From enrico.ferro@elsag.it Mon Jun 1 02:50:49 1998 From: Enrico Ferro Date: Mon Jun 1 02:50:53 PDT 1998 Subject: Fieldbus Foundation Protocol Stack under VxWorks Hi guys, is there anybody out there who knows if some company has developed a Fieldbus Foundation Protocol Stack running under VxWorks, possibly based on a 68K Motorola CPU? Thanks in advance, Enrico Ferro Elsag Bailey Hartmann & Braun S.p.A. Via Hermada 6, 16154 Genova - Italy Tel. 0039-10-6586222 or 6586200 Home 0039-10-6517797 Work E-mail Enrico.Ferro@elsag.it Home e-mail address Enrico.Ferro@usa.net From rcw@DynRes.com Mon Jun 1 11:39:00 1998 From: rcw@DynRes.com Date: Mon Jun 1 11:39:04 PDT 1998 Subject: binary semaphore in interrupt handler VXWORKS TORNADO MVME2604 sysAuxClk I'm having trouble getting an interrupt handler to work. The problem seems to be a semGive() in the handler. The routine works when I have that commented out but causes an immediate reboot with it in. Can anyone shed some light on this problem? By the way, I can get the higher level code to work by *manually* giving the semaphore. Here is my basic code: SEM_ID SchedulerRunSemaphore; int AuxTimerISRCounter = 0; int AuxTimerISR (int dummy) { #if 1 semGive(SchedulerRunSemaphore); #endif AuxTimerISRCounter++; return 0; } int SchedulerInit (void) { SchedulerRunSemaphore = semBCreate(SEM_Q_FIFO, SEM_EMPTY); #if 1 taskSpawn("tScheduler", // name 70, // priority NULL, // options 10000, // stack size (FUNCPTR) Scheduler, // function pointer NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL ); #endif #if 1 sysAuxClkDisable(); sysAuxClkConnect ((FUNCPTR)&AuxTimerISR, 0); sysAuxClkRateSet (SCHEDULER_CLOCK_RATE); sysAuxClkEnable (); #endif return 1; } int SchedulerRunCount = 0; void Scheduler ( void ) { int i; while (1) { SchedulerRunCount++; semTake(SchedulerRunSemaphore, WAIT_FOREVER); // wait for activation by clock } } From mikem@sdlabs.com Mon Jun 1 15:35:40 1998 From: Michael Morrison Date: Mon Jun 1 15:35:43 PDT 1998 Subject: Re: binary semaphore in interrupt handler the vxWorks Users Group Exploder wrote: > Submitted-by rcw@DynRes.com Mon Jun 1 11:39:00 1998 > Submitted-by: rcw@DynRes.com > > VXWORKS > TORNADO > > MVME2604 > sysAuxClk > > I'm having trouble getting an interrupt handler to work. The problem seems to be a semGive() in the handler. The routine works when I have that commented out but causes an immediate reboot with it in. > > Can anyone shed some light on this problem? > > By the way, I can get the higher level code to work by *manually* giving the semaphore. > > Here is my basic code: > > SEM_ID SchedulerRunSemaphore; > int AuxTimerISRCounter = 0; > > int AuxTimerISR (int dummy) > { > #if 1 > semGive(SchedulerRunSemaphore); > #endif > AuxTimerISRCounter++; > return 0; > } > > Is your timer ISR running before 'SchedulerInit' inits the semaphore? From yleduc@oceanet.fr Tue Jun 2 00:34:05 1998 From: Yves LEDUC Date: Tue Jun 2 00:34:09 PDT 1998 Subject: Tornado - vxworks I am a beginner in Tornado - VxWorks developpement. I works on MVME 162lx terget. Developpement are realise in C and ASM 68k. Somebody do you have same experience? Yes, please can you send me some C or ASM programs ? Thanks Yves LEDUC Industrial Computing Teacher LPT de la Salle NANTES - FRANCE yleduc@oceanet.fr From daemon@csg.lbl.gov Tue Jun 2 04:03:01 1998 From: daemon@csg.lbl.gov Date: Tue Jun 2 04:03:04 PDT 1998 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Tue Jun 2 04:02:59 PDT 1998 Subject: Re: c++ exceptions ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: c++ exceptions Date: Sun, 31 May 1998 18:22:22 +0100 From: Rachel-Louise Koktava Organization: KittyCrap Message-ID: <3KAfOBAOHZc1Ew6P@severins-child.demon.co.uk> References: <2remxguwzm.fsf@syseng.anu.edu.au> Hi thanks for the info, its always useful to know. Gcc 2.7.2 with Tornado is giving me such headaches and hassells. This would explain some of the code bloat since we have been using the objectspace stl which require exception handling. I agree about the template problems. WRS tell me that they will have a beta test compiler based on 2.8 at the end of a year with shippen planned for Q2 of 1999. They say this should address many of the problems of the 2.7.2 compiler, particulary on sparc-wrs-vxworks combinations. R-L > >---- >oisin hurley >p.s.e. - -- Rachel-Louise Koktava KittyCrap@severins-child.demon.co.uk "My path is strewn with cow pats from the Devils own Satanic Herd" --------------------------- End of New-News digest ********************** From ekrieg@eng.kns.com Tue Jun 2 05:30:54 1998 From: ekrieg@eng.kns.com (Eric Krieg) Date: Tue Jun 2 05:31:01 PDT 1998 Subject: Fieldbus Foundation Protocol Stack under VxWorks >From enrico.ferro@elsag.it Mon Jun 1 02:50:49 1998 >From: Enrico Ferro >Date: Mon Jun 1 02:50:53 PDT 1998 >Subject: Fieldbus Foundation Protocol Stack under VxWorks> >Hi guys, > >is there anybody out there who knows if some company has developed a >Fieldbus Foundation >Protocol Stack running under VxWorks, possibly based on a 68K Motorola CPU? > >Thanks in advance, > > >Enrico Ferro Enrico, Fieldbus Foundation is meant to run on any RTOS. In the 31kbs version you don't have to respond very quickly. If you use one of the higher performance MAU chips, your processor overhead is less. I don't think any of the protocol stack vendors implemented it in vxworks, but I don't believe a port would be that difficult, there is a relatively short list of OS interfaces needed to get a connection up depending on whether you want to make a slave or host. The stack vendors I know sold them at one time were: National Instruments, Ship Star and Smar. There is also a new product that tests if you conform to spec. Eric Krieg eric@voicenet.com =================== From ptp@mclean.sparta.com Tue Jun 2 05:54:13 1998 From: "Patrick T. Pinkowski" Date: Tue Jun 2 05:54:16 PDT 1998 Subject: VxWorks FDDI SENS MUX/END Device Driver For those interested in FDDI and the WindRiver SENS release (BSD 4.4), SPARTA has developed a VxWorks Device Driver for the Interphase 4511 PMC FDDI Card using the WindRiver SENS release and the MUX/END device driver model. The device driver is immediately availiable on the Motorola MV2604. SPARTA provides the source code with each copy of the device driver, so the driver can easily be ported to other platforms that support the WindRiver SENS release. Feel free to contact me if you need additional information. Regards Pat -- -------------------------------------------------- /\ Patrick T. Pinkowski, Senior Engineer /##\ ptp@mclean.sparta.com /####\ 703.448.1683 x228 /####/-- /####/---- SPARTA, Inc. -\####\__--- 7926 Jones Branch Drive, Suite 900 ---\##\ /---- McLean, Virginia 22102 ----\##\/\---- (703)448-0210 (Main) ----\####\-- (703)893-5494 (Facsimile) ----\####/ http://www.mclean.sparta.com ---/###/ -/###/ SPARTA \##/ ~~~~~~ \/ Pride In Performance From john.l.ahrens@boeing.com Tue Jun 2 09:23:45 1998 From: "John. L. Ahrens" Date: Tue Jun 2 09:23:49 PDT 1998 Subject: pc486 connect to target server? I can't seem to get my pc486 vxWorks target (running on a Dell P133) to connect to my target server on Sparc/Solaris. I get the following message: Killing this logfile viewer will not kill the Target Server. Current Log information are available in the file: /export/home/john/.wind/launchLog.vxworks1 Command: tgtsvr 136.240.44.106 -L -V -n vxworks1 -c /opt/wind/target/config/pc486/vxWorks.st tgtsvr (vxworks1@bigcheese): Tue Jun 2 09:16:44 1998 License request... authorized on host 'bigcheese'. Attaching backend... succeeded. Connecting to target agent... Error: rpccore backend client RPC: Timed out failed. Any suggestions as to what I might be doing wrong? TIA. -- I don't speak for my employer. | You can't have an impact Test Systems Research & Development | without expecting a collision. john.l.ahrens@boeing.com | From kimberly@spk.hp.com Tue Jun 2 09:32:47 1998 From: Kim Middleton Date: Tue Jun 2 09:32:58 PDT 1998 Subject: VxWorks FTP with Web Browsers? Hi - I've been experimenting with FTP and VxWorks. As you probably know, using a web browser (IE/Netscape whatever), you can put an ftp: address in the URL window in order to see a directory on your server. I can always ftp from a terminal window to my chassis, but I can't get it to work from my browser (which is the goal). Anyone know why or have any ideas? Kim Kim Middleton (formerly McCulla) Hewlett-Packard Company kimberly@spk.hp.com 24001 E. Mission Ave Voice: (509) 921-3817 Liberty Lake, WA 99019 FAX: (509) 921-3991 From bawilson@llnl.gov Tue Jun 2 11:37:57 1998 From: "Bruce A. Wilson" Date: Tue Jun 2 11:38:00 PDT 1998 Subject: Re: pc486 connect to target server? John, I've seen this behavior when I forgot to add the target name into the host's (bigcheese, in your case) /etc/hosts file, and/or the /etc/hosts.equiv (or /.rhosts). --Bruce Wilson >Submitted-by john.l.ahrens@boeing.com Tue Jun 2 09:23:45 1998 >Submitted-by: "John. L. Ahrens" > >I can't seem to get my pc486 vxWorks target (running on a Dell P133) to >connect to my target server on Sparc/Solaris. I get the following >message: >...snip... >tgtsvr (vxworks1@bigcheese): Tue Jun 2 09:16:44 1998 > License request... authorized on host 'bigcheese'. > Attaching backend... succeeded. > Connecting to target agent... > Error: rpccore backend client RPC: Timed out >failed. > >Any suggestions as to what I might be doing wrong? > >TIA. >-- >I don't speak for my employer. | You can't have an impact >Test Systems Research & Development | without expecting a collision. >john.l.ahrens@boeing.com | Bruce A. Wilson U.C. Lawrence Livermore National Lab P.O. Box 808 (L-054) Livermore, CA 94550 From cgrames@mdc.com Tue Jun 2 13:18:27 1998 From: Charlie Grames Date: Tue Jun 2 13:18:31 PDT 1998 Subject: binary semaphore in interrupt handler -Reply RCW, I took your test case, added the following to the beginning: #include "semLib.h" #include "sysLib.h" #include "taskLib.h" #define SCHEDULER_CLOCK_RATE 60 void Scheduler(void); and removed the comments (I made it C rather than C++). I then loaded your test case and ran it from WindSh with: SchedulerInit It ran without error (i.e., AuxTimerISRCounter and SchedulerRunCount both incremented at 60 Hz). How are you defining SCHEDULER_CLOCK_RATE? My choice of 60 Hz was arbitrary, but it cannot be less than 60 or more than 5000. Charlie Grames The Boeing Company (314) 233-1956 Charles.R.Grames@boeing.com >>> the vxWorks Users Group Exploder 06/01/98 01:39pm >>> Submitted-by rcw@DynRes.com Mon Jun 1 11:39:00 1998 Submitted-by: rcw@DynRes.com VXWORKS TORNADO MVME2604 sysAuxClk I'm having trouble getting an interrupt handler to work. The problem seems to be a semGive() in the handler. The routine works when I have that commented out but causes an immediate reboot with it in. Can anyone shed some light on this problem? By the way, I can get the higher level code to work by *manually* giving the semaphore. Here is my basic code: SEM_ID SchedulerRunSemaphore; int AuxTimerISRCounter = 0; int AuxTimerISR (int dummy) { #if 1 semGive(SchedulerRunSemaphore); #endif AuxTimerISRCounter++; return 0; } int SchedulerInit (void) { SchedulerRunSemaphore = semBCreate(SEM_Q_FIFO, SEM_EMPTY); #if 1 taskSpawn("tScheduler", // name 70, // priority NULL, // options 10000, // stack size (FUNCPTR) Scheduler, // function pointer NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL ); #endif #if 1 sysAuxClkDisable(); sysAuxClkConnect ((FUNCPTR)&AuxTimerISR, 0); sysAuxClkRateSet (SCHEDULER_CLOCK_RATE); sysAuxClkEnable (); #endif return 1; } int SchedulerRunCount = 0; void Scheduler ( void ) { int i; while (1) { SchedulerRunCount++; semTake(SchedulerRunSemaphore, WAIT_FOREVER); // wait for activation by clock } } From cgrames@mdc.com Tue Jun 2 17:08:51 1998 From: Charlie Grames Date: Tue Jun 2 17:08:54 PDT 1998 Subject: Auxiliary Clock Rate on MVME2604 VxWorks 5.3.1 MVME2604 1.1/4 I mistyped earlier when I said the rate range for the auxiliary timer on the MVME2604 was 60 to 5000 Hz. It is actually *40* to 5000 Hz. Sorry for the goof. Charlie Grames The Boeing Company (314) 233-1956 Charles.R.Grames@boeing.com From nicong@bigfoot.com Tue Jun 2 17:48:05 1998 From: mac Date: Tue Jun 2 17:48:08 PDT 1998 Subject: Any contracts available in SF Bay Area? My humble apologies to all for spamming the list with this, however, I'm looking for a job :) Anyone out there in the SF Bay Area looking for a VxWorks contractor? If so.. I'm free and looking. Regards, and again, my apologies :) brian -------------------------------------------------------- "It must be marvelous to be a man and just cheerfully assume I was superior. Better still, I could be an Irishman and have all the priveleges of being male without giving up the right to be wayward, temperamental and an appealing minority" Katherine Whitehorn From daemon@csg.lbl.gov Wed Jun 3 04:03:37 1998 From: daemon@csg.lbl.gov Date: Wed Jun 3 04:03:40 PDT 1998 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Wed Jun 3 04:03:35 PDT 1998 Subject: Re: block transfert on VME with a mv2301/vxWorks5.3.1 ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: block transfert on VME with a mv2301/vxWorks5.3.1 Date: Tue, 02 Jun 1998 18:35:04 GMT From: gstefkov@sprynet.com Organization: Deja News - The Leader in Internet Discussion Message-ID: <6l1gko$9r9$1@nnrp1.dejanews.com> References: <35740876.69E9@ipnsun5.in2p3.fr> I dont know exactly about the mv2301 board. However I have been working with a custom board that has a 604e, MPC106 and Universe II on it. Furthermore I dont use any board support packages or OSes - I write BSPs. In my case the 60X bus and PCI bus are not the same address. With MPC106 map A contiguous, PCI Address 0x80000000 maps to 60X address 0x0. Also 60X address 0xC0000000 maps to PCI address 0x0. I recommend a little more research on your memory map and address translations. Also beware the Tundra Universe devices. They have been giving my headaches for 2 years. Ive gotta wierd problem with the DMA right now. I'll post it when its resolved. You say the code doesn't work. Whats the Error? PCI Target abort? Try oring adressepci with 0x80000000 for the DLA reg. In article <35740876.69E9@ipnsun5.in2p3.fr>, grave wrote: > > Hi all vxWorkers ! > > I'm trying to do block transfert with my mv2301 board. > Here is the code I have done. > I'm in EXTENDED_VME and it seem that the CPU and PCI mapping are the > same. Can somebody help me to find what's the problem in my code ? > > Many thanks in advance, > > Xavier Grave > ---------------------------------- > #include "vxWorks.h" > #include "stdio.h" > #include "sysLib.h" > #include "config.h" > #include "universe.h" > > #ifndef UNIV_IN_LONG > #define UNIV_IN_LONG(adr,pVal) sysPciRead32((UINT32)(adr),(UINT32 > *)(pVal)); > #endif > > #ifndef UNIV_OUT_LONG > #define UNIV_OUT_LONG(adr,val) > sysPciWrite32((UINT32)(adr),(UINT32)(val)); > #endif > > IMPORT void sysOutByte (ULONG, UCHAR); > IMPORT UCHAR sysInByte (ULONG); > IMPORT void sysPciRead32 (UINT32, UINT32 *); > IMPORT void sysPciWrite32 (UINT32, UINT32); > > IMPORT void sysUniverseReset (void); > > void block_transfert(void) > { > UINT32 temp_data; > long *adressePci,test,i; > > sysUniverseInit(); > > UNIV_IN_LONG(UNIVERSE_DGCS,&temp_data); /* lecture du registre DMA */ > printf("DMA %x\n",temp_data); > UNIV_IN_LONG(UNIVERSE_DCTL,&temp_data); /* lecture du registre CTL */ > printf("CTL %x\n",temp_data); > UNIV_IN_LONG(UNIVERSE_DTBC,&temp_data); /* lecture du registre TBC */ > printf("TBC %x\n",temp_data); > UNIV_IN_LONG(UNIVERSE_DLA,&temp_data); /* lecture du registre DLA */ > printf("DLA %x\n",temp_data); > UNIV_IN_LONG(UNIVERSE_DVA,&temp_data); /* lecture du registre DVA */ > printf("DVA %x\n",temp_data); > > /* initialisation des registres pour le block transfert */ > UNIV_OUT_LONG(UNIVERSE_LINT_EN,0); /* met tout a zero dont DMA dans PCI > Interrupt Enable Register*/ > UNIV_OUT_LONG(UNIVERSE_DGCS,0) /* met tout a zero dont CHAIN et > INT_DONE dans DMA General Control/Status Register */ > temp_data = 0; > temp_data |= DCTL_VDW_32; > temp_data |= DCTL_VAS_A32; > temp_data |= DCTL_PGM_DATA; > temp_data |= DCTL_SUPER_SUP; > temp_data |= DCTL_VCT_EN; > /*temp_data |= DCTL_LD64EN;*/ > temp_data |= DCTL_L2V;/* direction VME vers PCI */ > UNIV_OUT_LONG(UNIVERSE_DCTL,temp_data); > > temp_data=1024*sizeof(long); > UNIV_OUT_LONG(UNIVERSE_DTBC,temp_data); > adressePci = (long *) cacheDmaMalloc(1024*sizeof(long)); > for(i=0;i<1024;i++) > { > adressePci[i]=0xdada; > } > printf("adresse memoire contigue %x\n",adressePci); > temp_data=(long)adressePci; > UNIV_OUT_LONG(UNIVERSE_DLA,temp_data); > temp_data=0x122a0000; > UNIV_OUT_LONG(UNIVERSE_DVA,temp_data); > > /* relecture des registres */ > UNIV_IN_LONG(UNIVERSE_DGCS,&temp_data); /* lecture du registre DMA */ > printf("DMA %x\n",temp_data); > UNIV_IN_LONG(UNIVERSE_DCTL,&temp_data); /* lecture du registre CTL */ > printf("CTL %x\n",temp_data); > UNIV_IN_LONG(UNIVERSE_DTBC,&temp_data); /* lecture du registre TBC */ > printf("TBC %x\n",temp_data); > UNIV_IN_LONG(UNIVERSE_DLA,&temp_data); /* lecture du registre DLA */ > printf("DLA %x\n",temp_data); > UNIV_IN_LONG(UNIVERSE_DVA,&temp_data); /* lecture du registre DVA */ > printf("DVA %x\n",temp_data); > > /* lancement block transfert */ > UNIV_IN_LONG(UNIVERSE_DGCS,&temp_data); /* lecture du registre DMA */ > temp_data |= DGCS_GO; > UNIV_OUT_LONG(UNIVERSE_DGCS,temp_data); > test=1; > while(test) > { > UNIV_IN_LONG(UNIVERSE_DGCS,&temp_data); > temp_data &= DGCS_ACT; > if (temp_data==0) test=0; > } > UNIV_IN_LONG(UNIVERSE_DGCS,&temp_data); > temp_data |= DGCS_DONE; > UNIV_OUT_LONG(UNIVERSE_DGCS,temp_data); > printf("%x\n",adressePci[0]); > printf("%x\n",adressePci[511]); > printf("%x\n",adressePci[1023]); > cacheDmaFree(adressePci); > } > - -----== Posted via Deja News, The Leader in Internet Discussion ==----- http://www.dejanews.com/ Now offering spam-free web-based newsreading --------------------------- End of New-News digest ********************** From malegoas@space.honeywell.com Wed Jun 3 05:08:53 1998 From: "Legoas, Marco A (FL51)" Date: Wed Jun 3 05:08:57 PDT 1998 Subject: GNU selective optimization VxWorks/Tornado: Does anyone know if there is a way to selectively enable and disable optimization within a single file?. I am trying to avoid braking this file into smaller ones and selectively optimize them. Thanks! Marco A. Legoas malegoas@space.honeywell.com From krueger@comsys.rockwell.com Wed Jun 3 06:12:26 1998 From: Bill Krueger Date: Wed Jun 3 06:12:30 PDT 1998 Subject: Ethernet card drivers Does anyone know of a vxworks driver for the 3com Etherlink 3c905/3c905b Fast Ethernet PCI network card. If not are there drivers for any cards other than the following: SMC Elite16 Ultra interface Eagle/Novell NE2000 interface 3COM EtherLink III interface INTEL EtherExpress interface SMC 91c9x Ethernet interface SMC Elite16 interface INTEL EtherExpress flash 32 Excelan Ethernet interface CMC Ethernet interface Intel Ether Express PRO100B PCI Thanks in advance. Bill Krueger From jcicon@tc.fluke.com Wed Jun 3 09:02:00 1998 From: Jim Cicon Date: Wed Jun 3 09:02:04 PDT 1998 Subject: ping crashes my app Using ping with my application causes bad things to happen. The post = mortum reveals that a task called 'pingTX' is in a suspended state and = has overflowed its stack. 'pingTX' seems to be spawned by the tornado = kernel as part of the ping process. Here is a stack check on it: pingTX stack info: size =3D 2984 high =3D 2984 margin =3D 0 What can I do? Is there a way to increase the stack size for this task? Jim Cicon From sgutman@globaloop.com Wed Jun 3 10:33:42 1998 From: "Shlomo Gutman" Date: Wed Jun 3 10:33:50 PDT 1998 Subject: Analog Modem Driver Hi, Does anyone know of a source code for Analog Modem Driver for VXWORKS or something that could be ported (Windows,Linux,PSOS...) ? Thank, Shlomo. From cgrames@mdc.com Wed Jun 3 15:08:13 1998 From: Charlie Grames Date: Wed Jun 3 15:08:21 PDT 1998 Subject: Flash Memory Mapping VxWorks 5.3.1 MVME2604 1.1/4 I am having trouble accessing my flash memory at address 0xff000000. Supposedly, I should be able to see all 5 MB of flash in both sets (soldered and socketed) starting at this address. However, I get all f's when I read from these locations. I CAN read the 1 MB flash region mapped to 0xfff00000, but this does me little good if I want to access more than 1 MB of flash. Has anyone else run into this problem, and how did you deal with it? I do have an entry in sysPhysMemDesc for the 16 MB flash region. Charlie Grames The Boeing Company (314) 233-1956 Charles.R.Grames@boeing.com From daemon@csg.lbl.gov Thu Jun 4 04:02:37 1998 From: daemon@csg.lbl.gov Date: Thu Jun 4 04:02:40 PDT 1998 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Thu Jun 4 04:02:35 PDT 1998 Subject: Host simulation on PC ? Subject: Re: VxWorks FTP with Web Browsers? Subject: How disconnect sysClk interrupts? ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Host simulation on PC ? Date: Wed, 03 Jun 1998 13:23:42 +0200 From: Yin Jie Organization: Alcatel Bell Telecom Message-ID: <3575323E.6626@sh.bel.alcatel.be> Hi, VixSim is a port of WxWorks RT system to the UNIX workstation. Host simulation seems not be supported on PC by VixSim ! Does anybody know is there some solutions of VxWorks to make host simulation possible on PC ? Best regards, Yin Jie _______________________________________________________________________ _ V Yin Jie email: yinj@sh.bel.alcatel.be - -----------------+ Dept. VH642 | A L C A T E L | Alcatel Bell Telephone +----------------+ F. Wellesplein 1 Voice : +32-3-240.7205 BELL TELEPHONE B-2018 Antwerpen, BELGIUM Fax : +32-3-240.9908 _______________________________________________________________________ --------------------------- Newsgroups: comp.os.vxworks Subject: Re: VxWorks FTP with Web Browsers? Date: 3 Jun 1998 13:14:00 GMT From: chc@nasa2.ksc.nasa.gov (Charles H. Chapman) Organization: NASA, Kennedy Space Center Message-ID: References: <3574292B.5792@spk.hp.com> On Tue, 02 Jun 1998 09:32:43 -0700, Kim Middleton wrote: >see a directory on your server. I can always ftp from a >terminal window to my chassis, but I can't get it to work >from my browser (which is the goal). > >Anyone know why or have any ideas? One possibility I guess is that your browser always logs in to FTP servers as "anonymous" and automatically sends your email address as the password as most FTP sites request. You might not have the FTP server on VxWorks setup to accept "anonymous" and your email address as a valid login. Chuck --------------------------- Newsgroups: comp.os.vxworks Subject: How disconnect sysClk interrupts? Date: Wed, 3 Jun 1998 15:58:16 +0100 From: "M. Bouri" Organization: EPFL Message-ID: <357556eb.0@epflnews.epfl.ch> Hello, Concerning periodically schedule a task by using sysClkConnect() function. I used it. sysClkConnect connect a routine to the sysClk interrupt for wich the frequency is ajusted by sysClkRateSet. But How to stop the periodically execution of this routine or how to disconnect this routine from the sysClk interrupts. What I did, is to connect another routine I named Nothing() which do nothing int Nothing(void) { return(0) } this effectively stops the last routine but when I unload the object module from the target and reload it (or reload another) , the system lost the connection with the target server when I try to spawn a task. the target server print the following message Error: rpccore backend client Timed out failed. and the widShell display this message : Target connection has been lost. Restarting shell... I thank you for any help. Mohamed. --------------------------- End of New-News digest ********************** From bob_anderson@mcg.mot.com Thu Jun 4 07:35:40 1998 From: Bob Anderson Date: Thu Jun 4 07:35:43 PDT 1998 Subject: RE: Flash Memory Mapping I'm not sure how to get at the flash from vxworks but the 5 meg of flash referenced below is not contiguous. There is a four meg bank located at 0xff000000 and the socketed 1 meg bank starts at 0xff800000. The 1 meg image of flash at 0xfff00000 is the first meg of the 4meg bank or the 1 meg bank depending on the setting of the J10 jumper on the board. Try accessing the banks of flash form PPCBUG to see if you may have a problem with the flash. Bob Anderson System Engineer Motorola Computer Group 410-312-5929 >---------- >From: vxwexplo@lbl.gov >Sent: Wednesday, June 3, 1998 6:08 PM >To: vxworks_users@csg.lbl.gov >Subject: Flash Memory Mapping > >Submitted-by cgrames@mdc.com Wed Jun 3 15:08:13 1998 >Submitted-by: Charlie Grames > >VxWorks 5.3.1 >MVME2604 1.1/4 > >I am having trouble accessing my flash memory at address 0xff000000. >Supposedly, I should be >able to see all 5 MB of flash in both sets (soldered and socketed) starting >at this address. >However, I get all f's when I read from these locations. > >I CAN read the 1 MB flash region mapped to 0xfff00000, but this does me >little good if I want to >access more than 1 MB of flash. Has anyone else run into this problem, and >how did you deal with >it? > >I do have an entry in sysPhysMemDesc for the 16 MB flash region. > >Charlie Grames >The Boeing Company >(314) 233-1956 >Charles.R.Grames@boeing.com > From mcneils@applique.dh.trw.com Thu Jun 4 08:44:53 1998 From: Sean McNeil Date: Thu Jun 4 08:44:56 PDT 1998 Subject: VxWorks PPC, GNU tools, and GNAT Ada95 expertise I've been working on VxWorks for a while now and have managed to = accumulate some knowledge in it's workings. I have worked through = several serious problems with the PPC environment - including support = for more than 32MB, compiler bugs in code generation, Ada95 support with = GNAT, gdb debugging... I'm pinging this group to see what the interests are in a sound = C/C++/Ada95 development environment for VxWorks. I am experienced in = Ada95, C, C++. I have worked with the team that produced GNAT. I've = helped solve key issues with the Linux and Windows versions of GNAT. = I'm currently working on a project using Ada95 and C targeting a PPC = VxWorks VME board. If this (GNAT support for VxWorks et al) is of any = interest to you, please send email. Sean McNeil sean@mcneil.com From Shaochen@aol.com Thu Jun 4 14:53:28 1998 From: Shaochen@aol.com Date: Thu Jun 4 14:53:31 PDT 1998 Subject: Help on memory leaking Hello, everyone, I am a beginner on vxWorks. Recently I wrote a program to store some data into memory (on target) from a file (on host), and later retrieve those data from memory into another file (on host). Basically I just used malloc() function to alloate memory which was freed later in the program using free() function. The data is about 1MB size. After running the module, about 1MB memory went missing so that the module can only be run once or twice before the whole system has to be rebooted. Why is that happening, I did free the memory which was allocated earlier. And the process was very slow especially retrieving the data out of memory, it took 5 minutes sometimes. why? Please help. Your help will be very much appreciated. I can forword souce code to you if needed. Daoshuo From froeber@BBN.COM Thu Jun 4 19:42:41 1998 From: Fred Roeber Date: Thu Jun 4 19:42:45 PDT 1998 Subject: Re: Help on memory leaking On Thu, 4 Jun 1998, Daoshuo wrote: > I am a beginner on vxWorks. Recently I wrote a program to store some data into > memory (on target) from a file (on host), and later retrieve those data from > memory into another file (on host). Basically I just used malloc() function to > alloate memory which was freed later in the program using free() function. The > data is about 1MB size. After running the module, about 1MB memory went > missing so that the module can only be run once or twice before the whole > system has to be rebooted. Why is that happening, I did free the memory which > was allocated earlier. And the process was very slow especially retrieving the > data out of memory, it took 5 minutes sometimes. why? My guess is that the files on the host are being accessed from the target using the "netdrv" driver rather than NFS. If you haven't done anything special on the target (like an nfsMount call) then file IO would default to using the netdrv mechanism which copies a file into memory when it is opened for reading. Until you close the file, that memory isn't freed up. Look into using nfsMount on the target to have your file system accessed using NFS. This only uses small memory buffers for IO and is usually faster. Fred | Fred J Roeber, BBN Systems & Technologies | | 4 John Clarke Road Middletown, RI 02842-5202 | | froeber@bbn.com 401-848-3548 | From Richard.Weiss@oen.siemens.de Fri Jun 5 01:50:10 1998 From: Weiss Richard Date: Fri Jun 5 01:50:14 PDT 1998 Subject: MUX/END support Hi, the SENS for Tornado Release 1.0 offers a new MUX/END interface. In our current project we are planning to use this END functionality, but WindRiver does not support our prossesor architecture yet (a R4750 from IDT). Does anybody know, how big the effort will be to integrate the MUX/END interface into a new BSP? Thanks in advance for any feedback. Richard From lars.antback@pullmax.se Fri Jun 5 02:20:09 1998 From: lars.antback@pullmax.se Date: Fri Jun 5 02:20:12 PDT 1998 Subject: RE: Flash Memory Mapping Hi ! I am currently trying to do the same thing on a MVME2603 1.1/4 with no success. I have the same problem when reading the addresses 0xff000000 and 0xff800000, it results in a lot of 0xff's. I my case I could not read the boot area 0xfff00000, it resulted in a lot of 0x00's. I have checked the initialization of the sysBatDesc and sysPhysMemDesc and found one contradiction between the sysBatDesc and the CPU manual. According to the CPU manual the system behaviour is not defined when one address is found in more than one BAT pair, see chapter 7.4.1 in "PowerPC 603 RISC Microprocessor User's Manual". After changing this setup to remove the overlaps the access to 0xfff00000 works as expected. Unfortunately I have not been able to apply the same fix to the base address of the FLASH banks. I hope this gives you some information for the solution of the problem. Plese let me know if you find a way to access the FLASH banks. I have posted a TSR to WindRiver about this and is waiting for an answer. Lars Antbaeck Pullmax AB Sweden ant@pullmax.se -------------------------------------------------------------------------- -- Submitted-by cgrames@mdc.com Wed Jun 3 15:08:13 1998 Submitted-by: Charlie Grames VxWorks 5.3.1 MVME2604 1.1/4 I am having trouble accessing my flash memory at address 0xff000000. Supposedly, I should be able to see all 5 MB of flash in both sets (soldered and socketed) starting at this address. However, I get all f's when I read from these locations. I CAN read the 1 MB flash region mapped to 0xfff00000, but this does me little good if I want to access more than 1 MB of flash. Has anyone else run into this problem, and how did you deal with it? I do have an entry in sysPhysMemDesc for the 16 MB flash region. Charlie Grames The Boeing Company (314) 233-1956 Charles.R.Grames@boeing.com From daemon@csg.lbl.gov Fri Jun 5 04:00:52 1998 From: daemon@csg.lbl.gov Date: Fri Jun 5 04:00:57 PDT 1998 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Fri Jun 5 04:00:51 PDT 1998 Subject: Re: Flash Memory Mapping ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: Flash Memory Mapping Date: Thu, 04 Jun 1998 12:48:41 GMT From: gstefkov@sprynet.com Organization: Deja News - The Leader in Internet Discussion Message-ID: <6l6539$t41$1@nnrp1.dejanews.com> References: If the board has a MPC106 or MPC105 there is room in the memory map for 16M of ROM from 0xFF000000(4G -16M) to 0xFFFFFFFF(4G-1). If you have 16M on your board, i would say that you ARE correctly reading fs at 0xFF000000. If you have XMb of flash, your flash would start at 4G - X. - -GS In article , Charlie Grames wrote: > > VxWorks 5.3.1 > MVME2604 1.1/4 > > I am having trouble accessing my flash memory at address 0xff000000. Supposedly, I should be > able to see all 5 MB of flash in both sets (soldered and socketed) starting at this address. > However, I get all f's when I read from these locations. > > I CAN read the 1 MB flash region mapped to 0xfff00000, but this does me little good if I want to > access more than 1 MB of flash. Has anyone else run into this problem, and how did you deal with > it? > > I do have an entry in sysPhysMemDesc for the 16 MB flash region. > > Charlie Grames > The Boeing Company > (314) 233-1956 > Charles.R.Grames@boeing.com > - -----== Posted via Deja News, The Leader in Internet Discussion ==----- http://www.dejanews.com/ Now offering spam-free web-based newsreading --------------------------- End of New-News digest ********************** From bobperry@bellatlantic.net Fri Jun 5 07:42:29 1998 From: "Robert L. Perry" Date: Fri Jun 5 07:42:33 PDT 1998 Subject: Re: MUX/END support Richard, Re: the END compliant framework for an unsupported architecture - you'll have to get the muxLib source (at least) and the ethernet family of routines (if you require them) from Wind River. We wrote an END HDLC driver for the 860 (which does not support an ethernet END, but does support SENS) and had a fair headache w/ the learning curve w/ both the driver and making it co-exist w/ the non-END ethernet. I believe that the MU/END interface would be easy to port if you can get your hands on it. Bob Perry >Hi, > >the SENS for Tornado Release 1.0 offers a new MUX/END interface. In our >current project we are planning to use this END functionality, but >WindRiver does not support our prossesor architecture yet (a R4750 from >IDT). > >Does anybody know, how big the effort will be to integrate the MUX/END >interface into a new BSP? > >Thanks in advance for any feedback. >Richard From daemon@csg.lbl.gov Sat Jun 6 04:03:24 1998 From: daemon@csg.lbl.gov Date: Sat Jun 6 04:03:27 PDT 1998 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Sat Jun 6 04:03:22 PDT 1998 Subject: Please help on connect to vxWorks IP Subject: Re: Evaluating VxWorks - Got questions!! ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Please help on connect to vxWorks IP Date: Fri, 05 Jun 1998 18:24:49 -0700 From: Ming Sun Message-ID: <35789A61.3F806A3B@lightera.com> Hi, I need to connect the vxWorks IP stack to my own driver s/w. I can find any pointer in my manuals. Does this cover in some BSP documents or other vxworks document? If you know what I should looking for please email me. Thanks, Ming Sun msun@lightera.com --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Evaluating VxWorks - Got questions!! Date: 06 Jun 1998 06:12:31 GMT From: wscriber@aol.com (WScriber) Organization: AOL http://www.aol.com Message-ID: <1998060606123100.CAA10910@ladder03.news.aol.com> References: <01bd8da9$d7059a60$e1d3b7c7@ajays-laptop> Lot's of CPU's are supported .... go to Wind River Systems product information at: http://www.wrs.com/products/html/bspweb.html and you will find the list of CPU's and Board Support Packages (BSP's) --------------------------- End of New-News digest ********************** From percheron@enertec.avicore.com Sun Jun 7 23:24:17 1998 From: "percheron cecile" Date: Sun Jun 7 23:29:25 PDT 1998 Subject: VxWorks-SCSI-Driver I use a SCSI Card (initiator mode) which has a NCR875chip. I drive my card throught a PCI bus on x86 architecture. I get some problem in initializing the system. Does any body gets a BSP well configured for that kind of application? I can't find what is wrong with my programmation. THANK YOU . From ashoka@teil.soft.net Mon Jun 8 07:00:04 1998 From: K Ashoka Date: Mon Jun 8 07:00:08 PDT 1998 Subject: VxWorks Hi Folks Could anybody explain the utility of netTask in VxWorks. And also pointers to where more information ( library calls ) can be found. Thanks & regards ---------------------------------------------------------------------- Ashoka K, Specialist - D & D, Tata Elxsi (India) Ltd. email : ashoka@teil.soft.net, Tel : 91-80-8410148, Fax 91-80-8410152 ----------------------------------------------------------------------- From cgrames@mdc.com Mon Jun 8 08:33:31 1998 From: Charlie Grames Date: Mon Jun 8 08:33:34 PDT 1998 Subject: RE: Flash Memory Mapping -Reply Lars, I am attaching a response I got from Mike McQuade on this topic. Mike had intended for it to get posted here, but he forgot to remove the trailer from my original message. It turns out that the Falcon chip must be set to enable flash access at address 0xff000000. It is not set that way by default. I twiddled per Mike's suggestions, and my problem was solved. After you read this, you may want to look at the MVME2600 Series Single Board Computer Programmer's Reference Guide. It has a complete description of the Falcon flash control registers. >>> "Mike McQuade" 06/03/98 07:45pm >>> I have seen things like this, in my case, it was a problem booting from flash with an image > 1 meg. I put the following code in bootInit.c: *(UINT *)(FALCON_BASE_ADRS + 0x50) = 0xff0b0006; This sets a bit in the falcon chip, which makes the 4 meg area of flash readable. The way the BSP comes, it only allows you to access the first 1 (one) meg of flash, you have to twiddle this bit to be able to read more than one meg. Try setting this bit and see if you can read your 4 megs of flash. Mike Charlie Grames The Boeing Company (314) 233-1956 Charles.R.Grames@boeing.com >>> the vxWorks Users Group Exploder 06/05/98 04:20am >>> Submitted-by lars.antback@pullmax.se Fri Jun 5 02:20:09 1998 Submitted-by: lars.antback@pullmax.se Hi ! I am currently trying to do the same thing on a MVME2603 1.1/4 with no success. I have the same problem when reading the addresses 0xff000000 and 0xff800000, it results in a lot of 0xff's. I my case I could not read the boot area 0xfff00000, it resulted in a lot of 0x00's. I have checked the initialization of the sysBatDesc and sysPhysMemDesc and found one contradiction between the sysBatDesc and the CPU manual. According to the CPU manual the system behaviour is not defined when one address is found in more than one BAT pair, see chapter 7.4.1 in "PowerPC 603 RISC Microprocessor User's Manual". After changing this setup to remove the overlaps the access to 0xfff00000 works as expected. Unfortunately I have not been able to apply the same fix to the base address of the FLASH banks. I hope this gives you some information for the solution of the problem. Plese let me know if you find a way to access the FLASH banks. I have posted a TSR to WindRiver about this and is waiting for an answer. Lars Antbaeck Pullmax AB Sweden ant@pullmax.se From daemon@csg.lbl.gov Mon Jun 8 11:32:13 1998 From: daemon@csg.lbl.gov Date: Mon Jun 8 11:32:16 PDT 1998 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Mon Jun 8 04:02:28 PDT 1998 Subject: Re: Connecting Windows 95 PPP to VxWorks (Answer and Question) ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: Connecting Windows 95 PPP to VxWorks (Answer and Question) Date: Mon, 08 Jun 1998 05:17:52 GMT From: cvenkat@giasdl01.vsnl.net.in Organization: Deja News - The Leader in Internet Discussion Message-ID: <6lfs60$v65$1@nnrp1.dejanews.com> References: <357ADA0D.33B3FE5F@netcom.ca> Why do you need the PPP tasks to always be running - why don't you simply let them die when a PPP session is finished ? What you need to do instead, IMHO, is to start a serial port listener task, and as soon as it detects a connect, close the serial port and call pppInit(). When the PPP session terminates, thereby killing the PPP tasks, restart the serial port listener task to detect the next connect and repeat the whole process. listenForConnections { while ( 1 ) { /*Use select() etc. to see if something came in on the serial port*/ /*If yes, read the port and see if its the Win95 magic string*/ /*If yes...*/ { /*close the serial port*/ pppInit(...); return; } } } pppDisconnectHook() { taskSpawn(...,listenForConnections,....) } In article <357ADA0D.33B3FE5F@netcom.ca>, kddoig@netcom.ca wrote: > > In order to connect up Windows 95 to VxWorks via PPP, I've had to add a > connect hook that strips the leading HELLO 1234556 messages that Win 95 > sends. > > In order to have the PPP daemon restarted after the PC closes the link, > I've also had to add a task that waits for my disconnect hook to tell it > to restart PPP. > > So 3 questions that I would appreciate if anyone could shed light on. > > 1. Does this hack/approach also work for Windows NT PPP. i.e. waiting > for some characters and sending a CR > > 2. Is there an easier way of re-starting the PPP session on the unit ? > > 3. Is there a technical note or patch from WindRiver that discusses how > to make VxWorks PPP work with Windows ? (shouldn't there be ? oops thats > #4) > > Source included, unsupported, for discussion or reuse, free of charge. > > Thanks, > Kel Doig > Cambrian Systems Corp > - -----== Posted via Deja News, The Leader in Internet Discussion ==----- http://www.dejanews.com/ Now offering spam-free web-based newsreading --------------------------- End of New-News digest ********************** From lemonas@syr.lmco.com Mon Jun 8 15:14:15 1998 From: "Andre S. Lemon" Date: Mon Jun 8 15:14:18 PDT 1998 Subject: Help building usrTime Lib? In building usrTime library into our vxWorks 5.3.1 system for the MVME- 2604, the generic instructions included with the distribution say: "7. Edit the LIB_EXTRA macro in MakeSkel to include usrTime.a, in whatever directory you built it." My question: Where is this MakeSkel file found? None of my vxworks colleagues have heard of this file. When I used unix "find" it returned multiple paths to different MakeSkel files. eg. /3rdparty/tornado1.0.1/target/unsupported/config/mv131/MakeSkel /3rdparty/tornado1.0.1/target/unsupported/config/mv141/MakeSkel /3rdparty/tornado1.0.1/target/unsupported/config/mv143/MakeSkel All were found under /3rdparty/tornado1.0.1/target/unsupported/config/some_dir/MakeSkel Nothing was found for the MVME-2604. The very next step is: "8. For VxWorks 5.1.1 and later: Edit the LIBS macro in $VW/h/make/makeTarget2 to place LIB_EXTRA ahead of the VxWorks libraries. This is due to a bug (SPR 3068) in usrLib.c that references localtime_r(), despite the the #undef INCLUDE_ANSI_TIME above. This step insures that our localtime_r() is referenced instead." My question is: Where is this makeTarget2 found or is it referring to in our case make.PPC604? In this case there is no LIBS or LIB_EXTRA in this file. Should I copy the appropriate lines from an existing file eg. make.PPC403, which has: LIBS = $(LIB_EXTRA) $(TGT_DIR)/lib/lib$(CPU)$(TOOL)vx.a Thanks in advance for your response Andre Lemon Please also reply to lemonas@syr.lmco.com From cgrames@mdc.com Tue Jun 9 07:16:28 1998 From: Charlie Grames Date: Tue Jun 9 07:16:32 PDT 1998 Subject: RE: Flash Memory Mapping -Reply VxWorks 5.3.1 MVME2604 1.1/4 Lars, I am attaching a response I got from Mike McQuade on this topic. Mike had intended for it to get posted here, but he forgot to remove the trailer from my original message. It turns out that the Falcon chip must be set to enable flash access at address 0xff000000. It is not set that way by default. I twiddled per Mike's suggestions, and my problem was solved. After you read this, you may want