From vxwexplo-errs@csg.lbl.gov Fri Sep 1 04:03:09 2000 From: Vxworks Exploder Date: Fri Sep 1 04:03:11 PDT 2000 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Fri Sep 1 04:03:04 PDT 2000 Subject: what is bootrom.hex? Subject: Target Shell Subject: Re: Target Shell Subject: Status of -fno-implicit-templates compiler flag Subject: Re: what is bootrom.hex? Subject: Re: Target Shell Subject: Re: what is bootrom.hex? Subject: Re: Target Shell Subject: Re: Target Shell Subject: working of functions in bootInit.c? Subject: Re: bootrom customization Subject: Building for VxSim Subject: Re: Building for VxSim Subject: Re: VxWorks and dual processors Subject: Re: AMD 79C973 driver (BSP Janich Klass ISA p7/103) Subject: Re: powerpc-wrs-vxworks and gcc-2.95.2 on Linux Subject: Re: VxWorks and dual processors Subject: Load Error "Relocation value does not fit in 24 bits" Subject: Re: network time protocol Subject: Re: working of functions in bootInit.c? Subject: BSP Validation Suite Question Subject: Re: Load Error "Relocation value does not fit in 24 bits" Subject: Re: Target Shell Subject: lptDevDelete() ?? Subject: Re: Vxsim File I/O with NT Subject: Macronix LAN chip Subject: Re: what is bootrom.hex? ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: what is bootrom.hex? Date: Thu, 31 Aug 2000 12:40:06 +0000 From: Brian Genisio Organization: Michigan State University Message-ID: <39AE5226.696ED2C8@orion-sim.com> I have figured out that I need to make a bootrom.hex to fix the problem of a large image file, but then what? Do I tell the VxWorks boot loader on the card to download that instead? Right now, it downloads the file vxWorks that the build procedure makes. Thanks, Brian --------------------------- Newsgroups: comp.os.vxworks Subject: Target Shell Date: Thu, 31 Aug 2000 08:43:57 -0400 From: Somesh Agrawal Organization: 3Com Message-ID: <39AE530D.8F1841DC@hotmail.com> Hi all, (Target: pentium BSP) I have included "INCLUDE_SHELL" in my config.h and then build my image. This way I am getting the shell prompt at the Target, but most of it's functionality is not there.I have included some other related macro's also but still not able to get it working properly.Any help will be appreciated. Thanks, S.A --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Target Shell Date: Thu, 31 Aug 2000 14:03:53 -0500 From: "Keith McHugh" Organization: Esat Net Customer Message-ID: <39ae5707.0@seat.s3two.ie> References: <39AE530D.8F1841DC@hotmail.com> Reply-To: "Keith McHugh" Yep, it is happening to me as well. I don't know either! Anyway, your work should not be hampered. Because Tornado Shell (Target server shell) should work properly irrespective of the Target shell! Keep your finger cross for more responses on the board in this regard! "Somesh Agrawal" wrote in message news:39AE530D.8F1841DC@hotmail.com... > Hi all, > > (Target: pentium BSP) > I have included "INCLUDE_SHELL" in my config.h and then build my image. > This way I am getting the shell prompt at the Target, but most of it's > functionality > is not there.I have included some other related macro's also but still > not able > to get it working properly.Any help will be appreciated. > > Thanks, > S.A > > > --------------------------- Newsgroups: comp.os.vxworks Subject: Status of -fno-implicit-templates compiler flag Date: Thu, 31 Aug 2000 14:56:49 +0200 From: Albert Wijnja Organization: MECO Equipment Engineers BV Message-ID: <39AE5611.CD515100@meco.nl> Reply-To: Albert.Wijnja@meco.nl Hello, We are doing a project where we use ACE/TAO for VxWorks. Until this morning we had major problems with getting things running. Finally, we found that the problems were caused by static member variables in template classes. We solved the problems by setting the compiler flag "-fno-implicit-templates". In the TAO documentation and makefiles we also found the following remark: ... else ifeq ($(CPU),PENTIUM) BINXSYM_NAME = xsymDec CFLAGS += -DCPU=PENTIUM TOOLENV = 386 else ifeq ($(CPU),I80486) #### Note: -fno-implicit-templates doesn't seem to work well with g++ #### cygnus-2.7.2-960126 for pc486 target. Some member functions don't #### get instantiated. Wind River knows about this problem (SPR 8483). BINXSYM_NAME = xsymDec CFLAGS += -DCPU=I80486 -m486 TOOLENV = 386 else ... For pc486 there seems to be problems. However, we are compiling for PENTIUM. Enabling the flag for PENTIUM seems to work. Does anyone know about the status of this flag for a PENTIUM target? Albert Wijnja --------------------------- Newsgroups: comp.os.vxworks Subject: Re: what is bootrom.hex? Date: Thu, 31 Aug 2000 08:22:25 -0500 From: Greg Willden Organization: Posted via Supernews, http://www.supernews.com Message-ID: <39AE5C11.EA1213CC@swri.edu> References: <39AE5226.696ED2C8@orion-sim.com> Brian Genisio wrote: > I have figured out that I need to make a bootrom.hex to fix the problem > of a large image file, but then what? Do I tell the VxWorks boot loader > on the card to download that instead? Right now, it downloads the file > vxWorks that the build procedure makes. > > Thanks, > Brian What you do next sort of depends on which board you are using. For the Motorola boards you have to change a jumper and boot from the PPCBug flash and download the proper bootrom file and then flash the appropriate memory. On boards from Synergy Microsystems you boot up vxworks, download an application provided in the BSP and run it. This application downloads the code for you and flashes the appropriate memory. These are the only two boards that I am familiar with... You should be able to find the required information in the documentation for the board you are using. HTH - -- Greg Willden gwillden@swri.edu Engineer Southwest Research Institute Ph: (210) 522-6241 Fax: (210) 522-2709 www.swri.edu --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Target Shell Date: Thu, 31 Aug 2000 08:24:54 -0500 From: Greg Willden Organization: Posted via Supernews, http://www.supernews.com Message-ID: <39AE5CA6.20246EC5@swri.edu> References: <39AE530D.8F1841DC@hotmail.com> Somesh Agrawal wrote: > Hi all, > > (Target: pentium BSP) > I have included "INCLUDE_SHELL" in my config.h and then build my image. > This way I am getting the shell prompt at the Target, but most of it's > functionality > is not there.I have included some other related macro's also but still > not able > to get it working properly.Any help will be appreciated. > Can you be more specific about what doesn't work? Usually there is an option in the kernel build that you are lacking before you can use all the functionality that you want from the Target shell. - -- Greg Willden gwillden@swri.edu Engineer Southwest Research Institute Ph: (210) 522-6241 Fax: (210) 522-2709 www.swri.edu --------------------------- Newsgroups: comp.os.vxworks Subject: Re: what is bootrom.hex? Date: Thu, 31 Aug 2000 13:51:50 +0000 From: Brian Genisio Organization: Michigan State University Message-ID: <39AE62F6.E5030403@orion-sim.com> References: <39AE5226.696ED2C8@orion-sim.com> <39AE5C11.EA1213CC@swri.edu> Ok, so when I make a new boot rom, it includes all my includes from MACH_EXTRA. Do I need to do this? Or should I leave these out and only compile it on the vxWorks image? More specifically, an object I have needs to be munched. So, if I leave it un-munched, vxWorks build is happy because it munches. If I leave it munched, bootrom.hex build is happy since it doesnt munch. But if the bootrom resides on the board, and only loads vxWorks from the host, it seems that all the MACH_EXTRA includes are not needed. Thanks, Brian Greg Willden wrote: > Brian Genisio wrote: > > > I have figured out that I need to make a bootrom.hex to fix the problem > > of a large image file, but then what? Do I tell the VxWorks boot loader > > on the card to download that instead? Right now, it downloads the file > > vxWorks that the build procedure makes. > > > > Thanks, > > Brian > > What you do next sort of depends on which board you are using. For the > Motorola boards you have to change a jumper and boot from the PPCBug flash > and download the proper bootrom file and then flash the appropriate memory. > On boards from Synergy Microsystems you boot up vxworks, download an > application provided in the BSP and run it. This application downloads the > code for you and flashes the appropriate memory. These are the only two > boards that I am familiar with... > You should be able to find the required information in the documentation for > the board you are using. > HTH > > -- > Greg Willden gwillden@swri.edu > Engineer > Southwest Research Institute > Ph: (210) 522-6241 > Fax: (210) 522-2709 > www.swri.edu --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Target Shell Date: Thu, 31 Aug 2000 10:01:12 -0400 From: Somesh Agrawal Organization: 3Com Message-ID: <39AE6528.84F93BB3@hotmail.com> References: <39AE530D.8F1841DC@hotmail.com> <39AE5CA6.20246EC5@swri.edu> In my case even "lkup help" command is not working. Thanks, Somesh Greg Willden wrote: > Somesh Agrawal wrote: > > > Hi all, > > > > (Target: pentium BSP) > > I have included "INCLUDE_SHELL" in my config.h and then build my image. > > This way I am getting the shell prompt at the Target, but most of it's > > functionality > > is not there.I have included some other related macro's also but still > > not able > > to get it working properly.Any help will be appreciated. > > > > Can you be more specific about what doesn't work? Usually there is an > option in the kernel build that you are lacking before you can use all the > functionality that you want from the Target shell. > > -- > Greg Willden gwillden@swri.edu > Engineer > Southwest Research Institute > Ph: (210) 522-6241 > Fax: (210) 522-2709 > www.swri.edu --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Target Shell Date: Thu, 31 Aug 2000 09:46:18 -0500 From: Greg Willden Organization: Posted via Supernews, http://www.supernews.com Message-ID: <39AE6FBA.20AF54C1@swri.edu> References: <39AE530D.8F1841DC@hotmail.com> <39AE5CA6.20246EC5@swri.edu> <39AE6528.84F93BB3@hotmail.com> This is a multi-part message in MIME format. - --------------C472E8BC1ABAC15D1B4AECA9 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Somesh Agrawal wrote: > In my case even "lkup help" command is not working. > > Thanks, > Somesh > It depends on the error message returned but it sounds like maybe you need to include the symbol table. Be advised when you include the symbol table on the host you may need to do symbol table synchronization if you want to switch back and forth between the Host shell and the Target shell. Read the documentation about the symbol table synchronization. - --------------C472E8BC1ABAC15D1B4AECA9 Content-Type: text/x-vcard; charset=us-ascii; name="gwillden.vcf" Content-Transfer-Encoding: 7bit Content-Description: Card for Greg Willden Content-Disposition: attachment; filename="gwillden.vcf" begin:vcard n:Willden;Greg x-mozilla-html:FALSE org:Southwest Research Institute;www.swri.edu version:2.1 email;internet:gwillden@swri.edu title:Engineer adr;quoted-printable:;;6220 Culebra Rd=0D=0A=0D=0A;San Antonio, TX 78238;;; fn:Greg Willden end:vcard - --------------C472E8BC1ABAC15D1B4AECA9-- --------------------------- Newsgroups: comp.os.vxworks Subject: working of functions in bootInit.c? Date: Thu, 31 Aug 2000 16:53:02 +0200 From: Matthias Gottwald Organization: Siemens AG Message-ID: <39AE714D.70103C5F@icn.siemens.de> Hi folks, I am working with Tornado 2 and generate a compressed Bootrom-Image (Bootstrap-Loader) by Project facilitiy and adapted Makefile/linker command file/ config.h.... I saw that the functions copyLongs() and fillLongs in bootInit.c produce a wrong source/destination address. The Copyloader-Routine is copied to RAM_LOW_ADRS. After this all not used memory is cleared. The Bootloader is copied to RAM_HIGH_ADRS. Are these addresses valid? I changed this file, so the addresses map to the right pointers. Can this be made or did I wrong? Best regards Matthias. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: bootrom customization Date: Thu, 31 Aug 2000 17:00:53 +0200 From: Matthias Gottwald Organization: Siemens AG Message-ID: <39AE7325.3CE8F19B@icn.siemens.de> References: <8oi8ch$c3o$1@nnrp1.deja.com> Hi, I am producing a Bootrom in Tornado2. You have to change the makefile in the BSP directory, to give extra defines. From the Menu: Build->Buil Boot ROM you can choose your BSP and the build the target. Best regards Matthias. egluska@my-deja.com wrote: > Is there a way to customize vxWorks bootrom using Tornado 2 project > facility? It seems that the project facility allows generation and > customization of vxWorks images, but if you want bootrom, you have > to work in the BSP directory under target/config using the local > makefile, and customization of vxWorks components within the bootom > is done in the "old style" makefile INCLUDE statements. > > Thanks > > Eran Gluska > PacketLight Networks > > Sent via Deja.com http://www.deja.com/ > Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Building for VxSim Date: Thu, 31 Aug 2000 16:12:22 +0100 From: "noSpam" Organization: Marconi Message-ID: <8olsip$oes$1@newsfeed.pit.comms.marconi.com> Hi, I am building an application for VxSim and I get the following error: ccsimpc -nostdlib -r -Wl,-X -Wl,@E:\depot\TornadoSim\MGCSim\prjObjs.lst E:\depot\TornadoSi m\H323Sim\SIMNTgnu\MGCH323.o E:\depot\TornadoSim\IsdnSim\SIMNTgnu\MGCIsdn.o E:\depot\Torna doSim\MegacoSim\SIMNTgnu\MGCMegaco.o E:\depot\TornadoSim\TrilliumSim\SIMNTgnu\TrilliumSim. a -o partialImage.o \\E\TornadoSim\host\x86-win32\i386-pc-mingw32\bin\ld.exe: partialImage.o: reloc overflow: 0x26440 > 0xffff partialImage.o: final close failed: File truncated make: *** [MGCSim.out] Error 0x1 Has anyone ever come across a similar problem, and a solution ? Many thanks in advance, Mark --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Building for VxSim Date: Thu, 31 Aug 2000 18:52:44 +0100 From: "Dave Korn" Organization: (Posted via) U-NET Internet Ltd. Message-ID: References: <8olsip$oes$1@newsfeed.pit.comms.marconi.com> >Hi, >I am building an application for VxSim and I get the following error: > >ccsimpc -nostdlib -r -Wl,-X -Wl,@E:\depot\TornadoSim\MGCSim\prjObjs.lst >E:\depot\TornadoSi >m\H323Sim\SIMNTgnu\MGCH323.o >E:\depot\TornadoSim\IsdnSim\SIMNTgnu\MGCIsdn.o E:\depot\Torna >doSim\MegacoSim\SIMNTgnu\MGCMegaco.o >E:\depot\TornadoSim\TrilliumSim\SIMNTgnu\TrilliumSim. >a -o partialImage.o >\\E\TornadoSim\host\x86-win32\i386-pc-mingw32\bin\ld.exe: partialImage.o: >reloc >overflow: >0x26440 > 0xffff >partialImage.o: final close failed: File truncated >make: *** [MGCSim.out] Error 0x1 > >Has anyone ever come across a similar problem, and a solution ? Quick! This is a job for -mlongcall! DaveK - -- They laughed at Galileo. They laughed at Copernicus. They laughed at Columbus. But remember, they also laughed at Bozo the Clown. