan 23 04:00:16 PST 1995 Subject: IP multicast in VxWorks ??? Subject: Re: Fast SCSI disk write/read ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: IP multicast in VxWorks ??? Date: Sun, 22 Jan 1995 14:21:01 GMT From: segrest@bobseg.enet.dec.com () Organization: Me and only Me Message-ID: <1995Jan22.142101.24919@datum.nyo.dec.com> Reply-To: segrest@bobseg.enet.dec.com () Sender: usenet@datum.nyo.dec.com (USENET News System) Greetings, Has any work been done to provide IP multicast support in VxWorks ??? I am working with someone that needs to embed an application currently implemented with Solaris into a rom loaded VxWorks environment. - -- Bob Segrest --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Fast SCSI disk write/read Date: Sun, 22 Jan 1995 21:58:30 GMT From: dat85715@indirect.com (D'Anne Thompson) Organization: Internet Direct, indirect.com Message-ID: References: <1995Jan20.101241.25243@eso.org> Sender: dat85715@bud.indirect.com >Does anyone know about a SCSI library/driver for fastest possible >data transfer to a SCSI disk? A file system is not needed. >We just want to dump data in "burst mode". > >-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-= >Peter Biereichel pbiereic@te1.hq.eso.org ph: ++49 89-32006-266 >ESO, European Southern Observatory fax:++49 89-3202362 >Karl-Schwarzschild-Str.2, D-85748 Garching, GERMANY >-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-= SCSI data rates are physically limited by the hardware at both ends, even more by the disk than the initiator cpu. If you are asking about software, then there really are only two choices I know of. 1. Stock VxWorks SCSI library. You can call the library routines scsiWrtSecs and scsiRdSecs directly to avoid any file system overhead. 2. Contact rtp.ltd.co.uk to get their upgraded SCSI library with full SCSI-2 command tagging/queing support. This REQUIRES a scsi-2 compliant disk that supports command tagging/queing. Command tagging/queing will help to reduce the software overhead in between data segments, since multiple requests can be queued up at the hardware level. If you are asking about hardware, then the slowest element controls the data rate. Generic Asynch SCSI transfers - Max rate 5MB/sec Synchronous SCSI-2 transfers - Max rate 10MB/sec Wide Synch SCSI-2 transfers - Max rate 20MB/sec All of the above are the hardware burst rates and do not include software overhead. As such the movement of data from user buffer to disk device (or vice versa) will be slower than the stated values. The only company I know of that can provide you with an SBC solution with a Wide Synchronous SCSI interface onboard is Synergy Microsystems. We have a VxWorks driver that works with the standard VxWorks SCSI Library and provides a continuous sustained data rate of 16.9212 MB/sec (average of both read/write) using the scsiRdSecs and scsiWrtSecs() interface. I work for Synergy. Our sales office is at 619-452-0020. D'Anne Thompson dat85715@indirect.com dee@ioinc.tucson.az.us --------------------------- End of New-News digest ********************** From mea@mclean.sparta.com Mon Jan 23 16:21:02 1995 From: mea@mclean.sparta.com (Mike Anderson) Date: Mon Jan 23 16:21:09 PST 1995 Subject: WORM Drive Access Greetings! Anyone out there in Netland have experience with WORM Drives and the raw filesystem? What I need to do is to seek out someplace on the media, write all of the data, then seek back to 0 to write a descriptor that tells me where I got the data, date/time and how big it is, etc. I'd like to have the descriptor at a known location, but I can't write it a-priori 'cause I don't know beforehand how much data I'll grab on this pass of the satellite. Regards, =============================================================================== __ Real-Time System Development, Integration, Training and Services //\\ // \\ Mike Anderson // /\ \\ Director, Real-Time Systems // / \ \\ SPARTA, Inc. Voice : (703) 448-0210 ext. 235 // \ \\ 7926 Jones Branch Drive FAX : (703) 734-3323 \\ \ // Suite 900 EMAIL : mea@mclean.sparta.com \\ \ / // McLean, VA 22102 \\ \/ // "Software development is like making \\ // a baby... You can't make a baby in one \\// month by impregnating nine women. -- "Pride in Performance" Some things just take time." =============================================================================== From daemon@vxw.ee.lbl.gov Tue Jan 24 04:00:30 1995 From: daemon@vxw.ee.lbl.gov Date: Tue Jan 24 04:00:39 PST 1995 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Tue Jan 24 04:00:20 PST 1995 Subject: Re: Restarting a task at a specified location? Subject: Re: IP multicast in VxWorks ??? Subject: Re: What happened to Telesystems SLW ? Subject: Re: VxSim Subject: Re: dual port (VME) memory on FORCE CPU-30 and CPU-40 Subject: Re: Rom emulators Subject: vxworks, errno=0x3d0002, OBJ_UNAVAILABLE, what gives? ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: Restarting a task at a specified location? Date: 23 Jan 1995 16:15:58 GMT From: johncoby@blkbox.com (John Cobarruvias) Organization: The Black Box, Houston, Tx (713) 480-2686 Message-ID: <3g0knu$i5k@news.blkbox.com> References: John R. Cobarruvias (cobarruvias@asd1.jsc.nasa.gov) wrote: I HAVE INCLUDED THE ANSWER TO MY QUESTION BELOW: > Bear with me folks, I am trying to understand someone elses design and code > from it.........(and worst of all, the designer is no longer in the > area......) > I have a task (TASK A) that does the following: > 1) spawned at initialization time > 2) performs task specific one time only initialization (such as ties a > Signal Handler to a RESET_SIGNAL) > 3) Then waits on a semaphore > 4) Upon getting the semaphore, performs its task > 5) Gives up the semaphore > 6) Goes to 3 (wait mode again) > But.....If the application receives a RESET interrupt, the ISR sends a > signal to TASK A, if it is active (not waiting for the semaphore). The > signal handler must then do the following: > 1) reset various parameters > 2) then instruct TASK A to give up the semaphore, and then restart at 3 > above, which is to wait for the semaphore again. > So the question is: How can the signal handler perform its processing, then > instruct TASK A to restart at the point of waiting for the semaphore > regardless of where in the code the task was "signaled" to stop? > Hey folks, any help would be appreciated! The signal handler runs in the tasks context, so if you want the task to give the semaphore, just have the tasks signal handler do the semGive. To get the task to sync up again, use longjmp. The following psuedo code is the idea (although I am typing it now, so it probably has lot's of problems - but it is the right idea): SEM_ID semId; jmp_buf jmpBuf; taskA () { intialize(); signal (signo, handler); /* actually, sigaction is better */ setjmp (jmpBuf); for (;;) { semTake(semId, WAIT_FOREVER); doStuff(); semGive (semId); } } handler () { semGive (semId); longjmp (jmpBuf, 1); } THANKS TO: marc@wrs.com > John R. Cobarruvias - -- John R. Cobarruvias Still trying to put together a sig file......... --------------------------- Newsgroups: comp.os.vxworks Subject: Re: IP multicast in VxWorks ??? Date: 23 Jan 1995 10:38:01 -0700 From: krussell@xmission.com (Kevin Russell) Organization: XMission Public Access Internet (801-539-0900) Message-ID: <3g0php$7c1@xmission.xmission.com> References: <1995Jan22.142101.24919@datum.nyo.dec.com> I implemented IP multicast in 5.1.1. I basically took the net sources from Wind River and merged it with the bsd sources from Steve Deering, the author of the Multicast RFC. His code is available via FTP from gregorio.standford.edu as vmtp-ip/ipmulticast.tar.Z. It took only about two days to implement the protocol level stuff. The hardware filtering requirements can be implemented if desired without a lot of additional effort at the driver level. I did not implement the multicast routing features. Kevin Russell 801-977-1638 segrest@bobseg.enet.dec.com wrote: : Greetings, : Has any work been done to provide IP multicast support in VxWorks ??? : I am working with someone that needs to embed an application currently : implemented with Solaris into a rom loaded VxWorks environment. : -- : Bob Segrest --------------------------- Newsgroups: comp.os.vxworks,comp.dcom.lans.ethernet Subject: Re: What happened to Telesystems SLW ? Date: 23 Jan 1995 19:22:40 GMT From: nelson@crynwr.crynwr.com (Russell Nelson) Organization: Crynwr Software Message-ID: References: In article hmp@frc2.frc.ri.cmu.edu (Henning Pangels) writes: A few years back, we purchased a wireless ethernet bridge (ARLAN 620) from Telesystems SLW, located at the time in Toronto, Canada. We've been pretty happy with the equipment, and were thinking of getting another unit for a new project - alas, they seem to have disappeared from the face of the earth! Would anyone happen to know their fate? They got bought out by Telxon, in Akron Ohio USA, and they've moved to Markham ONT. Their phone number is +1-905-475-7171. They say they have no email address. - -- - -russ http://www.crynwr.com/crynwr/nelson.html Crynwr Software | Crynwr Software sells packet driver support | ask4 PGP key 11 Grant St. | +1 315 268 1925 (9201 FAX) | What is thee doing about it? Potsdam, NY 13676 | What part of "Congress shall make no law" eludes Congress? --------------------------- Newsgroups: comp.os.vxworks Subject: Re: VxSim Date: 23 Jan 95 08:45:28 -0500 From: whiting_ms@corning.com (whiting_ms@corning.com (Matt Whiting)) Organization: Corning Incorporated Message-ID: <1995Jan23.084528.1@corning.com> References: <1995Jan20.200454.3862@picker.com> In article <1995Jan20.200454.3862@picker.com>, heiple@ct.picker.com (HEIPLE, HOLLY) writes: > Does anybody have any experience using VxSim? We are considering buying it > based on the literature but I am wondering how much a developer would really > use it. Why not just use the target? Any comments or words of wisdom? > > Thanks, > Holly Heiple > > > - -- We tried to use it recently during an evaluation of VxWorks, but found that it doesn't run under Solaris. This is documented in the release notes, but we didn't notice that and our sales guy knew we had Solaris and failed to mention it either. So just be sure that you have SunOS if you do plan to use VxSIM. Just using the target is OK assuming you have a target! We'd planned to use VxSIM as our target system was on back order. It could also be advantageous if you wanted to do development at multiple sites and each site didn't have a target sytem available. - ------------------------------------------------------------------------------- Matthew S. Whiting, P.E. | Corning Incorporated | (607) 974-6317 phone Sr. Project Engineer | HP-ME-03-073 | (607) 974-6752 fax whiting_ms@corning.com | Corning, NY 14831 | --------------------------- Newsgroups: comp.os.vxworks Subject: Re: dual port (VME) memory on FORCE CPU-30 and CPU-40 Date: 23 Jan 1995 19:35:37 GMT From: Shash Chatterjee Organization: Lockheed Missiles and Space Co. Message-ID: <3g10e9$o92@butch.lmsc.lockheed.com> References: <0098ABE6.0DE10C20.36@nrlvax.nrl.navy.mil> vilardebo@nrlvax.nrl.navy.mil wrote: > > > I have been trying to figure out how to set up dual port memory on the > FORCE CPU-30 and CPU-40. I want to have 512K bytes of memory set for > VME access (dual ported). I think this probably will require programming > the FGA and moving sysMemTop(). But I don't have much detail. > > If anyone has an example in either C or Ada it would be greately > appreciated. > > I am using the GNU C/C++, VxWorks 5.1, Compilier and the TARTAN v4.3.1 > Ada compilier for VxWorks. > > Thanks > Ken Vilardebo > Mnemonics, Inc. There's not much to this. First, make sure that dual-port RAM is enabled. The as-supplied BSP for VxWorks 5.1.1/CPU-30 enables DPR only for processor 0. Look in sysProcNumSet() in $(VW)/config/frc30/sysLib.c, and you'll see: /* enable dual porting of memory. It is disabled in sysHwInit() */ *FGA_ENAMCODE = FGA_ENAMCODE_EXTUSRDAT_RW | FGA_ENAMCODE_EXTSUPDAT_RW; Make sure that you move it outside the "if (procnum == 0) {} " block. The config.h file defines a macro "LOCAL_MEM_BUS_ADRS", which turns out to be 16M per processor starting at 0x1000000 (ie, the first processor is at 0x1000000, the second is at 0x2000000, etc). The same goes for the CPU-40. One thing to note, by default only EXTENDED (A32D32) Supervisory Data accesses are allowed. That's all there is to it. Hope this helps. +-----------------------------------------------------------------+ + INTERNET (SMTP): ChatterjeeS@lfwc.lockheed.com + + LFWC (SMTP): ChatterjeeS@esdi (ChatterjeeS@[134.189.222.82])+ + LFWC (MSMAIL): Sasvata.Chatterjee (CHTTRS) + +-----------------------------------------------------------------+ + Shash Chatterjee + + Electronic Systems Design & Integration + + Lockheed Fort Worth Company + + P.O. Box 748, MZ2273 VOICE/VOICE-MAIL: (817) 763-1495 + + Ft. Worth, TX 76101 FAX: (817) 777-2115 + +-----------------------------------------------------------------+ Shash --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Rom emulators Date: Tue, 24 Jan 1995 01:24:31 GMT From: dana@wrs.com (Dana Burd) Organization: Wind River Systems Message-ID: References: <9501201943.AA01518@lassen> Reply-To: dana@wrs.com Sender: news@wrs.com (News Manager) > Does anyone out there have experience with ROM emulators (good or bad)? Anyone > using XLNT system's products? We have a couple of netROMs from XLNT around WRS.... excellent sums it up. They are invaluable when doing a BSP... For debugging particularly sensisitive code, the communication channel (from target to host) works well. I hear that XLNT even has a VxWorks network driver that capitalizes on this feature, although it does not offer great bandwidth (someone mentioned ~40kB/sec, but don't quote me on that...). Cannot say whether it is worth the $$$ (I didn't shell out for it), but it has saved me serious amounts of time... - -Dana +--------------------------------------------------------------+ | Dana Burd Wind River Systems, Inc. | | Software