to look at the MVME2600 Series Single Board Computer Programmer's Reference Guide. It has a complete description of the Falcon flash control registers. >>> "Mike McQuade" 06/03/98 07:45pm >>> I have seen things like this, in my case, it was a problem booting from flash with an image > 1 meg. I put the following code in bootInit.c: *(UINT *)(FALCON_BASE_ADRS + 0x50) = 0xff0b0006; This sets a bit in the falcon chip, which makes the 4 meg area of flash readable. The way the BSP comes, it only allows you to access the first 1 (one) meg of flash, you have to twiddle this bit to be able to read more than one meg. Try setting this bit and see if you can read your 4 megs of flash. Mike Charlie Grames The Boeing Company (314) 233-1956 Charles.R.Grames@boeing.com >>> the vxWorks Users Group Exploder 06/05/98 04:20am >>> Submitted-by lars.antback@pullmax.se Fri Jun 5 02:20:09 1998 Submitted-by: lars.antback@pullmax.se Hi ! I am currently trying to do the same thing on a MVME2603 1.1/4 with no success. I have the same problem when reading the addresses 0xff000000 and 0xff800000, it results in a lot of 0xff's. I my case I could not read the boot area 0xfff00000, it resulted in a lot of 0x00's. I have checked the initialization of the sysBatDesc and sysPhysMemDesc and found one contradiction between the sysBatDesc and the CPU manual. According to the CPU manual the system behaviour is not defined when one address is found in more than one BAT pair, see chapter 7.4.1 in "PowerPC 603 RISC Microprocessor User's Manual". After changing this setup to remove the overlaps the access to 0xfff00000 works as expected. Unfortunately I have not been able to apply the same fix to the base address of the FLASH banks. I hope this gives you some information for the solution of the problem. Plese let me know if you find a way to access the FLASH banks. I have posted a TSR to WindRiver about this and is waiting for an answer. Lars Antbaeck Pullmax AB Sweden ant@pullmax.se From bobh@mbari.org Tue Jun 9 08:35:59 1998 From: Bob Herlien Date: Tue Jun 9 08:36:02 PDT 1998 Subject: Re: Help building usrTime Lib? the vxWorks Users Group Exploder wrote: > > Submitted-by lemonas@syr.lmco.com Mon Jun 8 15:14:15 1998 > Submitted-by: "Andre S. Lemon" > > In building usrTime library into our vxWorks 5.3.1 system for the MVME- > 2604, the generic instructions included with the distribution say: > > "7. Edit the LIB_EXTRA macro in MakeSkel to include usrTime.a, in > whatever > directory you built it." These directions were for VxWorks 5.1.1 (and 5.1). The "make" procedure and file structure changed considerably between 5.1 and 5.2 (or was it for 5.3?). In any case, although I have a version we're using in-house for Tornado, I haven't released it because I haven't documented it. And, as always, I'm swamped with a project unrelated to VxWorks right now. --------------------------------------------------------------- Bob Herlien "If we knew what we were doing, it would bobh@mbari.org not be called research, would it?" MBARI --Albert Einstein Monterey Bay Aquarium Research Institute From jcicon@tc.fluke.com Tue Jun 9 09:36:49 1998 From: Jim Cicon Date: Tue Jun 9 09:36:53 PDT 1998 Subject: RE: ping crashes my app You hit the nail straight on the head. We have implemented SENS and I am finishing up a frame relay ENDs. Pinging through my frame relay ends works most of the time, but sometimes crashes. I suspect some type of race condition or perhaps static data in the pingTx task. Jim Cicon ---------- From: Steve James[SMTP:ste@ply.wg.com] Sent: Tuesday, June 09, 1998 4:44 AM To: jcicon@tc.fluke.com Subject: Re: ping crashes my app the vxWorks Users Group Exploder wrote: > Submitted-by jcicon@tc.fluke.com Wed Jun 3 09:02:00 1998 > Submitted-by: Jim Cicon > > Using ping with my application causes bad things to happen. The post > = > mortum reveals that a task called 'pingTX' is in a suspended state and > = > has overflowed its stack. 'pingTX' seems to be spawned by the tornado > = > kernel as part of the ping process. Here is a stack check on it: > > pingTX stack info: > size =3D 2984 > high =3D 2984 > margin =3D 0 > > What can I do? Is there a way to increase the stack size for this > task? > --snip-- Hello, Your problem sounds suspiciously familiar. I haven't had a problem with ping itself, but I noticed only a few days ago that on our target tNetTask and tWdbTask are dangerously low on stack margin. It's down to about 100 bytes -- too close for comfort. I have a Support enquirey active on this. I've asked if these stack sizes can be increased. I suspect that this tight margin may be because we've just integrated SENS. We have an x86 target. What are you using? -- --Steve James, Senior Design Eng.--............................ --Wandel & Goltermann Ltd........--.+44 (0)1752 765367 (vmail). --Eurotech House, Burrington Way.--.+44 (0)1752 791101 (FAX)... --Plymouth, Devon. PL5 3LZ. U.K..--.+44 (0)1752 772773 (switcb) From jford@sadira.gb.nrao.edu Tue Jun 9 11:05:23 1998 From: John Ford Date: Tue Jun 9 11:05:28 PDT 1998 Subject: RE: ping crashes my app the vxWorks exploder said: > > > > Using ping with my application causes bad things to happen. The post > > = > > mortum reveals that a task called 'pingTX' is in a suspended state and > > = > > has overflowed its stack. 'pingTX' seems to be spawned by the tornado > > = > > kernel as part of the ping process. Here is a stack check on it: > > > > pingTX stack info: > > size =3D 2984 > > high =3D 2984 > > margin =3D 0 > > > > What can I do? Is there a way to increase the stack size for this > > task? > > --snip-- > > Hello, > Your problem sounds suspiciously familiar. I haven't had a problem with > ping itself, but I noticed only a few days ago that on our target > tNetTask and tWdbTask are dangerously low on stack margin. It's down to > about 100 bytes -- too close for comfort. I have a Support enquirey > active on this. I've asked if these stack sizes can be increased. > > I suspect that this tight margin may be because we've just integrated > SENS. We have an x86 target. What are you using? We saw the same thing with SENS integration. It is *Much* more stack-hoggish. In addition, pay close attention to what netStackDataPoolShow and netStackSysPoolShow tell you. We had to increase the default mbuf allocation by many times to satisfy our admittedly network-hungry system. Many things would fail, or just lock up, due to starvation of these resources. -- John Ford National Radio Astronomy Observatory Green Bank, WV 24944-0002 jford@nrao.edu From aguilar@popler.lansce.lanl.gov Tue Jun 9 13:13:13 1998 From: aguilar@popler.lansce.lanl.gov (Ron Aguilar) Date: Tue Jun 9 13:13:18 PDT 1998 Subject: MVME2300 Image Problem VxWorks users, I just received a trial version of WRS MVME2304 BSP. I am attempting to build a VxWorks image. When I attempt to make bootrom_uncmp for the Tornado window, I get the following error. ccppc -BD:\Tools\Tornado1.0.1/host/x86-win32/lib/gcc-lib/ -c -mstrict-align -ansi -nostdin c -O2 -fvolatile -fno-builtin -fno-for-scope -Wall -I/h -I. -ID:\Tools\Tornado1.0.1\targ et\config\all -ID:\Tools\Tornado1.0.1\target/h -ID:\Tools\Tornado1.0.1\target/src/config - ID:\Tools\Tornado1.0.1\target/src/drv -DCPU=PPC604 -DMV2300 -DTARGET_DIR="\"mv2304\"" -o version.o D:\Tools\Tornado1.0.1\target\config\all/version.c ldppc -X -N -e _romInit -Ttext 00200000 \ -o bootrom_uncmp romInit.o bootInit_uncmp.o version.o bootConfig.o \ sysALib.o sysLib.o dec21140.obj D:\Tools\Tornado1.0.1\target/lib/libPPC604gnuvx.a dec21140.obj: In function `dcStartOutput': dec21140.obj(.text+0x20a4): undefined reference to `m_freem' make.exe: *** [bootrom_uncmp] Error 0x1 Done The only place that a search points to m_freem or for that matter dcStartOutput is in the file dec21140.obj. Has anyone else experienced this error? If so, what is the solution? Thanks, 0000,0000,8080 0000,0000,8080 Ron Aguilar Los Alamos National Laboratory Manuel Lujan Jr. Neutron Scattering Center LANSCE-12, MS H805 Los Alamos, NM 87545 505-667-8369 505-665-2676 (fax) rgaguilar@lanl.gov From sojian@sdismtp.sdi.sharpwa.com Tue Jun 9 16:54:20 1998 From: Lena Sojian Date: Tue Jun 9 16:54:24 PDT 1998 Subject: Modem Drivers Hi, Does anyone know where I could find Modem Drivers for vxWorks? Thanks, Lena From iliab@telegate.co.il Wed Jun 10 00:16:41 1998 From: Ilia Bosis Date: Wed Jun 10 00:16:44 PDT 1998 Subject: Re: MVME2300 Image Problem This is a multi-part message in MIME format. --------------B61ACC5DFEE096442D83C175 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit the aguilar@popler.lansce.lanl.gov (Ron Aguilar) wrote: > VxWorks users, > > I just received a trial version of WRS MVME2304 BSP. > > I am attempting to build a VxWorks image. When I attempt > > to make bootrom_uncmp for the Tornado window, I get the > > following error. > [skip] > D:\Tools\Tornado1.0.1\target/lib/libPPC604gnuvx.a > > dec21140.obj: In function `dcStartOutput': > > dec21140.obj(.text+0x20a4): undefined reference to `m_freem' > > make.exe: *** [bootrom_uncmp] Error 0x1 > > Done > > The only place that a search points to m_freem or for that matter > dcStartOutput is > > in the file dec21140.obj. > > Has anyone else experienced this error? If so, what is the solution? > > Thanks, >Los Alamos National Laboratory > > Manuel Lujan Jr. Neutron Scattering Center > > LANSCE-12, MS H805 > > Los Alamos, NM 87545 > > 505-667-8369 > > 505-665-2676 (fax) > > rgaguilar@lanl.gov It looks like You use the old driver (from the previous BSP). The file dec21140.obj is the driver for the DEC21140 Fast Ethernet adapter chip, found on the MVME230x boards. I'd got exactly the same error when trying to build the SENS 4.4BSD VxWorks kernel with the old driver. In the upgrade process, the Wind River changed the memory-management functions names and included #define's, that renames the references to the old functions with the new ones. So obtain the new driver or re-compile the one, You have to put these #define's in work. Hope, this helps. Regards Ilia. --------------B61ACC5DFEE096442D83C175 Content-Type: text/x-vcard; charset=us-ascii; name="vcard.vcf" Content-Transfer-Encoding: 7bit Content-Description: Card for Ilia Bosis Content-Disposition: attachment; filename="vcard.vcf" begin: vcard fn: Ilia Bosis n: Bosis;Ilia org: Telegate adr: Haplada 7;;;Or-Yehuda;;60218;Israel email;internet: iliab@telegate.co.il title: S/W engineer tel;work: 03-5388-685 tel;fax: 03-5388634 x-mozilla-cpt: ;0 x-mozilla-html: TRUE version: 2.1 end: vcard --------------B61ACC5DFEE096442D83C175-- From percheron@enertec.avicore.com Wed Jun 10 01:24:40 1998 From: "=?ISO-8859-1?Q?C=E9cile_PERCHERON?=" Date: Wed Jun 10 01:24:45 PDT 1998 Subject: vxwork compile bootrom_uncmp Hello, I want to compile a new bootrom_uncmp. I get some problems. I get the compiling error=20 C:\TORNADO\target/lib/libI80486gnuvx.a(ncr810Lib.o)(.text+0x4521): undefined reference to `inOrderExecute' cc386 -BC:\TORNADO/host/x86-win32/lib/gcc-lib/ -c -m486 -ansi -nostdinc -= O -fvolatile -nos=20 tdlib -fno-builtin -fno-defer-pop -Wall -I/h -I. -IC:\TORNADO\target\config\ztAll -IC:\T=20 ORNADO\target/h -IC:\TORNADO\target/src/config -IC:\TORNADO\target/src/dr= v -DCPU=3DI80486 =20 C:\TORNADO\target\config\ztAll\bootInit.c=20 copy C:\TORNADO\target\config\ztAll\bootInit.c bootInit_uncmp.c 1 fichier(s) copi=82(s) cc386 -BC:\TORNADO/host/x86-win32/lib/gcc-lib/ -c -m486 -ansi -nostdinc -= O -fvolatile -nos=20 tdlib -fno-builtin -fno-defer-pop -Wall -I/h -I. -IC:\TORNADO\target\config\ztAll -IC:\T=20 ORNADO\target/h -IC:\TORNADO\target/src/config -IC:\TORNADO\target/src/dr= v -DCPU=3DI80486 =20 -DUNCOMPRESS bootInit_uncmp.c cc386 -BC:\TORNADO/host/x86-win32/lib/gcc-lib/ -m486 -ansi -nostdinc -O -fvolatile -nostdl=20 ib -fno-builtin -fno-defer-pop -I/h -I. -IC:\TORNADO\target\config\ztAl= l -IC:\TORNADO\ta=20 rget/h -IC:\TORNADO\target/src/config -IC:\TORNADO\target/src/drv -DCPU=3DI80486 -P -x as=20 sembler-with-cpp -c -o romInit.o romInit.s=20 cc386 -BC:\TORNADO/host/x86-win32/lib/gcc-lib/ -c -m486 -ansi -nostdinc -= O -fvolatile -nos=20 tdlib -fno-builtin -fno-defer-pop -Wall -I/h -I. -IC:\TORNADO\target\config\ztAll -IC:\T=20 ORNADO\target/h -IC:\TORNADO\target/src/config -IC:\TORNADO\target/src/dr= v -DCPU=3DI80486 =20 C:\TORNADO\target\config\ztAll\bootConfig.c cc386 -BC:\TORNADO/host/x86-win32/lib/gcc-lib/ -c -m486 -ansi -nostdinc -= O -fvolatile -nos=20 tdlib -fno-builtin -fno-defer-pop -Wall -I/h -I. -IC:\TORNADO\xwork compileget\config\ztAll -IC:\T=20 ORNADO\target/h -IC:\TORNADO\target/src/config -IC:\TORNADO\target/src/dr= v -DCPU=3DI80486 =20 -o version.o C:\TORNADO\target\config\ztAll/version.c ld386 -X -N -e _romInit -Ttext 00008000 \ -o bootrom_uncmp romInit.o bootInit_uncmp.o version.o bootConfig.o \ sysALib.o sysLib.o if_dc.obj ztutil.obj ztflash.obj ztflDrv.obj =20 C:\TORNADO\target/li=20 b/libI80486gnuvx.a C:\TORNADO\target/lib/libI80486gnuvx.a(ncr810Lib.o)(.text+0x4521): undefined reference to `inOrderExecute' make.exe: *** [bootrom_uncmp] Error 0x1 Do you have any idea?? From MSchrape@atomika.com Wed Jun 10 03:02:08 1998 From: Martin Schrape Date: Wed Jun 10 03:02:12 PDT 1998 Subject: full C++ support on VxWorks Hi, does anyone have some experience with one of these issues 1.) templates 2.) exceptions 3.) egcs 4.) STL 5.) System etc. from objectspace on VxWorks. -martin --------------------------------------------------------------- Dipl.-Phys. Martin Schrape schrape@atomika.com Senior Software Engineer Tel. +49 89 315 891 34 Atomika Instruments GmbH Fax +49 89 315 59 21 Bruckmannring 40 www.atomika.com 85764 Oberschleissheim/Munich, Germany --------------------------------------------------------------- "The technology brings a certain thrill to simple tasks."(Bill Gates) From daemon@csg.lbl.gov Wed Jun 10 04:02:39 1998 From: daemon@csg.lbl.gov Date: Wed Jun 10 04:02:42 PDT 1998 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Wed Jun 10 04:02:36 PDT 1998 Subject: Compressed Application Code ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Compressed Application Code Date: Tue, 09 Jun 1998 06:20:45 -0400 From: james littlefield Organization: Littlefield Associates Message-ID: <357D0C7D.70A@alum.mit.edu> Greetings vxWorkers! I am working on an applications which has a small flash subsystem which emulates an ata disk device. We would like to keep several different executable files on this "disk" and be able to bootload different applications at different times. Due to size limitation on the disk, it will be necessary to compress the executable files in a manner similar to that used for compressed bootroms. The normal vxworks bootload process seems to open a file handle then pass it to bootLoadModule() for which WRS does not provide source code(!). To employ compressed applications bootloaded from disk, it would seem that I need to modify bootLoadModule() to decompress the image file as it is read from disk. So my questions... 1. Is there a simpler/better way to handle multiple, compressed application images? 2. If not, is there a bootLoadMOdule() replacement with source that I could get so the decompression function can be added? Thanks for any suggestions Jim Littlefield Littlefield Associates --------------------------- End of New-News digest ********************** From jrowley@viasat.com Wed Jun 10 10:56:33 1998 From: "jrowley" Date: Wed Jun 10 10:56:37 PDT 1998 Subject: VxWorks Mailbox Interrupts Does anyone have any advice or examples of configuring/using mailbox interrupts. Generic hints would be useful however we are specifically using a Force 5V card running VxWorks 5.3.1. ----------------------------------------------------- Jeff Rowley Member, Technical Staff ViaSat jrowley@viasat.com 2290 Cosmos Ct Carslbad, CA 92009 ----------------------------------------------------- From rcw@DynRes.com Wed Jun 10 11:35:23 1998 From: rcw@DynRes.com Date: Wed Jun 10 11:35:26 PDT 1998 Subject: BSP help files under windows Tornado VxWorks Does anyone know the right way to access the BSP helpfiles that are supplied with the Windows version of Tornado? There is a bunch of stuff in \tornado\target\man\man1 and man2 but it looks like this stuff is not available through the Tornado help menu. Any ideas? From bwaite@cspi.com Wed Jun 10 11:58:52 1998 From: bwaite@cspi.com Date: Wed Jun 10 11:58:56 PDT 1998 Subject: CrossWind on Solaris 2.6 I was wondering if anyone in this group has run into the same problem with CrossWind that I have been encountering. I have sent this off to Wind River Support, but was hoping someone here may have worked through this crazy problem. When I am running Crosswind under Solaris 2.6 I cannot type commands correctly at the (gdb) prompt. When I try to type certain letters nothing happens. For example from the (gdb) prompt I can type 'hphphph' but I cannot type 'hhhhhh' or 'hppppp'. Also I cannot type commands such as 'add-symbol-file' because it will not accept the second 'd' in add. I have tested this with the installation off of the newest Tornado 1.0.1 CD (with Cumulative patch 1 installed) and also by installing the Culmulative Patch 1 from the Web site with the same results. I have also installed the Solaris 2.6 CrossWind Patch (TDK-12566.tar) and have the same problem. I also tested with and without the workaround for TSR# 29402, CrossWind running on Solaris drops keyboard input, to no avail. This problem makes CrossWind unusable for us at the moment. Any help would be great Thanks Brian Waite CSPI Customer Suppport From ads3000@phillipsaerospace.com Wed Jun 10 16:37:43 1998 From: Don McKenna Date: Wed Jun 10 16:37:47 PDT 1998 Subject: Announcement: vxWorks based Realtime Application now available. The ADS-3000 (v 1.90) is a COTS software package that fills the gaps between rapid prototyping, data acquisition, or simulation project, your Interface Control Document (ICD) or specification, and the physical I/O elements such as MIL-STD-1553, ARINC 629/429, Digital I/O, Analog I/O, GPIB, LVDT/RVDT, Synchro/Resolver. System developers use COTS tools to create simulations in C/C++, Fortran, ADA, or automated code generators. Simulations run under ADS-3000 deterministically and real-time in one or multiple VME chassis. vxWorks is the primary real-time OS used, and the API runs under various platforms such as Solaris 2.6, FreeBSD, etc. connected via standard networks like Ethernet, using standard TCP/IP protocol. The built in "connection editor" maps soft variable names to I/O elements. Data is time tagged (IRIG-B etc.), formatted and packaged to and from the hardware resources and mapped to software variables. Visualization and control is provided by Test Control Language (TeCL) that contains commands to maximize ADS-3000, with a full suite of strip charts, bar graphs, dials, gauges, boolean indicators, event detection, triggering, branching, and alarm features. TeCL provides access to system variables using ICD formatting and maps variables to displays and functions. Available as an option, DataSPY is a frame-based debug tool to display variables from real-time processes using TeCL. DataSPY runs concurrently on the host along side your simulation, data acquisition, process control, or other multi-processor multi-tasking application. No modifications are necessary, reducing overhead and preserving code integrity. This feature can be very important to developers under tightly controlled development environment such as MIL-STD-2167. Complex triggering and limit checking functions are provided for all variables. Notification of events can be visually displayed, and custom actions can be taken in real-time. Variable overwrites are also possible in a frame-based manner, allowing the user to interject static or dynamically changing values such as ramps, sine waves, pulses, fixed patterns, or user functions. DataSPY also allows multiple users to access data from the same system across a network. ADS-3000 is jointly developed and owned by: Tech S.A.T GmbH and /dev Software GmbH, Munich, Germany, and distributed in North America by Phillips Aerospace. ================================================= Contact Info: Don McKenna Phillips Aerospace 1100 S Jellick Ave. Industry, CA 91748, USA mailto:info@phillipsaerospace.com http://www.phillipsaerospace.com Phone:626-839-1777 Fax:626-839-1677 /dev Software Munich, Germany mailto:info@devsoft.com http://www.devsoft.com ================================================== From mumcu@venus.aselsan.com.tr Wed Jun 10 23:02:40 1998 From: Ahmet Mumcu Date: Wed Jun 10 23:02:47 PDT 1998 Subject: File System for VxWorks Hi, I am looking for a filesystem provided from a 3rd party company for Tornado environment. The DOS and Raw FS included with the OS seems to be very limited. PowerPC and/or 68K architecture support is needed. I will be glad to receive your recommendations. Thank you. ------------------------------------------------- Ahmet Mumcu ASELSAN Inc. MST Group /YMM PO Box 101 Yenimahalle Ankara 06172 TURKEY e-mail: mumcu@mst.aselsan.com.tr Tel: 90-312-3851900 ext 2633 Fax: 90-312-3545205 ------------------------------------------------- From GUY.G.C.CHATEL@RCM.thomson.fr Thu Jun 11 01:16:57 1998 From: GUY.G.C.CHATEL@RCM.thomson.fr Date: Thu Jun 11 01:17:01 PDT 1998 Subject: audio on cvme603 board Hello, (vxworks/tornado) I'am tying use the audio chip Crystal under vxworks 5.3 on a CVME603 board (provided by CETIA), but i don't know how to configure the Crystall and the PCI/ISA bridge to use DMA transferts between the PowerPc and the audio chip. The board reference manual is not really clear about the PCI/ISA bridge description, i only know that 7 DMA channels are available and nothing else. ------------------oooOooo----------------- Guy Chatel Thomson CSF-RCM 10 av. de la 1ere DFL 29283 BREST CEDEX tel: +33 298 312 601 fax: +33 298 312 443 email: guy.chatel@rcm.thomson.fr ------------------oooOooo----------------- From bits2@teil.soft.net Thu Jun 11 01:52:14 1998 From: BITS TRAINEES grp2 Date: Thu Jun 11 01:52:17 PDT 1998 Subject: Communication between two targets in Tornado Environment Hi everybody, I have Tornado installed over Sun Solaris. We use Vxsim for simulating the VxWorks environment. We are developing a network stack with IP and UDP. To test this product we need to communicate between two targets, both running our own network stack. One easy way is to make IP_OUTPUT of each of these entities write to a file and IP_INPUT to read from the file the other entity write into. By target, I mean two vxsim's with different processor number. 