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: VxWorks and dual processors Date: Thu, 31 Aug 2000 13:16:49 -0500 From: Andrew Johnson Organization: APS, Argonne National Laboratory Message-ID: <39AEA111.938839B4@aps.anl.gov> References: Den wrote: > > 1. Does VxWorks support dual processors? Do any other embedded OS's > support dual or multiple processors? Yes and no. I'm not aware of any RTOS that supports true Symmetric Multi-Processing in the way that Linux and Solaris do, but there is an extra-cost option called VxMP that makes it easier to write software asymmetric multi-processor systems. This is really aimed at bus-based architectures where each CPU has its own local RAM and ROM and access to an area of RAM which is shared. VxMP adds the ability to create shared semaphores and message queues and allocate RAM from that shared memory space for use by any CPU that can access it directly. There are also vxWorks Board Support Packages for some dual-CPU processor boards which pass I/O operations off to the second CPU to perform, but I don't think they could allow you to run normal vxWorks program code on the other CPU. > 2. Will a future version of VxWorks support 64-bit processors? Such as the > new Intel Itanium chip? That's a question for a WRS sales rep I suspect. - - Andrew - -- Every great idea appears crazy to start with. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: AMD 79C973 driver (BSP Janich Klass ISA p7/103) Date: Thu, 31 Aug 2000 19:26:17 GMT From: steve_ross@my-deja.com Organization: Deja.com - Before you buy. Message-ID: <8ombfu$83e$1@nnrp1.deja.com> References: <8muhpo$o3h$1@nnrp1.deja.com> <8mv1ea$56p$1@nnrp1.deja.com> In article <8mv1ea$56p$1@nnrp1.deja.com>, Rick Pierpont wrote: > Juno, > > Since that chipset isn't directly supported by the GUI project facility > (without a custom configlette) you need to make this modification in > your BSPs config.h file. > > 1) Exclude any END drivers in the project facility. > 2) By default INCLUDE_FEI is defined in config.h. You need to #undef > that line and #define INCLUDE_LN_97X_END instead. > > Now when you build your bootrom image and your VxWorks you will have > network support for that chipset. > > This is exactly the steps I had to take in order to use the same > chipset on my Pentium-based Jumptech/Adastra VNS-786 board. > > Hope this helps, > I made these modifcation and now when I build vxworks I get an error message: Pentiumpro\sysLib.c:670: undefined reference to `sysLan97xPciInit' This function lives in the file sysNetif.c. I've tried adding this to my projects workspace files, but this results in more errors when I try and build. What am I doing wrong? Regards, Steve Ross Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: powerpc-wrs-vxworks and gcc-2.95.2 on Linux Date: Thu, 31 Aug 2000 20:24:38 GMT From: karuottu@freenet.hut.fi (Kai Ruottu) Organization: Telia Finland Customer News Service Message-ID: <39aebeee.6716227@news.yhteys.mtv3.fi> References: <39AD6E21.FF4571D7@campio.com> Reply-To: karuottu@freenet.hut.fi On Wed, 30 Aug 2000 16:27:13 -0400, Vincent Hamrick wrote: >I am trying to build a cross-compiler for powerpc-wrs-vxworks using >gcc-2.95.2 in Linux and am having problems. I am getting the following >error while building the libiberty stuff. > >/usr/local/wind/gcc-2.95.2/gcc/xgcc -B/usr/local/wind/gcc-2.95.2/gcc/ >-B/usr/local/wind/powerpc-wrs-vxworks/bin/ -c -DHAVE_CONFIG_H -g -O2 -I. >-I./../../include choose-temp.c >In file included from >/usr/local/wind/gcc-2.95.2/gcc/include/types/vxANSI.h:45, > from /usr/local/wind/gcc-2.95.2/gcc/include/stdio.h:59, > from choose-temp.c:29: It seems that the build tried to fix the VxWorks headers and copied the fixed ones to the build directories... Weird indeed, I built for both Linux and Win32 hosts and for the same 'powerpc-wrs-vxworks' target today and no 'fixincludes' was run in either case... Perhaps I have some fix installed for this in my gcc-2.95.2 sources. >choose-temp.c:38: sys/file.h: No such file or directory >make[2]: *** [choose-temp.o] Error 1 > >Now, I pulled the include files from my Win NT partition >(Tornado\target\h) and there is no sys/file.h (or sys/param.h for that >matter). I'm curious whether these exist on the Sun platform, and this >is my problem. > >Any hints to what the problem is? I remember seeing somewhere that this >should work. Any help would be greatly appreciated. How 'standard' was your build method, what did you put into the 'configure' command? And what you did before running it? The expected 'standard' behaviour is: 1. build binutils, configuring with: --prefix=$prefix --target=powerpc-wrs-vxworks and having them installed into $prefix/powerpc-wrs-vxworks/bin (and .../lib/ldscripts) and $prefix/bin (with the name prefix 'powerpc-wrs-vxworks-') after running 'make install' 2. preinstall the VxWorks headers into $prefix/powerpc-wrs-vxworks/include 3. build GCC, configuring with: --prefix=$prefix --target=powerpc-wrs-vxworks The $prefix is the chosen install prefix, '/usr/local/wind' seems to be your choice... All kind of 'perverse' behaviour, like using the '--with-headers=...' to point to the headers to be 'fixed', may cause troubles. Some people have learned this to be useful for some cases and now suggest using it everywhere, but it isn't really suggested in the 'gcc/INSTALL' or in the GCC manual ("Installation / Cross-Compiler"). Installing the target headers to somewhere else than suggested in the instructions, and then using the '--with-headers' to repair the mistake, isn't very good practice... Another weird thing would be to take the Cygnus newlib sources and install them to the same directory structure with 'gcc', 'libiberty' etc., thinking that they must be used somehow (because the CrossGCC FAQ says so...). For VxWorks there isn't any use for newlib... It seems that your 'configure' now FINDS the 'sys/file.h' somewhere for some reason and thinks it belonging into your 'standard target headers'... Cheers, Kai --------------------------- Newsgroups: comp.os.vxworks Subject: Re: VxWorks and dual processors Date: Thu, 31 Aug 2000 16:31:31 -0500 From: Greg Willden Organization: Posted via Supernews, http://www.supernews.com Message-ID: <39AECEB3.619BB415@swri.edu> References: <39AEA111.938839B4@aps.anl.gov> This is a multi-part message in MIME format. - --------------F5BC3EACAFD8D6F08E9AB908 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Andrew Johnson wrote: > There are also vxWorks Board Support Packages for some dual-CPU processor > boards which pass I/O operations off to the second CPU to perform, but I > don't think they could allow you to run normal vxWorks program code on the > other CPU. Well actually I have a dual processor PPC board from Synergy Microsystems on which each processor has its own VxWorks kernel. Each processor is running essentially identical code. They share data through dual-ported ram set up as shared memory just the same way two boards would except the bus cycles are on board. (i.e. sharing data doesn't use the VME bus) - --------------F5BC3EACAFD8D6F08E9AB908 Content-Type: text/x-vcard; charset=us-ascii; name="gwillden.vcf" Content-Transfer-Encoding: 7bit Content-Description: Card for Greg Willden Content-Disposition: attachment; filename="gwillden.vcf" begin:vcard n:Willden;Greg tel;fax:(210) 522-2709 tel;work:(210) 522-6241 x-mozilla-html:FALSE org:Southwest Research Institute;www.swri.edu version:2.1 email;internet:gwillden@swri.edu title:Engineer adr;quoted-printable:;;6220 Culebra Rd=0D=0A=0D=0A;San Antonio, TX 78238;;; fn:Greg Willden end:vcard - --------------F5BC3EACAFD8D6F08E9AB908-- --------------------------- Newsgroups: comp.os.vxworks Subject: Load Error "Relocation value does not fit in 24 bits" Date: Thu, 31 Aug 2000 23:19:30 GMT From: "Lou Kale" Organization: EarthLink Inc. -- http://www.EarthLink.net Message-ID: <6KBr5.11372$K4.483872@newsread1.prod.itd.earthlink.net> I get an error "Relocation value does not fit in 24 bits" when I try to load an application by typing "ld Organization: Sanders-A Lockheed Martin Company Message-ID: <39AE54CC.14048F77@lmco.com> References: <4Ahr5.207632$Gh.4061788@news20.bellglobal.com> As Joachim points out, a basic NTP client is built into Tornado II. There's an earlier thread that covers this. deja.com should pick it out. It will set the time of the onboard clock once and only once. There is no clock discplining. I've been working on a port of the NTP v4 for vxWorks (should be version independant) and I can give you pointers if you'd like. If you want to use NTP v3, I think a stable port already exists. Email me directly. Bryan Dragan Ostojic wrote: > > Hi, > > I need information regarding the use of network time protocol (NTP). Is it > available for VxWorks and is it feasible using NTP to synchronize the > timeofday clock on the target board? > > Regards, > Dragan. - -- Bryan D. Lee Phone: 603/885-7025 Software Engineer & ELDP Fax : 603/885-3016 Sanders-A Lockheed Martin Company Pager: 603/423-5565 PO Box 868, PTP1-2361, Nashua, NH 03061 Email: bryan.d.lee@lmco.com --------------------------- Newsgroups: comp.os.vxworks Subject: Re: working of functions in bootInit.c? Date: Fri, 01 Sep 2000 00:12:57 GMT From: fywang@my-deja.com Organization: Deja.com - Before you buy. Message-ID: <8oms9r$shd$1@nnrp1.deja.com> References: <39AE714D.70103C5F@icn.siemens.de> Hi Matthias, Note: you cannot set RAM_HIGH_ADRS to LOCAL_MEM_SIZE I am using a BSP based on mbx860/ads860 BSP. Here is the configuration for my BSP: #define LOCAL_MEM_LOCAL_ADRS 0x00000000 /* Base of RAM */ #define LOCAL_MEM_SIZE 0x02000000 /* 32 Mbyte */ #define ROM_BASE_ADRS 0xFE000000 /* base address of ROM */ #define ROM_TEXT_ADRS (ROM_BASE_ADRS + 0x100) #define ROM_SIZE 0x00080000 /* 512K ROM space */ /* RAM address for ROM boot */ #define RAM_HIGH_ADRS (LOCAL_MEM_LOCAL_ADRS + 0x01C80000) /* RAM address for sys image */ #define RAM_LOW_ADRS (LOCAL_MEM_LOCAL_ADRS + 0x00010000) Wish this help, Tom Wang In article <39AE714D.70103C5F@icn.siemens.de>, Matthias Gottwald wrote: > Hi folks, > I am working with Tornado 2 and generate a > compressed Bootrom-Image (Bootstrap-Loader) > by Project facilitiy and adapted Makefile/linker command file/ > config.h.... > I saw that the functions copyLongs() and fillLongs in bootInit.c > produce a wrong source/destination address. > The Copyloader-Routine is copied to RAM_LOW_ADRS. > After this all not used memory is cleared. > The Bootloader is copied to RAM_HIGH_ADRS. > Are these addresses valid? > I changed this file, so the addresses map to the right pointers. > Can this be made or did I wrong? > > Best regards Matthias. Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: BSP Validation Suite Question Date: Fri, 01 Sep 2000 00:28:13 GMT From: gregmg@my-deja.com Organization: Deja.com - Before you buy. Message-ID: <8omt6a$th7$1@nnrp1.deja.com> I'm trying to run the VxWorks BSP Validation Suite (part of the BSP developer's toolkit) on a Motorola MCP750 CPU. No matter what I try, the timestamp test is always skipped. pkLib *is* getting recompiled and the target does have timestamp support. What am I missing? Any ideas? Thanks, Greg G. timeStamp : WARNING: There is no timestamp support on the target or pkLib.o is not re-compiled after adding timestamp support enable the timestamp timer : SKIP sysTimestamp() long-term test : SKIP sysTimestampLock() long-term test : SKIP disable the timestamp timer : SKIP re-enable after disable : SKIP Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Load Error "Relocation value does not fit in 24 bits" Date: Fri, 01 Sep 2000 03:07:54 GMT From: john_94501@my-deja.com Organization: Deja.com - Before you buy. Message-ID: <8on6hv$81e$1@nnrp1.deja.com> References: <6KBr5.11372$K4.483872@newsread1.prod.itd.earthlink.net> Hi, You have configured your board for more than 32Mb of memory (which is the maximum range of a relative branch on PPC). Due to differences in where the host loader and the target loader store the sections that are downloaded you can find that the target loader will place your code too far away for it to be able to make kernel calls. The history for all of this comes from the PPC EABI which specifies that function calls will be made using relative branches. VxWorks complies, which causes this kind of problem on boards with more memory :( There are some solutions to this though: 1) Compile your code with -mlongcall (which will cause the compiler to use 32 bit branches, so your code will be bigger and each branch a little slower). 2) Configure your board for 32Mb initially, then use memAddToPool to add in the extra memory. 