Development Engineer 1010 Atlantic Avenue | | dana@wrs.com Alameda, CA 94501 | +--------------------------------------------------------------+ --------------------------- Newsgroups: comp.os.vxworks Subject: vxworks, errno=0x3d0002, OBJ_UNAVAILABLE, what gives? Date: 24 Jan 1995 10:18:21 GMT From: pete@fof.nft.no (Pete J Story) Organization: nft Message-ID: <3g2k5d$irt@nftdf.fof.nft.no> Sender: pete@nftdf (Pete J Story) I am using sockets on vxworks. Occasionally I get a failure indication from write(2), and errno is then set to the magic number 0x3d0002, which most vxworks users will recognise from a task listing. This number comes from objLib, and means "S_objLib_OBJ_UNAVAILABLE". But what does it mean in terms of my write call? Has it worked, or has it not? If not, what am I supposed to do about it? Any ideas? pete@fof.nft.no PS, this is an *old* vxworks, 5.0.2b on sparc. I can't upgrade, because it is really vadsworks (for Ada) and there is no newer version for Ada usage. --------------------------- End of New-News digest ********************** From sergi@hpbpq11.bpo.hp.com Tue Jan 24 05:55:54 1995 From: Sergi Casas(QA Software testing) Date: Tue Jan 24 05:56:01 PST 1995 Subject: Re: comp.os.vxworks newsdigest > In article <1995Jan20.200454.3862@picker.com>, heiple@ct.picker.com (HEIPLE, HOLLY) writes: > > Does anybody have any experience using VxSim? We are considering buying it > > based on the literature but I am wondering how much a developer would really > > use it. Why not just use the target? Any comments or words of wisdom? > > > > Thanks, > > Holly Heiple > > > > > > > - -- > > We tried to use it recently during an evaluation of VxWorks, but found that it > doesn't run under Solaris. This is documented in the release notes, but we > didn't notice that and our sales guy knew we had Solaris and failed to mention > it either. So just be sure that you have SunOS if you do plan to use VxSIM. It works also pretty smooth on HP-UX 9.0 over Apollo 700 machines (PA-RISC). It is a quite comfortable development environment, specially for prototyping code before you have any real target. The only important tool it lacks is vxGdb: currently the only debugger is the standard HP-UX one, say "xdb". It just doesn't know anything on tasks nor windKernel structures, but in the first stages of code development it is usually enough. One alternative is to use evaluation boards of your target processor with ethernet connexion. It is quite fast (rebooting vxSim is almost immediate) and you have a real vxWorks with all the tools. It is usually much more expensive because each board (or target) allows only one engineer working at a time, so you'll often need to have more than one board (or target). We here use both approaches depending on the phase of the project and the kind of bugs we are looking for. -- |============================================================================| | Sergi Casas ##### / ##### Firmware Engineer | | Hewlett-Packard Company ### /_ _ ### | | Barcelona Division (BCD) ## / / / / ## e-mail: sergi@hp-bpo.bpo.hp.com | | Avda. Graells, 501 ## / / /_/ ## Phone : +34 3 582 1475 | | 08190 S. Cugat (Barcelona) ### / ### Fax : +34 3 582 2515 | | SPAIN ##### / ##### HP-Telnet: 712 1475 | |============================================================================| From jeffs@wrs.com Tue Jan 24 14:46:27 1995 From: Jeff Steinley Date: Tue Jan 24 14:46:33 PST 1995 Subject: EPC-7 & INDY 500 BSPs? Greetings. Has anyone created the VxWorks Board Support Package s/w for either of the following boards/workstations: 1) EPC-7 board from Radisys Corporation (80486-Based PC compatible VXIbus System Controller board) 2) Indy 500 workstation from Silicon Graphics (or similiar SGI workstation) Any information regarding the above BSPs would be greatly appreciated. Thanks, Jeff +-----------------------------------------------------------+ | Jeff Steinley Phone: (214) 386-8592 | | Field Application Engineer Fax : (214) 386-8594 | | Wind River Systems, Inc. Email: jeffs@wrs.com | | | | Mailing Address: 4201 Spring Valley Road, Suite 1400 | | Dallas, TX 75244 | +-----------------------------------------------------------+ From mumcu@venus.aselsan.com.tr Tue Jan 24 23:37:44 1995 From: Ahmet MUMCU Date: Tue Jan 24 23:37:51 PST 1995 Subject: ADA Tools for vxworks Does anyone have experience on the following ADA environments for vxworks? - Verdix Vadsworks - Tartan - Greenhill MULTI Any positive or negative feedback on performance, environment, support will be highly appreciated. -------------------------------------------------------------------- - AHMET MUMCU - ASELSAN INC. - PO BOX 101 YENIMAHALLE - ANKARA 06172 - TURKEY - - Fax: 90-312-3545205 - Tel: 90-312-3851900 - Email: mumcu@venus.aselsan.com.tr -------------------------------------------------------------------- From daemon@vxw.ee.lbl.gov Wed Jan 25 04:00:30 1995 From: daemon@vxw.ee.lbl.gov Date: Wed Jan 25 04:00:38 PST 1995 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Wed Jan 25 04:00:16 PST 1995 Subject: Re: Fast SCSI disk write/read Subject: Telneting to Non-Shell Subject: Re: Fast SCSI disk write/read Subject: Re: vxworks, errno=0x3d0002, OBJ_UNAVAILABLE, what gives? Subject: Re: vxworks, errno=0x3d0002, OBJ_UNAVAILABLE, what gives? ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: Fast SCSI disk write/read Date: Tue, 24 Jan 1995 19:17:26 GMT From: dlenrow@world.std.com (David R Lenrow) Organization: The World Public Access UNIX, Brookline, MA Message-ID: References: <1995Jan20.101241.25243@eso.org> While D'Anne Thompson provided a nice low level discussion of the limits on SCSI i/o, an important variable affecting throughput for vxworks (or any) applications is the i/o buffering algorithm. Reading/Writing multiples of SCSI block size, and coalescing adjacent I/O's can save any of: SCSI command exchanges SCSI arbitration cycles Head seeking/thrashing VME bus cycles VME interrupts i/o function call times etc. Experiments timing a 1MB read/write request compared against 1MB worth of small, misaligned, and discontiguous read/write requests may shock you. Also, for recording of bursty realtime data, large i/o buffers can allow you to exceed the peak rates of your i/o device if average throughput is managable. You may need to change your application to behave optimally as a SCSI client. I have previously implemented an i/o process which puts write requests in a multi megabyte FIFO buffer and asynchronously calls write for coalesced, aligned transfers. This setup allowed data with 4MB/sec. peak rates rates to be written to a 1.5 MB/sec tape drive and could buffer enough data to survive the tape drives head movement delay on startup (> 30 seconds). I think I know where you can buy source/binaries for a vxworks SCSI tape driver with similar mods if interested. Good Luck. - -drl --------------------------- Newsgroups: comp.os.vxworks Subject: Telneting to Non-Shell Date: 24 Jan 1995 20:02:33 GMT From: billag@b4pphff (Bill Gutknecht) Organization: Bell-Northern Research Message-ID: <3g3mcp$431@nrtphba6.bnr.ca> I'd like to remove the VxWorks shell for the production version of my product, and replace it with a menu based system. This seems to work well with devices connected to the console. I'd like to be able to telnet in as well ... but that doesn't seem to automagically work (surprise surprise). It was my impression that telnetd changed the global stdin/stdout/stderr to point to the pseudoterminal that telnetd is connected to, hence redirecting the shell. If that is so, why doesn't my menu get redirected? Am I totally off base here? Any help would be greatly appreciated ... Bill - ------------------------------------------------------------------------------ Bill Gutknecht "If I die, I will go before Crom and he will BNR/Northern Telecom ask me 'What is the Riddle of Steel?' If I Research Triangle Park, NC do not know it, he will cast me out of billag@bnr.ca Valhalla and laugh at me ... " - ------------------------------------------------------------------------------ --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Fast SCSI disk write/read Date: 24 Jan 1995 22:48:40 GMT From: friedl@cher.heurikon.com (Ted Friedl) Organization: Heurikon Corporation Message-ID: <3g4048$367@zebu.heurikon.com> References: <1995Jan20.101241.25243@eso.org> [D'Anne Thompson writes...] >2. Contact rtp.ltd.co.uk to get their upgraded SCSI library with full >SCSI-2 command tagging/queing support. This REQUIRES a scsi-2 >compliant disk that supports command tagging/queing. Command >tagging/queing will help to reduce the software overhead in between >data segments, since multiple requests can be queued up at the hardware >level. It should be said that command tagging/queuing is only useful in the case when two or more independent tasks are sharing the same device. For a single task beating on a disk, command tagging/queuing is nothing but overhead. >The only company I know of that can provide you with an SBC solution >with a Wide Synchronous SCSI interface onboard is Synergy Microsystems. > We have a VxWorks driver that works with the standard VxWorks SCSI >Library and provides a continuous sustained data rate of 16.9212 MB/sec >(average of both read/write) using the scsiRdSecs and scsiWrtSecs() >interface. I work for Synergy. Our sales office is at 619-452-0020. Heurikon provides 4 SBC solutions with VxWorks utilizing the NCR 53c720 wide/fast SCSI-DMA controller: the Nitro40 (68040-based), Nitro60 (68060-based), Laguna (MIPS R4000-based) and V960D (i960CF-based). Each SBC, with VxWorks, can provide a continuous sustained data rate of 20MB/sec (w/large data transfers) with local bus bandwidth to spare. Of course, most mechanical devices cannot sustain 20MB/sec but Heurikon can recommend devices that use the SCSI bus efficiently so additive throughput can be achieved. Ted Friedl Software Engineer Heurikon Corporation --------------------------- Newsgroups: comp.os.vxworks Subject: Re: vxworks, errno=0x3d0002, OBJ_UNAVAILABLE, what gives? Date: Wed, 25 Jan 95 10:26:12 GMT From: jpcalvez@ifremer.fr (Jean Pierre Calvez) Organization: Ifremer Message-ID: <1995Jan25.102612.11770@molene.ifremer.fr> References: <3g2k5d$irt@nftdf.fof.nft.no> Reply-To: jpcalvez@ifremer.fr Sender: news@molene.ifremer.fr --------------------------- Newsgroups: comp.os.vxworks Subject: Re: vxworks, errno=0x3d0002, OBJ_UNAVAILABLE, what gives? Date: Wed, 25 Jan 95 10:30:59 GMT From: jpcalvez@ifremer.fr (Jean Pierre Calvez) Organization: Ifremer Message-ID: <1995Jan25.103059.11948@molene.ifremer.fr> References: <3g2k5d$irt@nftdf.fof.nft.no> Reply-To: jpcalvez@ifremer.fr Sender: news@molene.ifremer.fr I encounter a similar phenomenon. I am testing a serial driver and get S_objLib_OBJ_TIMEOUT during a read. This is not an error generated by the driver. Has it to do with the system interface between user software and drivers? I am working with the release 5.1.1 of vxworks on a MVME147 CPU board. --------------------------- End of New-News digest ********************** From lfoster@astro.ge.com Wed Jan 25 05:00:27 1995 From: "Larry D. Foster" Date: Wed Jan 25 05:00:35 PST 1995 Subject: qsort problem Fellow VxWorks users, We are experiencing an intermittent problem using the qsort function. The routine that uses the qsort function is called during the start of our application. Sometimes (most of the time) our application starts without any problem, but there are times that the qsort function call causes an error. We get a "DATA ACCESS EXCEPTION..." and using trace we see that qsort is the problem. We are using qsort after we add an entry to sorted (with qsort) list; the entry is added at the bottom of the list. One important fact here is that the qsort has been called many times BEFORE we get the error. In our attempts to find out what is happening, we looked (dumped) at the memory used for the list that we're sorting. We see that the qsort call has worked (the list is sorted), but one of the entries has been truncated. We are sorting based on an ASCII string and when the entry gets truncated, the sorting is no longer correct (ie. we have one entry out of place). While we're not sure that the truncated entry is the problem, we're fairly certain that the qsort function is the root of our problem. We would appreciate any comments about this problem as well as hearing about other user's experience with the qsort function. Thanks, Larry Larry Foster Phone: (609) 490-7063 Martin Marietta email: lfoster@ge.astro.com Princeton, NJ From crispen@eight-ball.hv.boeing.com Wed Jan 25 07:23:02 1995 From: crispen@eight-ball.hv.boeing.com (Bob Crispen) Date: Wed Jan 25 07:23:09 PST 1995 Subject: Re: ADA Tools for vxworks Ahmet MUMCU asks: >Does anyone have experience on the following ADA environments for vxworks? > >- Verdix Vadsworks >- Tartan >- Greenhill MULTI > >Any positive or negative feedback on performance, environment, support >will be highly appreciated. Dunno about ADA (American Dental Association?) but I've used the VADSWorks Ada for years. Like it. Pretty near bulletproof. The only negative I can think of is that some time ago they distributed a lovely floating point math package for the 020/030, but it seems to have disappeared and been replaced by calls to libm.a routines. Hunt down that old math package. I hear tell that Verdix (now Rational) and Wind River have worked things out so that not a speck of VxWorks code has to be changed for VADSWorks; VADSWorks will now run seamlessly on top of VxWorks, which is good news since us Ada folks always seemed to be at least one generation behind C folks in what we could do in the operating system. +-------------------------------+--------------------------------------+ | Bob Crispen | 668: The neighbor of the Beast | | crispen@foxy.hv.boeing.com +--------------------------------------+ | (205) 461-3296 |Opinions expressed here are mine alone| +-------------------------------+--------------------------------------+ From greg@eng.nmr.varian.com Wed Jan 25 10:00:00 