1) Is there any other way of communicating between two tasks running on different targets? Thank you Prabhakaran.G From daemon@csg.lbl.gov Thu Jun 11 04:06:13 1998 From: daemon@csg.lbl.gov Date: Thu Jun 11 04:06:16 PDT 1998 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Thu Jun 11 04:06:11 PDT 1998 Subject: Re: about JestSend Subject: Re: Anybody using Motorola MBX860 board? ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: about JestSend Date: Wed, 10 Jun 1998 11:21:42 GMT From: cvenkat@giasdl01.vsnl.net.in Organization: Deja News - The Leader in Internet Discussion Message-ID: <6llq86$3lf$1@nnrp1.dejanews.com> References: <01bd943e$47dd2380$LocalHost@850155517> General information about JetSend is available at http://www.jetsend.hp.com. HP's developer's kit for JetSend contains a VxWorks port of the JetSend protocol stack and a few sample applications. The DK doesn't come cheap, though; the last time I heard, it was priced at US$15,000 . Venkat In article <01bd943e$47dd2380$LocalHost@850155517>, "Luis Vega" wrote: > > > Hello, > > Does anybody have already used JetSend, the new device-to-device > communication protocol of Hewlett-Packard ? Where can I get some > infomrtation about it ? > > Regards > > Luis Vega > > - -----== Posted via Deja News, The Leader in Internet Discussion ==----- http://www.dejanews.com/ Now offering spam-free web-based newsreading --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Anybody using Motorola MBX860 board? Date: Wed, 10 Jun 1998 17:48:15 -0400 From: Douglas Fraser Organization: Lucent Technologies Message-ID: <357EFF1F.6DC3395A@lucent.com> References: <01bd946d$75909e40$119c0ccb@jeremy> We are using the MBX860 (in large quantities). The BSP depends on you having the proper version of the MBX board. If you have a pre-letter release, you need to back up to the 0.7 release. If you have an 'A' or 'B' version of the board, you need the 0.8 release. The difference is that the early boards have a 32768 KHz xtal and the later use a 40Mhz/50Mhz ocs. clock. The UPM data is different for the two. You may find that the problems only occur when you have DIMM installed. I cannot provide you with my BSP info as a) Motorola would scream and b) Lucent Technologies would scream. Sorry. But if you think REALLY hard about it, (not that hard) it is possible to modify the contents of the .07 and .08 BSP release (only romInit.s changed) to support both families of the MBX860. For now, given your description of your failure, you should step back to BSP release 0.7. If you decide to merge the UPM, just examine the DPRAM configuration data after the maxI2cReadConfig (whatever) call to check the sysRefClkSpd (???) value. Is it 32768? 40000000? 50000000? There are definitions for these rates in mbx800.h Have fun and good luck! - -- Doug Fraser (The opinion's contained herein are mine alone, etc. etc....) --------------------------- End of New-News digest ********************** From don.blake@lmco.com Thu Jun 11 04:20:40 1998 From: "Donald R. Blake" Date: Thu Jun 11 04:20:44 PDT 1998 Subject: Re: CrossWind on Solaris 2.6 > When I am running Crosswind under Solaris 2.6 I cannot type commands > correctly at the (gdb) prompt. When I try to type certain letters nothing > happens. For example from the (gdb) prompt I can type 'hphphph' but I > cannot type 'hhhhhh' or 'hppppp'. Also I cannot > type commands such as 'add-symbol-file' because it will not accept the > second 'd' in add. I have tested this with the installation off of the > newest Tornado 1.0.1 CD (with Cumulative patch 1 installed) and also by > installing the Culmulative Patch 1 from the Web site with the same results. I've experienced similar symptoms running under SunOS 5.5.1 when the mouse cursor is outside the prompt window. With the cursor positioned within the prompt window, it works fine. Don -- Donald R. Blake Sr. Programmer Lockheed Martin Federal Systems 1802 State Route 17C Owego, NY 13827-3994 Phone: 607-751-2658 Fax: 607-751-2397 email: don.blake@lmco.com From mwicks@motown.lmco.com Thu Jun 11 04:28:44 1998 From: "Matthew J. Wicks" Date: Thu Jun 11 04:28:47 PDT 1998 Subject: Re: CrossWind on Solaris 2.6 Brian, I, too, have noticed this type of behavior from Crosswind. But I look at it as an inconvenience rather than a show stopper. I have found that I lose at least every other character when the mouse is in Crosswind's (upper) display pane but do not drop any characters when the mouse is in the (lower) command pane. HTH Matt Wicks Lockheed-Martin Government Electronics Systems Brian Waite wrote: > When I am running Crosswind under Solaris 2.6 I cannot type commands > correctly at the (gdb) prompt. When I try to type certain letters nothing > happens. For example from the (gdb) prompt I can type 'hphphph' but I > cannot type 'hhhhhh' or 'hppppp'. Also I cannot > type commands such as 'add-symbol-file' because it will not accept the > second 'd' in add. I have tested this with the installation off of the > newest Tornado 1.0.1 CD (with Cumulative patch 1 installed) and also by > installing the Culmulative Patch 1 from the Web site with the same results. > I have also installed the Solaris 2.6 CrossWind Patch (TDK-12566.tar) and > have the same problem. I also tested with and without the workaround for > TSR# 29402, CrossWind running on Solaris drops keyboard input, to no > avail. This problem makes CrossWind unusable for us at the moment. Any help > would be great > > Thanks > Brian Waite > CSPI Customer Suppport From kb+@andrew.cmu.edu Thu Jun 11 04:56:15 1998 From: Kevin Bradley Date: Thu Jun 11 04:56:19 PDT 1998 Subject: Re: audio on cvme603 board Excerpts from mail: 11-Jun-98 audio on cvme603 board by the U. G. Exploder@lbl.g > I'am tying use the audio chip Crystal under vxworks 5.3 on a CVME603 board > (provided by CETIA), but i don't know how to configure the Crystall and the > PCI/ISA bridge to use DMA transferts between the PowerPc and the audio chip. > The board reference manual is not really clear about the PCI/ISA bridge > description, i only know that 7 DMA channels are available and nothing else. I would try three things: 1) Look for a commercial audio driver, such as from OSS, for your system. This makes life *much* easier, as audio drivers are somewhat tricky to get right. 2) Check out Linux kernel source code for both the crystal driver and for programming DMA under PowerPC and PCI/ISA systems. There is a Linux implementation for PowerPC, so I expect it will have some relevance. 3) Call Crystal, and ask them for suggestions. They've probably done it before. Hope this helps... -- Kevin From BOOJ@kidd.Co.Za Thu Jun 11 05:32:31 1998 From: Johan Boot Date: Thu Jun 11 05:32:35 PDT 1998 Subject: stdarg Hi Tornado, vxWorks 5.3.1 I am having trouble using stdarg with my PowerPC 603e target. The routine I am using must take a numer of variables and pack it into a vector. The routine works correct on the first 8 arguments, then skips 3 and continues. If you had and solved this problem, please help me out! The routine is as follows: void Vec_Assign(TVec *Vec,...)/* List all the Elements */ { int i; /* Used to index vector. */ va_list ArgPtr; va_start(ArgPtr, *Vec); for (i = 0; i < Vec->Size; i++) Vec->Element[i] = va_arg(ArgPtr, double); /* end for */ va_end (ArgPtr); } /* end Vec_Assign */ with TVec defined as: typedef struct { double Element[MAT_MAX_SIZE]; int Size; } TVec; The following test is performed: void mytest(void) { int i; Mine.Size = 18; Assign(&Mine, (double)1.0, (double)2.0, (double)3.0, (double)4.0, (double)5.0, (double)6.0, (double)7.0, (double)8.0, (double)9.0, (double)10.0, (double)11.0, (double)12.0, (double)13.0, (double)14.0, (double)15.0, (double)16.0, (double)17.0, (double)18.0); for(i=0; i Date: Thu Jun 11 12:09:21 PDT 1998 Subject: MBX860/MPC860 and sysPhysMemDesc. ------------------------------------ Hello MPC (and PowerPC) users, In VxWorks sysLib.c the PCMCIA and Internal Memory map are marked as valid, writeable and not cached. Will the PowerPC do speculative reads on these areas? If there is any 'auto-increment' type hardware in any of these address ranges should the memory be marked as 'VM_STATE_GAURDED' as well? Or does the non-cached disable speculative reads. eg. .... if(need data) data = *auto_increment_hw; ... Here, the branch may not be evaluated immediately. However, the processor may have an empty load/store pipe and decide to read the pointer that is a memory mapped peice of hardware. If the read of the I/O has some side effect, this may cause some unwanted logic to happen. thanks in advance, Bill Pringlemeir ------------------------------------ vxWorks Tornado vxWorks (o)-) A snorkling cyclops. From roghol@lin.foa.se Fri Jun 12 03:42:11 1998 From: Roger Holmquist Date: Fri Jun 12 03:42:14 PDT 1998 Subject: Anybody have any MVME 2700 experience? Hello! We are about to buy a new general purpose VME-board running in the vxWorks/Tornado/Unix environment. Our main need is "speed" (for instance, fast enough FFT) and the general Motorola VME-features, serial ports, SCSI, Ethernet ,etc etc As far as I can understand it is not possible to mix 68K-boards and PPC-boards if you want to use the VxWorks multiprocessing features, WXMP ? I am thinking of MVME2700 but I can=B4t see that anybody is discussing that board here... Any suggestions? Greetings / Roger #------------------------------------------------------------------# # Roger Holmquist, FOA, National Defence Research Establishment # # sect 750, Olaus Magnus vag 42, Box 1165, 581 11 LINKOPING ------# #---------------------------- SWEDEN ------------------------------# # Phone: +46 13-378382 --------------------------------------------# # Fax: +46 13-378009 ----------------------------------------------# # Home: +46-13-214733 Mobile +46-706-250123 ----------------------# # Email: roghol@lin.foa.se ----------------------------------------# #------------------------------------------------------------------# From matthieu@elwood.laas.fr Fri Jun 12 05:11:52 1998 From: Matthieu Herrb Date: Fri Jun 12 05:11:55 PDT 1998 Subject: Re: stdarg You wrote (in your message from Fri 12) > > I am having trouble using stdarg with my PowerPC 603e target. The routine I am using must > take a numer of variables and pack it into a vector. The routine works correct on the first 8 > arguments, then skips 3 and continues. > > If you had and solved this problem, please help me out! I had this problem. It turned out to be a bug in the gcc version shipped by tornado. I solved it by installing egcs () as a cross-compiler (configured with --target=ppc-eabi) and I'm using it instead. I did report it to Wind River, but it doesn't seem to have made it into a SPR. Matthieu From cgrames@mdc.com Fri Jun 12 06:21:17 1998 From: Charlie Grames Date: Fri Jun 12 06:21:21 PDT 1998 Subject: Anybody have any MVME 2700 experience? -Reply Roger, We have been using the MVME2700 since the beginning of the year and have had good experience with it. The BSP for the 2700 is not yet available from Wind River, but Motorola supplies it as a patch to the MVME2604 BSP. For all practical purposes, the 2700 behaves just like the 2604. Integer performance is better than the 200 MHz 2604; floating-point performace is somewhat worse. Overall, we have noticed about 20% improvement with the 2700 over the 2604, mostly attributable to the larger L2 cache (1 MB v. 256 KB). Charlie Grames The Boeing Company (314) 233-1956 Charles.R.Grames@boeing.com >>> the vxWorks Users Group Exploder 06/12/98 05:42am >>> Submitted-by roghol@lin.foa.se Fri Jun 12 03:42:11 1998 Submitted-by: Roger Holmquist Hello! We are about to buy a new general purpose VME-board running in the vxWorks/Tornado/Unix environment. Our main need is "speed" (for instance, fast enough FFT) and the general Motorola VME-features, serial ports, SCSI, Ethernet ,etc etc As far as I can understand it is not possible to mix 68K-boards and PPC-boards if you want to use the VxWorks multiprocessing features, WXMP ? I am thinking of MVME2700 but I can=B4t see that anybody is discussing that board here... Any suggestions? Greetings / Roger #------------------------------------------------------------------# # Roger Holmquist, FOA, National Defence Research Establishment # # sect 750, Olaus Magnus vag 42, Box 1165, 581 11 LINKOPING ------# #---------------------------- SWEDEN ------------------------------# # Phone: +46 13-378382 --------------------------------------------# # Fax: +46 13-378009 ----------------------------------------------# # Home: +46-13-214733 Mobile +46-706-250123 ----------------------# # Email: roghol@lin.foa.se ----------------------------------------# #------------------------------------------------------------------# From patrick.keliher@wrs.com Fri Jun 12 11:48:35 1998 From: Patrick Keliher Date: Fri Jun 12 11:48:39 PDT 1998 Subject: Re: stdarg vxworks Just FYI, it's SPR 9934, but it hasn't made it into a release yet. Not speaking for Wind River as a Corporation, but just as an FAE with access to the SPR database... Pat Keliher At 05:11 AM 6/12/98 PDT, you wrote: >Submitted-by matthieu@elwood.laas.fr Fri Jun 12 05:11:52 1998 >Submitted-by: Matthieu Herrb > >You wrote (in your message from Fri 12) > > > > I am having trouble using stdarg with my PowerPC 603e target. The > routine I am using must > > take a numer of variables and pack it into a vector. The routine > works correct on the first 8 > > arguments, then skips 3 and continues. > > > > If you had and solved this problem, please help me out! > >I had this problem. It turned out to be a bug in the gcc version >shipped by tornado. I solved it by installing egcs >() as a cross-compiler >(configured with --target=ppc-eabi) and I'm using it instead. > >I did report it to Wind River, but it doesn't seem to have made it >into a SPR. > > > Matthieu From jonesboy@waverider.com Fri Jun 12 16:05:12 1998 From: Matt Walters Date: Fri Jun 12 16:05:15 PDT 1998 Subject: Multiple END driver strangeness vxworks I am loading two END drivers, the Motorola CPM Ethernet Driver and one of my own. I've loaded my driver successfully on its own (without the CPM driver), and it gets added to the muxTable, etc., but when I load my driver after the Motorola driver, it doesn't load. Neither muxDevLoad() or muxDevStart() fails, but endFindByName() returns a null. When I do a muxShow, the driver appears in the table, but if I specify my driver in the muxShow (ie: muxShow("x",0)), it returns a null. Any ideas? Thanks in advance. -=[Matt]=- +---+---+ Matt Walters, Embedded Systems @ WaveRider Communications (;;) F00FC7C8 - Catch fire and die horribly :) My views do not necessarily reflect those of WaveRider Communications in all cases, and I *have* been known to be wrong. Just not often. +---+---+ From daemon@csg.lbl.gov Sat Jun 13 04:00:32 1998 From: daemon@csg.lbl.gov Date: Sat Jun 13 04:00:36 PDT 1998 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Sat Jun 13 04:00:29 PDT 1998 Subject: PCI, PMC, mv2604 ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: PCI, PMC, mv2604 Date: Fri, 12 Jun 1998 18:48:41 GMT From: rcwagner@my-dejanews.com Organization: Deja News - The Leader in Internet Discussion Message-ID: <6lrt69$5pl$1@nnrp1.dejanews.com> VxWorks Tornado I am trying to get a mv2604 board running with a pmc card installed (Systran ScramNet)in addition to having another PMC carrier board (a Green Spring IPPC- 2636 [IP Carrier board]) connected via the PCI expansion connector. Does anyone have any sample code or tips for integrating either of these into VxWorks? Ron Wagner - -----== Posted via Deja News, The Leader in Internet Discussion ==----- http://www.dejanews.com/ Now offering spam-free web-based newsreading --------------------------- End of New-News digest ********************** From jonesboy@waverider.com Sat Jun 13 11:25:31 1998 From: Matt Walters Date: Sat Jun 13 11:25:34 PDT 1998 Subject: Fundamental flaw with SENS and multiple drivers vxworks users: I would like to warn you all of a serious problem with the WindRiver SENS stack. As you may recall, I sent a message to the exploder yesterday call "SENS multiple driver strangeness" or something of the like. I received a couple of emails saying that other people had experienced this same problem. So last night (and this morning), myself and a couple of co-workers decided to try to find out what was going on. For those that don't remember, we were trying to load the Motorola CPM Ethernet END driver, as well as our own END driver, and attach them both to the IP stack. The first driver loads and attaches properly, the second driver loads - and refuses to attach. After a little digging, it was determinded that endFindByName always returns a zero unless the name of the first driver is passed. Our driver was showing up in the muxShow table - but if we tried to go muxShow("x",0), it would return a null and display nothing. The TSR related to this problem is #107688. It states, in a nutshell, that endFindByName is broken. So my co-worker decided to figure out what endFindByName did. I won't go into the gory detail (if you want it, ask max@waverider.com what he did), but he replaced the function in the libPPC860gnuvx.a file with one that did what it was supposed to. So, according to the TSR, this should have fixed the problem. It turned out that the box blew up upon trying to load the second driver anyways. He continued to hack at it, and found that for some reason, the END linked list wasn't being updated properly when the second driver was loaded. In short, I guess - don't try to load a second END driver under SENS (well, where else would you try to load two END drivers, right). If you need two drivers, load one as a BSD4.4 driver and the other as an END driver. Let me make clear: SENS works great for a single END driver, and for (as far as I can tell) a single END driver and one or several BSD4.4 drivers. It's just broken as far as multiple END drivers is concerned. I hope this post helps someone out there, or perhaps makes someone stop scratching their head.... :) -=[Matt]=- +---+---+ Matt Walters, Embedded Systems @ WaveRider Communications (;;) F00FC7C8 - Catch fire and die horribly :) My views do not necessarily reflect those of WaveRider Communications in all cases, and I *have* been known to be wrong. Just not often. +---+---+ From ptp@mclean.sparta.com Sat Jun 13 13:12:38 1998 From: Patrick Pinkowski Date: Sat Jun 13 13:12:41 PDT 1998 Subject: Re: Fundamental flaw with SENS and multiple drivers --------------57ACC6CE9AD7EFE954E7BA66 Content-Type: text/plain; charset=iso-8859-1 Content-Transfer-Encoding: quoted-printable X-MIME-Autoconverted: from 8bit to quoted-printable by postal1.lbl.gov id NAA21892 Matt, I have seen a similar problem while developing a FDDI Device driver using SENS. I found the resolution to the problem. It was a bug in the endFindByName file. WRS sent me a bug fix for the problem on the Motorola MV2604. There were also edits that needed to be made in target/config/all/bootConfig.c and target/src/config/usrNetwork.c. I documented the items that needed fixing in my FDDI Device Driver document. I will copy the items from the documentation and paste them below. Don't mind the Section Numbers: 6.2.2 Target/config/all/bootConfig.c 6.2.2.1 Locate the definition of IP_MAX_UNITS, and change the definition of IP_MAX_UNITS from 1 to 2 (edits are in bold): #ifndef IP_MAX_UNITS /* * Interphase 4511 VxWorks Device Driver * Change IP_MAX_UNITS from it's default of 1 * to 2. IP_MAX_UNITS must be increased for * each SENS device that is included in the * system */ #define IP_MAX_UNITS 2 #endif 6.2.4 Target/src/config/usrNetwork.c 6.2.4.1 Locate the definition of IP_MAX_UNITS, and change the definition of IP_MAX_UNITS from 1 to 2 (edits are in bold): #ifndef IP_MAX_UNITS /* * Interphase 4511 VxWorks Device Driver * Change IP_MAX_UNITS from it's default of 1 * to 2. IP_MAX_UNITS must be increased for * each SENS device that is included in the * system */ #define IP_MAX_UNITS 2 #endif 6.2.7 Target/config/mv2604/Makefile 6.2.7.1 Create the extraObj directory in the BSP directory (i.e. $WIND_BASE/target/config/mv2604). Copy the muxLib-PPC604.o file to the extraObj subdirectory (edits are in bold): 6.2.7.2 Edit the target/config/mv2604/Makefile. Locate the definition for MACH_EXTRA. Remove the dec21140.obj from the right hand side of the equal sign and replace it with extraObj/muxLib-PPC604.o =85 MACH_EXTRA =3D extraObj/muxLib-PPC604.o =85 Regards Pat --------------57ACC6CE9AD7EFE954E7BA66 Content-Type: text/html; charset=iso-8859-1 Content-Transfer-Encoding: quoted-printable X-MIME-Autoconverted: from 8bit to quoted-printable by postal1.lbl.gov id NAA21892 Matt,