3) Link your application with the kernel image and boot it in one piece. For a little more information you might want to check these links out (you'll need a WindSurf account to access them): http://www.wrs.com/cgi-bin/csdocs/summaries/dispFAQ.cgi?104884.14 http://www.wrs.com/cgi-bin/csdocs/summaries/dispFAQ.cgi?122479.5 Hope that helps, John... In article <6KBr5.11372$K4.483872@newsread1.prod.itd.earthlink.net>, "Lou Kale" wrote: > I get an error "Relocation value does not fit in 24 bits" when I try to load > an application by typing > "ld > I'm using a PPC 604 with vxWorks 5.4 and Tornado 2. > > Loads done from Tornado 2 using a target server work ok. > > Does anyone know what the problem is? > > Regards, > -- > Lou Kale > KeNETec Inc. > 115 Hurley Road > Building 1B > Oxford, CT 06478 > Tel (203)-264-1888 x130 > Fax (203)-264-6724 > lou.kale@kenetec.com > > Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Target Shell Date: Fri, 01 Sep 2000 03:44:29 GMT From: "Dragan Ostojic" Organization: Sympatico Message-ID: References: <39AE530D.8F1841DC@hotmail.com> <39ae5707.0@seat.s3two.ie> Try to build the project using the tool in Tornado2 based on your BSP. Then configure VxWorks using the GUI tool so that it includes everything from the "target shell components" except "target shell script". Save the project, go to the directory where your .wpj file is generated and run wtxtcl %WIND_BASE%\host\src\hutils\configGen.tcl name.wpj. Then open prjParams.h file. Look into the #define INCLUDE_... options there. Compare these options with those that you have in your config.h file and I think you'll find what you're missing. Regards, Dragan. "Keith McHugh" wrote in message news:39ae5707.0@seat.s3two.ie... > Yep, it is happening to me as well. I don't know either! Anyway, your work > should not be hampered. Because Tornado Shell (Target server shell) should > work properly irrespective of the Target shell! > Keep your finger cross for more responses on the board in this regard! > > > "Somesh Agrawal" wrote in message > news:39AE530D.8F1841DC@hotmail.com... > > Hi all, > > > > (Target: pentium BSP) > > I have included "INCLUDE_SHELL" in my config.h and then build my image. > > This way I am getting the shell prompt at the Target, but most of it's > > functionality > > is not there.I have included some other related macro's also but still > > not able > > to get it working properly.Any help will be appreciated. > > > > Thanks, > > S.A > > > > > > > > --------------------------- Newsgroups: comp.os.vxworks Subject: lptDevDelete() ?? Date: Fri, 1 Sep 2000 10:47:36 +0200 From: "zbody" Organization: Slovenija OnLine - SiOL Message-ID: Hi! Does somebody know how to delete LPT device because as far as I can see there is no some kind of lptDevDelete() function. So if I need to delete LPT device (and I actually have to do this) how can I do this? Thanks! zbody --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Vxsim File I/O with NT Date: Fri, 01 Sep 2000 09:20:28 GMT From: hruesing@my-deja.com Organization: Deja.com - Before you buy. Message-ID: <8onsce$7c$1@nnrp1.deja.com> References: <8o6kgi$fvu$1@nnrp1.deja.com> In article <8o6kgi$fvu$1@nnrp1.deja.com>, jim.themis@genband.com wrote: According to WSR there is at the moment only a workaround available using the target server file system: > IDE: Tornado 2.0 > Component: tornado-vxsim > RTOS: VxWorks Version: 5.4 > Product Name: Tornado Release#: 2.0 Release Status: FCS > Products Affected: Tornado > Host: Intel x86/Pentium HostOs: HostOSRev: 4.0 sp3 > Arch Family: SIM NT Arch CPU: SIM NT BSP: Ver: > > DESCRIPTION > > The NT passFs device "host:" created by defining INCLUDE_NTPASSFS > in the SIM NT configuration, is inaccessible from both windsh and > target > shell. > > WORKAROUND > > The target server file system can be set up so that the target can > access the NT host file system. Select "Target Server File System" > in the "Target Server Properties" window of the target server > configurtion tool. Specify a Windows host file system. After launching > the target server, this file system can be accessed through the > device "/tgtsvr" on the target either directly from the shell or > programatically. > > I reviewed the archives but I could find anyone with a similar problem. > I am running VxSim on an NT box. I am trying to do file I/O via fopen, > fclose, fputs, etc. I am able to read files, but I am unable to create > and write files. I have tried adding "host:" to the beginning of my > path. I have enabled tsfs with r/w. I get no error when I perform > an fopen. However, when I execute an fputs(), I get the following > error: > ntPassFsLib: unknown ioctl = 0x14 > > I looked at errno and the value showed Unknown Request. > > Can anybody explain their set-up for reading and WRITE files under > VxSim. > > Thanks, > Jim Themis > > Sent via Deja.com http://www.deja.com/ > Before you buy. > Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Macronix LAN chip Date: Fri, 1 Sep 2000 11:38:22 +0200 From: "Bertram Tenner" Organization: Customer of UUNET Deutschland GmbH Message-ID: <39af7948$0$14980@businessnews.de.uu.net> Hello, I have a Macronix MX98725 LAN chip on a slot CPU in an Industrial x86 PC. Does anyone know which vxWorks network driver is compatible with this chip ? Thanks in advance Bertram --------------------------- Newsgroups: comp.os.vxworks Subject: Re: what is bootrom.hex? Date: Fri, 01 Sep 2000 06:30:10 -0700 From: Joe Durusau Organization: Lockheed Martin Corporation Message-ID: <39AFAF62.6EF395D9@lmco.com> References: <39AE5226.696ED2C8@orion-sim.com> You download it to the PROM on your target with a PROM burner, since it takes the place of the boot loader you have now. In almost every case you have to take the PROM out and load it with a sep machine, bu that may not be the case with your system. Personally, I would get a blank PROM and use that, so that in case of mistakes, the original is still intact. Speaking only for myself, Joe Durusau Brian Genisio wrote: > > I have figured out that I need to make a bootrom.hex to fix the problem > of a large image file, but then what? Do I tell the VxWorks boot loader > on the card to download that instead? Right now, it downloads the file > vxWorks that the build procedure makes. > > Thanks, > Brian --------------------------- End of New-News digest ********************** From vxwexplo-errs@csg.lbl.gov Fri Sep 1 06:12:02 2000 From: "Michael Lawnick" Date: Fri Sep 1 06:12:04 PDT 2000 Subject: (Fwd) TFFS on MV260x I wrote: >Hi folks, > >has anybody already integrated TFFS to on board flash of a MV230x, >MV240x, MV260x or MV270x ? >Could you share your code changes or help me to get on the road ? >I think vxWorks documentation is as simple as usually. Nobody answered :-(( Is there really nobody who has done this job ? Has anybody succeeded in integrating any other non-standard flash ? Regards, MfG Lawnick, SOFTEC GmbH ============================================== SOFTEC GmbH Tel +49-731-96600-0 Promenade 17 Fax +49-731-96600-23 D-89073 Ulm Michael Lawnick Germany lawnick@softec.de ============================================== From vxwexplo-errs@csg.lbl.gov Fri Sep 1 07:51:36 2000 From: "Patrick T. Pinkowski" Date: Fri Sep 1 07:51:39 PDT 2000 Subject: Software support for the BajaSpan and VxWorks 5.4 I have an Artesyn Bajaspan T1 Quad 860 processor board running VxWorks 5.4. I am attempting to configure the board to a single unframed T1 HDLC stream. Does anyone have software to configure the BajaSpan board to route the unframed T1 stream through the SCT (Dallas DS2152, Enhanced T1 Single-Chip Transceiver), through the TSI (Lucent T8105), to the 860's TDMa, and to an SCC port on the 860? Any assistance would be greatly appreciated. Thanks in advance, 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 vxwexplo-errs@csg.lbl.gov Sat Sep 2 04:03:08 2000 From: Vxworks Exploder Date: Sat Sep 2 04:03:11 PDT 2000 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Sat Sep 2 04:03:04 PDT 2000 Subject: Re: powerpc-wrs-vxworks and gcc-2.95.2 on Linux Subject: How the target Shell handles the console device driver ? Subject: Re: Adventnet SNMP agent for vxworks Subject: Re: Target Shell Subject: (Fwd) TFFS on MV260x Subject: Re: Target Shell Subject: [Q] how to use "ioDefPathSet"? Subject: Re: Creating a Project in T2 question Subject: Software support for the BajaSpan and VxWorks 5.4 Subject: Targetting COM ports 3 and 4 - need help Subject: US-NH: Portsmouth - Firmware Designers Subject: romStart Question: can I store stuff in USR_RESERVED_MEM Subject: romStart Question: can I store stuff in USR_RESERVED_MEM Subject: Problems with PCI ISR Subject: CrossWind Debugger Source Directories problem ... Subject: Need a few good VxWorks engineers in beautiful Santa Barbara ! Subject: Re: Green Hills Software Subject: Problems with WindView under Windows 2000. Subject: Re: Load Error "Relocation value does not fit in 24 bits" Subject: Re: How to work with PCI ? Subject: Re: Load Error "Relocation value does not fit in 24 bits" ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: powerpc-wrs-vxworks and gcc-2.95.2 on Linux Date: Fri, 1 Sep 2000 12:40:18 +0100 From: "Dave Korn" Organization: (Posted via) U-NET Internet Ltd. Message-ID: References: <39AD6E21.FF4571D7@campio.com> <39aebeee.6716227@news.yhteys.mtv3.fi> Kai Ruottu wrote in message <39aebeee.6716227@news.yhteys.mtv3.fi>... > > 2. preinstall the VxWorks headers into > $prefix/powerpc-wrs-vxworks/include You mean $prefix/powerpc-wrs-vxworks/sys-include. If the headers are there (which is where --with-headers copies them to at configure time) then they will be properly fixed. OTOH it doesn't really matter; I believe the gcc compiler, once built, will be fine with the old include files supplied by WRS; the only ones that need updating with the fixed versions are stdarg.h, vararg.h and va-ppc.h, since the varargs handling has changed substantially. Another point worth mentioning in reply to the original poster is that you don't need to worry about building all of the directories in the gcc source tree. After the first 'make all' gets as far as it can go, just chdir into the gcc-2.95.2/gcc subdir and do 'make all install' from that dir. There's been a lot of issues with the powerpc-wrs-vxworks target, but with some help from a guy called Rob Targosz I've more or less managed to get it working. I've figured out patches to a) enable the - -mlongcall option and b) prevent all those complaints about redefining big/little endian, and as soon as I've given it a couple of days testing and tidying up I will post a FAQ and patches here and perhaps a couple of other places as well. DaveK - -- They laughed at Galileo. They laughed at Copernicus. They laughed at Columbus. But remember, they also laughed at Bozo the Clown. --------------------------- Newsgroups: comp.os.vxworks Subject: How the target Shell handles the console device driver ? Date: Fri, 01 Sep 2000 11:48:38 GMT From: "JiHye, Sohn" Organization: Dacom Message-ID: Hello I'd like to know how the VxWorks target shell handles the console I/O. When a user try to communicate the target using RS-232C serial port or telent, he can get the shell prompt(->) in telnet terminal or serial comm. S/W like hyterterminal. I'm trying to develop my own monitor program like this scheme. First, I tried to create a file to control the ethernet device driver like the consoleFd. (- the File Descriptor to control the RS-232C serial port), but failed. I don't know how can I receive the monitoring request and send the response data by ethernet device driver or RS-232C serial port respectively. If you have any diffferent idea to implement this function, please let me know. Thanks for your help. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Adventnet SNMP agent for vxworks Date: Fri, 01 Sep 2000 11:49:03 GMT From: mathi@india.adventnet.com Organization: Deja.com - Before you buy. Message-ID: <8oo539$9hl$1@nnrp1.deja.com> References: <3999AE69.AF526782@yahoo.com> <8nemkb$cl5$1@nnrp1.deja.com> <399D42CC.23F50FAF@yahoo.com> <8nrfls$fti$1@nnrp1.deja.com> <8o1lub$otq$1@nnrp1.deja.com> Hi, Please find my answers below: In article <8o1lub$otq$1@nnrp1.deja.com>, pintu_12@my-deja.com wrote: > > > Hi, > > Apparently the advent net SNMP agent currently provides support > only for SNMP group counters, support for the Mib-2 variables would > be provided later. We support Mib-2 in our future release > > Also could any one of you let me know as to whether the Advent > Net SNMP agent would be compatible with TCP/IP stack of VxWorks > (as I am not very sure whether there SNMP agent could be integerated > with the VxWorks TCP/IP stack) Yes AdventNet SNMP agent can be integrated with VxWorks TCP/IP stack. For your information, we are doing 4.2 release second week of this month that will have more examples, tutorials, documents, etc. Thanks, Mathi. > > Randy, which SNMP agent of VxWorks did you try using - Envoy 8.0 > (better known as WindNet SNMP) or Envoy 9.1 ? An alternative to > Envoy could be the emanate SNMP agent. You have said that the > documentation for the wind river SNMP agent was good - which SNMP > agent are you reffering to - the Envoy 8.0 or the Envoy 9.1. I am > trying to use Envoy 9.1 and I am not finding the documentation to be > sufficient. In case you feel that the documentation for Envoy 8.0 is > good please let me know, > > Regards, > Piyush > > In article <8nrfls$fti$1@nnrp1.deja.com>, > randyj5573@my-deja.com wrote: > > My experiences with Wind River's SNMP agent has been extreemly > > frustrating. I have found that WRS tech support in other area's of > > Vxworks to be pretty good but their tech support for their SNMP agent > > to be almost non-existant. It seems that WRS's general tech suport has > > almost no experience with SNMP. They purchased their current agent > from > > a third party. WRS has since purchased the company that wrote their > > agent but there has yet to be any transfer of knowledge out to the > tech > > support people. WRS has fairly good general documentation about their > > SNMP agent but when it came time to write some code, I came up with > > questions on how some of their functions worked, what some of their > > internal structures looked like and questions about undocumented > > functions and macros that their mib compiler would use its generated > > code. I quickly came to the limit of what help their tech support > could > > provide and found myself teaching them so that maybe they could > > research a question to provide an answer. One question on traps took > > over three weeks to get an answer. The one thing that helped me was > > finding some programming examples on their web site. I was able to > > compile them and then step through them in order to figure out how > they > > work. > > > > Well enough venting. This was my 1st time at trying to extend an SNMP > > agent so I have no experience with other tool vendors to compare to. > > > > Regards, > > Randy Jones > > > > In article <399D42CC.23F50FAF@yahoo.com>, > > Bhavesh Patel wrote: > > > we plan to use both jmx agent and snmp embedded C agent, this is the > > reason > > > the architects here choose adventnet snmp toolkit. How are your > > experiences > > > with wind river's snmp agent? > > > Thanks for your response. > > > > > > bhavesh > > > randyj5573@my-deja.com wrote: > > > > > > > Why don't you use Wind River's SNMP agent? > > > > > > > > Randy Jones > > > > > > > > In article <3999AE69.AF526782@yahoo.com>, > > > > Bhavesh Patel wrote: > > > > > Hi, > > > > > > > > > > Has anybody used adventnet embedded/C snmp agent for vxworks > > > > platform? I > > > > > would like to know their experiences. Adventnet website does > not > > have > > > > > example/demo for vxworks platform. > > > > > > > > > > Thanks for any help. > > > > > > > > > > Bhavesh Patel > > > > > > > > > > > > > > > > > > Sent via Deja.com http://www.deja.com/ > > > > Before you buy. > > > > > > > > > > Sent via Deja.com http://www.deja.com/ > > Before you buy. > > > > Sent via Deja.com http://www.deja.com/ > Before you buy. > Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Target Shell Date: 31 Aug 2000 15:56:14 GMT From: litwin@izzy.jpl.nasa.gov (Todd Litwin) Organization: jpl.nasa.gov Message-ID: <8olv6u$cd3$1@nntp1.jpl.nasa.gov> References: <39AE530D.8F1841DC@hotmail.com> <39AE5CA6.20246EC5@swri.edu> <39AE6528.84F93BB3@hotmail.com> <39AE6FBA.20AF54C1@swri.edu> Try #define INCLUDE_CONFIGURATION_5_2 /* pre-tornado tools */ This should get you the target shell and all the pre-Tornado commands that you know and love. - -- Todd Litwin Jet Propulsion Laboratory (818) 354-5028 Todd.E.Litwin@jpl.nasa.gov --------------------------- Newsgroups: comp.os.vxworks Subject: (Fwd) TFFS on MV260x Date: Fri, 1 Sep 2000 14:49:16 +0200 From: "Michael Lawnick" Organization: Lawrence Berkeley National Laboratory Message-ID: <200009011248.OAA30424@net.softec.de> I wrote: >Hi folks, > >has anybody already integrated TFFS to on board flash of a MV230x, >MV240x, MV260x or MV270x ? >Could you share your code changes or help me to get on the road ? >I think vxWorks documentation is as simple as usually. Nobody answered :-(( Is there really nobody who has done this job ? Has anybody succeeded in integrating any other non-standard flash ? Regards, MfG Lawnick, SOFTEC GmbH ============================================== SOFTEC GmbH Tel +49-731-96600-0 Promenade 17 Fax +49-731-96600-23 D-89073 Ulm Michael Lawnick Germany lawnick@softec.de ============================================== --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Target Shell Date: Fri, 01 Sep 2000 09:15:21 -0400 From: Somesh Agrawal Organization: 3Com Message-ID: <39AFABE9.6788B068@hotmail.com> References: <39AE530D.8F1841DC@hotmail.com> <39AE5CA6.20246EC5@swri.edu> <39AE6528.84F93BB3@hotmail.com> <39AE6FBA.20AF54C1@swri.edu> <8olv6u$cd3$1@nntp1.jpl.nasa.gov> Thanks, for your help. Now everything is working fine. Thanks, S.A Todd Litwin wrote: > Try > > #define INCLUDE_CONFIGURATION_5_2 /* pre-tornado tools */ > > This should get you the target shell and all the pre-Tornado commands that you > know and love. > > -- > Todd Litwin > Jet Propulsion Laboratory > (818) 354-5028 > Todd.E.Litwin@jpl.nasa.gov --------------------------- Newsgroups: comp.os.vxworks Subject: [Q] how to use "ioDefPathSet"? Date: Fri, 1 Sep 2000 10:17:51 -0400 From: "Okehee Goh" Organization: UF Department of Computer & Information Science & Engineering Message-ID: <8oodt0$n9e$1@thunder.cise.ufl.edu> Hello, I'm trying to run ChaiVM(HP's Java Virtual Machine) on motorola MBX860 running VxWorks5.3.1 It failed to run it. The following steps are what I did. - -> ld 0,0,"slimrt_vxw.o" // load JVM value = 15336504 = 0xea0438 // set IO path to my machine oriole's path - -> ioDefPathSet("oriole:e:/jvmem/chaijvmvx/vxmbx860") value = -1 = 0xffffffff - -> putenv("CLASSPATH=.") // value = -1 = 0xffffffff - -> putenv("JAVART_MEMORY_SIZE=800000") value = 0 = 0x0 - -> vx_main "Test" //run main point value = -1 = 0xffffffff I don't know what is wrong with the setting of "ioDefPathSet". Could you tell me how to use this subroutine? Is there anybody who have done a porting a ChaiVM on VxWorks? If so, any kind of advice or comment will be great help to me. Thank you so much. ******************************************************************** Okehee Goh ogoh@cise.ufl.edu http://www.cise.ufl.edu/~ogoh RealTime System Lab. (T) 1-352-392-6875 CISE Dept. University Of Florida --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Creating a Project in T2 question Date: 1 Sep 2000 15:05:09 GMT From: johan@borksoft.xs4all.nl Organization: Borksoft Message-ID: <8oogj5$1s1$1@news1.xs4all.nl> References: Mike Kordik wrote: : I have an existing project layout on my hard drive. I am trying to convert : this to a project in T2. I added all the source files from the various : directories to the project. I also had T2 generate the external dependency : list. Because the header files are in a different directory the dependency : scan knows they exist but does not know where to find them. When the : makefile is generated the correct dependencies are there but make does not : know how to find them. How do I get T2 to find these files and include the : path to the header files in the makefile? Have a look at my website. You have to set the VPATH macro to the right value. Also have a look at the project-part of the website, as this describes the possibilty of using several sub-projects. Groeten, Johan - -- o o o o o o o . . . ___________________________________ o _____ || Johan Borkhuis | .][__n_n_|DD[ ====_____ | johan@borksoft.xs4all.nl | >(________|__|_[_________]_|________________________________| _/oo OOOOO oo` ooo ooo 'o!o!o o!o!o` === VxWorks FAQ: http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html === --------------------------- Newsgroups: comp.os.vxworks Subject: Software support for the BajaSpan and VxWorks 5.4 Date: Fri, 01 Sep 2000 10:36:32 -0400 From: "Patrick T. Pinkowski" Organization: Lawrence Berkeley National Laboratory Message-ID: <8ooh59$5bg$1@overload.lbl.gov> I have an Artesyn Bajaspan T1 Quad 860 processor board running VxWorks 5.4. I am attempting to configure the board to a single unframed T1 HDLC stream. Does anyone have software to configure the BajaSpan board to route the unframed T1 stream through the SCT (Dallas DS2152, Enhanced T1 Single-Chip Transceiver), through the TSI (Lucent T8105), to the 860's TDMa, and to an SCC port on the 860? Any assistance would be greatly appreciated. Thanks in advance, 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 --------------------------- Newsgroups: comp.os.vxworks Subject: Targetting COM ports 3 and 4 - need help Date: Fri, 1 Sep 2000 17:40:21 +0100 From: "Peter Deed" Message-ID: <967826453.9627.0.nnrp-14.c1ed3722@news.demon.co.uk> Hi I am running my application on an Arcom Media GX Board which uses a 233Mhz Cyriz chip. This board works with the Pentium BSP and contains 4 COM ports. I want to read and write to these ports but can only manage to access ports COM1 and COM2. I have added the relevant code to sysSerial.c and config.h but it makes no difference. WindRiver themselves tell me that it is probably more difficult than I think and I will probably have to allow COM1 and 3 to share one Interrupt Leven and COM2 and COM4 to share another and then do the detection of which COM port is relevant within ISR code. I do have enough interrupts for 4 ports but I realise I may not be allowed to handle them separately. Has anyone done interrupt sharing before ?? Many thanks to replies in advance Peter Deed Infocell Telecom ================== --------------------------- Newsgroups: comp.os.vxworks Subject: US-NH: Portsmouth - Firmware Designers Date: Fri, 01 Sep 2000 17:10:03 GMT From: cetaceannet@my-deja.com Organization: Deja.com - Before you buy. Message-ID: <8oonsq$4a$1@nnrp1.deja.com> Cetacean Networks is building the components needed for real-time applications to operate on the Internet. We are working towards the dream of a single network, able to support all forms of communications, becoming a reality. The company has a unique and powerful technology for achieving this goal. The product components include large IP switches and routers, network software and real-time appliances. e're currently looking for Firmware Designers. We’ve received $15 million in funding from CRV and Bessemer. We’re serious about IPO – no buy out. Hoping for Beta Release in 18 months. Sticking to 50 hour work weeks, open to some telecommuting. Great benefits: Medical, Dental, Vision, Flex Spending account, 401K. 15 vacation days, 8 holidays, 2 personal days and 5 sick days! Please contact me at the information below if you're interested in hearing more! Carrie Lussier Cetacean Networks Senior Recruiter 603. 766.6119 ddial 603.766.6190 fax clussier@cetaceannetworks.com Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: romStart Question: can I store stuff in USR_RESERVED_MEM Date: Fri, 1 Sep 2000 14:28:46 -0400 From: jim Organization: University of Waterloo Message-ID: References: <8oonsq$4a$1@nnrp1.deja.com> Hello, I have mad some modifications to the VxWorks boot code in romInit.s ... these modifications collect a small number of statistics on a few devices on my board. I now would like to store these statistics in a safe location ... currently I have decided to store these statistics in at the top of RAM in the area defined by the USR_RESERVED_MEM ... I do this because it is not zero'd by romStart.c at startup. Should it be safe to do this? I am hoping to access these "statistics's" later using a user task after the system is fully "awake". Cheers and Thanks in advance for any advice, Jim --------------------------- Newsgroups: comp.os.vxworks Subject: romStart Question: can I store stuff in USR_RESERVED_MEM Date: Fri, 01 Sep 2000 18:53:15 GMT From: "Jim" Organization: Magma Communications Ltd. Message-ID: Hello, I have mad some modifications to the VxWorks boot code in romInit.s ... these modifications collect a small number of statistics on a few devices on my board. I now would like to store these statistics in a safe location ... currently I have decided to store these statistics in at the top of RAM in the area defined by the USR_RESERVED_MEM ... I do this because it is not zero'd by romStart.c at startup. Should it be safe to do this? I am hoping to access these "statistics's" later using a user task after the system is fully "awake" Cheers and Thanks in advance for any advice, Jim --------------------------- Newsgroups: comp.os.vxworks Subject: Problems with PCI ISR Date: 1 Sep 2000 20:10:02 GMT From: tsikes@netcom.com (Terry Sikes) Organization: NETCOM / MindSpring Enterprises, Inc. Message-ID: <8op2eq$t6l$1@slb6.atl.mindspring.net> I'm having a problem connecting a PCI interrupt to my ISR. I'm using the Pentium BSP. I'm able to communicate with the card using I/O space, and the interrupt number I'm getting from config space matches that reported by the BIOS at boot time. I use the following code to hook up the ISR (myInterrupt is 10 at the moment, unshared): - ---snippet 1--- if (intConnect( INUM_TO_IVEC(INT_VEC_GET(myInterrupt)), (VOIDFUNCPTR) myIsr, (int) myIntSem) == ERROR) { logError("tm: intConnect failed"); } - ------ - ---snippet 2--- void myIsr(SEM_ID myIntSem) { semGive(myIntSem); } - ------ However, when I generate an interrupt using the card, I get the message "dosFsDevInit failed." on the console, and VxWorks appears to be locked up. It certainly appears as though the wrong ISR is getting called. Any advice would be greatly appreciated! Thanks, Terry - -- tsikes@netcom.com --------------------------- Newsgroups: comp.os.vxworks Subject: CrossWind Debugger Source Directories problem ... Date: Fri, 1 Sep 2000 17:12:27 -0400 From: "Alexander Povolotsky" Organization: Lucent Technologies, Columbus, Ohio Message-ID: <8op65u$63k@nntpa.cb.lucent.com> Crosswind Wind starts OK and reads symbols from the application module ... Unfortunately, after attaching to the task of interest, it shows me one of my source files in one of the directories, but not the one I want to see ... According to the documentation (gdb) dir "/some/fully/qualified/source/path" should set the source path to access the needed source file. It does sets it, as indicated by (gdb) show dir but, apparently, commmands (gdb) list filename.c:lineNumber or (gdb) list filename:functionName still fail .... changing directory to the one of interest prior to executing "list" (gdb) cd "/some/fully/qualified/source/path" did not help either ... so using commands (gdb) info source and (gdb) info sources showed that the source visibility is defined by the "compilation directory" - - documentation refers to it as '$cdir' string versus '$cwd' string , which refers to the "current working directory" We are using "Clearmake" in our multinested directories recursive "total" build process .... The shell we are using is "bash" .... Could I change setting of "cdir" ? - could I do it in .gdbinit ? (and if yes, what would be the correct syntax) ? In general, what is the solution for this problem (we are using Tornado 1.0.1 on Solaris 2.7 ) ? Best Regards, Alex Povolotsky --------------------------- Newsgroups: comp.os.vxworks Subject: Need a few good VxWorks engineers in beautiful Santa Barbara ! Date: Fri, 01 Sep 2000 23:16:00 GMT From: connected_systems@my-deja.com Organization: Deja.com - Before you buy. Message-ID: <8opdb0$q4i$1@nnrp1.deja.com> Connected Systems is a high tech startup in Santa Barbara, California nearing launch of our Internet messaging & voice appliance. Our products enable service providers to enhance their Broadband/Internet offerings with our new breed of high-value applications such as messaging, voice mail and beyond. We are growing rapidly and currently seeking experience in voice, CTI, messaging, networking, real-time systems for the following positions: · Software Engineers · Sr. Software Engineers Design and develop integrated applications, drivers on our IP-telephony platforms. C/C++ and networking experience required. Senior positions require RTOS and 5+ years of industry experience. VxWorks experience and familiarity with Unified Messaging, VoIP, LDAP, IMAP protocols desired. We offer competitive salaries, benefits, SEP program, and pre-IPO stock options. Please send your resumes to: jobs@connsys.com or Fax: (775) 205-6564 Check out our web site www.connectedsystems.com for more information. Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Green Hills Software Date: Fri, 01 Sep 2000 21:56:26 -0400 From: mcmorran@visi.net (Peter McMorran) Message-ID: <39b0638f$2$zpzbeena$mr2ice@news.visi.net> References: <8ogv5q$rpv$1@nnrp1.deja.com> In <8ogv5q$rpv$1@nnrp1.deja.com>, on 08/29/00 at 06:25 PM, gpbeck@my-deja.com said: >Is anyone using the Green Hills software to develop/debug their >VxWorks applications? We're using AdaMulti to port several Ada/Fortran/C Unix programs to embedded PowerPC. Naturally, we have to make some adaptations for VxWorks, but none related to the compiler system. We were tripped by one thing related to Fortran list-directed I/O. Most Unix compilers allow strings in a list-directed input file to be enclosed in double quotes, and we were in the habit of using these. AdaMulti sticks to the letter of the Fortran standard and requires single quotes. Since we have routines that painstakingly write these quotes, we had to do a little work to get this working. >If so, what is your opinion of the their software? Is it >reliable/stabile? How is their support? We found a bug in the Ada runtime which they fixed in a few weeks. They are now looking at a bug in the Fortran runtime which produces a memory leak. Support seems good, but these fundamental and easily reproduced bugs are surprising in a mature product. Aside from these two bugs, our programs run correctly, and they are over 100 K lines of code. So the compiler does seem to be reliable. The debugger is pretty neat, letting you grab multiple tasks at once. I personally dislike the AdaMulti GUI, using it only to access the debugger, or to find out the code for an option. We run the build program directly from make, creating the build files by cloning from existing ones. With a master makefile in the top directory and makefiles in each subdirectory, we can update everything with a single make command. The builder keeps track of all source dependencies, recompiling all affected programs when a C header, or Fortran include file, or Ada package spec is changed. Cheers, Peter - -- - ----------------------------------------------------------- mcmorran@visi.net (Peter McMorran) - ----------------------------------------------------------- --------------------------- Newsgroups: comp.os.vxworks Subject: Problems with WindView under Windows 2000. Date: Fri, 1 Sep 2000 21:23:53 -0700 From: Simon DiMaio Organization: ITServices, University of British Columbia Message-ID: Hi I have done a full install of Tornado 2.0 under Windows 2000. Upon running Tornado, I receive the following message: couldn't read file "TRY=localhost\host\resource\tcl\WindView.win32.tcl": no such file or directory Tornado runs fine, but WindView is not available. The base directory for my Tornado installation is C:\Tornado, so I don't know where the "TRY=localhost" has come from. Has anybody experienced a similar error message? Any ideas for a remedy would be greatly appreciated. Thanks in advance, Simon DiMaio - ------------------------------------------------------------------------ Simon DiMaio Email: simond@ece.ubc.ca Robotics and Control Group Dept. of Electrical and Computer Engineering Phone: (604)822-9215 University of British Columbia Fax : (604)822-5949 --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Load Error "Relocation value does not fit in 24 bits" Date: Sat, 02 Sep 2000 00:39:11 -0400 From: Al Johnston Organization: MindSpring Enterprises Message-ID: <39B0846F.E5FCCF83@mindspring.com> References: <6KBr5.11372$K4.483872@newsread1.prod.itd.earthlink.net> <8on6hv$81e$1@nnrp1.deja.com> > 1) Compile your code with -mlongcall (which will cause the compiler to > use 32 bit branches, so your code will be bigger and each branch a > little slower). Can I but in ... please? I ran across the "does not fit" a while back and the -mlongcall fixed it... But I have not seen any "32 bit branches"... And I have been looking for them. the Bx (BA, BLA) are the closest I have come. The have 26 bits. And that is where my question is. The doc says that these 26 bits (called the LI field) has two 0 bits concatenated to the little end, then the big end is "sign extended" and the result is used in calculating the effective address to jump to. But if you sign extend, then its a relative branch, not an absolute branch so the doc (32 bit prog env) seems to be contradicting itself. Can you straighten me out on this? The only thing I can come up is if the doc was WRONG when it said the LI field was sign extended. In that case, the field would just be a unsigned absolute 26 bit address ... and that gets you to abs. address 0 to 64 Meg. If the book is RIGHT, then you can get to abs address 0 to 32 Meg, and I have no idea what the cpu would do for the other half of the range.... give me a clue please. tnx > > 2) Configure your board for 32Mb initially, then use memAddToPool to add > in the extra memory. > > 3) Link your application with the kernel image and boot it in one piece. > > For a little more information you might want to check these links out > (you'll need a WindSurf account to access them): > > http://www.wrs.com/cgi-bin/csdocs/summaries/dispFAQ.cgi?104884.14 > http://www.wrs.com/cgi-bin/csdocs/summaries/dispFAQ.cgi?122479.5 > > Hope that helps, > > John... > In article <6KBr5.11372$K4.483872@newsread1.prod.itd.earthlink.net>, > "Lou Kale" wrote: > > I get an error "Relocation value does not fit in 24 bits" when I try > to load > > an application by typing > > "ld > > > I'm using a PPC 604 with vxWorks 5.4 and Tornado 2. > > > > Loads done from Tornado 2 using a target server work ok. > > > > Does anyone know what the problem is? > > > > Regards, > > -- > > Lou Kale > > KeNETec Inc. > > 115 Hurley Road > > Building 1B > > Oxford, CT 06478 > > Tel (203)-264-1888 x130 > > Fax (203)-264-6724 > > lou.kale@kenetec.com > > > > > > Sent via Deja.com http://www.deja.com/ > Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: How to work with PCI ? Date: Fri, 01 Sep 2000 09:12:41 +0200 From: "Dr. Peter Salewsky" Message-ID: <39AF56E9.1AF6538D@resolut.de> References: <8oif6g$5li$1@news.netvision.net.il> Ezra Koper schrieb: [...] > We are using a Pentium2 333Mhz PC platform. [...] > There is almost no ducementation on how to use PCI with VxWorks. > Any idea were i can find more information? > for Tornado II: Tornado/docs/vxworks/bsp/pcPentium.html (link sysLib, functions pci*) > 2. > I can't activate commands as sysPciDevShow. This command paper in > Widriver WindSurf WEB site for Technical Solution 130785.5 > so what might be the problem? I do not know this paper, but the above documentation does not contain sysPciDevShow; try pciDeviceShow instead. [...] Peter - -- Dr. Peter Salewsky phone +49 3493 743-20 resolut GmbH fax +49 3493 743-25 Antonienstr. 3 dialog@resolut.de D-06749 Bitterfeld, Germany www.resolut.de --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Load Error "Relocation value does not fit in 24 bits" Date: Sat, 02 Sep 2000 09:13:17 GMT From: john_94501@my-deja.com Organization: Deja.com - Before you buy. Message-ID: <8oqgb5$vru$1@nnrp1.deja.com> References: <6KBr5.11372$K4.483872@newsread1.prod.itd.earthlink.net> <8on6hv$81e$1@nnrp1.deja.com> <39B0846F.E5FCCF83@mindspring.com> Hi, In article <39B0846F.E5FCCF83@mindspring.com>, Al Johnston wrote: > > 1) Compile your code with -mlongcall (which will cause the compiler to > > use 32 bit branches, so your code will be bigger and each branch a > > little slower). > > Can I but in ... please? I ran across the "does not fit" a while > back and the -mlongcall fixed it... But I have not seen any > "32 bit branches"... And I have been looking for them. the Bx > (BA, BLA) are the closest I have come. The have 26 bits. And > that is where my question is. The doc says that these 26 bits > (called the LI field) has two 0 bits concatenated to the little > end, then the big end is "sign extended" and the result is used > in calculating the effective address to jump to. But if you > sign extend, then its a relative branch, not an absolute branch > so the doc (32 bit prog env) seems to be contradicting itself. The destination address is sign extended, and simply takes the place of the relative offset in the non-absolute addressing mode, so you get to jump to absolute addresses in the range -32Mb to +32Mb (give or take a few bytes). In real addresses, 0xfe000000 to 0x1ffffffc. Relative branches, of course, can therefore be +/- 32Mb from the location of the branch. All the absolute flag does is exchange current PC for zero in the expression. As for 32 bit branches, they are in fact a sequence of instructions that loads the address into a register (two instructions), moves that to either LR or CTR since they are the only registers you can branch to, and then branches. So the code might look like this: lis r0, HI(destination) addi r0, r0, LO(destination) mtlr r0 blrl (I think I got that right - no doubt somebody'll correct it if not :) HTH, John... Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- End of New-News digest ********************** From vxwexplo-errs@csg.lbl.gov Sat Sep 2 07:47:21 2000 From: "Tim Allen" Date: Sat Sep 2 07:47:23 PDT 2000 Subject: Re: TFFS on MV260x >>has anybody already integrated TFFS to on board flash of a MV230x, >>MV240x, MV260x or MV270x ? >>Could you share your code changes or help me to get on the road ? >>I think vxWorks documentation is as simple as usually. > >Nobody answered :-(( Since I haven't noticed any responses, this should stop you feeling lonely, but I don't think it will help! >Is there really nobody who has done this job ? >Has anybody succeeded in integrating any other non-standard flash ? I have integrated TFFS with a customer's flash programming library, which presented a non-standard interface. It wasn't as straightforward as the documentation would suggest. As well as some inconsistent function/structure definitions, I noticed that the Translation Layer was accessing the flash hardware directly and not going through the MTD layer as I expected. I did get it going though. Unfortunately, I don't think I know enough about the Mot boards/flash to help you specifically. Regards, Tim. -------------------- Tim Allen Ltd Real Time Embedded Software Consultancy From vxwexplo-errs@csg.lbl.gov Sun Sep 3 04:03:06 2000 From: Vxworks Exploder Date: Sun Sep 3 04:03:09 PDT 2000 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Sun Sep 3 04:03:03 PDT 2000 Subject: Re: powerpc-wrs-vxworks and gcc-2.95.2 on Linux Subject: #pragma pack and i960 Subject: Re: #pragma pack and i960 Subject: re: TFFS on MV260x Subject: Re: MPC8260 UPM Initialisation Subject: Re: romStart Question: can I store stuff in USR_RESERVED_MEM Subject: MUX and stackRcvRtn Subject: Re: #pragma pack and i960 ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: powerpc-wrs-vxworks and gcc-2.95.2 on Linux Date: Sat, 02 Sep 2000 11:58:46 GMT From: karuottu@freenet.hut.fi (Kai Ruottu) Organization: Telia Finland Customer News Service Message-ID: <39b0dc0e.1629563@news.yhteys.mtv3.fi> References: <39AD6E21.FF4571D7@campio.com> <39aebeee.6716227@news.yhteys.mtv3.fi> Reply-To: karuottu@freenet.hut.fi On Fri, 1 Sep 2000 12:40:18 +0100, "Dave Korn" wrote: >Kai Ruottu wrote in message <39aebeee.6716227@news.yhteys.mtv3.fi>... >> >> 2. preinstall the VxWorks headers into >> $prefix/powerpc-wrs-vxworks/include > >You mean $prefix/powerpc-wrs-vxworks/sys-include. No, the 'gcc/INSTALL' and the 'Using and Porting...", ie. the official GCC manual clearly says that one must use the 'include'. Please see "Installation / Cross-Compiler / Cross Headers" from the fine manual: - ----------------- a clip from the gcc-2.96 manual --------------- Cross-Compilers and Header Files If you are cross-compiling a standalone program or a program for an embedded system, then you may not need any header files except the few that are part of GNU CC (and those of your program). However, if you intend to link your program with a standard C library such as libc.a, then you probably need to compile with the header files that go with the library you use. When you have found suitable header files, put them in the directory /usr/local/target/include, before building the cross compiler. Then installation will run fixincludes properly and install the corrected versions of the header files where the compiler will use them. Provide the header files before you build the cross-compiler, because the build stage actually runs the cross-compiler to produce parts of libgcc.a. (These are the parts that can be compiled with GNU CC.) Some of them need suitable header files. - ----------------- clip ---------------------------------------- If you see something wrong in the GCC docs, please report them as bugs. The meaning of the directories may be changed only by changing the official GCC documents. If RMS would write that the right place for the target system headers is now the 'sys-include', then I would believe. The directoris people have learned to use in their installations during the GCC-history cannot be changed in a single day... The 'sys-include' was aimed for third-party headers which should be used instead of, or additionally with the standard system headers. GGC looks the 'sys-include' before the 'include' and this is why it is so suitable for the third-party stuff. And people normally don't want to mix the standard and the additional local headers, so a separate directory for the local ones was needed. But the 'sys-' in the name is more than a little misleading... The 'sys-include' goes with the name 'CROSS_INCLUDE_DIR' while building, and the 'include' with the name 'TOOL_INCLUDE_DIR'. The docs say about these two: - ----------------- clip ---------------------------------------- CROSS_INCLUDE_DIR is used only for a cross compiler. GNU CC doesn't install anything there. TOOL_INCLUDE_DIR is used for both native and cross compilers. It is the place for other packages to install header files that GNU CC will use. For a cross-compiler, this is the equivalent of /usr/include. When you build a cross-compiler, fixincludes processes any header files in this directory. - ----------------- clip ---------------------------------------- I think here is the big 'typo'... The words "It is the place for other packages to install header files that GNU CC will use." should be after the CROSS_INCLUDE_DIR, otherwise the explanation about "the equivalent of /usr/include" is bullshit. Who installs any third- party headers into '/usr/include'? Somehow I have forgotten to report this bug, although I have fixed it in my own docs... I remember the 'www.gnu.org' having some weird info about these things... BTW, the '--with-headers' is handy for pointing to the Linux kernel headers while building glibc-2.x for/with a cross-compiler... For the glibc-2.x they are 'third-party headers'... Using unnecessary options with configure is not good, the example in the GCC-manual uses only the '--target=...'