1995 From: greg@eng.nmr.varian.com (Greg Brissey x6951) Date: Wed Jan 25 10:00:07 PST 1995 Subject: Slip/PPP Dear vxPerts, I am look for anyone with experience in using slip and/or ppp between a SUN 4.1.3 or 2.4 and vxWorks 5.1.1 target. I have slip working but it is painfully slow :(, are there any ways of speeding it ? (if so how?) Would ppp be faster and/or better solution. I have a copy of ppp 1.3 for vxWorks. Where do I get a copy of ppp for the SUN 4.1.3 ? Any suggestion or comments would be greatly appreciated. Thanks in Advance, Greg Brissey greg.brissey@nmr.varian.com From stan@lassen.rti.com Wed Jan 25 17:01:46 1995 From: stan@lassen.rti.com (Stan Schneider) Date: Wed Jan 25 17:01:53 PST 1995 Subject: Re: Telneting to Non-Shell Hi Bill, >> I'd like to remove the VxWorks shell for the production >> version of my product, and replace it with a menu based >> system. This seems to work well with devices connected >> to the console. >> I'd like to be able to telnet in as well ... but that >> doesn't seem to automagically work (surprise surprise). >> It was my impression that telnetd changed the global >> stdin/stdout/stderr to point to the pseudoterminal that >> telnetd is connected to, hence redirecting the shell. >> If that is so, why doesn't my menu get redirected? >> Am I totally off base here? This is what rshell does: ioTaskStdSet(0, 0, sd); ioTaskStdSet(0, 1, sd); ioTaskStdSet(0, 2, sd); -- Stan =============================================================================== = = = = Stan Schneider = email: stan@rti.com = = Real-Time Innovations, Inc. = Phone: (408) 720-8312 = = 954 Aster, Sunnyvale, CA 94086 = Fax: (408) 720-8419 = = = = =============================================================================== From daemon@vxw.ee.lbl.gov Thu Jan 26 04:00:30 1995 From: daemon@vxw.ee.lbl.gov Date: Thu Jan 26 04:00:39 PST 1995 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Thu Jan 26 04:00:20 PST 1995 Subject: Re: vxworks, errno=0x3d0002, OBJ_UNAVAILABLE, what gives? Subject: Re: What happened to Telesystems SLW ? Subject: GUI solution for VxWorks Subject: Re: vxworks, errno=0x3d0002, OBJ_UNAVAILABL Subject: Network Interface Ethernet COntrollers Subject: DCE on vxWorks? Subject: Re: Fast SCSI disk write/read Subject: Re: ADA Tools for vxworks Subject: Re: Host Linux ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: vxworks, errno=0x3d0002, OBJ_UNAVAILABLE, what gives? Date: Wed, 25 Jan 95 11:39:05 GMT From: jpcalvez@ifremer.fr (Jean Pierre Calvez) Organization: Ifremer Message-ID: <1995Jan25.113905.15162@molene.ifremer.fr> References: <3g2k5d$irt@nftdf.fof.nft.no> Reply-To: jpcalvez@ifremer.fr Sender: news@molene.ifremer.fr A correction to my precedent post. My error was due to a timeout occuring on a semaphore in the driver. Errors on semaphore cause messages of the kind: "S_objLib_XXXXX". I have constated this on version 5.0.1 and version 5.1.1. --------------------------- Newsgroups: comp.os.vxworks,comp.dcom.lans.ethernet Subject: Re: What happened to Telesystems SLW ? Date: Wed, 25 Jan 1995 15:37:21 GMT From: ramey@gnat.den.mmc.com (Mat Ramey) Organization: Martin Marietta Astronautics, Denver Message-ID: References: Sender: news@news2.den.mmc.com (News Admin) In article hmp@frc2.frc.ri.cmu.edu (Henning Pangels) writes: > A few years back, we purchased a wireless ethernet bridge (ARLAN 620) > from Telesystems SLW, located at the time in Toronto, Canada. We've > been pretty happy with the equipment, and were thinking of getting > another unit for a new project - alas, they seem to have disappeared > from the face of the earth! Would anyone happen to know their fate? We recently bought an ARLAN 620 ourselves ... I have the following for their new address: Telesystems SLW 85 Scarsdale Road, Suite 201 Don Mills, Ontario, Canada (905) 475-7171 > In looking for an alternative, I'd be interested in hearing what > people are using these days in the area of wireless networking. Any > recommendations? I had seriously considered buying Proxim's RangeLan2 Ethernet Bridge. It is a 2.4 GHz spread-spectrum device vs. Arlan's 928 Mhz spread-spectrum, and it was also less expensive at the time; however, it was still under Beta-testing at the time (about 1 yr ago?) ... maybe they're ready to go by now! Proxim Inc. 295 North Barnardo Ave. Mountain View, CA 94043 tel: (415) 960-1630 or (800) 229-1630 fax: 415.964.5181 Also, some notes I had once extracted from the comp.sys.robotics FAQ: Motorola Radio-Telephone Systems Group, Arlington Heights, ILL tel: 708.632.5000 AltairNet: 18GHz-based system design for wireless, indoors networking. The boxes are fairly large, about the size of a shoebox, and are relocatable but not portable. Problem is that is really isn't for mobile applications. Area is really like swiss cheese. Not a problem for fine adjustment in stationary applications but a big problem for mobile devices. [I talked to: Koz Geiss @ 708.576.8216] NCR sells the WaveLAN, which has about a 1Mbit/sec data rate. Not exactly "ethernet", but interfaceable to most networks using MS-DOS boxes as routers. Tetherlink in California is experimenting with a 2Mbit/sec cellular system that is designed for roving portables. [Need address] Cylink 310 N. Mary Avenue Sunnyvale, CA 94086 tel: 408.735.5800 fax: 408.735.6643 AirLink - series of highspeed synch or async modems to 256kb/s. Interfaces include V.11, RS-232, EIA-530. Spread spectrum device operating in 902-928Mhz range. - -- Mat Ramey Martin Marietta Astronautics ramey@den.mmc.com P.O. Box 179 Mail Stop 4372 ,,, ph: 303-977-4339 Denver, Colorado 80201 (- -) ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~oOOo~(_)~oOOo~~ --------------------------- Newsgroups: comp.os.vxworks Subject: GUI solution for VxWorks Date: 25 Jan 1995 18:09:19 GMT From: stevet@an.hp.com (Steve Toro) Organization: Hewlett Packard Andover Site Message-ID: <3g644f$r8o@hpaneqb4.an.hp.com> I am looking for a Graphical User Interface solution for VxWorks that is not VxWindows (that is, not based on X/Motif), but provides the primitives for building a GUI and takes less resources (memory and CPU). Does any one know about 3rd parties that have ported something like GFX or Metawindows or something similar into VxWorks? I am very interested to know about this. Thank you in advance for your help. Regards, Steve --------------------------- Newsgroups: comp.os.vxworks Subject: Re: vxworks, errno=0x3d0002, OBJ_UNAVAILABL Date: 25 Jan 1995 18:24:43 GMT From: bonaker@amasd.anatcp.rockwell.com (Joe Bonaker) Organization: Rockwell Message-ID: <3g651b$2b0@huron.eel.ufl.edu> References: <1995Jan25.113905.15162@molene.ifremer.fr> Reply-To: bonaker@amasd.anatcp.rockwell.com pete@fof.nft.no (Pete J Story) writes: >I am using sockets on vxworks. Occasionally I get a failure indication >from write(2), and errno is then set to the magic number 0x3d0002, which >most vxworks users will recognise from a task listing. This number comes >from objLib, and means "S_objLib_OBJ_UNAVAILABLE". But what does it >mean in terms of my write call? Has it worked, or has it not? If not, >what am I supposed to do about it? jpcalvez@ifremer.fr (Jean Pierre Calvez) responds: >I encounter a similar phenomenon. I am testing a serial driver and get >S_objLib_OBJ_TIMEOUT during a read. This is not an error generated by the driver. >Has it to do with the system interface between user software and drivers? [Next post....] >A correction to my precedent post. My error was due to a timeout occuring on a >semaphore in the driver. > >Errors on semaphore cause messages of the kind: "S_objLib_XXXXX". >I have constated this on version 5.0.1 and version 5.1.1. This example illustrates one of my frustrations with VxWorks' well-intended, module-level descriptive errnos: "too much of a good thing," as the underlying, inter-task objects' errors are propogated to callers. This works great in the simple case where each error results in perror() printouts to reveal the root of the problem. But in cases where some errors are considered "soft," e.g., timeouts or NO_WAIT contention, this approach burdens the driver/library service provider (or more likely, its users) with enumeration of all possible descriptive errors. In general, the VxWorks man pages are remiss in defining errnos and associated conditions. The driver or library developer must "hunt and peck" to determine exactly what errnos are generated by the underlying semaphore/queue objects under certain failure scenarios, else propogate these undocumented errors to the users. We developed a serial driver which sat on a mutual exlusion semaphore. A non-blocking read/write operation's semTake(2) errnos had to be caught by the driver and mapped into more conventional EWOULDBLOCK or EIO errnos to allow the driver user to take appropriate action based on the errno. Based on available VxWorks documentation, our driver developer had no idea that S_objLib_OBJ_UNAVAILABLE would be the errno resultant from a NO_WAIT semTake() when the sempaphore was busy. Fortunately, the VxWorks shell made it fairly painless to stimulate the semaphore busy condition and discover the resultant errno. Guess I'm lobbying for better enumeration and description of errnos in the VxWorks man pages. This would allow a driver/library developer the option of propogating errnos or mapping them to more conventional, Unix errnos. Joe B. --------------------------- Newsgroups: comp.os.vxworks Subject: Network Interface Ethernet COntrollers Date: 25 Jan 1995 18:54:23 GMT From: jorge@b4pph131.bnr.ca (Jorge Rodriguez contractor) Organization: Bell-Northern Research, Ottawa, Canada Message-ID: <3g66ov$438@brtph500.bnr.ca> I am interested in any suggestions for Ethernet Controllers used with a Vxworks based system. I am in the process of identifying possible controllers for my system and would like to hear suggestions from previous experience. Any comments would be greatly appreciated, specially advantages/disadvantages of a particular controller. Anxiously awaiting, Jorge Rodriguez Bell Northern Research (919) 991-4638 Email:Jorge@bnr.ca --------------------------- Newsgroups: comp.os.vxworks Subject: DCE on vxWorks? Date: Tue, 24 Jan 95 11:53:58 PDT From: Miller_DF@Corning.com Message-ID: <1995Jan24.115551.715@corning.com> The version of vxWorks which we are using currently supports the Sun version of remote procedure calls. Is anyone aware of a version of vxWorks which supports OSF DCE remote procedure calls? Thanks. Don Miller Corning, Inc. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Fast SCSI disk write/read Date: 25 Jan 1995 16:46:32 -0600 From: buchele@arlut.utexas.edu (Steven Buchele) Organization: Applied Research Labs - The University of Texas at Austin Keywords: SCSI Message-ID: <3g6kc8$k7a@csdsun1.arlut.utexas.edu> References: <1995Jan20.101241.25243@eso.org> <3g4048$367@zebu.heurikon.com> Sender: Steven W. Buchele >Heurikon provides 4 SBC solutions with VxWorks utilizing the NCR 53c720 >wide/fast SCSI-DMA controller: the Nitro40 (68040-based), Nitro60 >(68060-based), Laguna (MIPS R4000-based) and V960D (i960CF-based). >Each SBC, with VxWorks, can provide a continuous sustained data rate >of 20MB/sec (w/large data transfers) with local bus bandwidth >to spare. > >Of course, most mechanical devices cannot sustain 20MB/sec but Heurikon >can recommend devices that use the SCSI bus efficiently so additive >throughput can be achieved. > >Ted Friedl >Software Engineer >Heurikon Corporation Yes, but you need to drive the software drivers from AP Labs. The drivers that Heurikon supplys are from VxWorks 5.02 and won't work with any release later (such as the current 5.1.1 or 5.2). Using Heurikon 4220 Cougar board VxWorks on a MV167:33MHz AP Labs aio software driver Seagate 2GB Baraccuda we have seen raw file system performance in the range of 4.7MB/sec (using 1MB buffers). BTW The Heurikon + AP Labs combo has been quite easy to configure and get going. Steven Buchele Applied Research Labs Austin, Texas --------------------------- Newsgroups: comp.os.vxworks Subject: Re: ADA Tools for vxworks Date: 25 Jan 1995 22:46:03 GMT From: djohnson@space.honeywell.com (David Johnson) Organization: Honeywell, Inc. Message-ID: References: <9501250737.AA05575@lbl.gov> In article <9501250737.AA05575@lbl.gov>, Ahmet MUMCU wrote: > Does anyone have experience on the following ADA environments for vxworks? > > - Verdix Vadsworks > - Tartan > - Greenhill MULTI > Honeywell Space Systems of Clearwater Florida has used the Vadsworks environment and Ada for several aerospace projects. To start with, I would send an email to Thomas Smith "smith@space.honeywell.com". Tom will either know the answer or be able to point you in the right direction. David Johnson "djohnson@space.honeywell.com" --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Host Linux Date: Thu, 26 Jan 1995 00:58:24 GMT From: hjb@netcom.com (Hwa-Jin Bae) Organization: Peaceful Star, Oakland, CA Message-ID: References: In article , Matthew H. Gerlach wrote: >Hello fellow embedded hackers, > >According to the FAQ, Wind River supports the following hosts: >Sun3, Sun4, HP9000, IBM RS-6000, DEC, SGI, and MIPS. > >Has it ever been considered or tried to use a PC running Linux as the host >for vxworks development. It seems to me the only real requirement of the >host is that it supports various TCP/IP protocols and that one can build >the appropriate cross-compiler variant of the GNU tools. > this is true. you can roll your own compiler tools. i have built m68k x-devel tools using GNU stuff before on linux. i used to use it to compile my realtime kernel for a wirewrapped 68k prototype i have. the binaries for the tools should still be available in: cnuce_arch.cnr.it (131.114.1.10) /pub/micros/gnu-motorola - -- Hwa-Jin Bae (hjb@netcom.com) Peaceful Star, Oakland, CA --------------------------- End of New-News digest ********************** From mea@mclean.sparta.com Thu Jan 26 06:52:42 1995 From: mea@mclean.sparta.com (Mike Anderson) Date: Thu Jan 26 06:52:50 PST 1995 Subject: Re: comp.os.vxworks newsdigest Greetings! > From: buchele@arlut.utexas.edu (Steven Buchele) > [ extra deleted...] > Yes, but you need to drive the software drivers from AP Labs. The > drivers that Heurikon supplys are from VxWorks 5.02 and won't work > with any release later (such as the current 5.1.1 or 5.2). > > Using > Heurikon 4220 Cougar board > VxWorks on a MV167:33MHz > AP Labs aio software driver > Seagate 2GB Baraccuda > > we have seen raw file system performance in the range of 4.7MB/sec > (using 1MB buffers). > > BTW The Heurikon + AP Labs combo has been quite easy to configure and > get going. I think Steve is confusing Heurikon with Interphase. Heurikon doesn't make the Cougar, Interphase does. However, I believe that Steve's message taken with D'Anne's and Ted's regarding SCSI speeds is significant. My own experiences with drives such as the Seagate Hawk 2 (ST31200N, ST32400N) as well as the Barracuda and the (now defunct) DEC DSP series drives, shows quite a bit of fluctuation in performance from drive to drive and depending on the technique for reading and writing to the media. In using the VxWorks I/O system (read/write) with 1 MB buffers, you can typically expect to see 3.5-5 MBytes/sec sustained (using Sync SCSI-2). I'm still waiting on my SCSI analyzer to show up so I can't tell you much about the SCSI-Fast or Fast/Wide interfaces. However, when you drop down to low-level scsiRdSecs/scsiWrtSecs (with the proper SCSI controller chip) your data rates can jump up by a factor of 2-4x. The apparent "slowness" of the I/O system is one of the prices we pay for portability/maintainability. Most programmers understand "read" and "write" when they see it in the code. We know from our "C" manuals how the data stream pointers are updated. All of this is reasonably well-defined, but carries with it the burden of overhead. One of the really nice things about VxWorks is that we have the ability to bypass all of the overhead and go right to the metal. The low-level hooks are usually alread provided by WRS or by one of us when we write the driver in the first place. Unfortunately, in many cases we must sacrifice things like portability and trade off development time (and money) to get the speed. As they say, "good, cheap, fast. Pick any two" ;-). Regards, =============================================================================== __ Real-Time System Development, Integration, Training and Services //\\ // \\ Mike Anderson // /\ \\ Director, Real-Time Systems // / \ \\ SPARTA, Inc. Voice : (703) 448-0210 ext. 235 // \ \\ 7926 Jones Branch Drive FAX : (703) 734-3323 \\ \ // Suite 900 EMAIL : mea@mclean.sparta.com \\ \ / // McLean, VA 22102 \\ \/ // "Software development is like making \\ // a baby... You can't make a baby in one \\// month by impregnating nine women. -- "Pride in Performance" Some things just take time." =============================================================================== From mea@mclean.sparta.com Thu Jan 26 06:52:48 1995 From: mea@mclean.sparta.com (Mike Anderson) Date: Thu Jan 26 06:53:01 PST 1995 Subject: Re: Slip/PPP Greetings! > Submitted-by: greg@eng.nmr.varian.com (Greg Brissey x6951) > > Dear vxPerts, > > I am look for anyone with experience in using slip and/or ppp > between a SUN 4.1.3 or 2.4 and vxWorks 5.1.1 target. > > I have slip working but it is painfully slow :(, > are there any ways of speeding it ? (if so how?) > > Would ppp be faster and/or better solution. > I have a copy of ppp 1.3 for vxWorks. > Where do I get a copy of ppp for the SUN 4.1.3 ? > I currently use SLIP from my Opus SPARCard 5 (a 95 MHz uSPARC 2 under SunOS 4.1.3) to an ISA-bus based 68302 running VxWorks. The best speed I can get out of the Sun Serial ports is 38.4K. That speed is usable for most work, but it's pretty slow in comparison to Ethernet :-(. To get faster than 38.4, I have had to resort to the SBus based Aurora 400SX 4-port, high-speed serial board for the Unix side. That board can handle up to 115 KBps with no problems. Then I boosted the speed of the 68302 and can run at 78 KBPs without a problem. An alternative would be to use a PC with 16550 UARTS running SLIP. The PC can also support the 115 KBPs data rates. PPP won't help you all that much. It does have some facilities for header compression ala CSLIP and that would be of help if your problems required mostly interactive, character-by-character access. However, I got the impression that raw serial port speed is your problem. One other technique that I have used successfully is to use another VxWorks system with an Ethernet as the interface to the main net and use /tyCo/1 (or what ever you have) as the SLIP front end. You can then tune the speeds of the serial link as appropriate for your application. HTH, =============================================================================== __ Real-Time System Development, Integration, Training and Services //\\ // \\ Mike Anderson // /\ \\ Director, Real-Time Systems // / \ \\ SPARTA, Inc. Voice : (703) 448-0210 ext. 235 // \ \\ 7926 Jones Branch Drive FAX : (703) 734-3323 \\ \ // Suite 900 EMAIL : mea@mclean.sparta.com \\ \ / // McLean, VA 22102 \\ \/ // "Software development is like making \\ // a baby... You can't make a baby in one \\// month by impregnating nine women. -- "Pride in Performance" Some things just take time." =============================================================================== From dan@radguard.co.il Thu Jan 26 09:15:24 1995 From: Dan Frommer Date: Thu Jan 26 09:15:30 PST 1995 Subject: Stack modifications Folks, I'd like to make a couple of minor modifications to the TCP/IP stack, namely to the modules ip_input and ip_output. Since I don't have access to the source code and wouldn't like to pay the hefty price tag, I'd like to replace the above modules with modified ones based on Tahoe sources. Has anyone attempted to go through a similar exercise? Can anyone shed some light on the differences between the original Tahoe code and the one used in VxWorks? Any inputs or insights will be highly appreciated! Thanks, -- Dan Frommer Voice: +972-3-6459592 RADGuard, Ltd. Fax: +972-3-6480859 Email: dan@radguard.co.il From Doug.Owens.CHEM#u#235#u#05@quickmail.llnl.gov Thu Jan 26 09:38:31 1995 From: "Doug Owens" Date: Thu Jan 26 09:38:37 PST 1995 Subject: Unsubscribe Subject: Time:9:49 AM OFFICE MEMO Unsubscribe Date:1/20/95 I was subscribed to the exploder for owens@lodtm.llnl.gov. Please unsubscribe - the system is gone and the mail is backing up. Thanks Doug Owens From dwampl@atl.com Thu Jan 26 09:51:03 1995 From: dwampl@atl.com (Dean Wampler (dwampl@atl.com)) Date: Thu Jan 26 09:51:09 PST 1995 Subject: Re: GUI solution for VxWorks > I am looking for a Graphical User Interface solution for VxWorks that > is not VxWindows (that is, not based on X/Motif), but provides > the primitives for building a GUI and takes less resources (memory and CPU). > Does any one know about 3rd parties that have ported something like > GFX or Metawindows or something similar into VxWorks? > I am very interested to know about this. Thank you in advance for your help. Has anyone tried using Tcl/Tk with Vxworks? This is X-based, of course, but I believe it may be slightly smaller and faster than Motif (because it doesn't use Xt or Motif libraries). Dean +-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-+ | Dean Wampler, Ph.D. email: dwampl@atl.com | | Advanced Technology Laboratories | | MS: 264 office: (206) 487-7871 | | 22100 Bothell Highway S.E. fax: (206) 486-5220 | | Bothell, WA 98041-3003 | |-----------------------------------------------------------------| | "I feel your pain...." =:O] | +-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-+ From wynia@iroc.arc.nasa.gov Thu Jan 26 11:03:37 1995 From: "James P. Wynia" Date: Thu Jan 26 11:03:45 PST 1995 Subject: unsubscribe unsubscribe -- ----------------------------< Jim P. Wynia >----------------------------- Sterling Software @ NASA Ames Research Center/ Phone: (415) 604-4797 MS 221-6 / FAX: (415) 604-0071 Moffett Field, CA 94035-1000 /email: wynia@viper.arc.nasa.gov ------------------------------------------------------------------------- From daemon@vxw.ee.lbl.gov Fri Jan 27 07:08:58 1995 From: daemon@vxw.ee.lbl.gov Date: Fri Jan 27 07:09:07 PST 1995 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Fri Jan 27 04:00:43 PST 1995 Subject: VxWorks device driver for 8/4 mill EXABYTE tape unit (SCSI interface) Subject: Re: IP multicast in VxWorks ??? Subject: GNU GDB for VxWorks Subject: Re: Any experience with 68060 boards? Subject: Re: Power PC compilers Subject: fopen on local SCSI doesn't ignore CR Subject: Re: comp.os.vxworks newsdigest Subject: What does "workQPanic" mean? ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: VxWorks device driver for 8/4 mill EXABYTE tape unit (SCSI interface) Date: Thu, 26 Jan 1995 16:17:16 GMT From: adaless@relay.nswc.navy.mill (Anthony DAlessandro dep.F32) Organization: NSWC DAHLGREN VA Keywords: I would like any information on creating a vxworks SCSI device driver for EXABYTE tape drives . Message-ID: <1995Jan26.161716.10051@relay.nswc.navy.mil> Sender: jporter@jasper (Jackie Porter (F32)) --------------------------- Newsgroups: comp.os.vxworks Subject: Re: IP multicast in VxWorks ??? Date: 26 Jan 1995 16:57:25 GMT From: sarkar@aesir.mit.edu (ss) Organization: m Message-ID: <3g8k9l$81c@senator-bedfellow.MIT.EDU> Reply-To: sarkar@aesir.mit.edu In message <1995Jan22.142101.24919@datum.nyo.dec.com>, segrest@bobseg.enet.dec.com () writes: > >Greetings, > >Has any work been done to provide IP multicast support in VxWorks ??? > >I am working with someone that needs to embed an application currently >implemented with Solaris into a rom loaded VxWorks environment. > >-- >Bob Segrest > I would also like to receive copies of any email on this topic. thanks serge sarkar --------------------------- Newsgroups: comp.os.vxworks Subject: GNU GDB for VxWorks Date: 26 Jan 1995 11:36:17 -0600 From: dvernon@fohnix.metronet.com (David Vernon) Organization: Texas Metronet, Internet for the Individual 214-705-2901 (info) Message-ID: <3g8mih$1hc@fohnix.metronet.com> I just read this in the VxWorks FAQ: >> With a little bit of hacking, regular GDB works just fine. Personally, >> I find GUI to a debugger gets in the way of real work. I use GNU >> Emacs GDB interface which works well and can be easily customized. Can anyone detail for me the "little bit of hacking" that is required to get GDB to work with VxWorks? Thanks for any input. Dave Vernon dvernon@metronet.com --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Any experience with 68060 boards? Date: 26 Jan 1995 19:54:40 GMT From: synergy@nic.cerf.net (John E. Clark) Organization: CERFnet Dial n' CERF Customer Message-ID: <3g8um0$7g2@news.cerf.net> References: <9411171915.AA28760@glenvcr.glenayre.com> In article <9411171915.AA28760@glenvcr.glenayre.com> jtosey@glenayre.com (Joseph Tosey [4572]) writes: +167 with copyback caching. They have made some significant +changes to the caching model in the 060 which cripples it from +our point of view. The 040 supports snooping to maintain cache +coherency, but the 060 does not. While snooping assists the programer in writing code, snooping slows the memory down on all DMA transfers that must be snooped. While code is executing out of cache, a peripheral device may be hanging out to dry on snooped cylces. +As a result our application, which does not know about caching +modes at all, can no longer run in copyback caching mode. We +tried to put in cache flushes around disk and serial i/o +operations, but they happen very frequently as I/O request +rates climb. We are currently using the 060 in writethrough +mode, and our processor utilization tools tell us that is +reducing performance a full 50%. Obviously I don't know about your application, nor enviornment. However, one trick is to use the MMU to map the same physical memory to two different virtual address regions. One region is marked as copy-back, the other no-cache. Io transfers are done via the no-cache region and 'code' is executed out of the CBC region. Just as obvious, if code is part of the I/O transfer, then a 'cinva' needs to be done before execution starts. As the region would have been filled by acesses to the non-cached zone, but code is excuted out of the CBC zone. +The point is that the 68060 affects applications, not just the +OS. Your application has to understand the caching model if it +writes to buffers which are DMAed. This was not necessary for +the 040 and other earlier processors. Well, here 'cache ignorance' was easier to overlook, but there was a 'hit'. But with ETHERNET at about a Meg/sec. and SCSI often bursty, but overall lower than say 1-2 Meg/sec. one could overlook the 'hit'. But now with faster SCSI, 100 Meg-a-bit ETHERNET, 155 Meg-a-bit ATM, and Fiber of one sort or another all starting at 100 Meg-a-bit, and going up, the 'hit' is going to be more pronounced. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Power PC compilers Date: 26 Jan 1995 20:03:21 GMT From: synergy@nic.cerf.net (John E. Clark) Organization: CERFnet Dial n' CERF Customer Message-ID: <3g8v69$7i3@news.cerf.net> References: <9501180316.AA16552@sunshine.labs.tek.com> In article <9501180316.AA16552@sunshine.labs.tek.com> johncal@sunshine.labs.tek.com (John Calvin) writes: + +What compiler are people using for Power PC's? + + +GNU I've been advised, is 8 to 10 months away from a Power-PC compiler. +Can anybody confirm this? Uh? The current GNU binutils, and gcc are capable of generating PPC code to produce an elf binary output. It's not exactly 'straightforward' as say 'configure --host=sunos4 --target==power-pc', but it is do-able. As for 'optimization', beats me.... The gcc-ppc was built on the rs6000 processor work, so if it was optimized for that architecture, it may have some beenfit in the ppc. Then there's gdb... but who uses debuggers anyway, really? --------------------------- Newsgroups: comp.os.vxworks Subject: fopen on local SCSI doesn't ignore CR Date: Thu, 26 Jan 1995 22:52:32 GMT From: geger@phantom.den.mmc.com (George Eger) Organization: Martin Marietta Astronautics Group Keywords: fopen SCSI Message-ID: <1995Jan26.225232.5686@den.mmc.com> Followup-To: Opening local SCSI Sender: news@den.mmc.com (News Admin) I had posted a message about why: fgets(&(filename[0]), SOME_MAX, stdin); NewFD = fopen(&(filename[0]), "r"); worked on a remote Unix directory, but failed on a local SCSI disk. Thanks to Alan Nishioka (alann@abekas.com), I found my problem. Turns out that the fopen() on a local DOS filesystem is sensitive to the NEWLINE at the end of the string. So by trimming out the ending NEWLINE, I can open the local files from my program. I guess that the mechanism to open a remote file on my Unix host must trim out (or maybe Unix ignores) the ending NEWLINE that fgets() puts in the string buffer. Thanks to all, GWE - -- ||========================================================================== ||George Eger / geger@den.mmc.com || Voice - (303) - 971 - 6974 || ||Integ. Fault Tolerant Avionics || Fax - (303) - 977 - 1145 || ||Space Launch Systems || MS T320 || ||Martin Marietta Astronautics || P.O. Box 179, Denver CO 80201 || ||========================================================================== ||We are at a cusp - between the past when humans were more reliable than || ||computers and the future when computers are more reliable than humans. || ||========================================================================== ||All opinions (however truthful or misinformed) are my own. || ||========================================================================== --------------------------- Newsgroups: comp.os.vxworks Subject: Re: comp.os.vxworks newsdigest Date: 27 Jan 1995 02:03:32 GMT From: friedl@cher.heurikon.com (Ted Friedl) Organization: Heurikon Corporation Message-ID: <3g9k9k$8nh@zebu.heurikon.com> References: <9501261458.AA01071@locutus> In article <9501261458.AA01071@locutus>, Mike Anderson wrote: > My own experiences with drives >such as the Seagate Hawk 2 (ST31200N, ST32400N) as well as the >Barracuda and the (now defunct) DEC DSP series drives, shows quite >a bit of fluctuation in performance from drive to drive and depending >on the technique for reading and writing to the media. I have been checking drives as they pass through my hands and taking notes when possible. The following results are from the second (2nd) 20 seconds of reading 524,288 byte buffers sequentially from the outside track toward the inside (i.e. w/512-byte block device, read LBA 0-1023, 1024-2047, ...). All drives were tested with a 100ns period and 8 req/ack offset with their factory-default mode pages. Some wide drives were also measured running 8-bit... 8-bit (or default) disc / comp = ratio sustained_rate / bus_time = burst_rate Ciprico AS6701 358 358 1 9.38 MB/s 96.4 % 9.72 MB/s Fuji M2694ESA 903 146 6 3.79 MB/s 41.3 % 9.20 MB/s HP C2247 0 145 0 3.79 MB/s 99.4 % 3.82 MB/s HP C3323A 0 184 0 4.77 MB/s 99.3 % 4.80 MB/s Qntm Atlas XP32150 18225 279 65 7.26 MB/s 78.5 % 9.25 MB/s Qntm Empire 1080SWD 883 147 6 3.82 MB/s 39.0 % 9.79 MB/s Qntm Empire 540S 878 146 6 3.80 MB/s 38.8 % 9.79 MB/s Qntm Lightning 540S 4972 142 35 3.69 MB/s 56.0 % 6.60 MB/s Seagate ST12550WD 705 204 3 5.35 MB/s 56.9 % 9.39 MB/s Seagate ST31200W 647 162 3 4.21 MB/s 44.5 % 9.46 MB/s (MB = 1,000,000 bytes!) 16-bit disc / comp = ratio sustained_rate / bus_time = burst_rate Ciprico AS6701 4+1 665 665 1 17.42 MB/s 93.5 % 18.63 MB/s DEC DSP3107LW 3082 182 16 4.74 MB/s 41.7 % 11.34 MB/s Qntm Empire 1080SWD 608 147 4 3.82 MB/s 27.4 % 13.91 MB/s Seagate ST12550WD 1170 208 5 5.43 MB/s 30.9 % 17.55 MB/s Seagate ST31200W 841 162 5 4.21 MB/s 23.7 % 17.73 MB/s (MB = 1,000,000 bytes!) The more efficient drives have better additive performance when sharing the same bus. For example two Quantum Empires 540S's can sustain 7.43 MB/s while two comparable HP C2247's can only sustain 4.14 MB/s. On the other hand, drives that tend to grab the bus and not let go (like the HP C2247) look better than others sharing the bus when the others are the more polite variety. Ted Friedl Software Engineer Heurikon Corporation --------------------------- Newsgroups: comp.os.vxworks Subject: What does "workQPanic" mean? Date: Thu, 26 Jan 1995 23:59:14 GMT From: claus@pepiiu1.SLAC.Stanford.EDU (Ric Claus) Organization: Stanford Linear Accelerator Center Message-ID: Reply-To: claus@slac.stanford.edu (Ric Claus) Sender: news@unixhub.SLAC.Stanford.EDU I have been plagued by the message workQPanic: Kernel work queue overflow. after running some 7 or 8 application codes in the context of a "shell", all of which I wrote. The application codes handle DSP processor boards that all run the same code. Thus, the interrupt timing is pretty bursty, but no higher than 5 to 10 at "a time" (< 100 nS window?). The shell provides the ISR and packages the interrupt messages to be sent to the application's task level handler. (Perhaps the ISR takes too long?) When invoking an application, it is loaded from a Unix disk through ethernet. Starting it causes two tasks to be created successively. When the first finishes and returns, the second is started. It sure seems like VxWorks is running out of some resource, but I can't figure out what. When I checkStack after every application invocation, I see that all stacks have been used to no more than half the alloted memory. The INTERRUPT stack never uses more than 268 bytes out of 1000. I see no signs of anything being corrupted anywhere. Perhaps task descriptors aren't being recycled? When I memShow after every application invocation, I see that memory usage is well bounded and nowhere close to the edge. Memory usage is application dependent, but if I invoke the same application as the one before the previous application, memory usage goes back to the same values. As an aside, I was not expecting this to be true because I understood that there was a memory leak associated with taskSpawns. Unless memShow does not properly account for all the memory used, I see no leaks due to tasks being created and properly finishing (i.e. not taskDeleted, by me at least). It's hard to look around after workQPanic occurs because it causes a reboot. I'm also not sure where to look. I've looked in the VxWorks manuals for what this message might mean, but have come up empty handed. It would be nice if it (and a bunch of other things) were in the index... The hardware is a 25 MHz Force 40 with 4 Meg and an Eagle 02 (ethernet and VSB) running version 5.1.1 VxWorks. Thanks in advance for any light anyone can shed on this. Ric Claus (415.926.2697) --------------------------- End of New-News digest ********************** From @icineca.cineca.it:sandri@deis58.cineca.it Fri Jan 27 12:50:44 1995 From: sandri@deis58.cineca.it (Mario Sandri) Date: Fri Jan 27 12:50:51 PST 1995 Subject: Unsubscribe Dear Sirs, please unsubscribe me. Thanks Mario Sandri From bordua@csg.lbl.gov Fri Jan 27 13:13:47 1995 From: bordua@csg.lbl.gov (Michael Bordua) Date: Fri Jan 27 13:13:53 PST 1995 Subject: Re: Unsubscribe Unsubscribed. Mike Bordua MGBordua@lbl.gov VxWorks Exploder Administrator Computer Systems Group Lawrence Berkeley Laboratory From daemon@vxw.ee.lbl.gov Sat Jan 28 04:00:32 1995 From: daemon@vxw.ee.lbl.gov Date: Sat Jan 28 04:00:46 PST 1995 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Sat Jan 28 04:00:21 PST 1995 Subject: Re: Info wanted on off-the-shelf MIL-SPEC embedded controller boards Subject: Re: dual port (VME) memory on FORCE CPU-30 and CPU-40 Subject: 5.1 qsort Subject: Can SLIP use any non block device driver ? Subject: Re: GNU GDB for VxWorks Subject: Re: 5.1 qsort Subject: Re: GNU GDB for VxWorks Subject: Re: How can I test for existence of a file? Subject: Re: GNU GDB for VxWorks Subject: Re: GNU GDB for VxWorks Subject: Re: GNU GDB for VxWorks Subject: Removable Media for VME + VxWorks Subject: Re: Slip/PPP Subject: msgQLib Subject: Re: VxSim Subject: scsiIoctl() takes 300 microseconds? ------------------------------------------------------- Newsgroups: comp.os.vxworks,comp.sys.m68k,alt.industrial.computing,sci.electronics,rec.radio.amateur.space Subject: Re: Info wanted on off-the-shelf MIL-SPEC embedded controller boards Date: 27 Jan 1995 12:31:34 GMT From: aranders@kosepc01.delcoelect.com (Alan Anderson) Organization: Delco Electronics Corporation, Kokomo, Indiana Message-ID: <3gap36$p9f@kocrsv08.delcoelect.com> References: <3fi0d6$hbr@CUBoulder.Colorado.EDU> Reply-To: aranders@kosepc01.delcoelect.com (Alan Anderson) Pardon the severely off-topic note, but I just had to respond. In , lacovara@ollie.jsc.nasa.gov (R. C. Lacovara) writes: >[note deleted, .sig retained] >Texas is the largest and grandest State of the United States. It retains the ^^^^^^^ >right to fly its flag at the same height as that of the United States. Texas >must not be abbreviated, like NJ or RI: NJ fits between Houston and San >Antonio, with RI to spare. Al Watch out, Texas, or Alaska might split in two and make you the *third* largest state! = === === === = = = === === === === = = === = = = === = = === = Alan Anderson | Ignorance can be fixed, (I speak AT, not FOR, Delco Electronics) | but stupidity is permanent. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: dual port (VME) memory on FORCE CPU-30 and CPU-40 Date: Mon, 23 Jan 1995 18:02:30 GMT From: healy@nosc.mil (Mike Healy) Organization: NCCOSC RDT&E Division, San Diego, CA Message-ID: <1995Jan23.180230.11551@nosc.mil> References: <0098ABE6.0DE10C20.36@nrlvax.nrl.navy.mil> Sender: news@nosc.mil On CPU-30 and CPU-40 you need to set some registers in the FGA to specify the address to reference the shared memory across the VME bus. Check the FGA manual. sysMemTop uses the following macro from frc40.h: /* PIT macro to return local memory size in bytes */ #define FRC40_MEM_SIZE \ (0x2000000 >> ((*PIT_PBDR (FRC40_PIT2_BASE_ADRS)) & 0x7)) from config.h for CPU-40, vxWorks 5.1: /* * Local-to-Bus memory address values: * The CPU-40 dual port memory address is under software control and may be * changed by programming the FGA-002. For more information on the user * programmable options of the Force FGA-002 see the CPU-40 or FGA-002 manual. * * To determine the actual memory size use sysMemTop(). The constant * LOCAL_MEM_SIZE is used to configure the MMU in sysLib.c. For optimal use * of cache, build VxWorks with LOCAL_MEM_SIZE set to the size of the * CPU-40's local RAM */ #define LOCAL_MEM_LOCAL_ADRS 0x00000000 #define LOCAL_MEM_BUS_ADRS (0x01000000*(sysProcNumGet()+1)) /* 16M*proc# */ #define LOCAL_MEM_SIZE 0x00400000 /* minimum memory available */ I think if you adjust both of these MEM_SIZE values you'll end up with memory at the top of the address space of the boards that will be protected from vxWorks use. Mike Healy healy@nosc.mil --------------------------- Newsgroups: comp.os.vxworks Subject: 5.1 qsort Date: Wed, 25 Jan 1995 16:40:05 GMT From: healy@cod.nosc.mil (Mike Healy) Organization: NCCOSC RDT&E Division, San Diego, CA Message-ID: <1995Jan25.164005.15811@nosc.mil> Sender: news@nosc.mil We have run into a problem with qsort on our 5.1 CPU-40 SGI version. It is returning corrupt data. We have no problem with qsort from our 5.1.1 CPU-40 Sun version. Is there a fixed version for the SGI available somewhere? Thanks Mike Healy healy@nosc.mil --------------------------- Newsgroups: comp.os.vxworks Subject: Can SLIP use any non block device driver ? Date: Fri, 27 Jan 1995 08:36:33 GMT From: lee@dradeb.demon.co.uk (Lee) Organization: DRA Bedford Message-ID: Sender: news@demon.co.uk (Usenet Administration) We have two VME systems connected by ARINC429 and would like to run TCP/IP over the link. If we implemented a non block device driver (ie open,read,write, ioctrl etc) for the 429 board could the SLIP driver use it. Any information, advice or links would be great. martin@drabed.demon.co.uk --------------------------- Newsgroups: comp.os.vxworks Subject: Re: GNU GDB for VxWorks Date: Fri, 27 Jan 1995 14:13:01 GMT From: clegg@ssd.fsi.com Organization: Flightsafety International, Broken Arrow, OK Message-ID: <1995Jan27.141301.19732@dev1.ssd.fsi.com> References: <3g8mih$1hc@fohnix.metronet.com> Reply-To: clegg@ssd.fsi.com Sender: news@dev1.ssd.fsi.com In <3g8mih$1hc@fohnix.metronet.com>, dvernon@fohnix.metronet.com (David Vernon) writes: >I just read this in the VxWorks FAQ: > >>> With a little bit of hacking, regular GDB works just fine. Personally, >>> I find GUI to a debugger gets in the way of real work. I use GNU >>> Emacs GDB interface which works well and can be easily customized. > >Can anyone detail for me the "little bit of hacking" that is required >to get GDB to work with VxWorks? > I would be interested in this also. Thanks R. Clegg _____________________________ __ _ _ |__|_|_|==========__ [=== / \ | \ FlightSafety ) | \ SSD ) |===|===\ ) |_\_|___|\=========) [___________________] \ \/ / \ \/ / \/ / \ \/ The best safety device / /\ /\ \ in any aircraft is a ___/ /\_\___/_/\ \__ well trained pilot. ____[____________________]_____ clegg@ssd.fsi.com --------------------------- Newsgroups: comp.os.vxworks Subject: Re: 5.1 qsort Date: 27 Jan 1995 16:42:27 GMT From: mwette@mr-ed.jpl.nasa.gov (Matt Wette) Organization: Jet Propulsion Laboratory Message-ID: <3gb7pj$17j@grover.jpl.nasa.gov> References: <1995Jan25.164005.15811@nosc.mil> In article <1995Jan25.164005.15811@nosc.mil>, healy@cod.nosc.mil (Mike Healy) writes: |> |> We have run into a problem with qsort on our 5.1 |> CPU-40 SGI version. It is returning corrupt data. |> We have no problem with qsort from our 5.1.1 CPU-40 |> Sun version. Is there a fixed version for the SGI |> available somewhere? qsort in 5.1 is broken. You can get a version of qsort from the GNU glibc distribution, or from the Cygnus newlib distribution, which is included below. Matt /* FUNCTION <>---sort an array INDEX qsort ANSI_SYNOPSIS #include void qsort(void *<[base]>, size_t <[nmemb]>, size_t <[size]>, int (*<[compar]>)(const void *, const void *) ); TRAD_SYNOPSIS #include qsort(<[base]>, <[nmemb]>, <[size]>, <[compar]> ) char *<[base]>; size_t <[nmemb]>; size_t <[size]>; int (*<[compar]>)(); DESCRIPTION <> sorts an array (beginning at <[base]>) of <[nmemb]> objects. <[size]> describes the size of each element of the array. You must supply a pointer to a comparison function, using the argument shown as <[compar]>. (This permits sorting objects of unknown properties.) Define the comparison function to accept two arguments, each a pointer to an element of the array starting at <[base]>. The result of <<(*<[compar]>)>> must be negative if the first argument is less than the second, zero if the two arguments match, and positive if the first argument is greater than the second (where ``less than'' and ``greater than'' refer to whatever arbitrary ordering is appropriate). The array is sorted in place; that is, when <> returns, the array elements beginning at <[base]> have been reordered. RETURNS <> does not return a result. PORTABILITY <> is required by ANSI (without specifying the sorting algorithm). Supporting OS subroutines required: <>, <>, <>, <>, <>, <>, <>. */ /* * qsort.c * Original Author: G. Haley * * Sorts an array of nmemb objects, the initial member of which is pointed to * by base. The size of each object is specified by size. The contents of the * array are sorted in ascending order according to a comparison function * pointed to by compar, which is called with two arguments pointing to the * objects being compared. The function shall return an integer less than, * equal to or greater than zero if the first argument is considered to be * respectively less than, equal to or greater than the second. If two members * compare as equal, their order in the sorted array is unspecified. */ #include #include #include static int _DEFUN (find_pivot, (base, i, j, size, compar), _PTR base _AND int i _AND int j _AND size_t size _AND int (*compar) ()) { _PTR first_key; _PTR next_key; int k, res; first_key = (_PTR) (((char *) base) + (i * size)); next_key = first_key; for (k = i + 1; k <= j; k++) { next_key = (_PTR) (((char *) next_key) + size); res = (*compar) (next_key, first_key); if (res > 0) return k; else if (res < 0) return i; } return -1; } static void _DEFUN (swap, (base, i, j, size), _PTR base _AND int i _AND int j _AND size_t size) { #ifdef __GNUC__ _PTR temp = __builtin_alloca (size); #else static _PTR temp = NULL; static size_t max_size = 0; #endif _PTR elem1, *elem2; #ifndef __GNUC__ if (size > max_size) { temp = realloc (temp, size); max_size = size; } #endif elem1 = (_PTR) (((char *) base) + (i * size)); elem2 = (_PTR) (((char *) base) + (j * size)); memcpy (temp, elem1, size); memcpy (elem1, elem2, size); memcpy (elem2, temp, size); } static int _DEFUN (partition, (base, i, j, pivot_index, size, compar), _PTR base _AND int i _AND int j _AND int pivot_index _AND size_t size _AND int (*compar) ()) { int left, right; left = i; right = j; do { swap (base, left, right, size); if (pivot_index == left) pivot_index = right; else if (pivot_index == right) pivot_index = left; while (compar ((_PTR) (((char *) base) + (left * size)), (_PTR) (((char *) base) + (pivot_index * size))) < 0) left++; while (compar ((_PTR) (((char *) base) + (right * size)), (_PTR) (((char *) base) + (pivot_index * size))) >= 0) right--; } while (left <= right); return left; } static void _DEFUN (inside_qsort, (base, i, j, size, compar), _PTR base _AND int i _AND int j _AND size_t size _AND int (*compar) ()) { int pivot_index, mid; if ((pivot_index = find_pivot (base, i, j, size, compar)) != -1) { mid = partition (base, i, j, pivot_index, size, compar); inside_qsort (base, i, mid - 1, size, compar); inside_qsort (base, mid, j, size, compar); } } _VOID _DEFUN (qsort, (base, nmemb, size, compar), _PTR base _AND size_t nmemb _AND size_t size _AND int (*compar) ()) { inside_qsort (base, 0, nmemb - 1, size, compar); } - -- matthew.r.wette@jpl.nasa.gov --------------------------- Newsgroups: comp.os.vxworks Subject: Re: GNU GDB for VxWorks Date: 27 Jan 1995 15:39:18 GMT From: bowman@waterloo.hp.com (Don Bowman) Organization: H-P Panacom Div, Waterloo, ON Canada Message-ID: <3gb436$2f8@hppadbk.waterloo.hp.com> References: <3g8mih$1hc@fohnix.metronet.com> David Vernon (dvernon@fohnix.metronet.com) wrote: : I just read this in the VxWorks FAQ: : >> With a little bit of hacking, regular GDB works just fine. Personally, : >> I find GUI to a debugger gets in the way of real work. I use GNU : >> Emacs GDB interface which works well and can be easily customized. : Can anyone detail for me the "little bit of hacking" that is required : to get GDB to work with VxWorks? Just run configure and make. The magic for us was: configure --target vxworks960 hppa1.1-hpux - -- bowman@waterloo.hp.com ** 519-883-3019 ** http://www.waterloo.hp.com/~bowman --------------------------- Newsgroups: comp.os.vxworks Subject: Re: How can I test for existence of a file? Date: Fri, 27 Jan 1995 16:26:04 +0000 From: markE@baesema1.demon.co.uk (Mark Endacott) Organization: BAeSEMA Message-ID: <791223964snz@baesema1.demon.co.uk> References: Reply-To: markE@baesema1.demon.co.uk Sender: usenet@demon.co.uk In article gamin@ireq-robot.hydro.qc.ca "Martin Boyer" writes: > I am trying to test for the existence of a file (in an NFS-mounted > partition) under VxWorks 5.1.1. > > Testing the return value of the following functions gives me enough > information, but it also prints an error message on the console if the > file doesn't exist (which I don't want). > > stat(filename, &st) > => OK if the file exists > => Error if not (but also prints an error message) > > open(hostconfig, READ, 0) > => a file descriptor if the file exists > => Error if not (but also prints an error message) > > I was looking for something like the UNIX "access" function, but > couldn't find it. I'd be willing ro "redirect standard error to > /dev/null" if that would get rid of the error message and if I knew > how to do that in VxWorks. > Don't know about the 'access' function although there may be one, however redirection of i/o is easy. vxWorks uses Fd's 0,1 and 2 as the default stdin, stdout and stderror, these can be manipulated with the routines ioGlobalStdSet, ioGlobalStdGet,ioTaskStdGet and ioTaskStdSet. The difference between them being the task routines only change the stdio for the task and not all tasks. All you have to do is call the following ioTaskStdSet ( task_id,3,new_fd); Where 'new_fd' is the fd of a null file. There may already be a 'null' Fd already opened by vxWorks, I can't remember and I don't currently have access to our target hardware that runs vxWorks. Try typing iosFdShow at the console and see what it gives you in terms of open Fd's. One thing I am unsure of is, if the 'error handler' that is generating the message on your console is tied to the task's I/O or the Global I/O. If its global you can use the global routines however this will route ALL stderror traffic to NULL. - -- Mark Endacott --------------------------- Newsgroups: comp.os.vxworks Subject: Re: GNU GDB for VxWorks Date: 27 Jan 1995 18:20:11 GMT From: synergy@nic.cerf.net (John E. Clark) Organization: CERFnet Dial n' CERF Customer Message-ID: <3gbdgr$6qf@news.cerf.net> References: <3g8mih$1hc@fohnix.metronet.com> In article <3g8mih$1hc@fohnix.metronet.com> dvernon@fohnix.metronet.com (David Vernon) writes: +I just read this in the VxWorks FAQ: + +>> With a little bit of hacking, regular GDB works just fine. Personally, +>> I find GUI to a debugger gets in the way of real work. I use GNU +>> Emacs GDB interface which works well and can be easily customized. + +Can anyone detail for me the "little bit of hacking" that is required +to get GDB to work with VxWorks? The GDB that one can pick up from prep.ai.mit.edu has in it an RPC interface to vxworks. I don't know how straight forward it would be to included that in the build for your local gdb. I don't know if there's a convienent 'configure' option or what. Also, there may be a difference between the current supported vxworks RPC interface and the one that's in the prep distribution. There are a couple of other 'remote' debugging interfaces there as well. One of which will go over a serial line... for the really minimally inclined... --------------------------- Newsgroups: comp.os.vxworks Subject: Re: GNU GDB for VxWorks Date: Fri, 27 Jan 1995 21:46:30 GMT From: hjb@netcom.com (Hwa-Jin Bae) Organization: Peaceful Star, Oakland, CA Message-ID: References: <3g8mih$1hc@fohnix.metronet.com> <3gbdgr$6qf@news.cerf.net> In article <3gbdgr$6qf@news.cerf.net>, John E. Clark wrote: > >The GDB that one can pick up from prep.ai.mit.edu has in it an RPC >interface to vxworks. I don't know how straight forward it would be >to included that in the build for your local gdb. I don't know if >there's a convienent 'configure' option or what. Also, there may >be a difference between the current supported vxworks RPC interface >and the one that's in the prep distribution. recent versions of GDB that has vxworks target "configure" will build and work fine on machines that are known to be supported, but YMMV. it won't do anything unless you buy the vxworks portion of the RPC/debug interface stub, which is not included in GDB (costs big bucks). so much for the free software... > >There are a couple of other 'remote' debugging interfaces there as >well. One of which will go over a serial line... for the really >minimally inclined... > this option is not "minimally inclined". this actually works much better, IMHO. vxgdb never worked properly every time i attempted to use it in the past and most only for task/app debugging (can't do real low level debugging), full of bugs in the debugger itself, etc. but YMMV, again. - -- Hwa-Jin Bae (hjb@netcom.com) Peaceful Star, Oakland, CA --------------------------- Newsgroups: comp.os.vxworks Subject: Re: GNU GDB for VxWorks Date: Fri, 27 Jan 1995 22:01:31 GMT From: hjb@netcom.com (Hwa-Jin Bae) Organization: Peaceful Star, Oakland, CA Message-ID: References: <3g8mih$1hc@fohnix.metronet.com> In article <3g8mih$1hc@fohnix.metronet.com>, David Vernon wrote: >I just read this in the VxWorks FAQ: > >>> With a little bit of hacking, regular GDB works just fine. Personally, >>> I find GUI to a debugger gets in the way of real work. I use GNU >>> Emacs GDB interface which works well and can be easily customized. > >Can anyone detail for me the "little bit of hacking" that is required >to get GDB to work with VxWorks? > >Thanks for any input. > >Dave Vernon >dvernon@metronet.com > get latest GDB source distribution. i used 4.8 (old but compatible with DUEL extention language). read installation doc and build for proper target and host pair using configure script. you may or may not have to hack on the config files. a lot of times, things dont build right unless you're using extremely vanilla setup. you just have to go around and fix various files that are missing/ or not compiling. you also have to decide whether you're gonna use RPC based (like vxgdb) app debugging or plain old serial remote debugging (better, IMHO). the former requires you to buy expensive little remote module from WRS. the latter does not. just look at remote-*.c files and pick the closes one that might work with your setup. i usually pick m68k-stub.c and hack on it to make it work with vxworks. hook up over null modem and you have a real remote debugger that can do low-level debugging as well as application debugging! - -- Hwa-Jin Bae (hjb@netcom.com) Peaceful Star, Oakland, CA --------------------------- Newsgroups: comp.os.vxworks Subject: Removable Media for VME + VxWorks Date: 27 Jan 1995 16:41:43 -0600 From: buchele@arlut.utexas.edu (Steven Buchele) Organization: Applied Research Labs - The University of Texas at Austin Message-ID: <3gbsr7$7kd@csdsun1.arlut.utexas.edu> We are looking for some sort of removeable media for our project so that the sponsor can easily/quickly take home data, but does not have to wait for the whole hard drive to be archived. Our system is a MV167 running VxWorks with a couple of SCSI 2 differental Elite 9 GB drives controlled by an Interphase 4220 Cougar running the AP Labs aio/ados drivers. FYI Results of our Interphase 4220 + AP Labs aio/ados drivers Test Raw file system: 4.7 MB/sec 1MB buffer DOS file system: 3.8 MB/sec 1MB buffer The system MV167 + VxWorks 5.1.1 + Interphase 4220 + Seagate Hawk 1GB + aio/ados drivers. Steve Buchele Applied Research Labs buchele@arlut.utexas.edu --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Slip/PPP Date: Fri, 27 Jan 1995 22:15:08 GMT From: hjb@netcom.com (Hwa-Jin Bae) Organization: Peaceful Star, Oakland, CA Message-ID: References: <9501251803.AA10068@eng.nmr.varian.com> In article <9501251803.AA10068@eng.nmr.varian.com>, Greg Brissey x6951 wrote: >Dear vxPerts, > >I am look for anyone with experience in using slip and/or ppp >between a SUN 4.1.3 or 2.4 and vxWorks 5.1.1 target. > >I have slip working but it is painfully slow :(, >are there any ways of speeding it ? (if so how?) > yeah... no doubt it's slow. i honestly don't think anyone really did any work on that SLIP driver ever since i hacked it up at WRS. like, nobody bothered to add a simple VJcompression stuff onto it. sigh. >Would ppp be faster and/or better solution. >I have a copy of ppp 1.3 for vxWorks. >Where do I get a copy of ppp for the SUN 4.1.3 ? > well... WRS should by now have a PPP done by an outside help. if it's done right, it should have VJ compressions. it prolly is about 10 times bigger than the slip driver, and i'm not sure of actual performance characteristics. Don Brooks are you there ? ;-) PPP for SUN can be gotten everywhere on internet. ftp.uu.net, for one. >Any suggestion or comments would be greatly appreciated. > >Thanks in Advance, > >Greg Brissey >greg.brissey@nmr.varian.com > --------------------------- Newsgroups: comp.os.vxworks Subject: msgQLib Date: 27 Jan 1995 09:59:00 GMT From: kauer@aut.alcatel.at (Helmut Kauer) Organization: Alcatel Austria AG Keywords: intertask