I have seen a similar problem while developing a FDDI Device driver using SENS.   I found the resolution to the problem.  It was a bug in the endFindByName file. WRS sent me a bug fix for the proble= m on the Motorola MV2604.

There were also  edits that needed to be made in target/config/al= l/bootConfig.c and target/src/config/usrNetwork.c.  I documented the items that nee= ded fixing in my FDDI Device Driver document.  I will copy the items fro= m the documentation and paste them below.  Don't mind the Section Numb= ers:

6.2.2 Target/config/all/bootConfig.c
6.2.2.1 Locate the definition of IP_MAX_UNITS, and change the definit= ion of IP_MAX_UNITS from 1 to 2 (edits are in bold):

#ifndef IP_MAX_UNITS
/*
 * Interphase 4511 VxWorks Device Driver
 * Change IP_MAX_UNITS from it's default of 1
 * to 2.  IP_MAX_UNITS must be increased for
 * each SENS device that is included in the
 * system
 */
#define IP_MAX_UNITS 2
#endif

6.2.4 Target/src/config/usrNetwork.c
6.2.4.1 Locate the definition of IP_MAX_UNITS, and change the definit= ion of IP_MAX_UNITS from 1 to 2 (edits are in bold):

#ifndef IP_MAX_UNITS
/*
 * Interphase 4511 VxWorks Device Driver
 * Change IP_MAX_UNITS from it's default of 1
 * to 2.  IP_MAX_UNITS must be increased for
 * each SENS device that is included in the
 * system
 */
#define IP_MAX_UNITS 2
#endif

6.2.7 Target/config/mv2604/Makefile
6.2.7.1 Create the extraObj directory in the BSP directory (i.e. $WIN= D_BASE/target/config/mv2604). Copy the muxLib-PPC604.o file to the extraObj subdirectory (edits are in bold):