. A newbie starting only with that is better than starting with ten options and not understanding anything about them... This seems to be just the same kind of issue as was with the CONFIG.SYS'es and AUTOEXEC.BAT's in DOS-machines -- people copied things there from their friends because they said them being 'useful' in their machines... > If the headers are there (which is where --with-headers copies them >to at configure time) then they will be properly fixed. So you would trust the 'fixincludes' more than yourself ? Funny but just the 'gcc/fixinc' subdir has caused the most problems with the current GCC snapshots... I haven't seen it working for any target under Linux and cannot understand why the GCC build tries to fix the headers from the Cygnus newlib-1.8.2, Linux-glibc-2.x or from any up-to-date header-collections. Very old systems with their original headers like SCO Xenix or SunOS 4.1.x may have non-ANSI headers, but who says that they have remained unfixed, probably the cross-compiler to be built for them is already the tenth in the series... Keeping all the target headers in two places, the original unfixed ones in one and the fixed-fixed ones in another (after fixincludes one must always check the sanity of the fixes!) sounds somehow unsane, so most people have fixed the original headers long time ago and disable the fixincludes... Perhaps the VxWorks headers still live in the '80's, at least the famous '#define NULL (void *(0))' was there. Of course all C++ compiles puke when seeing this... Some people say that the NULL-problems are now fixed in newer VxWorks headers. >I believe the gcc compiler, once built, will be fine with the old include >files supplied by WRS; the only ones that need updating with the >fixed versions are stdarg.h, vararg.h and va-ppc.h, since the varargs >handling has changed substantially. Those VxWorks standard headers I have seen, included C++ headers like 'new.h' and 'typeinfo' in them,, but at least the 'typeinfo' isn't present in the older ones.. And then they had some C-headers, just the same as will come with GCC, 'stddef.h' and 'stdarg.h'. The normal practice is to try to use the standard target ones with C and to use the GCC ones with C++. Anyway GCC uses the target headers while building, but uses the GCC ones after installation. So there are all kind of headers clashes while building and another after installation. Something must be done, so any instructions about how to fix the standard 'stdarg.h' and 'stddef.h' are really needed.... >Another point worth mentioning in reply to the original poster is that >you don't need to worry about building all of the directories in the gcc >source tree. After the first 'make all' gets as far as it can go, just chdir >into the gcc-2.95.2/gcc subdir and do 'make all install' from that dir. Yes, if needing GCC, one needs to build only 'gcc'... The 'make all-gcc' takes care about this. Combining libiberty and libstdc++ with GCC wasn't perhaps very good, now people don't understand having a 'fully working compiler' when the build says "The compiler cannot make executables" (no C-library yet or no fully defined target yet). Ie. mixing compiling to compiling + linking is very common... >There's been a lot of issues with the powerpc-wrs-vxworks target, >but with some help from a guy called Rob Targosz I've more or less >managed to get it working. I've figured out patches to a) enable the >-mlongcall option and b) prevent all those complaints about redefining >big/little endian, and as soon as I've given it a couple of days testing >and tidying up I will post a FAQ and patches here and perhaps a >couple of other places as well. What about all the generated multilibs with AIX, Linux, SysV etc. calling conventions? And all the little-endian variations... I saw a new tmake_file, 'gcc/config/rs6000/t-vxworks', being necessary instead of the current 't-ppcbare' or something, which now defines all kind of perhaps unnecessary variations. I used only the '-msoft-float' to multilibs. But shouldn't there be some CPU-specific variations too? Or is the 'libgcc.a' totally unused with VxWorks? Has VxWorks its own speedy basic soft-float routines (fadd/fsub/fmul/fdiv) and the slow ones in 'fp-bit.c' aren't used and so don't need any optimizations for different PowerPC-CPU's ? Cheers, Kai --------------------------- Newsgroups: comp.os.vxworks Subject: #pragma pack and i960 Date: Sat, 02 Sep 2000 07:52:41 -0500 From: Steve Bakke Organization: Airnews.net! at Internet America Message-ID: <3B29B7E6452D52CD.CD8C08CDBEAC22C0.EBD8CD27DF40F6B6@lp.airnews.net> I'm trying to use #pragma pack on the i960 and it appears it doesn't do anything. #pragma pack 1 struct TEST { short t1; long t2; }; The offset of t2 is 4 not 2 as it should be. We use the same code with x86 and all is fine. The "crack" VxWorks documentation doesn't mention much about #pragma, but the Intel cc960 compiler manual says they work just fine. Any ideas? Steve --------------------------- Newsgroups: comp.os.vxworks Subject: Re: #pragma pack and i960 Date: Sat, 02 Sep 2000 13:58:29 GMT From: "telia" Organization: Enitel Internet Public Access Message-ID: <9I7s5.7119$Dxe.197278208@news.telia.no> References: <3B29B7E6452D52CD.CD8C08CDBEAC22C0.EBD8CD27DF40F6B6@lp.airnews.net> Steve Bakke wrote in message news:3B29B7E6452D52CD.CD8C08CDBEAC22C0.EBD8CD27DF40F6B6@lp.airnews.net... > I'm trying to use #pragma pack on the i960 and it appears > it doesn't do anything. > > #pragma pack 1 > struct TEST { > short t1; > long t2; > }; > > The offset of t2 is 4 not 2 as it should be. We use the same > code with x86 and all is fine. > > The "crack" VxWorks documentation doesn't mention much about > #pragma, but the Intel cc960 compiler manual says they work just fine. > > Any ideas? > > > Steve try compiler switch -mold-align It worked for us ( a long time ago ) --------------------------- Newsgroups: comp.os.vxworks Subject: re: TFFS on MV260x Date: Sat, 2 Sep 2000 15:41:00 +0100 From: "Tim Allen" Organization: Lawrence Berkeley National Laboratory Message-ID: <8or5h7$t1n$1@overload.lbl.gov> >>has anybody already integrated TFFS to on board flash of a MV230x, >>MV240x, MV260x or MV270x ? >>Could you share your code changes or help me to get on the road ? >>I think vxWorks documentation is as simple as usually. > >Nobody answered :-(( Since I haven't noticed any responses, this should stop you feeling lonely, but I don't think it will help! >Is there really nobody who has done this job ? >Has anybody succeeded in integrating any other non-standard flash ? I have integrated TFFS with a customer's flash programming library, which presented a non-standard interface. It wasn't as straightforward as the documentation would suggest. As well as some inconsistent function/structure definitions, I noticed that the Translation Layer was accessing the flash hardware directly and not going through the MTD layer as I expected. I did get it going though. Unfortunately, I don't think I know enough about the Mot boards/flash to help you specifically. Regards, Tim. - -------------------- Tim Allen Ltd Real Time Embedded Software Consultancy --------------------------- Newsgroups: comp.os.vxworks Subject: Re: MPC8260 UPM Initialisation Date: Sat, 2 Sep 2000 08:46:19 -0700 From: "Davis" Organization: SBC Internet Services Message-ID: References: <3981ab00.340821029@192.168.1.11> <8lse5d$ogr$1@nnrp1.deja.com> I think romInit.s is the place to initialize CS's, assuming you use EST8260BSP with vxWorks. Davis wrote in message news:8lse5d$ogr$1@nnrp1.deja.com... > In article <3981ab00.340821029@192.168.1.11>, > John Keightley wrote: > > I'm looking for help setting up UPM state machines on an EST8260 > > system - specifically UPMC and CS8 to access our own memory mapped > > subsystem. We can initialise this OK via VisionICE probe - now need to > > do this as part of the VxWorks initialisation sequence. > > > > I've tried setting this up in sysHwInit() - writing to BR8,OR8 but > > then get a crash when attempting to write the first RAM location > > (after loading MDR) after setting the OP code bit in MCMR > > > > Does anyone have sample code or experience with this? > > 1. Did you check the other bits in MCMR, such as BSEL and RFEN ? > 2. Are you sure the memory area you set BR8 to isn't already covered > by another (previously initialized) chip select and that is the one > that's firing ? > 3. Are you in C or assembly ? Are you sure you're writing a 32-bit > value to MCMR ? > > I believe the 860FADS BSP has assembly code to init the UPM on the 860 > if you can get a hold of that... > > hth, > Dan Gold gold@sdsu.edu > > > Sent via Deja.com http://www.deja.com/ > Before you buy. > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: romStart Question: can I store stuff in USR_RESERVED_MEM Date: Sat, 02 Sep 2000 09:58:19 -0700 From: "Michael R. Kesti" Organization: MK Associates Message-ID: <39B131AB.CF1C60A4@gv.net> References: Reply-To: mkesti@gv.net Jim wrote: >I have mad some modifications to the VxWorks boot code in romInit.s ... >these modifications collect a small number of statistics on a few devices >on my board. I now would like to store these statistics in a safe location >... currently I have decided to store these statistics in at the top of >RAM in the area defined by the USR_RESERVED_MEM ... I do this because it >is not zero'd by romStart.c at startup. > >Should it be safe to do this? Yes, that should work very well. I've made similar use of USR_RESERVED_MEM. - -- ======================================================================== Michael Kesti | "And like, one and one don't make | two, one and one make one." mkesti@gv.net | - The Who, Bargain --------------------------- Newsgroups: comp.os.vxworks Subject: MUX and stackRcvRtn Date: Sat, 02 Sep 2000 17:16:12 GMT From: nicenice@my-deja.com Organization: Deja.com - Before you buy. Message-ID: <8orckg$sq0$1@nnrp1.deja.com> Hello. I have written two simple MUX protocols. One is snarf type protocol and the other is a type defined in rfc1700. The MUX calls stackRcvRtn of snarf protocol, and stackRcvRtn returns FALSE so that the other protocol can receive. But the MUX never calls stackRcvRtn in the other protocol. Could you tell me what is wrong ? At the time, muxShow command shows my two protocols, IP4.4 arp and IP4.4. Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: #pragma pack and i960 Date: Sat, 02 Sep 2000 16:24:33 +0100 From: molochai Message-ID: <4d3bc8f749%molochai@vapour-trail.demon.co.uk> References: <3B29B7E6452D52CD.CD8C08CDBEAC22C0.EBD8CD27DF40F6B6@lp.airnews.net> <9I7s5.7119$Dxe.197278208@news.telia.no> In message <9I7s5.7119$Dxe.197278208@news.telia.no> "telia" wrote: > > Steve Bakke wrote in message > news:3B29B7E6452D52CD.CD8C08CDBEAC22C0.EBD8CD27DF40F6B6@lp.airnews.net... > > I'm trying to use #pragma pack on the i960 and it appears > > it doesn't do anything. > > > > #pragma pack 1 > > struct TEST { > > short t1; > > long t2; > > }; > > > > The offset of t2 is 4 not 2 as it should be. We use the same > > code with x86 and all is fine. > > You probably need a align directive as well. (Assuming you're using GCC that is). Forcing alignment of 2 will probably do the trick. This can be applied to the individual structure declarations using the __attribute keyword, > > The "crack" VxWorks documentation doesn't mention much about > > #pragma, but the Intel cc960 compiler manual says they work just fine. You either need the Tornado GNU Tools manual (GCC Section) or alternatively any distribution of GCC will have the documentation with it somewhere. --------------------------- End of New-News digest ********************** From vxwexplo-errs@csg.lbl.gov Mon Sep 4 04:03:08 2000 From: Vxworks Exploder Date: Mon Sep 4 04:03:10 PDT 2000 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Mon Sep 4 04:03:04 PDT 2000 Subject: PCI card in a mtx603 board Subject: opcode 1111 Subject: Re: opcode 1111 Subject: STLport 4.1 Beta 2 is available Subject: Re: romStart Question: can I store stuff in USR_RESERVED_MEM Subject: Re: opcode 1111 Subject: Re: Problems with PCI ISR Subject: Re: Vxsim File I/O with NT Subject: Re: building vxWorks using Tornado II Subject: Re: building vxWorks using Tornado II Subject: Re: Increasing SIMNT memory Subject: 82543GC Gigabit Ethernet MAC ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: PCI card in a mtx603 board Date: Thu, 24 Aug 2000 17:03:48 +0100 From: "Kai Mattern" Organization: Heidelberger Druckmaschinen AG Message-ID: <8o3de1$4p4$1@news.hdpp.de> I included this text as a .txt data because last time i posted my format went crasy please help if you can Hi all, im at my diploma and realy need help. I have to write a driver under VxWorks for a Printer. But the biggest problem i have know for about 2 and a half week is to get the interface card running. Thats what i did up to know - ---------------------------------------------------------------------------- - ------------------------ The configuration header of the card. I need 4 Base Adresses mem maped - Pci Base Adress 0 for Memory Mapped Configuration Register 64kbyte io maped - Pci Base Adress 1 for I/O Memory Mapped Configuration Register 64kbyte mem maped - Pci Base Adress 0 for Local Adress Space 0 256MByte mem maped - Pci Base Adress 1 for Local Adress Space 1 64kbyte these are located in the header from 10h down to 1Ch So i saw i cant use autoconfig because my BSP dont have it and pciDevConfig is only for devices with one memory bar. So i did it the hard way and added the following to sysHwInit function in sysLib.c if ((pciFindDevice(0x112B, 0x6304, 0, & pciBusNo, & pciDevNo, & pciFuncNo) != ERROR)) { pciConfigOutLong(pciBusNo, pciDevNo, pciFuncNo, 0x10, 0x01000000 + 0x10010000); (1) pciConfigOutLong(pciBusNo, pciDevNo, pciFuncNo, 0x14, PCI_IO_ADRS + BASE_ADR_1); (2) pciConfigOutLong(pciBusNo, pciDevNo, pciFuncNo, 0x18, 0x01000000 + 0x0); (3) pciConfigOutLong(pciBusNo, pciDevNo, pciFuncNo, 0x1C, 0x01000000 + 0x10000000); (4) pciConfigOutWord(pciBusNo, pciDevNo, pciFuncNo, 0x04, 0x0007 ); (5) } (2) The i/o mapped i put to PCI_IO_ADRS plus BASE_ADR_1 (its assigned from me in mv2600.h as 0xa0000). The PCI_IO_ADRS is also assigned in mv2600.h and its value is 0x01000000 /* base of PCI I/O address */ (3) PCI_MEM_ADRS 0x01000000 /* base of PCI MEM address */ plus 0x0 offset range is 256MByte (4) this is 256MByte above (3) and is 64kbyte long (1) this is 64kbyte above (4) and is 64kbyte long (5) is only command register..i took the falue out of the docu i got about the board. Next i took a look to the memory and found that standard is only 16MB space memory for PCI. #define CPU_PCI_MEM_SIZE 0x10000000 /* 16 meg */ /*you can see i changed it to 256MB*/ also i searched the memory but didnt found anything above 0x0C1000000 so there is enoght space for my 256MB (Raven and Falcon ship starting at FEF80000 but thats high enough. What i know did (because i thought thats all i need to do is writing to the card. I have a led on the card who tells me if reset register is low or high. The Reset register is located in (4) with an Offset of 0xC0 So i wrote the following code to toggle the led on and off sysPciOutLong(0xC1000000 + 0x10000000 + 0xC0, run); (run is 0 or 1..i tried both..) 0xC1000000 is the starting adress of mem space seen from CPU plus the offset from the Register in mem space plus offset inside the register. Here is the memory mapping from raven chipset MPC Address Range PCI Address Range Definition - ----------------- ----------------- ----------------- 80000000 BF7FFFFF 00000000 3F7FFFFF ISA/PCI I/O space C0000000 FCFFFFFF 00000000 3CFFFFFF ISA/PCI Mem space w/MPIC FD000000 FDFFFFFF 00000000 00FFFFFF ISA/PCI Mem space FE000000 FE7FFFFF 00000000 007FFFFF ISA/PCI I/O space So if i got this right....if i write to adress 81000000 im on the PCI Bus at Adress 01000000 and so on. But im unable to toggel the led. Nor am i able to write to the memory. Did i something wrong ? Do i have to do something else ? What do i need to do to toggle the led ?