communication, binary data, message queue Message-ID: <3gag54$5oo@ccsun5.aut.alcatel.at> Reply-To: kauer@aut.alcatel.at Hi all, For intertask communication we want to use the message queue library (msgQLib), but want to transfer binary data. The library seems to accept only character strings and stops at a \0 character!! Has anybody an idea, how to handle this problem?? Thanks Helmut !! --------------------------- Newsgroups: comp.os.vxworks Subject: Re: VxSim Date: Sat, 28 Jan 1995 01:52:37 GMT From: davem@micom.com (David Monachello) Organization: Micom Communications Corp. Message-ID: <1995Jan28.015237.8960@micom.com> References: <1995Jan20.200454.3862@picker.com> <1995Jan23.084528.1@corning.com> Sender: news@micom.com whiting_ms@corning.com (Matt Whiting) (whiting_ms@corning.com) wrote: : In article <1995Jan20.200454.3862@picker.com>, heiple@ct.picker.com (HEIPLE, HOLLY) writes: : > Does anybody have any experience using VxSim? We are considering buying it : > based on the literature but I am wondering how much a developer would really : > use it. Why not just use the target? Any comments or words of wisdom? : > : > Thanks, : > Holly Heiple : > : > : > : -- : We tried to use it recently during an evaluation of VxWorks, but found that it : doesn't run under Solaris. This is documented in the release notes, but we : didn't notice that and our sales guy knew we had Solaris and failed to mention : it either. So just be sure that you have SunOS if you do plan to use VxSIM. : Just using the target is OK assuming you have a target! We'd planned to use : VxSIM as our target system was on back order. It could also be advantageous if : you wanted to do development at multiple sites and each site didn't have a : target sytem available. hi, Do you know if wind river plans to rectify the solaris problem? We have used vxsim very successfully in the development of our product and we plan to continue to use it for debugging and simulation even after the project has been fully ported to the target hardware. Using vxsim also allows us to run multiple copies of our product on a single sparc station to test the interaction between them. There does seems to be a few problems with vxsim however. Signals seem to behave strangely and excluding certain vxworks features causes unresolved references. bye, dave monachello --------------------------- Newsgroups: comp.os.vxworks Subject: scsiIoctl() takes 300 microseconds? Date: 27 Jan 1995 18:14:59 GMT From: edh@antlabw1.svl.trw.com (ED) Organization: svl.trw.com Message-ID: <3gbd73$86p@gatekeeper.svl.trw.com> I'm using vxWorks 5.1, the beta version of the SCSI-2 library, and a board with an NCR 710 SCSI chip. I have to use the scsiIoctl() command to send commands to my SCSI device. However, it takes 300 microseconds to do send one -- timed by sending out a hardware pulse just before and after the scsiIoctl(). And I need it to run significantly faster than that. I've tried experimenting with some of the options, but I can't find anything that helps. Has anyone else seen this problem? - -- Ed Haymore edh@antlabw1.esl.com --------------------------- End of New-News digest ********************** From daemon@vxw.ee.lbl.gov Sun Jan 29 04:00:26 1995 From: daemon@vxw.ee.lbl.gov Date: Sun Jan 29 04:00:33 PST 1995 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Sun Jan 29 04:00:16 PST 1995 Subject: Anyone got DEC VxWorks V3.1: ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Anyone got DEC VxWorks V3.1: Date: 29 Jan 1995 13:23:15 +1100 From: jb@werple.mira.net.au (John Birrell) Organization: werple public-access unix, Melbourne Message-ID: <3geu6j$bnm@werple.mira.net.au> We have been trying to obtain DEC VxWorks V3.1 from Digital for months now. The most we can get is the SPD which tells us that the AXP board support kit is part of V3.1. Digital delivered us OSF/1 V3.0 Layered Products dated Sep 94, but this contained V3.0 which only has the Motorola board support kit. Has anyone received V3.1? If so, how did you get it? Are you using the AXP board support kit? If so are there any problems that we can avoid if we ever get it? As you may gather, I'm annoyed. John Birrell CIMlogic Pty Ltd --------------------------- End of New-News digest ********************** From daemon@vxw.ee.lbl.gov Mon Jan 30 04:00:26 1995 From: daemon@vxw.ee.lbl.gov Date: Mon Jan 30 04:00:33 PST 1995 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Mon Jan 30 04:00:16 PST 1995 Subject: Re: Info wanted on off-the-shelf MIL-SPEC embedded controller boards ------------------------------------------------------- Newsgroups: comp.os.vxworks,comp.sys.m68k,alt.industrial.computing,sci.electronics,rec.radio.amateur.space Subject: Re: Info wanted on off-the-shelf MIL-SPEC embedded controller boards Date: Sun, 29 Jan 1995 13:20:42 UNDEFINED From: lacovara@ollie.jsc.nasa.gov (R. C. Lacovara) Organization: GeoControl Systems, Incorporated Message-ID: References: <3fi0d6$hbr@CUBoulder.Colorado.EDU> <3gap36$p9f@kocrsv08.delcoelect.com> In article <3gap36$p9f@kocrsv08.delcoelect.com> aranders@kosepc01.delcoelect.com (Alan Anderson) writes: >From: aranders@kosepc01.delcoelect.com (Alan Anderson) >Subject: Re: Info wanted on off-the-shelf MIL-SPEC embedded controller boards >Date: 27 Jan 1995 12:31:34 GMT >Pardon the severely off-topic note, but I just had to respond. >In , >lacovara@ollie.jsc.nasa.gov (R. C. Lacovara) writes: >>[note deleted, .sig retained] >>Texas is the largest and grandest State of the United States. It retains the > ^^^^^^^ >>right to fly its flag at the same height as that of the United States. Texas >>must not be abbreviated, like NJ or RI: NJ fits between Houston and San >>Antonio, with RI to spare. Al >Watch out, Texas, or Alaska might split in two and make you the *third* >largest state! I believe that I have made this point in the past, but... I'm not sure that actual Alaskans ever discuss dividing their state. But Texans do not, even as a joke. Dem's fight'n words, pard'ner... In any event, I also maintain that the apparent size of Alaska is due to a consipriacy of map-makers and their projections. Hop-along-Bob... R. C. Lacovara, Ph. D. GeoControl Systems voice: 713 333 2561 Houston, Texas 77058 Please be advised that there exists a probability P that any opinions expressed above are not shared by GeoControl or NASA, at least for 1-(1/Avogadro's number) < P < 1. --------------------------- End of New-News digest ********************** From Glenn_Tanaka@milstar.sp.trw.com Mon Jan 30 14:15:36 1995 From: "Glenn Tanaka" Date: Mon Jan 30 14:15:45 PST 1995 Subject: IO Driver for ACL-II+ Hi there, I am using VxWorks For Windows 5.1 on a X86 target. I was wondering if anyone out there has a driver for an Advanced Communication Link (ACL) II+ serial I/O card. I would appreciate any help. Thank you, Glenn Tanaka TRW Space & Electronics Email: glenn.tanaka@trw.com Ph#: (310) 812.1461 From maden@aco.psi.ch Tue Jan 31 01:27:23 1995 From: "David Maden (PSI, CH-5232 Villigen-PSI)" Date: Tue Jan 31 01:27:31 PST 1995 Subject: RE: Anyone got DEC VxWorks V3.1 John Birrell asks: > Has anyone received V3.1 [of DEC's VxWorks]? Yes, we have. We took delivery of an AXPvme64 earlier this month (January) and then discovered that AXP support was not in the VxWorks Vn3.0 which was the version on our OSF/1 distribution CDROM (OSF/1 Vn 3.0 Rev. 347, I think). Frantic calls to DEC eventually produced a casette of VxWorks Vn3.1 which we installed. Low and behold, AXP support was in it, the make files successfully made a system image and we eventually booted VxWorks over Ethernet into the AXPvme (after getting all the BOOTP/TFTP setup right on the OSF/1 host). Since then, we have received a CDROM of the VxWorks Vn3.1 distribution. I should add that DEC were very helpful in getting us through this startup phase. Hope this helps, David Maden, Paul Scherrer Institute, Villigen, Switzerland. BTW, I wrote a note to document my fight to get the AXPvme booted for the first time which might be useful to others (I am new to the Unix mentality and hence, probably, the fight and hence, probably, the note will not be that useful to others!). Just in case, though, it is accessible on the international DECnet via VAXnotes as Note 29.1 in conference 20550::TIPS_N_TRICKS. The note is a bit long (there's lots of console log reproduced for completeness' sake) to publish here. If you want a copy and can't get it via VAXnotes, e-mail me. From klf@redifon.demon.co.uk Tue Jan 31 03:37:46 1995 From: Kieren Feakes Date: Tue Jan 31 03:37:54 PST 1995 Subject: Re: MVME167 memory problem Hi there, We are using VxWorks version 5.1.1 with a Motorola MVME 167-33A board. If an attempt is made to write to any part of a specific 8 kbyte block of memory near the top of DRAM a bus error occurs (eg by using the VxWorks "m" command). The address of the 8 kbyte block appears at a constant offset from the top of the on-board memory; with 4 Mbytes of DRAM, the block's address is 0x3E6000..0x3E7FFF, with 16 Mbytes the address is 0xFE6000..0xFE7FFF. Does anybody know the reason for this or a way to make this area writable ? Regards, Kieren Feakes ################################################################################ Redifon Technology Phone 0372-376677 Fax 0372-379480 Email klf@redfion.demon.co.uk ################################################################################ From daemon@vxw.ee.lbl.gov Tue Jan 31 04:00:27 1995 From: daemon@vxw.ee.lbl.gov Date: Tue Jan 31 04:00:34 PST 1995 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Tue Jan 31 04:00:18 PST 1995 Subject: Re: What does "workQPanic" mean? Subject: ISA/PCI Synchronous Serial Cards Subject: Re: GNU GDB for VxWorks ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: What does "workQPanic" mean? Date: 30 Jan 1995 13:04:09 -0500 From: mroske@rapnet.sanders.lockheed.com (9240) Organization: Lockheed Sanders Message-ID: <3gj9mp$f6g@rapnet.sanders.lockheed.com> References: Ric Claus (claus@pepiiu1.SLAC.Stanford.EDU) wrote: : workQPanic: Kernel work queue overflow. : after running some 7 or 8 application codes in the context of a "shell", all of : which I wrote. The application codes handle DSP processor boards that all run : It sure seems like VxWorks is running out of some resource, but I can't figure : out what. When I checkStack after every application invocation, I see that all : stacks have been used to no more than half the alloted memory. The INTERRUPT : stack never uses more than 268 bytes out of 1000. I see no signs of anything : being corrupted anywhere. Perhaps task descriptors aren't being recycled? : When I memShow after every application invocation, I see that memory usage is : well bounded and nowhere close to the edge. Memory usage is application : dependent, but if I invoke the same application as the one before the previous : application, memory usage goes back to the same values. As an aside, I was not : expecting this to be true because I understood that there was a memory leak : associated with taskSpawns. Unless memShow does not properly account for all : the memory used, I see no leaks due to tasks being created and properly : finishing (i.e. not taskDeleted, by me at least). : It's hard to look around after workQPanic occurs because it causes a reboot. : I'm also not sure where to look. I've looked in the VxWorks manuals for what : this message might mean, but have come up empty handed. It would be nice if it : (and a bunch of other things) were in the index... : The hardware is a 25 MHz Force 40 with 4 Meg and an Eagle 02 (ethernet and VSB) : running version 5.1.1 VxWorks. We're running the same board with the same VxWorks. I have seen the same error when I try to run a large program directly within the shell. I found out that using the 'sp' command to spawn the program as a task eliminated the error. This gives a default stack size. If you are using taskSpawn to start your application, you might want to try using sp and see if it works differently. Let me know how it works out, because we may run into the same problem when our application is fully developed... Mike Roske mroske@mailgw.sanders.lockheed.com --------------------------- Newsgroups: comp.os.linux.networking,comp.sys.ibm.pc.hardware.networking,comp.os.vxworks Subject: ISA/PCI Synchronous Serial Cards Date: 30 Jan 1995 20:54:00 GMT From: smeg@bnr.ca (Maarten Koning) Organization: Bell-Northern Research Canada Keywords: sync, serial, PCI, ISA Message-ID: <3gjjl8$7ug@bcarh8ab.bnr.ca> Reply-To: smeg@bnr.ca Hello, I am looking for manufacturers of PCI-based or ISA-based synchronous serial interface cards for use in a PC. If you know of any companies that manufacture the interface types listed below, could you please mail me their names and/or phone numbers. Also, if you have any information or opinion on the quality/price/performance of those cards, I would be interested to hear about that too. I am looking for info on synchronous V.11, V.24, V.35, T1/DS1 and E1 cards. I will summarize the results to these groups in one week. Many thanks... Maarten Koning (smeg@bnr.ca) ps. I am *not* looking for information on ISDN, async, ATM or HIPPI cards. pps. I am *not* looking for information on cards with other buses such as VME, SBUS etc. ppps. Please do not send me 'me too' requests, unless I fail to post the summary! ;-) - -- //include1 pgm=disclaimer,parm='my opinions only' Maarten Koning | Internet: smeg@bnr.ca | Phone: (613) 763-8796 BNR Ltd. | This space for rent | FAX: (613) 763-2626 --------------------------- Newsgroups: comp.os.vxworks Subject: Re: GNU GDB for VxWorks Date: 31 Jan 1995 05:19:54 GMT From: synergy@nic.cerf.net (John E. Clark) Organization: CERFnet Dial n' CERF Customer Message-ID: <3gkh9q$2dm@news.cerf.net> References: <3g8mih$1hc@fohnix.metronet.com> <3gbdgr$6qf@news.cerf.net> In article hjb@netcom.com (Hwa-Jin Bae) writes: +In article <3gbdgr$6qf@news.cerf.net>, +John E. Clark wrote: +> +>There are a couple of other 'remote' debugging interfaces there as +>well. One of which will go over a serial