6.2.7.2 Edit the target/config/mv2604/Makefile.  Locate the defin= ition for MACH_EXTRA.  Remove the dec21140.obj from the right hand side of the equal sign and replace it with extraObj/muxLib-PPC604.o
=85
MACH_EXTRA      =3D   extraObj/mux= Lib-PPC604.o
=85

Regards
Pat --------------57ACC6CE9AD7EFE954E7BA66-- From yleduc@oceanet.fr Sun Jun 14 22:54:55 1998 From: Yves LEDUC Date: Sun Jun 14 22:55:00 PDT 1998 Subject: ASM on VxWorks I am begenner. I want write some program in ASM 680x0 (MVME162LX). Problems are: - Makefile to assemble ASM code with .o execute on VxWorks and debug with GDB. - Call C function by ASM code. - Call ASM function by C code. - Interrupt routines in ASM. Anyone help me? I am an Industrial Computing Teacher I am looking for contact with teacher who work with Tornado VxWorks MVME162 for information exchange. Thanks in advance. Yves LEDUC STS Informatique Industrielle LPT de la SALLE 44000 NANTES CEDEX 1 FRANCE yleduc@oceanet.fr From John_W_Cosgrove@res.raytheon.com Mon Jun 15 03:10:55 1998 From: John_W_Cosgrove@res.raytheon.com Date: Mon Jun 15 03:10:59 PDT 1998 Subject: workQPanic: Kernel work queue overflow vxWorks VxWorks VXWORKS Fellow vxWorks people, I am debugging a system that has worked fine up til now, but am now experiencing the dreaded "workQPanic: Kernel work queue overflow". I am of the notion to just up the kernel queue size, but do not know where to do that. If someone could point me in the right direction.... Thanks John Cosgrove Raytheon (401)842-4167 (desk) (401)842-5660 (lab) jwc_NOSPAM@ssd.ray.com Pls remove the _NOSPAM to respond to my messages. From daemon@csg.lbl.gov Mon Jun 15 04:01:27 1998 From: daemon@csg.lbl.gov Date: Mon Jun 15 04:01:31 PDT 1998 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Mon Jun 15 04:01:24 PDT 1998 Subject: What exactly is Look! ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: What exactly is Look! Date: 15 Jun 1998 02:37:42 GMT From: obrien1966@aol.com (OBrien1966) Organization: AOL http://www.aol.com Message-ID: <1998061502374200.WAA11478@ladder03.news.aol.com> New to VxWorks and started debugging code and have seen mention of a tool called "Look!" .. Can anyone tell me what this is exactly and where do I get it from ? Thanks, Brooks --------------------------- End of New-News digest ********************** From martin.klaper@crypto.ch Mon Jun 15 04:46:12 1998 From: Klaper Martin Date: Mon Jun 15 04:46:15 PDT 1998 Subject: vxWorks Tornado romable version PCI bus i-960 Target: cyclone IQ80960Rx Evaluation Platform with SCSI Ethernet IQ Module Board problem: BSP romable does not initialize properly Several vxWorks targets can be compiled: - The version with download of the vxWorks library via the ftp is running fine. - The standalone version seems not initialize the PCI bus / Ethernet Controller correctly. In this standalone (romable) version not all registers of the ethernet controller are visible or are correctly initialized. The PCI primary bus is not visible. The PCI secondary bus is supposedly not initialized. We already checked that the functions usNetInit and sysPciInit are called; Can anyone give us a hint, where to look next. Thank you. Martin Klaper, Crypto AG > -------------------------------------------------------------------------- Martin Klaper MAILTO:Martin.Klaper@crypto.ch > Crypto AG http://www.crypto.ch > P.O. Box phone: +41 41 749 75 22 > CH-6301 Zug fax: +41 41 741 22 72 > -------------------------------------------------------------------------- > From Shaochen@aol.com Mon Jun 15 06:38:04 1998 From: Shaochen@aol.com Date: Mon Jun 15 06:38:08 PDT 1998 Subject: Help on setting a register address and initialising i2c registers Hello, Everyone: We have a mbx821 board running vxWorks just fine, now my collegues have designed a device which has a register address for example 0xf0000000, but mbx821 won't recognize it. Every time we access it, we get a bus error. I was told that I need to add something in sysPhyMemDesc[] table in sysLib.c, but how? The device (I mentioned above) is using i2c bus, some registers need initialising, anyone has any C code to do this? A few lines will do. (which library I need include?) Please help if you can. Thanks in advance. DaoShuo From cruz_nojunk@xyplex.com Mon Jun 15 07:15:04 1998 From: Roger Cruz Date: Mon Jun 15 07:15:08 PDT 1998 Subject: Re: workQPanic: Kernel work queue overflow vxWorks VxWorks VXWORKS My understanding of this is that the queue is fixed by its design and therefore can't be changed. I can't remember the exact details of when we looked at this, but I think it can only be 64 entries because they use a byte as the queue index and each entry is 8 bytes. We got around this error by getting rid of the netQJobAdd (or something similar to that which uses the work queue) and creating our own queue & task. I hope this helps Roger PS. Remove _nojunk from email address -------------- Fellow vxWorks people, I am debugging a system that has worked fine up til now, but am now experiencing the dreaded "workQPanic: Kernel work queue overflow". I am of the notion to just up the kernel queue size, but do not know where to do that. If someone could point me in the right direction.... -- Roger Cruz cruz_nojunk@xyplex.com Xyplex Networks w: 978-952-4783 295 Foster Street f: 978-952-4887 Littleton, MA 01460 From Ioan.marusca@edc.ge.com Mon Jun 15 08:23:37 1998 From: "Marusca, Ioan (ED&C,Pwr Mgt,UR)" Date: Mon Jun 15 08:23:41 PDT 1998 Subject: RE: MBX860/MPC860 and sysPhysMemDesc. Bill, MPC860 does not read data in advance from memory if that area is marked as not cacheable. Only data and instructions from cacheable memory may be read before are needed. The PPC860 manual explains more about Load/Store Instruction issue and serialization at page 6-25,6-26. ========================================================= Ioan Marusca General Electric Multilin 215 Anderson Avenue, Markham, Ontario, L6E1B3, Canada Email: MaruscaI@edc.ge.com Tel:(905)-201-2141 ========================================================= >Hello MPC (and PowerPC) users, > >In VxWorks sysLib.c the PCMCIA and Internal Memory map >are marked as valid, writeable and not cached. Will >the PowerPC do speculative reads on these areas? > >If there is any 'auto-increment' type hardware in any of >these address ranges should the memory be marked as >'VM_STATE_GAURDED' as well? Or does the non-cached >disable speculative reads. > >eg. > .... > if(need data) > data = *auto_increment_hw; > ... >Here, the branch may not be evaluated immediately. However, >the processor may have an empty load/store pipe and decide to >read the pointer that is a memory mapped peice of hardware. If >the read of the I/O has some side effect, this may cause some >unwanted logic to happen. > >thanks in advance, >Bill Pringlemeir >------------------------------------ >vxWorks Tornado vxWorks (o)-) A snorkling cyclops. > From rajankapur@cbgw2.lucent.com Mon Jun 15 09:09:53 1998 From: rajankapur@lucent.com Date: Mon Jun 15 09:09:56 PDT 1998 Subject: vxworks question Hi, Does anybody know if there is a vxworks port available for ncurses 4.2 in the public domain. Has anybody done any work on this and could they shed some light on what is involved in making ncurses 4.2 work with vxworks. Thanks. Rajan Kapur From robert_smith@mcg.mot.com Mon Jun 15 12:06:58 1998 From: Robert Smith Date: Mon Jun 15 12:07:01 PDT 1998 Subject: Re: Anybody have any MVME 2700 experience? Roger: The MVME2700 supports atomic RMW operations over the VME bus and therefore should be able to co-exist with 68K boards that also support RMW. This is a relatively recent development for Motorola PPC boards with the "PowerPlus" architecture (i.e., MVME2300/2600/2700, etc.). The MVME2700 is not supported directly by WindRiver yet. However, the preliminary BSP is available and may be obtained through your local Motorola Computer Group Systems Engineer. To use RMW, undef ANY_BRDS_IN_CHASSIS_NOT_RMW in "config.h". I have recently concluded stress testing of VxMP with eight MVME2700 boards running at 266MHz in the chassis. In my test, using atomic RMW resulted in a 12-18% improvement over using the pseudo-atomic method (locking the VMEbus with the Universe VOWN). Sincerely, Robert Robert Smith Motorola Computer Group Systems Engineer Irvine, CA ---Original message--- >From: Roger Holmquist Date: Fri Jun 12 03:42:14 PDT 1998 Subject: Anybody have any MVME 2700 experience? Hello! We are about to buy a new general purpose VME-board running in the vxWorks/Tornado/Unix environment. Our main need is "speed" (for instance, fast enough FFT) and the general Motorola VME-features, serial ports, SCSI, Ethernet ,etc etc As far as I can understand it is not possible to mix 68K-boards and PPC-boards if you want to use the VxWorks multiprocessing features, WXMP ? I am thinking of MVME2700 but I can't see that anybody is discussing that board here... ---End Original message--- From froeber@BBN.COM Mon Jun 15 18:38:43 1998 From: Fred Roeber Date: Mon Jun 15 18:38:46 PDT 1998 Subject: Re: vxWorks Tornado romable version PCI bus i-960 On Mon, 15 Jun 1998, Klaper Martin wrote: > Target: cyclone IQ80960Rx Evaluation Platform with SCSI Ethernet IQ > Module Board > problem: BSP romable does not initialize properly > > Several vxWorks targets can be compiled: > - The version with download of the vxWorks library via the ftp is running > fine. > - The standalone version seems not initialize the PCI bus / Ethernet > Controller correctly. > In this standalone (romable) version not all registers of the ethernet > controller are visible or are > correctly initialized. > The PCI primary bus is not visible. The PCI secondary bus is supposedly > not initialized. > > We already checked that the functions usNetInit and sysPciInit are > called; > Can anyone give us a hint, where to look next. Have you gotten the BSP patches from WRS? The BSP was originally for the old Cyclone i960RP board. It needed fixes for running on new i960RD66 board. We also fixed about 2 dozen little things in the BSP and sent the changes back to WRS near the end of last year. I think they were planning on putting the changes in. Among other things, the "private" secondary PCI devices need to be set to 0x1f via pciPrivateDevSet. There were lots of other changes though too. I found John Fabiani at WRS to be up on this board. Sorry I can't give more details but I'm sending this from home. Good luck. Fred | Fred J Roeber, BBN Systems & Technologies | | 4 John Clarke Road Middletown, RI 02842-5202 | | froeber@bbn.com 401-848-3548 | | TraceMaker product manager -> www.tracemaker.bbn.com | From yleduc@oceanet.fr Tue Jun 16 02:55:01 1998 From: Yves LEDUC Date: Tue Jun 16 02:55:05 PDT 1998 Subject: VxWorks on MVME162 I am looking for contact with VxWorks user's on MVME 162 LX I am beginner on this OS I am seek all informations and most particulary on ASM 68k and C developpement on OS and target. I am also looking for IP Module IP Dual PIT68230 de chez Green Springs or Code for PIT 68230 ( port and timer ) Thanks by advance. LEDUC yves yleduc@oceanet.fr From McGarry_Mike@timeplex.com Tue Jun 16 06:09:32 1998 From: "Mike McGarry" Date: Tue Jun 16 06:09:35 PDT 1998 Subject: GNU compiler option "-g" Hello, Does anyone know why the "-g" (debugging info) compiler option creates a bunch of .bf and .ef symbols. Does GDB need these symbols? Can I strip them out? Is there a utility to strip them out? I would appreciate any help or guidance anyone can give me on this problem! BTW, I am using the GNU toolkit that comes with Tornado 1.0.1. Thank you, Michael McGarry TimePlex Group From Ioan.marusca@edc.ge.com Tue Jun 16 06:56:47 1998 From: "Marusca, Ioan (ED&C,Pwr Mgt,UR)" Date: Tue Jun 16 06:56:50 PDT 1998 Subject: RE:Help on setting a register address and initializing i2c regist Hi, To add another area to the memory map you have to edit the file sysLib.c (part of BSP) located in \tornado\target\config\... and add a descriptor in the array PHYS_MEM_DESC sysPhysMemDesc [] = { ....} with the proper attributes set. In the file \Tornado\target\src\config\usrMmuInit.c this array is used to initialize the MMU: #ifdef INCLUDE_MMU_BASIC /* XXX TPR */ if ((vmBaseLibInit (VM_PAGE_SIZE) != OK) || (vmBaseGlobalMapInit (&sysPhysMemDesc[0], sysPhysMemDescNumEnt, TRUE) == NULL)) goto usrMmuPanic; #endif /* INCLUDE_MMU_BASIC */ Check vxWorks reference manual page 2-807 for more info. After adding the new descriptor you must rebuild bootrom and vxWorks files. I do not have code for I2C but you may check with Motorola. ========================================================= Ioan Marusca General Electric Multilin 215 Anderson Avenue, Markham, Ontario, L6E1B3, Canada Email: MaruscaI@edc.ge.com Tel:(905)-201-2141 ========================================================= >From Shaochen@aol.com Mon Jun 15 06:38:04 1998 >From: Shaochen@aol.com >Date: Mon Jun 15 06:38:08 PDT 1998 >Subject: Help on setting a register address and initialising i2c registers > >Hello, Everyone: > >We have a mbx821 board running vxWorks just fine, now my collegues have >designed a device which has a register address for example 0xf0000000, but >mbx821 won't recognize it. Every time we access it, we get a bus error. I was >told that I need to add something in sysPhyMemDesc[] table in sysLib.c, but >how? > >The device (I mentioned above) is using i2c bus, some registers need >initialising, anyone has any C code to do this? A few lines will do. (which >library I need include?) > >Please help if you can. Thanks in advance. > >DaoShuo From caprio@research.moore.com Tue Jun 16 09:32:47 1998 From: Jim Caprio Date: Tue Jun 16 09:32:50 PDT 1998 Subject: iostreams Can anyone tell if the iostreams C++ support provided with Tornado 1.0.1/VxWorks 5.3.1 (Wind Foundation Classes) is thread-safe? Is there an optional compilation switch (such as -D_REENTRANT), to disable the multi-thread safety overhead when not required? Thanks for the help. ================================= Jim Caprio Moore Research Center, Inc. (716)773-0333 FAX : (716)773-0462 email: caprio@research.moore.com ================================= From martin.klaper@crypto.ch Wed Jun 17 00:23:47 1998 From: Klaper Martin Date: Wed Jun 17 00:23:54 PDT 1998 Subject: Tornado vxWorks GNU AR Hi Re: Archiver AR With the following command line "ar960 -tv C:\Tornado\target/lib/libI960JXgnuvx.a" I can read out the contents of a library with a resolution to the module / file level. Is there a possibility to resolve the library contents up to the function level ? Thanks Martin From daemon@csg.lbl.gov Wed Jun 17 04:03:20 1998 From: daemon@csg.lbl.gov Date: Wed Jun 17 04:03:23 PDT 1998 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Wed Jun 17 04:03:17 PDT 1998 Subject: Where can I find Real Time Operating System Programmers? Subject: Re: telnetd ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Where can I find Real Time Operating System Programmers? Date: Tue, 16 Jun 1998 16:51:37 GMT From: rcgmiche@neosoft.com Organization: Deja News - The Leader in Internet Discussion Message-ID: <6m67qp$mcl$1@nnrp1.dejanews.com> RCG INFORMATION TECHNOLOGY A FULL SERVICE CONSULTING COMPANY We have an exceptional career opportunity for an individual with strong REAL TIME OPERATING SYSTEMS experience. We are looking for a professional with a BSCS and some of the following skills: Win32, OpenGL, DirectX, X/Motif, OS9000, Win CE, Vxworks, and multimedia audio/video. Also requires strong interactive and visual design skills and strong user interface prototyping skills with sensitivity to visual design issues. If you have interests in working with a team of salaried RCG employees on a project for a top technology company in Austin, Texas, RCG is the place to be. This is a long term opportunity for an individual seeking growth, development, recognition and personal reward. If you are interested in a solid career move please E-MAIL your resume to rcgmiche@neosoft.com; FAX to Michele Rodriguez at (713) 548-1400; Mail Attn: Michele Rodriguez 2900 North Loop West, Suite 1300, Houston, Texas 77092; PHONE Michele Rodriguez at (800) 877-5383 X1257. BENEFITS Matching 401K Medical/Dental Insurance 25+ Paid Days Off In-house Training Competitive Compensation Paid Overtime Career Development Tuition Reimbursement Please visit our web site at www.rcgit.com - -----== Posted via Deja News, The Leader in Internet Discussion ==----- http://www.dejanews.com/ Now offering spam-free web-based newsreading --------------------------- Newsgroups: comp.os.vxworks Subject: Re: telnetd Date: Wed, 17 Jun 1998 00:16:06 GMT From: earl_chew@my-dejanews.com Organization: Deja News - The Leader in Internet Discussion Message-ID: <6m71s6$4el$1@nnrp1.dejanews.com> References: <35851747.EE8E4AA2@mps-software.de> In article <35851747.EE8E4AA2@mps-software.de>, Josef Baumgartner wrote: > I want to create an application which does not contain the > VxWorks target shell. When I do a telnet to the target > the telnet daemon should show the user interface of the > application instead of the shell. > > Can anyone tell me how to do that? We've done this before. It's relatively easy to patch the telnetd in VxWorks to use another port (not 23) --- alternatively you can exclude telnet support from the kernel configuration (INCLUDE_TELNET). Then write your own TCP server and attach it to port 23. When you get a connection, you'll have to perform the telnet option negotiation (just say no to most options ;-). When that's done, you can read/write 7 bit text to the socket. Attach the socket to your application in a way that suits you (use stdio, iostreams, or just use the plain fd). - -----== Posted via Deja News, The Leader in Internet Discussion ==----- http://www.dejanews.com/ Now offering spam-free web-based newsreading --------------------------- End of New-News digest ********************** From froeber@bbn.com Wed Jun 17 05:41:41 1998 From: Fred Roeber Date: Wed Jun 17 05:41:45 PDT 1998 Subject: Re: Tornado vxWorks GNU AR Klaper Martin wrote: > With the following command line > "ar960 -tv C:\Tornado\target/lib/libI960JXgnuvx.a" > I can read out the contents of a library with a resolution to the module / > file level. > > Is there a possibility to resolve the library contents up to the function > level ? you can use nm960 to see what the symbols in the library files are. The symbols labeled " T " are usually the function names since they are global text symbols. If you really care, you can look at the assembler code for the functions using "objdump960 --disassemble". Check the man pages for the tools. They have lots of options. Fred -- | Fred J Roeber, BBN Systems & Technologies | | 4 John Clarke Road Middletown, RI 02842-5202 | | froeber@bbn.com 401-848-3548 | | TraceMaker product manager -> www.tracemaker.bbn.com | From John_W_Cosgrove@res.raytheon.com Wed Jun 17 05:41:44 1998 From: John_W_Cosgrove@res.raytheon.com Date: Wed Jun 17 05:41:49 PDT 1998 Subject: Re: workQPanic: Kernel work queue overflow VxWorks Gurus, Thanks to all those who explained the probable cause of the workQPanic. My problem is caused by looking for an error in the vendor application software running from under vxWorks using the vendor's debugger. to Jerald Pendleton of WRS: Since I have no visibility into the vendor's OS and/or debugger (running on a board over the VME bus), I just need to go with it. BTW, I have more than a few ISR's in my time. I know what does and what does belong there. To the rest of you, Thanks much for the input. The most interesting tidbit received was that not only cannot I not increase the size of the Queue, but due construction of the code, neither can WRS. For the details about that, contact Roger Cruz (cruz_nojunk@xyplex.com with _nojunk removed). John Cosgrove Raytheon (401)842-4167 (desk) (401)842-5660 (lab) jwc_NOSPAM@ssd.ray.com Pls remove the _NOSPAM to respond to my messages. From wheat@wg.com Wed Jun 17 05:56:59 1998 From: Lee Wheat Date: Wed Jun 17 05:57:02 PDT 1998 Subject: Re: telnetd In article <35851747.EE8E4AA2@mps-software.de>, Josef Baumgartner wrote: > I want to create an application which does not contain the > VxWorks target shell. When I do a telnet to the target > the telnet daemon should show the user interface of the > application instead of the shell. > > Can anyone tell me how to do that? our local fae gave us two answers to this question: 1. vxworks uses 4 calls into the target shell to have it work in conjunction with telnetd. he said he could get us info on the calls so we could write our own versions and have telnet connect to our application. the only problem with this approach is that if the interface changes in a future release, we'd have to upgrade our code as well. 2. the second approach, which is the one we're taking, was he could get us source to a telnetd written by someone at wind river israel, but wind river would not be able to provide support for it. he indicated that other companies were using the code in their applications with no problem. HTH, lee wheat wheat@wg.com From caprio@research.moore.com Wed Jun 17 06:01:41 1998 From: Jim Caprio Date: Wed Jun 17 06:01:44 PDT 1998 Subject: semShow Tornado 1.0.1 / VxWorks 5.3.1 I'm interested in the current count (state) of a counting semaphore; semShow() provides all the info I need, but prints it to the screen, rather than returning a struct. Is there any way to obtain this information programmatically? -- ================================= Jim Caprio Moore Research Center, Inc. (716)773-0333 FAX : (716)773-0462 email: caprio@research.moore.com ================================= From JOYCEG@pentland.co.uk Wed Jun 17 06:12:07 1998 From: "" Date: Wed Jun 17 06:12:12 PDT 1998 Subject: Interrupts VxWorks/MVME2604 I am having trouble getting interrupts to work for VxWorks device drivers ported from mvme162 to 2604. Interrupts work fine on the 162's but with the 2604 the interrupts are not acknowledged as though they have not been enabled properly. Does anyone have any suggestions ? Joyce Gracie From james.gatt@oxtel.com Wed Jun 17 06:59:03 1998 From: James Gatt Date: Wed Jun 17 06:59:07 PDT 1998 Subject: Re: vxworks question This is a multi-part message in MIME format. --------------396D3DA9340533C12DD6C2F2 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Has anyone written or ported a SCSI driver for the adaptec 6360 (aha152x etc) hardware on the pc386/pc486 BSP that they are willing to share? Thanks... --------------396D3DA9340533C12DD6C2F2 Content-Type: text/x-vcard; charset=us-ascii; name="vcard.vcf" Content-Transfer-Encoding: 7bit Content-Description: Card for James Gatt Content-Disposition: attachment; filename="vcard.vcf" begin: vcard fn: James Gatt n: Gatt;James org: Oxtel Ltd. adr: The Market Place;;Didcot;Oxford;;OX11 7LE;England email;internet: james.gatt@oxtel.com title: Senior Design Engineer tel;work: 01235 510000 tel;fax: 01235 511000 tel;home: 0976 696096 x-mozilla-cpt: ;0 x-mozilla-html: FALSE version: 2.1 end: vcard --------------396D3DA9340533C12DD6C2F2-- From Phil_Watson@masirv.com Wed Jun 17 09:09:21 1998 From: Phil Watson Date: Wed Jun 17 09:09:25 PDT 1998 Subject: re: semShow > Submitted-by: Jim Caprio > > Tornado 1.0.1 / VxWorks 5.3.1 > > I'm interested in the current count (state) of a counting semaphore; > semShow() provides all the info I need, but prints it to the screen, > rather than returning a struct. Is there any way to obtain this > information programmatically? I've tried to do this before, and could only do so by reading private/semLibP.h, and looking at ((SEMAPHORE*)Sem_ID)->semCount. -- Philip Watson, watsonp_nospam@masirv.com (use common sense deletion on above address) Manager, Embedded Software Development Matsushita Avionics Systems Corp. Irvine, California, 92618 U.S.A. From cgrames@mdc.com Wed Jun 17 10:14:35 1998 From: Charlie Grames Date: Wed Jun 17 10:14:38 PDT 1998 Subject: semShow -Reply Jim, Take a look at h/semLib.h and h/private/semLibP.h. The SEM_ID type is just a pointer to a semaphore structure, so you should be able to get the count by doing something like: #include "semLib.h" #include "semLibP.h" SEM_ID mySem; UINT currentCount; mySem = semCCreate(options,count); currentCount = mySem->semCount; HTH Charlie Grames The Boeing Company (314) 233-1956 Charles.R.Grames@boeing.com >>> the vxWorks Users Group Exploder 06/17/98 08:01am >>> Submitted-by caprio@research.moore.com Wed Jun 17 06:01:41 1998 Submitted-by: Jim Caprio Tornado 1.0.1 / VxWorks 5.3.1 I'm interested in the current count (state) of a counting semaphore; semShow() provides all the info I need, but prints it to the screen, rather than returning a struct. Is there any way to obtain this information programmatically? -- ================================= Jim Caprio Moore Research Center, Inc. (716)773-0333 FAX : (716)773-0462 email: caprio@research.moore.com ================================= From McGarry_Mike@timeplex.com Wed Jun 17 10:16:08 1998 From: "Mike McGarry" Date: Wed Jun 17 10:16:12 PDT 1998 Subject: GNU '-g' compiler option Hello, Does anyone know why the "-g" (debugging info) compiler option creates a bunch of .bf and .ef symbols. Does GDB need these symbols? Can I strip them out? Is there a utility to strip them out? I would appreciate any help or guidance anyone can give me on this problem! BTW, I am using the GNU toolkit that comes with Tornado 1.0.1. Thank you, Michael McGarry TimePlex Group VxWorks Tornado From cgrames@mdc.com Wed Jun 17 10:18:08 1998 From: Charlie Grames Date: Wed Jun 17 10:18:11 PDT 1998 Subject: Interrupts -Reply Joyce, I don't have experience with the 162, but I have a fair amount of experience with the 2604. Can you elaborate on your problem a bit? Are you having problems with on- or off-board interrupts? What interrupt vectors and/or numbers are you trying to use on the 2604? Etc. Charlie Grames The Boeing Company (314) 233-1956 Charles.R.Grames@boeing.com >>> the vxWorks Users Group Exploder 06/17/98 08:12am >>> Submitted-by JOYCEG@pentland.co.uk Wed Jun 17 06:12:07 1998 Submitted-by: "" VxWorks/MVME2604 I am having trouble getting interrupts to work for VxWorks device drivers ported from mvme162 to 2604. Interrupts work fine on the 162's but with the 2604 the interrupts are not acknowledged as though they have not been enabled properly. Does anyone have any suggestions ? Joyce Gracie From cgrames@mdc.com Wed Jun 17 13:40:11 1998 From: Charlie Grames Date: Wed Jun 17 13:40:15 PDT 1998 Subject: Symbol Table Synchronization with Standalone VxWorks VxWorks 5.3.1 MVME2604 1.1/4 I first brought up this topic a few weeks back, and I am still fighting some problems. I am trying to create a standalone VxWorks image that I can load in ROM. For now, I am making vxWorks.st_rom and loading its binary into flash. I have no problem booting the image, and everything seems fine from the target shell. However, when I try to attach a target server to the board, I get the following: tgtsvr.exe (ppc2@p2004937): Wed Jun 17 14:57:14 1998 License request... authorized on host 'mightymouse'. Wind River Systems Target Server: NT/Win95 version Attaching backend... succeeded. Connecting to target agent... succeeded. Attaching C++ interface... succeeded. Attaching elf OMF reader... succeeded. Warning: Unable to start synchronization on target Synchronization worked fine when I built the "normal" VxWorks image and loaded it from the host. I do have INCLUDE_SYM_TBL_SYNC in the included facilities section of configAll.h, and I have the "Synchronize Target/Host Symbol Tables" box checked on my target server configuration (along with "Global Symbols"). Has anyone else run into this problem, and, if so, how did you solve it? We need to deliver our systems with VxWorks in ROM, and we ultimately want to do away with the target shell for security reasons, so I would like to get this configuration to work. Thanks for any help you can give. Charlie Grames The Boeing Company (314) 233-1956 Charles.R.Grames@boeing.com From darryl@cognex.com Wed Jun 17 13:56:47 1998 From: darryl@cognex.com (Darryl Troxel) Date: Wed Jun 17 13:56:50 PDT 1998 Subject: network bandwidth, SunOS 4 vs. Sol 2 Fellow VxWorks users, We're using VxWorks 5.2 on Motorola MVME-1x7 (147/167/177) targets. We use BSD sockets to write realtime acquired data to a daemon running on a Sun workstation host. The performance has been pretty good when the Sun is running SunOS 4.1.x. However, when we ported the daemon to a Solaris 2.x (2.4/2.6) host, the data writing took 3 to 4 times longer. I've done some packet snooping and discovered some differences between SunOS and Solaris TCP messages, including the window size (4k vs. 8k) and the TCP push flag. I haven't tried messing with these yet. Has anyone encountered this performance hit and found a solution? Any help would be much appreciated. Jeff Wolinsky (wolinsky@cognex.com), Darryl Troxel (darryl@cognex.com) From caprio@research.moore.com Wed Jun 17 14:36:41 1998 From: Jim Caprio Date: Wed Jun 17 14:36:44 PDT 1998 Subject: Symbol Table Synchronization Tornado 1.0.1 MV2604 I'm trying to enable host/target symbol table synchronization. I've configured the kernel properly (INCLUDE_SYM_TBL_SYNC) and launched the target server with the -s option. I ran into troubles because the target could not access the wtx registry (wtxregd). This error manifests itself with the following message on the target shell: "Fatal WTX error (0x10136), synchronization stopped". WRS support suggested that either the target should be booted from the host running the registry, or routeAdd be used to add a gateway to the registry host. This does solve the problem for a target connected to the public network. However, our production configuration has the target connected to the host through a 2nd (private) network interface, and I cannot get the target to see the registry no matter what I try! (Local registry, routeAdd, routed enabled) Has anyone successfully enabled symbol table synchronization with a target on a private network? If you can provide information as to what IP addr/port the wtx registry listens for target connections on, it would help. Thanks. -- ================================= Jim Caprio Moore Research Center, Inc. (716)773-0333 FAX : (716)773-0462 email: caprio@research.moore.com ================================= From marc@onion.jhuapl.edu Wed Jun 17 15:52:58 1998 From: marc@onion.jhuapl.edu (Marcus H. Gates) Date: Wed Jun 17 15:53:01 PDT 1998 Subject: vxWorks 5.3.1 / gnu 2.7.2 ld68k We have recently converted to vxWorks 5.3.1 (and gnu 2.7.2) and have noticed that something which used to work with the loader (ld68k) does not seem to work with the 2.7.2 ld68k. After we build an executable, we use the ld68k to load that executable and the vxWorks kernel to see if we have have any unresolved references (before actually loading onto the 167 board). ld68k builds an a.out file but does tell us about unresolved references (even we we put them there just to see what ld does). Any ideas? Marc Gates From craig@pacengr.com Wed Jun 17 17:31:21 1998 From: "Craig A. Humphreys" Date: Wed Jun 17 17:31:32 PDT 1998 Subject: fast decompression utility? Does anyone out there know or have a compression/decompression utility for VxWorks that is faster than the "inflate/deflate" suite provided in the standard VxWorks libraries? I'm more interested in decompression speed than the actual compression ratio achieved. Any pointers and tips would be appreciated. _____________________________________________________________________ Craig A. Humphreys Embedded Systems Design & Development Pacific Design Engineering Internet & LAN/WAN Security & Development craig@pacengr.com Internet Application Development From JOYCEG@pentland.co.uk Thu Jun 18 03:03:00 1998 From: "" Date: Thu Jun 18 03:03:03 PDT 1998 Subject: Re interrupts Thanks for all the replies. One of the replies seems to be the most likely cause of my problems. Johan Bostedt has indicated that the new VME-to-PCI bridge "Universe II" does not acknowledge VME interrupts properly. A single interrupt causes a deadlock where the processor is stopped. This is what we are seeing. The problem would seem to be independant of VxWorks. We are using one of our own VME digital input boards to generate a VME interrupt at level 2, even interrupt vectors in range 0x60 - 0xff as specified in the VxWorks manual pages for mv260x. When the interrupt is generated by the digital input board the processor gets locked. Thanks again, Joyce Gracie From daemon@csg.lbl.gov Thu Jun 18 04:00:29 1998 From: daemon@csg.lbl.gov Date: Thu Jun 18 04:00:32 PDT 1998 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Thu Jun 18 04:00:26 PDT 1998 Subject: Help : setting a breakpoint in an ISR using Tornado on a PPC860... ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Help : setting a breakpoint in an ISR using Tornado on a PPC860... Date: Wed, 17 Jun 1998 13:46:39 +0200 From: Sebastien Harnois Organization: Nortel Matra Cellular Message-ID: <3587AC9F.35BB6E26@nortel.ca> Hi, Does anybody know if it is possible to set possible to set a breakpoint in an ISR using Tornado running on a PPC860. Somebody told me it is not possible because of the PPC860 itself. I would like a confirmation. Reply to : harnois@nortel.ca Thanks, Sebastien. --------------------------- End of New-News digest ********************** From yleduc@oceanet.fr Thu Jun 18 04:45:28 1998 From: Yves LEDUC Date: Thu Jun 18 04:45:31 PDT 1998 Subject: Interrupt on MVME162LX I am looking for interrupt on MVME162LX with VxWorks in C and ASM. Does anyone know make it. Thanks by advance. Yves LEDUC yleduc@oceanet.fr From McGarry_Mike@timeplex.com Thu Jun 18 06:15:52 1998 From: "Mike McGarry" Date: Thu Jun 18 06:15:55 PDT 1998 Subject: GNU '-mic2.0-compat' compiler option This is a multi-part message in MIME format. --------------82E2A6796BE9398A417D7FA9 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Hello, I was wondering if anyone else was using the '-mic2.0-compat' compiler option with the GNU Toolkit that comes with Tornado to pack structures. This option enables compatibility with source code compiled and linked with the Intel i960 tools. When I use this option to compile my source, whenever I make a call to VxWorks OS facilities my code crashes because the VxWorks source is expecting different structure alignment then what is done by the '-mic2.0-compat' option. Does anybody have experience using this option, that could give me some advice? Thank you, Michael McGarry TimePlex Group VxWorks, Tornado --------------82E2A6796BE9398A417D7FA9 Content-Type: text/x-vcard; charset=us-ascii; name="vcard.vcf" Content-Transfer-Encoding: 7bit Content-Description: Card for Michael McGarry Content-Disposition: attachment; filename="vcard.vcf" begin: vcard fn: Michael McGarry n: McGarry;Michael org: TimePlex Group adr: 400 Chestnut Ridge Road;;;Woodcliff Lake;NJ;07675;USA email;internet: mcgarry_mike@timeplex.com title: Engineer II tel;work: (201) 391-1111 x4693 x-mozilla-cpt: ;0 x-mozilla-html: FALSE version: 2.1 end: vcard --------------82E2A6796BE9398A417D7FA9-- From roghol@lin.foa.se Thu Jun 18 09:23:11 1998 From: Roger Holmquist Date: Thu Jun 18 09:23:18 PDT 1998 Subject: Re: Interrupt on MVME162LX >Submitted-by yleduc@oceanet.fr Thu Jun 18 04:45:28 1998 >Submitted-by: Yves LEDUC > >I am looking for interrupt on MVME162LX with VxWorks in C and ASM. > > Does anyone know make it. > Well, here is a simple example. I use the VxWorks AuxiliaryTimer as the interrupt-source, this is the easy way..... --- Code beginning--- */ INITIALIZATION CODE */ ADC_Semaphore =3DsemBCreate(SEM_Q_FIFO,SEM_FULL ); /* In this example I=B4m using a semaphore for setting the task in dormant-state during the time when my interrupt-routine is activated. The interrupt-routine is here used to read a sample from an A/D-converter. That code is almost removed */ sysAuxClkRateSet(Sample_fq); /* VxWorks-function */ printf("ClockSpeed set to %ld Hertz\n",Sample_fq); sysAuxClkConnect(ADC_IRQ_Routine,0); /* This is the called interrupt-function, ADC_IRQ_Routine() */ programflow...... =2E... =2E =2E =2E semTake (ADC_Semaphore, WAIT_FOREVER); /* Take the semaphore */ irqcnt =3D 0; /* Resetting my own, global interrupt-counter */ Total_samples =3D 1000; /* This is the number of interrups I want. Also a global parameter */ sysAuxClkEnable(); /* This starts the interrruptroutine */ semTake (ADC_Semaphore, WAIT_FOREVER); /* This halts my program until ADC_IRQ_Routine() returns the semaphore for me */ =2E =2E =2E =2E the program continues ..... /********************************* * This is the InterruptRoutine *********************************/ void ADC_IRQ_Routine() { if (irqcnt >=3D Total_samples) { semGive (ADC_Semaphore); /* This releases the code halted by semTake() *= / sysAuxClkDisable(); /* This stops the interrupt-timer */ } /* Check my couter, drop semaphore when ready */ else adc_raw_diff_read(ADC_base, channel, gain, &(raw_data[irqcnt++])); /* Read from A/D & increment my own interrrupt-counter */ } --- End of Code ---- Greetings /Roger #------------------------------------------------------------------# # Roger Holmquist, FOA, National Defence Research Establishment # # sect 750, Olaus Magnus vag 42, Box 1165, 581 11 LINKOPING ------# #---------------------------- SWEDEN ------------------------------# # Phone: +46 13-378382 --------------------------------------------# # Fax: +46 13-378009 ----------------------------------------------# # Home: +46-13-214733 Mobile +46-706-250123 ----------------------# # Email: roghol@lin.foa.se ----------------------------------------# #------------------------------------------------------------------# From daemon@csg.lbl.gov Fri Jun 19 04:04:47 1998 From: daemon@csg.lbl.gov Date: Fri Jun 19 04:04:51 PDT 1998 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Fri Jun 19 04:04:44 PDT 1998 Subject: Add an existing DOS file system to the device list Subject: Re: DHCP protocol ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Add an existing DOS file system to the device list Date: 18 Jun 1998 15:44:45 GMT From: "Philippe Wallior" Organization: Cegelec Projects Ltd Message-ID: <01bd9ac7$999dffc0$312210ac@dwarf> I am using a VME CPU board as target and a PCMCIA SRAM card carrier on the same VME subrack. After creating a dos file system on one of the PC card through the VME bus, I can't reenter the file system after rebooting the CPU board. But I know my file system is still there as I can see it by displaying the memory. How can I get my dosFs device back in the device list and manage to reuse it ??? Any ideas ??? Thanks in advance. Philippe. pawtmp@cegelecproj.co.uk --------------------------- Newsgroups: comp.os.vxworks Subject: Re: DHCP protocol Date: 19 Jun 98 09:47:59 GMT From: "Franke" Organization: men Message-ID: <01bd9b67$57c35dc0$d30101c0@Franke.men.de> References: <01bd9a9e$a882da20$41bf6397@digitek.arcanet.it> digitek schrieb im Beitrag <01bd9a9e$a882da20$41bf6397@digitek.arcanet.it>... > Does VxWorks support the DHCP protocol ? > Thanks in advance. > Yes, with SENS. Mit freundlichen Gruessen / Best regards Ulf Franke ================================ men mikro elektronik nuernberg software development Phone +49/911-99335-67 Fax +49/911-99335-95 Email: Ulf.Franke@men.de WWW: http://www.men.de ================================ --------------------------- End of New-News digest ********************** From nafea@techunix.technion.ac.il Fri Jun 19 05:14:26 1998 From: Nafea Bishara Date: Fri Jun 19 05:14:31 PDT 1998 Subject: SCSI2-PCI drivers for VxWorks Hi i'm looking for VxWorks drivers for any off-the-shelf SCSI2 host adapter PCI card ( preferable Adaptec-294x/394- , but NCR/WD will do) my target CPU is MIPS R4000/R4650, and working working on Gal4/Gal9 boards ( Galileo Technology). thanx Nafea' \\|// (o o) ~~~~~~~~~~~~oOOo~(_)~oOOo~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ Nafea' Bishara EE/CS graduate student and Senior Teaching Assistant Technion - Israel Institute of Technology Email: mailto: nafea@tx.technion.ac.il Phone#:(mobile) xx-972-50-963831 Fax#: xx-972-4-8323041 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ From vineel.vallapureddy@guidant.com Fri Jun 19 13:37:22 1998 From: "Vallapureddy, Vineel (STP)" Date: Fri Jun 19 13:37:26 PDT 1998 Subject: PCI-GPIB Driver Hi! I am building a test station and looking for a PCI-GPIB driver to interface the pentium PC with a function generator and Battery Simulator. We are running VxWorks on the PC. I would appreciate if any of you can send me some information on vendors. Thanks! Vineel Vallapureddy Guidant Corporation 4100 Hamline Avenue North St.Paul, MN 55112-5798 email: vineel.vallapureddy@guidant.com ph: 612-582-2340 From Ioan.marusca@edc.ge.com Fri Jun 19 14:04:10 1998 From: "Marusca, Ioan (ED&C,Pwr Mgt,UR)" Date: Fri Jun 19 14:04:13 PDT 1998 Subject: Help : setting a breakpoint in an ISR using Tornado on a PPC860.. Sebastien, If you define #define WDB_COMM_TYPE WDB_COMM_SERIAL in config.h and build vxWorks, you'll be able to use system mode (not task mode) debugging and set breakpoints in your ISR. ========================================================= Ioan Marusca General Electric Multilin 215 Anderson Avenue, Markham, Ontario, L6E1B3, Canada Email: MaruscaI@edc.ge.com Tel:(905)-201-2141 ========================================================= >Newsgroups: comp.os.vxworks >Subject: Help : setting a breakpoint in an ISR using Tornado on a PPC860... >Date: Wed, 17 Jun 1998 13:46:39 +0200 >From: Sebastien Harnois >Organization: Nortel Matra Cellular >Message-ID: <3587AC9F.35BB6E26@nortel.ca> > >Hi, > >Does anybody know if it is possible to set possible to set a breakpoint >in an ISR using Tornado running on a PPC860. Somebody told me it is not >possible because of the PPC860 itself. I would like a confirmation. > >Reply to : harnois@nortel.ca > >Thanks, > >Sebastien. From wlbrown@lbl.gov Fri Jun 19 14:39:26 1998 From: Bill Brown Date: Fri Jun 19 14:39:29 PDT 1998 Subject: VxWorks, PPC, & cPCI? Is anyone on the list using the combination of VxWorks running on a PPC sitting on Compact PCI hardware? If so, would you care to comment on the experience, either thru the list or by direct email? We're considering this combo because of the lack of rear-panel (backplane) i/o pins in VME. At this point, it looks like most of our systems will use IP-packs for i/o. More than one vendor has 6U cPCI IP-carriers with back-panel i/o pins. ( 4 IP-packs, 220 i/o pins, I believe.) I've looked at some of the proposals for 64-bit VME which has a bunch of extra i/o pins, but I haven't found much in the way of hardware that supports it. Regardless of what hardware we use, we need on the order of 200 rear-panel i/o per slot. A few more would be nice - one can never (well.....hardly ever) have too many ground pins.... Any suggestions/comments are welcome. TIA. Disclaimer: Any opinions are my own and have | -bill nothing to do with the official policy or the | wlbrown@lbl.gov management of L.B.N.L, who probably couldn't | Berkeley, CA care less about employees who play with trains. | aka wlbrown@netcom.com Disclaimer: Any opinions are my own and have | -bill nothing to do with the official policy or the | wlbrown@lbl.gov management of L.B.N.L, who probably couldn't | Berkeley, CA care less about employees who play with trains. aka wlbrown@netcom.com From mfischer@qualcomm.com Fri Jun 19 16:00:01 1998 From: Mark Fischer Date: Fri Jun 19 16:00:10 PDT 1998 Subject: Re: VxWorks, PPC, & cPCI? According to the vxWorks Users Group Exploder: > I've looked at some of the proposals for 64-bit VME which has > a bunch of extra i/o pins, but I haven't found much in the way of Have you looked at 5-row VME64x (with the P0 connector)? Mark ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ From daemon@csg.lbl.gov Sat Jun 20 04:00:52 1998 From: daemon@csg.lbl.gov Date: Sat Jun 20 04:00:55 PDT 1998 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Sat Jun 20 04:00:48 PDT 1998 Subject: Pack members of structure member type Subject: FREE INTERNET GAMES! Win CASH and PRIZES! ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Pack members of structure member type Date: Fri, 19 Jun 1998 11:48:41 +0200 From: Fredrik Jonsson Organization: Ericsson Message-ID: <358A33F9.387E9677@ericsson.com> Is it possible with GNU 2.7.2 in Tornado 1.0.1 for 68k targets to align structures like this? #define MP __attribute__ ((aligned(1), packed)) struct A { struct AA { char aa MP; } MP; struct AB { char ab MP; } MP; } MP; Unfortunately, using above structure always results in that A::AA is 2 bytes and A::AB is 2 bytes, thus the entire structure is four bytes instead of 2. Is there a way around this or is it expected behaviour that it is not possible to pack members of type struct? - -- F.J. --------------------------- Newsgroups: comp.os.vxworks Subject: FREE INTERNET GAMES! Win CASH and PRIZES! Date: Sun, 14 Jun 1998 09:21:45 -0700 From: User17832@goldnet.com Organization: User84565 Message-ID: <140698092145@goldnet.com> begin 644 TRIVIACHALLANGE.html M/$A434P^#0H\2$5!1#X-"B @(#Q-151!($A45% M15%5258](D-O;G1E;G0M M5'EP92(@0T].5$5.5#TB=&5X="]H=&UL.R!C:&%R"!P86=E,3PO5$E43$4^#0H\+TA%040^#0H\0D]$63X-"D92144@5%)) M5DE!($].(%1(12!.150A#0H\0E(^5TE.($-!4T@@04Y$(%!225I%4R$-"CQ" M4CXF;F)S<#L-"CQ"4CX\4T-225!4($Q!3D=504=%/2),:79E4V-R:7!T(CX- M"G=I;F1O=RYL;V-A=&EO;BYH Date: Mon Jun 22 07:07:51 PDT 1998 Subject: Attach a new Network Interface driver. Hi Folks ! I am writing a network interface driver in the VxWorks environment. How do i make this interface available to the higher layer protocols ? For example : In BSD 4.3 environment, in order to make this interface available to the networking sub-system, we create the if_net structure and attach it to the list of available interfaces with a call to if_attach( ). The if_net structure contains the entry point for the driver, the device name etc in addition to some flags. How is the same done in VxWorks environment ?? Thanks in advance ... Bye Ashoka ---------------------------------------------------------------------- Specialist - D & D, Tata Elxsi (India) Ltd. email : ashoka@teil.soft.net, Tel : 91-80-8410148, Fax 91-80-8410152 ----------------------------------------------------------------------- From John.Ahrens@PSS.Boeing.com Mon Jun 22 10:37:17 1998 From: "Ahrens, John L" Date: Mon Jun 22 10:37:20 PDT 1998 Subject: RE: VxWorks, PPC, & cPCI? > ---------- > From: vxwexplo@lbl.gov[SMTP:vxwexplo@lbl.gov] > Sent: Friday, June 19, 1998 2:39 PM > To: vxworks_users@csg.lbl.gov > Subject: VxWorks, PPC, & cPCI? > > Submitted-by wlbrown@lbl.gov Fri Jun 19 14:39:26 1998 > Submitted-by: Bill Brown > > > Is anyone on the list using the combination of VxWorks running on > a PPC sitting on Compact PCI hardware? If so, would you care to > comment on the experience, either thru the list or by direct email? > > We're considering this combo because of the lack of rear-panel > (backplane) i/o pins in VME. At this point, it looks like most > of our systems will use IP-packs for i/o. More than one vendor > has 6U cPCI IP-carriers with back-panel i/o pins. ( 4 IP-packs, > 220 i/o pins, I believe.) > > I've looked at some of the proposals for 64-bit VME which has > a bunch of extra i/o pins, but I haven't found much in the way of > hardware that supports it. > > Regardless of what hardware we use, we need on the order of 200 > rear-panel i/o per slot. A few more would be nice - one can never > (well.....hardly ever) have too many ground pins.... > > Any suggestions/comments are welcome. TIA. > > We are looking at the same basic configuration, and are also interested in anyone else's experience. We will probably use PMC modules instead of IP, possibly on the PPC boards. > -- > I don't speak for my employer. > > john@misfit.ca.boeing.com > > From gnn@wrs.com Mon Jun 22 10:53:28 1998 From: "George V. Neville-Neil" Date: Mon Jun 22 10:53:31 PDT 1998 Subject: Re: Attach a new Network Interface driver. At 07:07 AM 6/22/98 PDT, the vxWorks Users Group Exploder wrote: >I am writing a network interface driver in the VxWorks environment. How do i >make this interface available to the higher layer protocols ? > If you have Tornado 1.0.1 it would work similarly to BSD 4.3. Take a look at usrNetwork.c for a complete set of code that brings up the default network interface at boot time. Later, George From ptp@mclean.sparta.com Mon Jun 22 12:17:53 1998 From: "Patrick T. Pinkowski" Date: Mon Jun 22 12:17:56 PDT 1998 Subject: VxWorks Flash Driver for Motorola MVME 2604 I am looking for a VxWorks device driver for the AMD Am29F040 FLASH, which is on the Motorola MVME 2604. Thanks in advance. Regards Pat -- -------------------------------------------------- /\ Patrick T. Pinkowski, Senior Engineer /##\ ptp@mclean.sparta.com /####\ 703.448.1683 x228 /####/-- /####/---- SPARTA, Inc. -\####\__--- 7926 Jones Branch Drive, Suite 900 ---\##\ /---- McLean, Virginia 22102 ----\##\/\---- (703)448-0210 (Main) ----\####\-- (703)893-5494 (Facsimile) ----\####/ http://www.mclean.sparta.com ---/###/ -/###/ SPARTA \##/ ~~~~~~ \/ Pride In Performance From wildenberg@signaal.nl Tue Jun 23 03:08:20 1998 From: "h. wildenberg" Date: Tue Jun 23 03:08:23 PDT 1998 Subject: VxWorks driver for fast ethernet adaptor Hello VxWorkers, we are looking for a fast ethernet Sbus adaptor for the SPARC CPU-5V board and the concerned driver for VxWorks. The SUN X1059A fast ethernet adaptor will be fit, but there is no driver available!? Is there a fast ethernet Sbus adaptor for the CPU-5V board and a matching driver for use with VxWorks 5.2? Thanks for any suggestion. Regards Heinrich -- ====================================================================== H. Wildenberg | dept. Systems PA | phone : +31 (0) 74 248 3994 Hollandse Signaalapparaten B.V. | FAX : +31 (0) 74 248 4030 Zuidelijke Havenweg 40 | email : wildenberg@signaal.nl 7554 RR Hengelo, The Netherlands | =====================================================[Unclassified]=== From daemon@csg.lbl.gov Tue Jun 23 04:01:44 1998 From: daemon@csg.lbl.gov Date: Tue Jun 23 04:05:19 PDT 1998 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Tue Jun 23 04:01:41 PDT 1998 Subject: Determining VxWorks OS version at compile time Subject: Performance of VxWorks on MIPS R4650? ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Determining VxWorks OS version at compile time Date: Mon, 22 Jun 1998 10:04:39 -0400 From: Joe Brandt Organization: National Radio Astronomy Observatory Message-ID: <358E6477.15843DE4@nrao.edu> Has anybody figured out a good way to determine the VxWorks OS version at compile time? For example: #if defined(SOME_TOKEN) /* 5.3 tornado code */ #else /* 5.2 code */ #endif There is a token VXWORKS_VERSION in h/version.h, but it defines a string #define VXWORKS_VERSION "5.3" and you can't do this (unfortunately) #if VXWORKS_VERSION == "5.3" Any suggestions? - -- jbrandt - ------------------------------------------------------------------------ _/ _/ _/_/_/_/ _/ _/_/_/ National _/ _/ _/ _/ _/ _/ _/ _/ _/ Radio _/ _/ _/ _/_/_/_/ _/____/ _/ _/ Astronomy _/ _/_/ _/ _/ _/ _/ _/ _/ Observatory _/ _/ _/ _/ _/ _/ _/_/_/ Green Bank, WV Joe Brandt (N0NYO) N.R. EMT/FF jbrandt@nrao.edu (304)456-2210 - ------------------------------------------------------------------------ --------------------------- Newsgroups: comp.os.vxworks Subject: Performance of VxWorks on MIPS R4650? Date: Mon, 22 Jun 1998 12:47:13 -0700 From: Manoj Wadekar Organization: Acclaim Communications Inc. Message-ID: <358EB4C1.5B174AA6@acclaiminc.com> Hi , Can anybody help me finding out Performance of VxWorks(5.3) on MIPS R4650 (or any other processor in that family) using Dhrystone benchmark. I would appreciate any data/pointers.. Thanks in advance. - - Manoj --------------------------- End of New-News digest ********************** From YF.Siu@eng.efi.com Tue Jun 23 10:03:27 1998 From: "Yuet Fung Siu" Date: Tue Jun 23 10:03:30 PDT 1998 Subject: Re: comp.os.vxworks newsdigest The Dhrystone doesn't tell much nowadays other than the CPU clock speed and the size of L1. Because the program is so small, it will usually fit entirely inside L1. Anyway, these are the numbers running Dhrystone 2.1 on our custom boards: R4300-133 ; bus 66MHz --------------------- Register option selected? NO Microseconds for one run through Dhrystone: 6.9 Dhrystones per Second: 144508.7 VAX MIPS rating = 82.247 R4700-133 ; bus 66MHz --------------------- Register option selected? NO Microseconds for one run through Dhrystone: 6.3 Dhrystones per Second: 158982.5 VAX MIPS rating = 90.485 R5000-200 ; bus 66MHz; 512KB sync L2 ------------------------------------ Register option selected? NO Microseconds for one run through Dhrystone: 4.1 Dhrystones per Second: 242718.4 VAX MIPS rating = 138.144 YF + --------------------------- + + Newsgroups: comp.os.vxworks + Subject: Performance of VxWorks on MIPS R4650? + Date: Mon, 22 Jun 1998 12:47:13 -0700 + From: Manoj Wadekar + Organization: Acclaim Communications Inc. + Message-ID: <358EB4C1.5B174AA6@acclaiminc.com> + + Hi , + + Can anybody help me finding out Performance of VxWorks(5.3) + on MIPS R4650 (or any other processor in that family) using + Dhrystone benchmark. I would appreciate any data/pointers.. + + Thanks in advance. + + - - Manoj + + From James_M_Flynn@res.raytheon.com Wed Jun 24 05:58:53 1998 From: James_M_Flynn@res.raytheon.com Date: Wed Jun 24 05:58:56 PDT 1998 Subject: mv2604 Scsi Xfer to VME bus hangs Hello All, I am using vxworks 5.3.1 with the MV2604 BSP 1.1/4. My application loads a file from one of several locations (SCSI port, ethernet, or external VME Bus Flash Memory) to a VME memory. If I set my source for the file to be either the ethernet or the VME Flash Memory, the file is sent to the appropriate VME memory (I use the read (file_descriptor, pointer_to_vme_memory, size) subroutine. However, when I use the SCSI port as the source of the file, the program hangs in the read routine. I know the SCSI file is good, because if I change the destination from VME Bus memory to local memory, the file is transferred ok! It seems as though I have some VME and SCSI resources incorrectly allocated, or perhaps incorrectly sharing resources. I have not changed much from the original configurations in the BSP. I also have a kernel developed under BSP version 1.1/3 that works OK. Can anyone shed any light on this problem? Thanks Jim From amolkho@lannet.com Wed Jun 24 07:26:51 1998 From: "Adi molkho DVP-TA" Date: Wed Jun 24 07:26:55 PDT 1998 Subject: MPC860 MMU Package Hello vxWorks Users Does anyone know if there is a full mmu package which dynamically configures the MPC860 MMU TLB every task switch Thanks Adi +---------------------------------+ Adi Molkho Lannet a company of Madge Networks amolkho@lannet.com 972-3-6458351 From rcooper@mdc.com Wed Jun 24 07:41:15 1998 From: Rob Cooper Date: Wed Jun 24 07:41:18 PDT 1998 Subject: Removable Harddrives( Single Partition ) I am new to the vxWorks environment and still have allot to learn. I have been tasked to incorporate a SCSI drive onto a target PPC 2700 board. I have accomplished this with a fixed hard drive. ( I have had some problems with corrupted FAT tables, a known problem ) However I am having different problems with a SyJet 1.5G removable drive. 1. I need to call "scsiPhysDevCreate" function twice to get a valid SCSI_PHYS_DEV pointer. ( That is the first time I call the function it returns a NULL pointer and if I call it again, with out resetting the scsiBus(CLUE), I get a valid pointer. ) 2. It would appear that vxWorks allows only one partition to be set for a removable hard drive? What am I missing? I this a feature? From daemon@csg.lbl.gov Wed Jun 24 08:35:11 1998 From: daemon@csg.lbl.gov Date: Wed Jun 24 08:35:15 PDT 1998 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Wed Jun 24 04:00:56 PDT 1998 Subject: Re: comp.os.vxworks newsdigest ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: comp.os.vxworks newsdigest Date: Tue, 23 Jun 1998 12:45:02 -0700 From: Manoj Wadekar Organization: Acclaim Communications Inc. Message-ID: <359005BE.288480EF@acclaiminc.com> References: <199806231703.KAA27302@mission.eng.efi.com> - --------------331547F85FF77E68951BAB36 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Thanks for the response. Yuet Fung Siu wrote: > The Dhrystone doesn't tell much nowadays other than > the CPU clock speed and the size of L1. Because the program > is so small, it will usually fit entirely inside L1. Anyway, > these are the numbers running Dhrystone 2.1 on our custom > boards: > I agree. But interestingly vxWorks provides 50% of MIPS performance as that of the performance of the CPU spec. e.g. > R4700-133 ; bus 66MHz > --------------------- > Register option selected? NO > Microseconds for one run through Dhrystone: 6.3 > Dhrystones per Second: 158982.5 > VAX MIPS rating = 90.485 > For this processor without vxWorks running, and on manufacturer's boot monitor I can observe 140 VAX MIPS rating. As Dhrystone is very small programme, I would expect it to fit in cache and provide same dhrystone numbers as on non-multitasking OS (boot-monitor). Unless the programme is not getting cached... Any inputs will be very much appreciated. Thanks, - - Manoj - --------------331547F85FF77E68951BAB36 Content-Type: text/html; charset=us-ascii Content-Transfer-Encoding: 7bit Thanks for the response.
 
 

Yuet Fung Siu wrote:

The Dhrystone doesn't tell much nowadays other than
the CPU clock speed and the size of L1. Because the program
is so small, it will usually fit entirely inside L1. Anyway,
these are the numbers running Dhrystone 2.1 on our custom
boards:
 
I agree. But interestingly vxWorks provides 50% of MIPS performance
as that of the performance of the CPU spec. e.g. 
R4700-133 ; bus 66MHz
- ---------------------
Register option selected?  NO
Microseconds for one run through Dhrystone:     6.3
Dhrystones per Second:                        158982.5
VAX MIPS rating =     90.485
For this processor without vxWorks running, and on manufacturer's
boot monitor I can observe 140 VAX MIPS rating.  As Dhrystone is
very small programme, I would expect it to fit in cache and provide
same dhrystone numbers as on non-multitasking OS (boot-monitor).
Unless the programme is not getting cached...

Any inputs will be very much appreciated.

Thanks,
- Manoj
  - --------------331547F85FF77E68951BAB36-- --------------------------- End of New-News digest ********************** From cgrames@mdc.com Wed Jun 24 10:50:44 1998 From: Charlie Grames Date: Wed Jun 24 10:50:47 PDT 1998 Subject: Symbol Table Synchronization with vxWorks.st_rom VxWorks 5.3.1 MVME2604 1.1/4 Many thanks for the plethora of responses I received regarding problems I was having trying to get symbol table synchronization working with vxWorks.st_rom. Bo Wang of Lucent Technologies pointed out that when you build vxWorks.st_rom, you must also build vxWorks.st and use that as the file path under "Core File and Symbols" in the target server configuration. That solved my problem. Thanks again. Charlie Grames The Boeing Company (314) 233-1956 Charles.R.Grames@boeing.com From brett.smith@wg.com Wed Jun 24 11:14:37 1998 From: brett.smith@wg.com Date: Wed Jun 24 11:14:41 PDT 1998 Subject: localtime vs. localtime_r Is it legal to use the localtime function that RETURNS a pointer to a "broken-down" date structure? Does the function allocate this structure? Does this cause a memory leak? Or, should the localtime_r function always be used, where the caller supplies the struct? I didn't see anything in the man pages... Brett Smith vxworks From R.Kirnum@dialogic.com Wed Jun 24 11:15:32 1998 From: "Kirnum, Robert" Date: Wed Jun 24 11:15:35 PDT 1998 Subject: File Format Conversion We are planning on using HP tools for tracing, and inverse assembling. Unfortunately, the file formats they support do not include those generated by the Tornado / ARM toolset. This would limit the symbolic and emulation capabilities of the tools. Can anyone suggest a possible solution such as a format conversion utility. HP supports Elf/Dwarf, IEEE-695, OMF86/286/386, OMF96, TICOFF, and TICOFF2. Thanks. _______________________________________________________ D I A L Bob Kirnum - Project Engineer _|_|_|O Dialogic Corp., 1515 Rt. 10, Parsippany, NJ 07054 _|_|_|G mailto:R.Kirnum@dialogic.com _|_|_|I Desk - 973-993-3000 Ext. 6589 Fax - 973-993-5916 _|_|_|C Get the Dialogic Edge at http://www.dialogic.com Get The OpenSwitch Edge at http://openswitch.dialogic.com _______________________________________________________ Believe in Macintosh? Check out http://www.evangelist.macaddict.com/ From YF.Siu@eng.efi.com Wed Jun 24 11:32:22 1998 From: "Yuet Fung Siu" Date: Wed Jun 24 11:32:25 PDT 1998 Subject: Re: comp.os.vxworks newsdigest Thanks for bringing out this interesting point. Actually, although I'm running VxWorks when I did the measuremnt, interrupts were disabled for the entire test period. The timing was obtained from the CPU internal counter instead of the timer tick interrupts. So technically speaking, VxWorks wasn't involved in the test and the environment should be very close to a boot-monitor. If you can provide details like Dhrystone version, compile options, CPU/memory specs, I would be very much appreciated. Thanks, YF + I agree. But interestingly vxWorks provides 50% of MIPS performance + as that of the performance of the CPU spec. e.g. + + > R4700-133 ; bus 66MHz + > --------------------- + > Register option selected? NO + > Microseconds for one run through Dhrystone: 6.3 + > Dhrystones per Second: 158982.5 + > VAX MIPS rating = 90.485 + > + For this processor without vxWorks running, and on manufacturer's + boot monitor I can observe 140 VAX MIPS rating. As Dhrystone is + very small programme, I would expect it to fit in cache and provide + same dhrystone numbers as on non-multitasking OS (boot-monitor). + Unless the programme is not getting cached... + + Any inputs will be very much appreciated. + + Thanks, + - - Manoj + + From sblachma@aoc.nrao.edu Wed Jun 24 11:43:53 1998 From: Steve Blachman Date: Wed Jun 24 11:43:58 PDT 1998 Subject: fortran and vxWorks I am interested in fortran cross compilers running on Sun/solaris, linux, or NT/95 with 68040 (mv167) as the target. So far, I have found only Green Hills and g77. Are there other possibilities that anyone knows about? I am in the process of investigating the Green Hills OASYS compiler. Has anyone used it like this? g77 uses an f2clib which I have not tried to port to vxWorks yet. Has anyone done this? Does it seem like a viable solution, especially for the long term? A one time use of f2c is not a viable solution since we are porting a third party package and we have to cope with updates. Thanks for your input, Steve Blachman NRAO From chip@Cerent.com Wed Jun 24 13:14:41 1998 From: Chip Roberson Date: Wed Jun 24 13:14:44 PDT 1998 Subject: Re: localtime vs. localtime_r (vxWorks) Bret, It depends on your use. If you are not worried about re-entrancy, then you can use localtime. Chip -- Charles S. Roberson This message may contain two attachments: My Business vCard and My S/MIME Digital ID From ashoka@teil.soft.net Wed Jun 24 21:38:34 1998 From: "Ashoka K." Date: Wed Jun 24 21:38:39 PDT 1998 Subject: Driver development document Hi, What is the documentation of VxWorks, where we can find info about writing network device drivers. Thanks Ashoka ----------------------------------------------------------------------------- Ashoka K. Specialist, D & D, Tata Elxsi (India) Ltd. email: ashoka@teil.soft.net Tel : 91-80-8410148 ------------------------------------------------------------------------------ From DaoShuo@aol.com Thu Jun 25 01:12:18 1998 From: DaoShuo@aol.com Date: Thu Jun 25 01:12:22 PDT 1998 Subject: help on using I2c driver to set up registers Hello, Everyone: We have a mbx821 board running vxWorks, now we have a another board plugg= ed=0Ainto the mbx board using I2c from the mbx board.=A0 I was told that = the I2c=0Adriver is in the BSP.=A0 How can I use this I2c driver to talk = to our device and=0Aset up registers for our device? Anyone experienced o= n mbx821 and I2c,=A0 please=0Ahelp, every little help will be appreciated= . Thanks in advance DaoShuo From Shaochen@aol.com Thu Jun 25 03:15:04 1998 From: Shaochen@aol.com Date: Thu Jun 25 03:15:08 PDT 1998 Subject: help on using I2c driver to set up registers Hello, Everyone: We have a mbx821 board running vxWorks, now we have a another board plugg= ed=0Ainto the mbx board using I2c from the mbx board.=A0 I was told that = the I2c=0Adriver is in the BSP.=A0 How can I use this I2c driver to talk = to our device and=0Aset up registers for our device? Anyone experienced o= n mbx821 and I2c,=A0 please=0Ahelp, every little help will be appreciated= . Thanks in advance DaoShuo=0A From mschnarc@lannet.com Thu Jun 25 03:38:19 1998 From: "Miki schnarch DVP-TA" Date: Thu Jun 25 03:38:22 PDT 1998 Subject: Diagnostic tools for MPC860/850 Hi We are considering to use the MPC8BUG software which is a monitor debugger for the Motorola MPC860 PowerQUICC processor device. We would like use it's diagnostic mode tests for diagnostic of our target board, in order to do Chip tests, ram tests , flash tests , UART tests , Ethernet port test etc. for a MPC860 target board. I would like to know if someone knows of other diagnostic tools for the MPC860/850 processor that have such capabilities ? I would like to know if someone used the MPC8BUG it in that way ? How can i write my own diagnostic scripts and to use them with the MPC8BUG ? We are planning to use the motorola evaluation board MPC8xxFADS with the MPC850(DE) cpu, we would like to know if there is a VxWorks BSP for that configuration ? If not, does someone knows what are the BSP changes that need to be done to the MPC860 ADS evaluation board in order to use it with the MPC8xxFADS ? Any help is extremely appreciated. thanks, miki. ======================================================= Miki Schnarch Email : mschnarc@madge.com LAN Division, Voice : 972-3-6457669 Madge Networks (Israel) Ltd. Fax : 972-3-6487146 Atidim Technology Park, Bldg. 3, Tel-Aviv 61131, Israel. ======================================================= From R.Kirnum@dialogic.com Thu Jun 25 08:02:46 1998 From: "Kirnum, Robert" Date: Thu Jun 25 08:03:04 PDT 1998 Subject: grom & objcopy Can anyone suggest a replacement for the now-unsupported grom utility. We were using grom960 to generate a binary file from multiple coff files. It allows for address specification for each coff input file, padding with a user specified value, start and size specification for resulting image. The Tornado replacement, objcopy, claims to handle the same functionality, but does not work. I can only get it to generate a binary file for one input coff file. Unless I get a 'working' version of objcopy, I will have to create a utility to combine the binary files like grom does. Thanks, Bob _______________________________________________________ D I A L Bob Kirnum - Project Engineer _|_|_|O Dialogic Corp., 1515 Rt. 10, Parsippany, NJ 07054 _|_|_|G mailto:R.Kirnum@dialogic.com _|_|_|I Desk - 973-993-3000 Ext. 6589 Fax - 973-993-5916 _|_|_|C Get the Dialogic Edge at http://www.dialogic.com Get The OpenSwitch Edge at http://openswitch.dialogic.com _______________________________________________________ Believe in Macintosh? Check out http://www.evangelist.macaddict.com/ From gnn@wrs.com Thu Jun 25 09:28:57 1998 From: "George V. Neville-Neil" Date: Thu Jun 25 09:29:00 PDT 1998 Subject: Re: Driver development document At 09:38 PM 6/24/98 PDT, the vxWorks Users Group Exploder wrote: >Submitted-by ashoka@teil.soft.net Wed Jun 24 21:38:34 1998 >Submitted-by: "Ashoka K." > >Hi, > What is the documentation of VxWorks, where we can find info about >writing >network device drivers. This is in the BSP porting kit. Later, George From nsandhu@ca.newbridge.com Thu Jun 25 09:33:21 1998 From: Navjeet Sandhu Date: Thu Jun 25 09:33:25 PDT 1998 Subject: MPC860T FEC help Is anybody knows, what type of support does Vxworks provides to support Fast Ethernet Controller (FEC) of MPC860T Navjeet From kkauper@draper.com Thu Jun 25 11:04:40 1998 From: Kris Kauper Date: Thu Jun 25 11:04:44 PDT 1998 Subject: C++ and VxWorks VxWorks I'm considering using C++ and Tornado for an upcoming project, however, I am unsure as to how VxWorks handles C++ objects. Can an object's member function (either virtual or non-virtual) be spawned as a task? And, if so, does the member function then have all of the access privileges of a normal member function? Are there any special considerations I should be aware of? Thanks for any and all help. Examples or source code snippets would be greatly appreciated. Kris Kauper C.S. Draper Laboratory 555 Technology Square, M/S 18 Cambridge, MA 02139 kkauper@draper.com 617-258-1590 617-258-3858 (fax) From pat.shaw@canada.cdev.com Thu Jun 25 13:48:24 1998 From: "Shaw, Pat" Date: Thu Jun 25 13:48:31 PDT 1998 Subject: VxWORKS Inflate Hi: I am having trouble inflating images stored in ROM during the boot process. I was wondering if the compression software links with any outside symbols ( eg: _romInit ... ) to inflate. I have a custom Pentium Card with FLASH mapped into memory after my RAM. This allows me to run from flash at about the same speed as RAM. The problem is that if I try to create a special compressed version in FLASH that gets copied to RAM it fails during the inflation. Is there a special trick involved in getting this to fly ? Thanks Patrick Shaw pat.shaw@canada.cdev.com From R.Kirnum@dialogic.com Thu Jun 25 15:00:52 1998 From: "Kirnum, Robert" Date: Thu Jun 25 15:00:56 PDT 1998 Subject: Piping and make Can anyone explain why the following 'crashes' when executed with in a Makefile: deflate < o/tmp.bin > o/tmp.Z The equivalent instructions executed from the command line (NT) works just fine: deflate < o\tmp.bin > o\tmp.Z I'm using GNU make (Cygnus) along with the NT Tornado toolset (x86-win32) for the i960 target. The same make file works just fine under Unix (Sparc / Solaris / i960). tia Bob _______________________________________________________ D I A L Bob Kirnum - Project Engineer _|_|_|O Dialogic Corp., 1515 Rt. 10, Parsippany, NJ 07054 _|_|_|G mailto:R.Kirnum@dialogic.com _|_|_|I Desk - 973-993-3000 Ext. 6589 Fax - 973-993-5916 _|_|_|C Get the Dialogic Edge at http://www.dialogic.com Get The OpenSwitch Edge at http://openswitch.dialogic.com _______________________________________________________ Believe in Macintosh? Check out http://www.evangelist.macaddict.com/ From jford@sadira.gb.nrao.edu Fri Jun 26 05:08:26 1998 From: John Ford Date: Fri Jun 26 05:08:29 PDT 1998 Subject: [jford@sadira.gb.nrao.edu: Re: C++ and VxWorks] Dang. Forgot to defeat the spam filters... > Submitted-by kkauper@draper.com Thu Jun 25 11:04:40 1998 > Submitted-by: Kris Kauper > > VxWorks > > I'm considering using C++ and Tornado for an upcoming project, however, I > am unsure as to how VxWorks handles C++ objects. Can an object's member > function (either virtual or non-virtual) be spawned as a task? And, if so, > does the member function then have all of the access privileges of a normal > member function? Are there any special considerations I should be aware of? > Thanks for any and all help. Examples or source code snippets would be > greatly appreciated. > This is a "main" function for one of our device managers. You see how we wrap the method call with a static object, and pass the object to the taskSpawn call. I`ve seen other methods. ... static void AccelerometerSeqTask(AccelerometerMgr * obj ) { obj->sequence(); } static void AccelerometerMonTask(AccelerometerMgr * obj ) { obj->sample(); } static void AccelerometerMgrTask( char * name, int panel) { char buf[80]; AccelerometerMgr *obj = new AccelerometerMgr( name); // Start up the data reader task // We do it here to be sure all the goodies are created first strcpy(buf,obj->getName()); strcat(buf, "Mon"); int MonTaskId = taskSpawn(buf, 79, VX_FP_TASK, 20000, (FUNCPTR)AccelerometerMonTask, (int) obj,1,2,3,4,5,6,7,8,9); // unused args if(MonTaskId == ERROR ) { cerr << "Could not spawn " << obj->getName() << " ACCELEROMETER Manager Monitor" << endl; } strcat(buf, "Seq"); int SeqTaskId = taskSpawn(buf, 150, VX_FP_TASK, 20000, (FUNCPTR)AccelerometerSeqTask, (int) obj,1,2,3,4,5,6,7,8,9); // unused args if(SeqTaskId == ERROR ) { cerr << "Could not spawn " << obj->getName() << " ACCELEROMETER Manager Monitor" << endl; } // Now just run the panel server server->run(); delete (server); } -- John Ford National Radio Astronomy Observatory Green Bank, WV 24944-0002 jford@nrao.edu From sl065st@uni-duisburg.de Fri Jun 26 07:00:38 1998 From: stevens Date: Fri Jun 26 07:00:42 PDT 1998 Subject: gateway configuration in vxworks ! Hi VxWorkers, There is a HTTP-Server running on our VxWorks board using TCP-Sockets. The HTTP-Server works fine if there is no gateway between Browser and the board. The board is behind a Gateway (Proxy-Server) and the access from the internet should be possible. The target is added in the gateway=B4s routing tables, but a ping from the internet can not reach th= e target or is not answered. The Gateway routes in both directions. Adding the Gateway in the Ethernet boot options has no effect. How can the board be reached from Internet users ? Is routeAdd( ) at runtime of the software a solution ? Thanks in advance, Torsten -- *************************************************************************= * fraunhofer institute of microelectronic circuits and systems department sat name torsten stevens mailto:stevens@ims.fhg.de address finkenstrasse 61 phone +0049/203/3783-238 47057 duisburg fax +0049/203/3783-266 germany http://www.fhg.de/depts/ims-du-e.html *************************************************************************= * From caprio@research.moore.com Fri Jun 26 08:26:33 1998 From: Jim Caprio Date: Fri Jun 26 08:26:36 PDT 1998 Subject: Re: C++ and VxWorks VxWorks To my knowledge, only a static member function can be input to taskSpawn, which precludes virtual members. To effect the behavior of spawning a non-static member function (that is, one which 'knows' its specific object identity), the object's 'this' pointer must be input as an argument to taskSpawn, to be passed along to the threaded method. Internally, that method will cast the argument back into the correct object type, and thereby have access to the specific object state. It's a bit awkward, but does work. So for example, in the class definition: class ActiveObject { static int ThreadedMethod(void* objThis); // castable to FUNCPTR int TID_; // the task id ... }; The call to taskSpawn looks something like: TID = taskSpawn( name_, priority_, options_, stackSize_, (FUNCPTR)ThreadedMethod, // entry point int(this), // arg1 to ThreadedMethod ... // add'l arguments ) Inside of ThreadedMethod, cast the arg1 (int) back to its proper type: ActiveObject::ThreadedMethod(void* objThis) { ActiveObject* This = (ActiveObject*)objThis; ... // now can operate on specific object, including calling member fncs } > I'm considering using C++ and Tornado for an upcoming project, however, I > am unsure as to how VxWorks handles C++ objects. Can an object's member > function (either virtual or non-virtual) be spawned as a task? And, if so, > does the member function then have all of the access privileges of a normal > member function? Are there any special considerations I should be aware of? > Thanks for any and all help. Examples or source code snippets would be > greatly appreciated. ================================= Jim Caprio Moore Research Center, Inc. (716)773-0333 FAX : (716)773-0462 email: caprio@research.moore.com ================================= From daemon@csg.lbl.gov Sat Jun 27 04:00:57 1998 From: daemon@csg.lbl.gov Date: Sat Jun 27 04:02:56 PDT 1998 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Sat Jun 27 04:00:55 PDT 1998 Subject: Re: C++ and VxWorks ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: C++ and VxWorks Date: Fri, 26 Jun 1998 10:07:55 -0400 From: Van_D_Nguyen <@res.raytheon.com> Organization: Raytheon Electronic Systems Message-ID: <3593AB3A.DEE969C9@res.raytheon.com> References: <3.0.3.32.19980625140419.00a264d0@pop.draper.com> If you have Wind Foundation Classes, you can use VXWTask class to spawn tasks. We found out the hard way that you need to pass the pointer to the object in order for it to work properly (not documented anywhere). For example: you want to spawn a task running member function called "Run" of the object "MyObj", following the documentation, you would do: VXWTask* pMyTask = new VXWTask("NewTask", MY_TASK_PRIORITY, 0, MY_TASK_STACK_SIZE, (FUNCPTR)pMyObj->Run, 0,0,0,0,0,0,0,0,0,0); where pMyObj is a pointer to MyObj This will not work. You need to do this: VXWTask* pMyTask = new VXWTask("NewTask", MY_TASK_PRIORITY, 0, MY_TASK_STACK_SIZE, (FUNCPTR)pMyObj->Run, (int)pMyObj, 0,0,0,0,0,0,0,0,0,0); Van Kris Kauper wrote: > VxWorks > > I'm considering using C++ and Tornado for an upcoming project, however, I > am unsure as to how VxWorks handles C++ objects. Can an object's member > function (either virtual or non-virtual) be spawned as a task? And, if so, > does the member function then have all of the access privileges of a normal > member function? Are there any special considerations I should be aware of? > Thanks for any and all help. Examples or source code snippets would be > greatly appreciated. > > > Kris Kauper > C.S. Draper Laboratory > 555 Technology Square, M/S 18 > Cambridge, MA 02139 > kkauper@draper.com > 617-258-1590 > 617-258-3858 (fax) --------------------------- End of New-News digest ********************** From iliab@telegate.co.il Sun Jun 28 00:28:51 1998 From: Ilia Bosis Date: Sun Jun 28 00:28:54 PDT 1998 Subject: Re: gateway configuration in vxworks This is a multi-part message in MIME format. --------------CAD007C4D170ECB26FA1BF02 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit > Submitted-by stevens Fri Jun 26 07:00:38 1998 > > Hi VxWorkers, > > The board is behind a Gateway (Proxy-Server) and the access > from the internet should be possible. The target is added in the > gateway's routing tables, but a ping from the internet can not reach the > target or is not answered. The Gateway routes in both directions. > > How can the board be reached from Internet users ? > Is routeAdd( ) at runtime of the software a solution ? > > Thanks in advance, > Torsten > > -- Two points to check: 1. If Yours gateway is configured as the HTTP proxy server or as an IP router, too? HTTP proxy doesn't forward the ICMP, so the pings have no chances to pass through. HTTP, on another hand, have to pass OK. 2. If the gateway is an IP router, You must define the default route in the target's routing table as a pointer to this gateway. Use the line, like this: routeAdd("0.0.0.0", "Your_gateway_IP_address"); Hope, this helps. Regards Ilia. --------------CAD007C4D170ECB26FA1BF02 Content-Type: text/x-vcard; charset=us-ascii; name="vcard.vcf" Content-Transfer-Encoding: 7bit Content-Description: Card for Ilia Bosis Content-Disposition: attachment; filename="vcard.vcf" begin: vcard fn: Ilia Bosis n: Bosis;Ilia org: Telegate adr: Haplada 7;;;Or-Yehuda;;60218;Israel email;internet: iliab@telegate.co.il title: S/W engineer tel;work: 03-5388-685 tel;fax: 03-5388634 x-mozilla-cpt: ;0 x-mozilla-html: TRUE version: 2.1 end: vcard --------------CAD007C4D170ECB26FA1BF02-- From Friedrich.Scheurer@t-online.de Sun Jun 28 02:20:51 1998 From: Friedrich.Scheurer@t-online.de (F. Scheurer) Date: Sun Jun 28 02:20:56 PDT 1998 Subject: Once again (!) VxWorks 5.3: Problems with Booting SPARC Hello, several times i posted the following message to the VxWorks UserGroup and didn't see it in the mails that i get from the exploder. What's wrong with my posting? ----- Begin Included Message ----- >From zfssrr Thu Jun 25 00:07:25 1998 To: vxwexplo@lbl.gov Subject: VxWorks 5.3: Problems with Booting SPARC Hi VxWorkers, we start new with VxW 5.3 and want to go on using our Themis SPARC 10HS. We boot over ethernet and the first phase works fine (nothing changed to 5.1). The second phase (VxWorks) breaks: The flags parameter is set to 0x80 and the following information occures on the console during VxWorks booting: Attaching network interface fn0... done Subnet mask 0xffffff00 Attaching network interface lo0... done Loading... 541840 + 31152 + 36612 Starting at 0x20000... (System does not show any activity) What i can see (Solaris - snoop) during loading and starting is the following break of the TFTP transmission: .... zfshost -> zfstarget TFTP Data block 1122 (512 bytes) zfstarget -> zfshost TFTP Ack block 1122 zfshost -> zfstarget TFTP Data block 1123 (512 bytes) zfstarget -> zfshost TFTP Ack block 1123 zfshost -> zfstarget TFTP Data block 1124 (512 bytes) zfstarget -> zfshost TFTP Error: not defined .... What mistake do i make? What's wrong? Where can i search? Under VxW 5.1.1 everything worked fine. I didn't get any support by the local provider and the producer. ----- End Included Message ----- Thanks in advance F. Scheurer -- +-------------------------------------------------+ | | | Dipl.-Ing. F. Scheurer | | | | c/o Zentrum Fertigungstechnik Stuttgart | | Nobelstrasse 15 | | D-70569 Stuttgart | | | | Telefon +49 711 13162-41(-0) | | Fax +49 711 13162-11 | | E-Mail scheurer@zfs.rus.uni-stuttgart.de | | zfssrr@zfssun6.tz.rus.uni-stuttgart.de | | WWW http://zfs.rus.uni-stuttgart.de | | | +-------------------------------------------------+ | | | privat: | | Telefon +49 7161 686362 | | Fax +49 7161 686301 | | E-Mail Friedrich.Scheurer@t-online.de | | | +-------------------------------------------------+ From ToWang@GI.com Mon Jun 29 07:15:00 1998 From: "Wang, Tom (HT-EX)" Date: Mon Jun 29 07:15:05 PDT 1998 Subject: RE: C++ and VxWorks VxWorks, The taskSpawn() accept function entry_point as the parameter to spawn the task. The function name has to be resolved when compiling and the function has to be accessible when calling the taskSpawn(). In the C++, there are class object and the encoded function name for the member function. To integrate these two, I create C_API layer. The C_API layer, extern C function, is used to wrap the member function and pass the C function name into the taskSpawn(). You got to take care of object creation, i.e. either from static class object or have the first C_API create the object for the rest of the member functions. This is the approach that I used for the projects on the Tornado, and it is easy to implement. (This layer works as a bridge between C and C++, and everything fit nicely together.) HTH. -Tom email: s.wang@ieee.org > > I'm considering using C++ and Tornado for an upcoming project, however, > I > > am unsure as to how VxWorks handles C++ objects. Can an object's member > > function (either virtual or non-virtual) be spawned as a task? And, if > so, > > does the member function then have all of the access privileges of a > normal > > member function? Are there any special considerations I should be aware > of? > > Thanks for any and all help. Examples or source code snippets would be > > greatly appreciated. > > From ajitn@future.futsoft.com Mon Jun 29 23:57:56 1998 From: ajitn Date: Mon Jun 29 23:57:59 PDT 1998 Subject: bare necessities : RTOS VxWorks VxWorkers 1. What are the factors in designing a high performance real time system . 2. What are the optimisations in coding for real time systems in C any pointer to the above welcome Thanx in advance, Ajit. From BOOJ@kidd.Co.Za Tue Jun 30 01:17:17 1998 From: Johan Boot Date: Tue Jun 30 01:17:21 PDT 1998 Subject: PPC603 caches Hi, Tornado 1.0.1 vxWorks 5.3.1 I am using a custom BSP (based on the MV1300 BSP) for a PPC603. I would like to enable the caches of the device without using the MMU. Is this possible? I have included the following in config.h: #define INCLUDE_CACHE_SUPPORT #undef USER_I_CACHE_ENABLE #undef USER_I_CACHE_MODE #define USER_I_CACHE_MODE CACHE_COPYBACK #undef USER_D_CACHE_ENABLE #undef USER_D_CACHE_MODE #define USER_D_CACHE_MODE CACHE_COPYBACK #undef USER_B_CACHE_ENABLE This should allow initialisation of the caches in usrConfig.h, as described in the documentation. I then use cacheEnable to enable the caches in my code. I execute a test routine and measure the time taken using tickGet. Enabling or disabling the caches have no impact on the execution time. What more must be done? Any help will be appreciated. Regards, Johan Johan Boot Software Engineer PPG, Kentron RSA. From rtp.co.uk!jge@rtp.co.uk Tue Jun 30 01:47:29 1998 From: John Efstathiades Date: Tue Jun 30 01:48:28 PDT 1998 Subject: Any IrDA and USB stack recommendations? Hello, Has anybody used and can recommend an infra-red protocol stack for use with VxWorks? Ideally, I am looking for a system that has been pre-ported to, or written for, VxWorks, although any suggestions would be welcome. My aim is get an IR link up and running as quickly as possible. Sorry, I can't give h/w details at this stage, but it will meet the IrDA 1.0 recommendations. I am also interested in USB stacks - same criteria as above, h/w meets USB OHCI 1.0 standard. TIA, John ---------------------------------------------------------------- John Efstathiades Real Time Products Ltd e-mail: jge@rtp.co.uk Chancery House, Tel: +44 (0)121 234 6600 8 Edward Street, Fax: +44 (0)121 234 6611 Birmingham. B1 2RX. England. Web: www.rtp.co.uk ---------------------------------------------------------------- From kb+@andrew.cmu.edu Tue Jun 30 05:07:00 1998 From: Kevin Bradley Date: Tue Jun 30 05:07:03 PDT 1998 Subject: Re: PPC603 caches Excerpts from mail: 30-Jun-98 PPC603 caches by the U. G. Exploder@lbl.g > I then use cacheEnable to enable the caches in my code. I execute a > test routine and measure the time taken using tickGet. Enabling or > disabling the caches have no impact on the execution time. What more > must be done? I don't know anything about PPC caches, or how the BSP works with them, but I think I can comment on observing cache effects in execution time. Two things: a) We had a bad PC motherboard that effectively had no L2 cache. Code went 6-8 times slower (L1->main memory), e.g. timer interrupt went from 30 us to 130 us. b) Measuring time with tickGet() is suspect if the code is short and the timer tick is slow. If your code, for example, normally runs for 10 ms and then slows down to 20 ms and if your timer tick is the default 60 Hz, then observing a difference is difficult, as it depends on the synchronization between your task start and the number of clock ticks. I suggest finding a *much* higher resolution timer, something on the microsecond range. I believe PPC has cycle count measurers on it, or access to a secondary clock of high frequency that can be read. Either that or write a routine that's *really* long, like 5 seconds with caches on, and you can observe several seconds delay with caches off. It is possible to do memory measurements using lmbench or HBench; the code is relatively simple to port from Unix to VxWorks. These give memory delays in accessing a large portion of memory in various strides, from which it is possible to derive memory hierarchy information. If you have access to a high resolution timer, then this is the best for determining if your caches are on or off. I don't have the URLs with me at the moment, but I'll dig them up and post them (along with a pointer to my ported-to-VxWorks code for HBench, for Pentiums -- it uses the Pentium clock cycle counter for a timer. It should be easy to port this to the PPC high-res timer). -- Kevin From David.L.Fuson@boeing.com Tue Jun 30 08:13:45 1998 From: David Fuson Date: Tue Jun 30 08:13:48 PDT 1998 Subject: vxWorks I originally started using dual SPARC VME cards from FORCE and VxWorks 5.1 for code development. After a time I decided to upgrade to version 5.2 of VxWorks. I experienced numerous problems with VxWorks 5.2 executing on a SPARC computer. First, there was no debugger available for "C" compiled code, "The VxWork Support informed that there was NO "C" debugger developed, and that if I wanted the debugger I would have to pay all the development cost. Second, the "C" compiler X-Reference output showed that incorrect addresses was being calculated for terms I was storing in shared memory. Third, this presented numerous for me, since I was using ADA, C and two SPARC computers executing on a shared memory. However, the VxWorks OS seemed to work properly with for versions 5.1 and 5.2 using only Ada code. p.s. Did you make sure that you completely recompiled your code? Have your .cshrc or .login file to reflect using the new VxWorks? A handy test would be to see if you could boot the vendor's default OS. Also, check to see if your EPROM defaults have been correctly changed. From smcfarr@coppermountain.com Tue Jun 30 12:24:08 1998 From: Shawn McFarr Date: Tue Jun 30 12:24:12 PDT 1998 Subject: I2O experience with VxWorks (IxWorks) I am interested to learn more about using IxWorks, Is it worth to design it into a project? How fast is it? How big is it?... Any real experience that you could share with is appreciated in advance. Thanks, Shawn McFarr smcfarr@coppermountain.com Copper Mountain Networks, Inc. 619.453.8799 x108 3931 Sorrento Valley Blvd. 619.453.9244 (FAX) San Diego, CA 92121 http://www.coppermountain.com From JWhisler@alphatronix.com Tue Jun 30 12:56:07 1998 From: JWhisler@alphatronix.com Date: Tue Jun 30 12:56:14 PDT 1998 Subject: Desktop/Server Management on VxWorks I'm looking for Desktop Management software (e.g. LANDesk Client) that runs on VxWorks. Has anyone ported LANDesk to VxWorks? Does anyone know of another desktop management client or "DMI service provider" available for VxWorks? Or, in general, any suggestions for platform managment on VxWorks platforms? Thanks John Whisler From MSchrape@atomika.com Tue Jun 30 13:58:14 1998 From: Martin Schrape Date: Tue Jun 30 13:58:18 PDT 1998 Subject: Tornado tools Hi, does anyone know if there are some general purpose Tornado tools available on the net. -martin