line... for the really +>minimally inclined... +> + +this option is not "minimally inclined". this actually works much better, +IMHO. vxgdb never worked properly every time i attempted to use it in the +past and most only for task/app debugging (can't do real low level +debugging), full of bugs in the debugger itself, etc. but YMMV, again. My 'minimally inclined' was refering to a system sans 'ethernet'. As I have never used vxgdb, but rather the 'scratch head' until bald method with vxworks, until finally someone who liked it came along. And low and behold, it's just as well.... --------------------------- End of New-News digest ********************** From mitchell@harpo.aaec.com Tue Jan 31 05:56:26 1995 From: mitchell@harpo.aaec.com Date: Tue Jan 31 05:56:33 PST 1995 Subject: Re: What does "workQPanic" mean? >Ric Claus (claus@pepiiu1.SLAC.Stanford.EDU) wrote: > > >: workQPanic: Kernel work queue overflow. > >: after running some 7 or 8 application codes in the context of a "shell", all >of >: which I wrote. The application codes handle DSP processor boards that all >run > If memory serves correct, this error will be generated if something leaves IACK asserted even after the appropriate IACK cycle is run. This could happen if your DSP card asserts an interrupt and requires some special handshake other than an IACK (such as writing a control register). Some designs two sets of jumpers for interrupts: one selects which IRQ line to drive, the other which IACK code to respond to. If these two settings are inconsistent, the board will leave IACK asserted (and you may get spurrious interrupt exceptions - but they might not be printed before the reboot). Try disabling DSP interrupts. If your DSP boards can also be VME masters, then life can get more complicated. You may want try running your tasks with the DSPs idle or running something "safe" if the DSP boards normally do VME writes (I had DSP boards write to both of a MV147's mailbox registers to signal different events). >: It sure seems like VxWorks is running out of some resource, but I can't >figure >: out what. When I checkStack after every application invocation, I see that >all >: stacks have been used to no more than half the alloted memory. The INTERRUPT >: stack never uses more than 268 bytes out of 1000. I see no signs of anything >: being corrupted anywhere. Perhaps task descriptors aren't being recycled? The work queue is a very low level resource that is involved with the handling of intterupt events, before they are placed in the INTERRUPT context, which is why you didn't see anything wrong. If overflows when things happen too fast for interrupts to be serviced. A high intterupt rate (like a timer interrupting at a 10KHz rate) will also cause this error. > >: When I memShow after every application invocation, I see that memory usage is >: well bounded and nowhere close to the edge. Memory usage is application >: dependent, but if I invoke the same application as the one before the >previous >: application, memory usage goes back to the same values. As an aside, I was >not >: expecting this to be true because I understood that there was a memory leak >: associated with taskSpawns. Unless memShow does not properly account for all >: the memory used, I see no leaks due to tasks being created and properly >: finishing (i.e. not taskDeleted, by me at least). > My understanding is that memory "leaks" will occur if the task is spawned with the context save option, which saves the task info on purpose (so its not really a leak). They also occur if string constants are used on the shell command line. -> sp myTask,"Hello World\n" will "leak" 13 bytes for the string storage (only a pointer to the string goes on the task's stack). >: The hardware is a 25 MHz Force 40 with 4 Meg and an Eagle 02 (ethernet and >VSB) >: running version 5.1.1 VxWorks. Check for VSB interrupts too while your're at it (assuming DSP boards are VSB equipped) - though I don't think anybody uses them. Happy Hunting! Andrew Mitchell | Internet: mitchell@aaec.com Atlantic Aerospace Electronics | tel: (617) 890-4200 470 Totten Pond Road | fax: (617) 890-0224 Waltham MA 02154 USA | Opinions are my own, not AAEC's From daly@aries.sbc.anl.gov Tue Jan 31 06:33:34 1995 From: Bob Daly Date: Tue Jan 31 06:33:41 PST 1995 Subject: RE: Anyone got DEC VxWorks V3.1 I would appreciate info. on booting AXPvme. Thanks for your help Bob Daly Argonne National Lab return address rtd@aps.anl.gov From lia!lia.com!glenn@netcom.com Tue Jan 31 17:05:03 1995 From: glenn@lia.com (Glenn Herteg) Date: Tue Jan 31 17:05:11 PST 1995 Subject: VxWorks/Solaris mounting problem I am experiencing a problem at a client site, and would like to know if anyone else has seen this. The environment is: VxWorks ver 4.0.2 Kernel: PSOS ver 4.1 PHILE ver 2.11 The client has recently attached Solaris 2.3 machines to his network, previously containing just SunOS 4 machines. He has long been able to export filesystems from the VxWorks machine and mount them under SunOS 4. But when he tries to mount them under Solaris 2, he gets an error message like "NFS service not responding". The problem is all the more mysterious because he is still able to mount the filesystem under SunOS 4, and is able to ping and rlogin to the VxWorks machine from Solaris 2; so basic connectivity is assured. We have already run into and fixed the problem with root being, by default, in too many groups (/etc/group; > 8) under Solaris 2, which yielded messages about client authentication failures. But now we are stumped. Any clues out there? Glenn Herteg IA Corporationi glenn@lia.com From thor@thor.atd.ucar.edu Tue Jan 31 23:00:26 1995 From: thor@thor.atd.ucar.edu (Rich Neitzel) Date: Tue Jan 31 23:00:34 PST 1995 Subject: Monthly VxWorks archive posting This is the monthly posting showing the current holdings in the VxWorks Software Archive. To get more detailed infomation send email to: vxworks_archive@ncar.ucar.edu The message body must read: send index send index from vx ------------------------------------------------ VxWorks sources: total 13650 lrwxrwxrwx 1 thor other 5 Dec 16 17:31 README -> index -rw-r--r-- 1 thor other 22132 Sep 20 07:52 ansi.p1 -rw-r--r-- 1 thor other 22717 Sep 20 07:52 ansi.p2 -rw-r--r-- 1 thor other 24174 Sep 20 07:52 ansi.p3 -rw-r--r-- 1 thor other 8108 Sep 20 07:52 ansi.patch1 -rw-r--r-- 1 thor other 37126 Sep 20 07:52 ansilib01 -rw-r--r-- 1 thor other 18913 Sep 20 07:52 ansilib02 -rw-r--r-- 1 thor other 2671 Sep 20 07:52 benchmarks -rw-r--r-- 1 thor other 7168 Sep 20 07:52 bitcnt -rw-r--r-- 1 thor other 11437 Sep 20 07:52 c++builtin.shar -rw-r--r-- 1 thor other 22330 Sep 20 07:52 c++headers.p1 -rw-r--r-- 1 thor other 22775 Sep 20 07:52 c++headers.p2 -rw-r--r-- 1 thor other 29052 Sep 20 07:52 camaclib1 -rw-r--r-- 1 thor other 25095 Sep 20 07:52 camaclib2 -rw-r--r-- 1 thor other 31005 Sep 20 07:52 camaclib3 -rw-r--r-- 1 thor other 37770 Sep 20 07:52 cbench.shar -rw-r--r-- 1 thor other 7371 Sep 20 07:52 cntsem_class.shar -rw-r--r-- 1 thor other 5853 Sep 20 07:52 crc.shar -rw-r--r-- 1 thor other 8917 Sep 20 07:52 deadman.shar -rw-r--r-- 1 thor other 41669 Sep 20 07:52 dhrystones01 -rw-r--r-- 1 thor other 19170 Sep 20 07:52 dirlib01 -rw-r--r-- 1 thor other 25681 Sep 20 07:52 dt1451 -rw-r--r-- 1 thor other 4905 Sep 20 07:53 dup.shar -rw-r--r-- 1 thor other 5944 Sep 20 07:52 fcompress.shar -rw-r--r-- 1 thor other 11561 Sep 20 07:53 flags_class.shar -rw-r--r-- 1 thor other 88109 Sep 20 07:53 flash.zip -rw-r--r-- 1 thor other 44762 Sep 20 07:53 force.p1 -rw-r--r-- 1 thor other 40154 Sep 20 07:53 force.p2 -rw-r--r-- 1 thor other 80491 Sep 20 07:53 force.shar -rw-r--r-- 1 thor other 2453 Sep 20 07:53 gcc+68040 -rw-r--r-- 1 thor other 6106 Sep 20 07:53 getdate -rw-r--r-- 1 thor other 9774 Sep 20 07:53 hkv30extintutil.shar -rw-r--r-- 1 thor rsf 24394 Jan 31 15:04 index -rw-r--r-- 1 thor other 2694 Sep 20 07:53 ivecalloc.shar -rw-r--r-- 1 thor other 35245 Sep 20 07:53 joblib2.p1 -rw-r--r-- 1 thor other 18110 Sep 20 07:53 joblib2.p2 -rw-r--r-- 1 thor other 9079 Sep 20 07:53 lclflag.shar -rw-r--r-- 1 thor other 174829 Sep 20 07:53 lexyacc.tar.gz drwxr-xr-x 2 thor other 512 Sep 20 07:53 libX11 -rw-r--r-- 1 thor rsf 14059 Nov 4 16:03 libg++-2.6.1.patch -rw-r--r-- 1 thor rsf 1302 Nov 23 10:34 libgcc2-2.6.2.patch -rw-r--r-- 1 thor rsf 1862 Dec 29 10:50 libgcc2-2.6.3.patch -rw-r--r-- 1 thor rsf 1222 Nov 4 16:03 libio-2.6.1.patch -rw-r--r-- 1 thor rsf 277 Nov 4 16:03 libobjc.patch drwxr-xr-x 2 thor other 512 Sep 20 07:53 libx11 -rw-rw-r-- 1 ftp other 3515 Nov 2 15:29 loadmeter.shar -rw-r--r-- 1 thor other 10399 Sep 20 07:53 math.shar -rw-r--r-- 1 thor other 11950 Sep 20 07:53 math2 -rw-r--r-- 1 thor other 26655 Sep 20 07:53 monitor.shar -rw-r--r-- 1 thor other 18733 Sep 20 07:53 msgque_class.shar -rw-r--r-- 1 thor other 829713 Sep 20 07:53 ntpv3.1.tar.gz -rw-r--r-- 1 thor other 1010176 Sep 20 07:53 ntpv3.tar.gz -rw-r--r-- 1 thor other 1082 Sep 20 07:53 objc.patch -rw-r--r-- 1 thor other 19422 Sep 20 07:53 ping01 -rw-r--r-- 1 thor other 20494 Sep 20 07:53 pipe.shar -rw-r--r-- 1 thor other 15418 Sep 20 07:53 poolLib -rw-r--r-- 1 thor other 15418 Sep 20 07:53 poollib -rw-r--r-- 1 thor other 13204 Sep 20 07:53 ring.shar -rw-r--r-- 1 thor other 6614 Sep 20 07:53 semCnt -rw-r--r-- 1 thor other 6614 Sep 20 07:53 semcnt -rw-r--r-- 1 thor rsf 48659 Dec 2 10:25 snmp2.01 -rw-r--r-- 1 thor rsf 48658 Dec 2 10:25 snmp2.02 -rw-r--r-- 1 thor rsf 48666 Dec 2 10:25 snmp2.03 -rw-r--r-- 1 thor rsf 48723 Dec 2 10:25 snmp2.04 -rw-r--r-- 1 thor rsf 48869 Dec 2 10:25 snmp2.05 -rw-r--r-- 1 thor rsf 48632 Dec 2 10:25 snmp2.06 -rw-r--r-- 1 thor rsf 48658 Dec 2 10:25 snmp2.07 -rw-r--r-- 1 thor rsf 48652 Dec 2 10:25 snmp2.08 -rw-r--r-- 1 thor rsf 48714 Dec 2 10:25 snmp2.09 -rw-r--r-- 1 thor rsf 48615 Dec 2 10:25 snmp2.10 -rw-r--r-- 1 thor rsf 49185 Dec 2 10:25 snmp2.11 -rw-r--r-- 1 thor rsf 49214 Dec 2 10:25 snmp2.12 -rw-r--r-- 1 thor rsf 49203 Dec 2 10:25 snmp2.13 -rw-r--r-- 1 thor rsf 49239 Dec 2 10:25 snmp2.14 -rw-r--r-- 1 thor rsf 49176 Dec 2 10:25 snmp2.15 -rw-r--r-- 1 thor rsf 49216 Dec 2 10:25 snmp2.16 -rw-r--r-- 1 thor rsf 49182 Dec 2 10:25 snmp2.17 -rw-r--r-- 1 thor rsf 49188 Dec 2 10:25 snmp2.18 -rw-r--r-- 1 thor rsf 49189 Dec 2 10:25 snmp2.19 -rw-r--r-- 1 thor rsf 49200 Dec 2 10:25 snmp2.20 -rw-r--r-- 1 thor rsf 48646 Dec 2 10:25 snmp2.21 -rw-r--r-- 1 thor rsf 48671 Dec 2 10:25 snmp2.22 -rw-r--r-- 1 thor rsf 49229 Dec 2 10:25 snmp2.23 -rw-r--r-- 1 thor rsf 49235 Dec 2 10:25 snmp2.24 -rw-r--r-- 1 thor rsf 49268 Dec 2 10:25 snmp2.25 -rw-r--r-- 1 thor rsf 48682 Dec 2 10:25 snmp2.26 -rw-r--r-- 1 thor rsf 49223 Dec 2 10:25 snmp2.27 -rw-r--r-- 1 thor rsf 48676 Dec 2 10:25 snmp2.28 -rw-r--r-- 1 thor rsf 48818 Dec 2 10:25 snmp2.29 -rw-r--r-- 1 thor rsf 49308 Dec 2 10:25 snmp2.30 -rw-r--r-- 1 thor rsf 49294 Dec 2 10:25 snmp2.31 -rw-r--r-- 1 thor rsf 48634 Dec 2 10:26 snmp2.32 -rw-r--r-- 1 thor rsf 48708 Dec 2 10:26 snmp2.33 -rw-r--r-- 1 thor rsf 49055 Dec 2 10:26 snmp2.34 -rw-r--r-- 1 thor rsf 48724 Dec 2 10:26 snmp2.35 -rw-r--r-- 1 thor rsf 48736 Dec 2 10:26 snmp2.36 -rw-r--r-- 1 thor rsf 48667 Dec 2 10:26 snmp2.37 -rw-r--r-- 1 thor rsf 48626 Dec 2 10:26 snmp2.38 -rw-r--r-- 1 thor rsf 48653 Dec 2 10:26 snmp2.39 -rw-r--r-- 1 thor rsf 14122 Dec 2 10:26 snmp2.40 -rw-r--r-- 1 thor other 2308 Sep 20 07:53 ss1.bnch -rw-r--r-- 1 thor other 41196 Sep 20 07:53 stevie01 -rw-r--r-- 1 thor other 35279 Sep 20 07:53 stevie02 -rw-r--r-- 1 thor other 35278 Sep 20 07:53 stevie03 -rw-r--r-- 1 thor other 35012 Sep 20 07:53 stevie04 -rw-r--r-- 1 thor other 34502 Sep 20 07:53 stevie05 -rw-r--r-- 1 thor other 37476 Sep 20 07:53 stevie06 -rw-r--r-- 1 thor other 30073 Sep 20 07:53 stevie07 -rw-r--r-- 1 thor other 31562 Sep 20 07:53 stevie08 -rw-r--r-- 1 thor other 37360 Sep 20 07:53 stevie09 -rw-r--r-- 1 thor other 20662 Sep 20 07:53 stevie10 -rw-r--r-- 1 thor other 25717 Sep 20 07:53 stevie11 -rw-r--r-- 1 thor other 28075 Sep 20 07:53 stevie12 -rw-r--r-- 1 thor other 31852 Sep 20 07:53 stevie13 -rw-r--r-- 1 thor other 16225 Sep 20 07:53 string.shar -rw-r--r-- 1 thor other 8424 Sep 20 07:53 syslog.shar -rw-r--r-- 1 thor other 15096 Sep 20 07:53 task_class.shar -rw-r--r-- 1 thor other 16171 Sep 20 07:53 taskmon.shar -rw-rw-r-- 1 thor other 417608 Jan 31 14:43 tclvx7.0v5.tar.gz -rw-r--r-- 1 thor other 10523 Sep 20 07:53 tod.shar -rw-r--r-- 1 thor other 19912 Sep 20 07:53 tp41.shar -rw-r--r-- 1 thor other 25790 Sep 20 07:53 ty335.shar -rw-r--r-- 1 thor other 44504 Sep 20 07:53 usrTime.shar -rw-r--r-- 1 thor other 44504 Sep 20 07:53 usrtime.shar -rw-r--r-- 1 thor other 4945 Sep 20 07:53 veclist01 -rw-r--r-- 1 thor other 25814 Sep 20 07:53 vtape.shar -rw-r--r-- 1 thor other 43671 Sep 20 07:53 vwcurses01 -rw-r--r-- 1 thor other 40180 Sep 20 07:53 vwcurses02 -rw-r--r-- 1 thor other 38308 Sep 20 07:53 vwcurses03 -rw-r--r-- 1 thor other 31181 Sep 20 07:53 vwcurses04 -rw-r--r-- 1 thor other 31798 Sep 20 07:53 vwcurses05 -rw-r--r-- 1 thor other 31459 Sep 20 07:53 vwcurses06 -rw-r--r-- 1 thor other 24279 Sep 20 07:53 vwcurses07 -rw-r--r-- 1 thor rsf 4973 Nov 4 16:01 vx_cplusplus -rw-r--r-- 1 thor other 40485 Sep 20 07:53 vxrouted01 -rw-r--r-- 1 thor other 40512 Sep 20 07:53 vxrouted02 -rw-r--r-- 1 thor other 40482 Sep 20 07:53 vxrouted03 -rw-r--r-- 1 thor other 40472 Sep 20 07:53 vxrouted04 -rw-r--r-- 1 thor other 40526 Sep 20 07:53 vxrouted05 -rw-r--r-- 1 thor other 40540 Sep 20 07:53 vxrouted06 -rw-r--r-- 1 thor other 40552 Sep 20 07:53 vxrouted07 -rw-r--r-- 1 thor other 22675 Sep 20 07:53 vxrouted08 -rw-r--r-- 1 thor other 29720 Sep 20 07:53 vxrsh.p1 -rw-r--r-- 1 thor other 26002 Sep 20 07:53 vxrsh.p2 -rw-r--r-- 1 thor other 13713 Sep 20 07:53 vxrsh.p3 -rw-r--r-- 1 thor other 4702 Sep 20 07:53 wdog_class -rw-r--r-- 1 thor other 40860 Sep 20 07:53 xmodem.01 -rw-r--r-- 1 thor other 40463 Sep 20 07:53 xmodem.02 -rw-r--r-- 1 thor other 40403 Sep 20 07:53 xmodem.03 -rw-r--r-- 1 thor other 40451 Sep 20 07:53 xmodem.04 -rw-r--r-- 1 thor other 40721 Sep 20 07:53 xmodem.05 -rw-r--r-- 1 thor other 9333 Sep 20 07:53 xmodem.06