` Please help if you can thanks in advance Kai Mattern Kai_Mattern@web.de begin 666 done work.txt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rganization: Deutsches Elektronen-Synchrotron (DESY) Message-ID: <8onvs7$7e1$1@claire.desy.de> Hello, Could anyone clarify a question what opcode 1111 is? The diagnostic message is: Unimplemented opcode 1111 Program Counter: 0x00540012 Status Register: 0x3008 ... Thank you in advance, Alexei. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: opcode 1111 Date: Sun, 03 Sep 2000 21:53:30 -0400 From: Joe Durusau Message-ID: <39B3009A.600F018@bellsouth.net> References: <8onvs7$7e1$1@claire.desy.de> The error message it telling you that an error in your code caused the computer to attempt to execture an opcode 1111, but since no such code exists, the processor raised an exception of some sort. These problems are usually caused by stack overflows or exceeding the size of arrays. (At least these are the most popular causes.) Unfortunately, there are many more possibilities, far too many to list. Speaking only for myself, Joe Durusau Dmitrovskiy Alexei wrote: > > Hello, > > Could anyone clarify a question what opcode 1111 is? > The diagnostic message is: > > Unimplemented opcode 1111 > Program Counter: 0x00540012 > Status Register: 0x3008 > ... > > Thank you in advance, > Alexei. --------------------------- Newsgroups: comp.unix.aix,gnu.g++.help,ibm.software.vacpp.compiler,comp.os.vxworks Subject: STLport 4.1 Beta 2 is available Date: Mon, 04 Sep 2000 02:57:07 GMT From: boris_fomitchev@my-deja.com Organization: Deja.com - Before you buy. Message-ID: <8ov31u$l35$1@nnrp1.deja.com> Hi there, STLport 4.1 Beta #2 is available at www.stlport.org. Please find corresponding ChangeLog portion below : - -------------------------------------------- * _algobase.h : fixed __STL_DECLARE_COPY_TRIVIAL definition (thanks to Charles Burfoot for the report) * _algobase.h : added copy_backward specializations for builtin types for compilers w/o partial spec. * _char_traits.h : fixed zero-length check for copy (thanks to Michael Tsirkin) * msvc_warnings_off.h : fixed "and" typo (thanks to Charles Burfoot for the report) * num_put_float.cpp : fixed rounding for "g" format (thanks to Bernd Mohr for the report) * Initial port of iostreams to xlC 5.0 added (thanks to Jack Andrews) * stl_sunpro.h : a fix for Forte config (Thanks to Alex Vanic for the report) * old_hp/iterator.h : fixed iterator_category()/etc. import * _threads.h : __stl_atomic_t used instead of "unsigned long" (thanks to Alex Vanic for the report) * using/iomanip : old-streams path used for native iomanip.h (thanks to Michael Tsirkin for the report). * _bitset.h , _bitset.c : direct STL_THROW replaced with __stl_throw_xxx calls (thanks to Michael Tsirkin for the report) * range_errors.h/cpp : added functions to support throwing of overflow_error and invalid_argument * _threads.h : enganced HP-specific guards for HP-UX 11 (thanks to Matthew Collins for the report) * _string.h : compare() bugfix (thanks to Anton Sergeev) * moneypunct_byname_w.cpp : rolled back wrong bugfix (thanks to Ken) * _vector.h, _deque.h, _list.h : nonstandard versions of push_back ()/push_front() moved under __STL_NO_ANACHRONISMS guard (thanks to Ed Brey) * _tree.c, _list.c, _slist.c, _debug.c : compile-time improvements * _tree.h/c, _list.h/c, _slist.h/c, _vector.h/c, _string.h/c : using directive for data members changed to explicit "this->". * _complex.h : signatures of some methods changed to take value_type instead of const value_type&, for better conformance (thanks to Petr Ovchenkov) * Bunch of fixes to compile with gcc on HP-UX 10, 11 (thanks to Petr Ovchenkov) * Code duplication for __STL_INLINE_MEMBER_TEMPLATES removed -- corresponding methods are always inlined * stl_sunpro.h : fix for "-compat" mode (thanks to Vijay Ramachandran) * pthread.h : new wrapper added * Added custom prolog/epilog for HP aCC to select kthreads on HP-UX 11 (Thanks to Michael Tsirkin) * : fixed throw specification for bad_alloc (has to be inherited) * Macro __STL_NATIVE_CPP_RUNTIME_INCLUDE_PATH introduced to specify separate path for , , . (thanks to Bryan Byrnes) Sincerely, - -Boris. Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: romStart Question: can I store stuff in USR_RESERVED_MEM Date: Mon, 04 Sep 2000 09:17:41 +0530 From: Vijay Kumar Peshkar Organization: Ye 'Ol Disorganized NNTPCache groupie Message-ID: <39B31B5D.13220071@cisco.com> References: <39B131AB.CF1C60A4@gv.net> Reply-To: vpeshkar@cisco.com This is a multi-part message in MIME format. - --------------CAC42ACCDE78AB803BC97282 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Jim, Yes it works. I did the same in one of my earlier project to store some flags related to Built-in Self Test which I was performing immediately after boot. Vijay "Michael R. Kesti" wrote: > Jim wrote: > > >I have mad some modifications to the VxWorks boot code in romInit.s ... > >these modifications collect a small number of statistics on a few devices > >on my board. I now would like to store these statistics in a safe location > >... currently I have decided to store these statistics in at the top of > >RAM in the area defined by the USR_RESERVED_MEM ... I do this because it > >is not zero'd by romStart.c at startup. > > > >Should it be safe to do this? > > Yes, that should work very well. I've made similar use of USR_RESERVED_MEM. > > -- > ======================================================================== > Michael Kesti | "And like, one and one don't make > | two, one and one make one." > mkesti@gv.net | - The Who, Bargain - --------------CAC42ACCDE78AB803BC97282 Content-Type: text/x-vcard; charset=us-ascii; name="vpeshkar.vcf" Content-Transfer-Encoding: 7bit Content-Description: Card for Vijay Kumar Peshkar Content-Disposition: attachment; filename="vpeshkar.vcf" begin:vcard n:Peshkar;Vijay x-mozilla-html:FALSE org:Wipro Technologies,K2, Bangalore;Wishing you blue skies, a couple of daisies, some fresh air and a smile! adr:;;;;;; version:2.1 email;internet:vijay.peshkar@wipro.com title:Senior Software Engineer fn:Vijay Kumar Peshkar end:vcard - --------------CAC42ACCDE78AB803BC97282-- --------------------------- Newsgroups: comp.os.vxworks Subject: Re: opcode 1111 Date: Mon, 4 Sep 2000 07:58:17 +0200 From: "Martin Raabe" Organization: T-Online Message-ID: <8ovdu4$em1$15$1@news.t-online.com> References: <8onvs7$7e1$1@claire.desy.de> Hello Alexei, what CPU are you using? 68K. Here are two hooks to implement own Opcodemacros, which are starting with the opcode highbyte 0xAx or 0xFx. If there is an opcode found, where no opcode macro is assigned to, there is a "Line A" or "Line F" trape or exception raised by the 68K CPU. Maybe you did hit this. If you did not intend to assign a new opcodemacro, like having a SoftWare Floatingpoint unit, then usually wrong pointers or stacks can cause this. HIH (Hope It Helped) Ciao - -- Martin Raabe > Dmitrovskiy Alexei wrote: > > > > Hello, > > > > Could anyone clarify a question what opcode 1111 is? > > The diagnostic message is: > > > > Unimplemented opcode 1111 > > Program Counter: 0x00540012 > > Status Register: 0x3008 > > ... > > > > Thank you in advance, > > Alexei. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Problems with PCI ISR Date: Mon, 04 Sep 2000 06:32:59 GMT From: Thomas Rahn Organization: MacroSystem Message-ID: <8ovfmr$1ps$1@nnrp1.deja.com> References: <8op2eq$t6l$1@slb6.atl.mindspring.net> Hi Terry, from your code snippets I get the impression you do not enable the interrupts (something like sysIntEnablePic(IRQ_LVL)) Maybe this helps Bye Thomas Rahn Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Vxsim File I/O with NT Date: Mon, 04 Sep 2000 09:55:27 +0200 From: Bart Gouw Organization: Medtronic Inc. Message-ID: <39B3556F.330769F@vitatron.com> References: <8o6kgi$fvu$1@nnrp1.deja.com> <8onsce$7c$1@nnrp1.deja.com> Reply-To: vitatron@gwgouw.myweb.nl I ran into the same problem a few months ago. I discovered that all "host:" I/O actions failed. This is an VxWorks 5.4 bug. You must use "/tgtsvr" instead, as a work-around (this was told to me by the VxWorks support desk). So an 'C' fopen command may look like this: fopen ("/tgtsvr/test.txt", "a");. The files are written in the directory defined in the VxSim Target Server command line arguments for example "-V -B wdbpipe -R D:/VxWorks/FileIO -RW". All files are written in "D:/VxWorks/FileIO" (or sub-directories) Bart Gouw hruesing@my-deja.com wrote: > In article <8o6kgi$fvu$1@nnrp1.deja.com>, > jim.themis@genband.com wrote: > > According to WSR there is at the moment only a workaround available > using the target server file system: > > > IDE: Tornado 2.0 > > Component: tornado-vxsim > > RTOS: VxWorks Version: 5.4 > > Product Name: Tornado Release#: 2.0 Release Status: FCS > > Products Affected: Tornado > > Host: Intel x86/Pentium HostOs: HostOSRev: 4.0 sp3 > > Arch Family: SIM NT Arch CPU: SIM NT BSP: Ver: > > > > DESCRIPTION > > > > The NT passFs device "host:" created by defining INCLUDE_NTPASSFS > > in the SIM NT configuration, is inaccessible from both windsh and > > target > > shell. > > > > WORKAROUND > > > > The target server file system can be set up so that the target can > > access the NT host file system. Select "Target Server File System" > > in the "Target Server Properties" window of the target server > > configurtion tool. Specify a Windows host file system. After > launching > > the target server, this file system can be accessed through the > > device "/tgtsvr" on the target either directly from the shell or > > programatically. > > > > > I reviewed the archives but I could find anyone with a similar > problem. > > I am running VxSim on an NT box. I am trying to do file I/O via > fopen, > > fclose, fputs, etc. I am able to read files, but I am unable to > create > > and write files. I have tried adding "host:" to the beginning of my > > path. I have enabled tsfs with r/w. I get no error when I perform > > an fopen. However, when I execute an fputs(), I get the following > > error: > > ntPassFsLib: unknown ioctl = 0x14 > > > > I looked at errno and the value showed Unknown Request. > > > > Can anybody explain their set-up for reading and WRITE files under > > VxSim. > > > > Thanks, > > Jim Themis > > > > Sent via Deja.com http://www.deja.com/ > > Before you buy. > > > > Sent via Deja.com http://www.deja.com/ > Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: building vxWorks using Tornado II Date: Thu, 24 Aug 2000 22:42:35 +0200 From: "Martin Raabe" Organization: T-Online Message-ID: <8o41hu$7kk$15$1@news.t-online.com> References: <8o3s7c$3qm$1@overload.lbl.gov> Hello Tom, there is a description of the component writtten in CDL (Component Decription Language). Here are the headers, libraries, compiler switches and other stuff defined. HIH Martin Raabe --------------------------- Newsgroups: comp.os.vxworks Subject: Re: building vxWorks using Tornado II Date: 24 Aug 2000 20:48:42 GMT From: johan@borksoft.xs4all.nl Organization: Borksoft Message-ID: <8o41na$a5s$1@news1.xs4all.nl> References: <8o3s7c$3qm$1@overload.lbl.gov> Tom Miller wrote: : 1) How do I know what library these components are coming from? say: : \target\lib\objPENTIUMgnuvx vs \target\lib\obj80386gunvx? That depends on the toolchain you choose. Look in de buildspec properties window, you can find your architecture there, or otherwise look in the Makefile. : 2) How do I know which "C" source files are associated with a given : component that I'm including? See the properties of the included component, it is the item CONFIGLET. : 3) If I want to modify a given components source file to make it fit my : particular hardware, how do I specify to use my version of the file instead : of the one from the library? If you make sure that the code is part of your project it will always be included first. Or (if your own code is in a library) add your library to the LIBS macro Groeten, Johan - -- o o o o o o o . . . ___________________________________ o _____ || Johan Borkhuis | .][__n_n_|DD[ ====_____ | johan@borksoft.xs4all.nl | >(________|__|_[_________]_|________________________________| _/oo OOOOO oo` ooo ooo 'o!o!o o!o!o` === VxWorks FAQ: http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html === --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Increasing SIMNT memory Date: 24 Aug 2000 20:50:29 GMT From: johan@borksoft.xs4all.nl Organization: Borksoft Message-ID: <8o41ql$a5s$2@news1.xs4all.nl> References: <2vap5.1225$R2.107383306@nr1.ottawa.istar.net> Said Abdosh wrote: : D:\TORNADO\target\config\simpc>vxworks /r 12345 <- this in KB. But it did : not increased the memory. The size is in bytes, not in kBytes. The manual is wrong here. Groeten, Johan - -- o o o o o o o . . . ___________________________________ o _____ || Johan Borkhuis | .][__n_n_|DD[ ====_____ | johan@borksoft.xs4all.nl | >(________|__|_[_________]_|________________________________| _/oo OOOOO oo` ooo ooo 'o!o!o o!o!o` === VxWorks FAQ: http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html === --------------------------- Newsgroups: comp.os.vxworks Subject: 82543GC Gigabit Ethernet MAC Date: Thu, 24 Aug 2000 20:50:30 GMT From: eliasson@my-deja.com Organization: Deja.com - Before you buy. Message-ID: <8o41qf$h33$1@nnrp1.deja.com> Hello, Is anyone developing a driver for the 82543GC Gigabit Ethernet MAC for VxWorks? Thanx in advance. Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- End of New-News digest ********************** From vxwexplo-errs@csg.lbl.gov Mon Sep 4 04:26:25 2000 From: "Ramachandramurthy, Bharat (Bharat)** CTR **" Date: Mon Sep 4 04:26:27 PDT 2000 Subject: Enquiry Can this question be answered:- 1. Does vxWorks support opening multiple instances of the same file. if yes then is there any provisions for LOCKING the file. If LOCKING is not supported then how do we prevent the file from being used by the other users when one user is manipulating it. if not does that mean no other user can operate on the file until and unless the current user finishes. If so could u explain how this is done . Can mail me back to bharathramchandra@yahoo.com/br41@lucent.com regards bharath From vxwexplo-errs@csg.lbl.gov Tue Sep 5 04:03:07 2000 From: Vxworks Exploder Date: Tue Sep 5 04:03:09 PDT 2000 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Tue Sep 5 04:03:04 PDT 2000 Subject: sizeof() is type "long" on m68k but "int" on ppc604 Subject: Re: opcode 1111 Subject: Enquiry Subject: benchmark data sheets for various processor families Subject: SIMD instructions on Pentium Subject: MPC860P + MII Subject: Re: VxWorks and dual processors Subject: Re: TrueFFS Subject: Re: Enquiry Subject: Driver Questions Subject: Re: benchmark data sheets for various processor families Subject: no symbols available after download by kernel Subject: PCMCIA (ATA card) and MBX821 Subject: CodeTest or (ScopeProfile+MemScope) Subject: Re: Green Hills Software ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: sizeof() is type "long" on m68k but "int" on ppc604 Date: Mon, 04 Sep 2000 13:15:50 MET From: mathar@mpia-hd.mpg.de (Richard Mathar) Organization: Max-Planck-Institut fuer Astronomie Keywords: C Message-ID: <2000Sep4.131550@sun18> Reply-To: mathar@mpia-hd.mpg.de I am trying to compile the C-line printf("%d\n",sizeof(double)) ; and find that sizeof() is assumed to be of type "long" if I use the cc68k, but of type "int" (what I'd have expected) for the ccppc, both for the ./host/parisc-hpux9/lib/gcc-lib/powerpc-wrs-vxworks/cygnus-2.7.2-960126 and ./host/parisc-hpux9/lib/gcc-lib/m68k-wrs-vxworks/cygnus-2.7.2-960126 with VxWorks 5.3 . vxTyptesBase.h has the lines #ifdef __GNUC_TYPEOF_FEATURE_BROKEN_USE_DEFAULT_UNTIL_FIXED__ #undef _TYPE_size_t #define _TYPE_size_t typedef __typeof(sizeof(0)) size_t #else #ifndef _TYPE_size_t #define _TYPE_size_t typedef unsigned int size_t #endif #endif Where is the actual decision made that sizeof() needs "%ld" with printf() on MC68040 but "%d" on PPC604? Richard J Mathar, mathar@mpia-hd.mpg.de --------------------------- Newsgroups: comp.os.vxworks Subject: Re: opcode 1111 Date: Mon, 4 Sep 2000 12:51:35 +0100 From: "Dave Korn" Organization: (Posted via) U-NET Internet Ltd. Message-ID: References: <8onvs7$7e1$1@claire.desy.de> Dmitrovskiy Alexei wrote in message <8onvs7$7e1$1@claire.desy.de>... >Hello, > >Could anyone clarify a question what opcode 1111 is? >The diagnostic message is: > >Unimplemented opcode 1111 >Program Counter: 0x00540012 >Status Register: 0x3008 >... This is an M68k series exception, caused when the CPU attempts to execute an opcode in the 0xF??? range. These opcodes are reserved for the floating point or MMU coprocessor and so are an error for the CPU. I can think of three possible reasons why you might see this exception: 1) Your CPU has no floating point coprocessor, but you forgot to give the compiler the -msoft-float flag, so it included opcodes for the FPU coprocessor in your program. 2) Your CPU has no MMU and your program tried to execute a command for the MMU, which also behaves like a coprocessor. 3) Your call stack or program counter, or perhaps a function pointer, became corrupted and your program jumped into random memory space. hth, DaveK - -- They laughed at Galileo. They laughed at Copernicus. They laughed at Columbus. But remember, they also laughed at Bozo the Clown. --------------------------- Newsgroups: comp.os.vxworks Subject: Enquiry Date: Mon, 4 Sep 2000 16:56:15 +0530 From: "Ramachandramurthy, Bharat (Bharat)** CTR **" Organization: Lawrence Berkeley National Laboratory Message-ID: <8p03nn$9hu$1@overload.lbl.gov> Can this question be answered:- 1. Does vxWorks support opening multiple instances of the same file. if yes then is there any provisions for LOCKING the file. If LOCKING is not supported then how do we prevent the file from being used by the other users when one user is manipulating it. if not does that mean no other user can operate on the file until and unless the current user finishes. If so could u explain how this is done . Can mail me back to bharathramchandra@yahoo.com/br41@lucent.com regards bharath --------------------------- Newsgroups: comp.os.vxworks Subject: benchmark data sheets for various processor families Date: Mon, 04 Sep 2000 12:32:27 GMT From: ehudfl@my-deja.com Organization: Deja.com - Before you buy. Message-ID: <8p04or$ncd$1@nnrp1.deja.com> Hello all In WindRiver web site Mark Armitage, from Wind River Systems, published an article at Spring 1998 whos name is: Rehosting Code Written for UNIX on VxWorks at the address: http://www.wrs.com/windword/html/8.rehosting.html In that article he wrote that in VxWorks, switching between tasks takes a few microseconds, and he continue by writing that Wind River publishes benchmark data sheets for various processor families. I tried to search in the site but I havn't found those benchmark data sheets. Where can I find those benchmark data sheets (are they available somewhere in the net?) Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: SIMD instructions on Pentium Date: Mon, 04 Sep 2000 13:20:54 GMT From: lourens@mecalc.co.za (Lourens Geldenhuys) Organization: Mecalc (Pty) Ltd Message-ID: <39b3a11a.631125861@modem.mecalc> Reply-To: lourens@mecalc.co.za Does anyone know if the version of the GNU compiler/assembler that ships with Tornado 2.0 supports the SIMD (Single Instruction Multiple Data) instructions for the Pentium processor? Regards, Lourens ================================================================ Lourens Geldenhuys Tel: +27 (0)12 665 1480 Project Engineer Fax: +27 (0)12 665 1495 Mecalc (Pty) Ltd e-mail: lourens@mecalc.co.za 86 Oak Avenue, Highveld Technopark, Centurion, South Africa ================================================================ --------------------------- Newsgroups: comp.os.vxworks,comp.arch.embedded,comp.sys.powerpc.tech Subject: MPC860P + MII Date: Mon, 4 Sep 2000 16:17:05 +0100 From: "Jeremy Pye" Organization: Customer of Energis Squared Message-ID: <7ce0p8.39l.ln@gate.endusis.com> Can anyone help I am design a 10/100Mb ethernet interface on a MPC860P using the MII/FEC interface. Is there a standard way of connecting an AMD NetPHY-1LP ? Are there vxWorks drivers available ? cjp@endusis.com --------------------------- Newsgroups: comp.os.vxworks Subject: Re: VxWorks and dual processors Date: Mon, 04 Sep 2000 17:17:45 +0200 From: Gilbert Cabillic Organization: Irisa - Campus de Beaulieu - RENNES Message-ID: <39B3BD19.2FFCD025@irisa.fr> References: <39AEA111.938839B4@aps.anl.gov> Andrew Johnson wrote: > > Den wrote: > > > > 1. Does VxWorks support dual processors? Do any other embedded OS's > > support dual or multiple processors? > > Yes and no. I'm not aware of any RTOS that supports true Symmetric > Multi-Processing in the way that Linux and Solaris do, Andrew and Den, I believe that the RTEMS system can be run on a dual pentium. (http://www.rtems.com/). But I don't know if it's true symmetric (only one operating system on the processors). I have never used it. > There are also vxWorks Board Support Packages for some dual-CPU processor > boards which pass I/O operations off to the second CPU to perform, but I > don't think they could allow you to run normal vxWorks program code on the > other CPU. Yes. We have a dual pentium vxworks target. There is one vxworks on each processor and we have vxMp library to share memory space. Every component of the kernel can be activated on each processor. Gilbert. // http://www.irisa.fr/solidor/work/scratchy.html --------------------------- Newsgroups: comp.os.vxworks Subject: Re: TrueFFS Date: Mon, 04 Sep 2000 16:08:19 GMT From: "Mark Bentley" Message-ID: References: <39a4022a@news.barak.net.il> Yaron, Yeah, Wind River support is real unsupportive about TFFS. I have been waiting for 2 weeks on a problem on it... A couple things: 1) You have to write a lot of the stuff, like the code to access the actual flash. So, even if they say it is supported, you may no be betting a simple "plug and play" sort of driver. 2) We have never found anyone at WindRiver who seemed to know how it works and could really answer our questions. I have been told flat out by Wind River (un)support that it will not run in the configuration that we have it running in (2 16 bit flash parts side by side for a 32 bit wide bus). 3) Just so you know it is a b---- to debug! They do things like any error code that you try to return, will be changed to the ERROR define. Also, there are cases where it will bring down the whole system if you upset it rather than return an error code at all (like opening an already open file). 4) Just a little help if you do use it... The format command, might, or it might not... If you want to format do a full erase of the entire area. If the TFS becomes corrupted, the format will often succeed, but not actually write anything new and not actually fix the TFFS! So, keep that in mind. It is a eral pain when you are trying to get it all going... Hope that clarifies a bit... Mark Y.H wrote in message <39a4022a@news.barak.net.il>... >Does anybody knows if est860mdprpo board support TrueFFS. WindRiver support >, didn't come up with a clear answer . >Thanks > >Yaron > > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Enquiry Date: Mon, 4 Sep 2000 18:40:33 +0100 From: "Dave Korn" Organization: (Posted via) U-NET Internet Ltd. Message-ID: References: <8p03nn$9hu$1@overload.lbl.gov> Ramachandramurthy, Bharat (Bharat)** CTR ** wrote in message <8p03nn$9hu$1@overload.lbl.gov>... >Can this question be answered:- >1. Does vxWorks support opening multiple instances of the same file. Yes. >if yes > then is there any provisions for LOCKING the file. No (AFAIK). > If LOCKING is not >supported then how do we prevent the file from being used by the >other users when one user is manipulating it. Use a semaphore of your own to control access to it. hth, DaveK - -- They laughed at Galileo. They laughed at Copernicus. They laughed at Columbus. But remember, they also laughed at Bozo the Clown. --------------------------- Newsgroups: comp.os.vxworks Subject: Driver Questions Date: Mon, 4 Sep 2000 14:43:41 -0400 From: "Tim Stanka" Organization: Eastman Kodak Company Message-ID: <8p0pl3$g9j$1@news.news.kodak.com> I just designed my first VXWORKS driver. My designed was OKed but I have not coded it up yet. I have a couple of questions: 1) How could I implement marking a device as "open for reading" or "open for writing" so a device could only be open for reading with one fd. I wanted to use a section of the device descriptor to store information about how a device was opened. (RD, WR, RDWR etc). This was easy to do in my xxOpen() function but when I started to design xxClose() I realized I needed an association between an fd and the open mode in order to correctly unmark the device for RD, WR etc. I want to look at the flags in xxOpen() and mark the device accordingly. ie, if the flag is O_RDONLY I would mark the device for reading. In xxClose() I want to be able to "unmark" the device but xxOpen and xxClose do not have access to the fd table. Also, an fd is not fed to these functions. Without the fd to open mode association I do not know how to properly "unmark" the device 2) The VXWORKS device descriptor (DEV_HDR) contains three fields: name, driver number and dl. The dl field is a type DL_NODE. This is a self-referencing structure with two fields: typedef struct dl_node { dl_node *prev; dl_node * next; } DL_NODE; I'm not sure why they did this. I would have a "void *data;" data field either in DEV_HDR or DL_NODE so a user could connect a custom data structure to the device descriptor. Has anyone used the device descriptor to store device unique information? Thanks, Tim Stanka --------------------------- Newsgroups: comp.os.vxworks Subject: Re: benchmark data sheets for various processor families Date: Mon, 04 Sep 2000 22:15:29 GMT From: john_94501@my-deja.com Organization: Deja.com - Before you buy. Message-ID: <8p16tr$qdv$1@nnrp1.deja.com> References: <8p04or$ncd$1@nnrp1.deja.com> > Where can I find those benchmark data sheets (are they available > somewhere in the net?) No, but they can be obtained from your local sales representative. You just need to ask... John... Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: no symbols available after download by kernel Date: Tue, 5 Sep 2000 08:14:33 +0200 From: "Markus MItterer" Organization: KPNQwest customer news service Message-ID: <_70t5.3$237.385@nreader1.kpnqwest.net> Hello VxWorkers, My problem is the following: I`m working on a PPC (mv2303) and my booting sequence during development was to load the Kernel automaticly (per FTP) and to download all additional modules with the Tornado2 IDE. SymbolTables were getting successfully synchronized. Now I extended my kernel to download a module at bootup with option LOAD_GLOBAL_SYMBOLS. When I use the telnet shell everything is working properly but with my host shell under Tornaod2 I didn`t see the added symbols (only the system-symbols). The synchronization task (tSymSync) is still running. Any ideas??? regards - -- Markus Mitterer Software-Development, STK mailto: markus_mitterer@sbu1.storkgroup.com --------------------------- Newsgroups: comp.os.vxworks Subject: PCMCIA (ATA card) and MBX821 Date: Tue, 05 Sep 2000 10:15:07 +0200 From: Emmanuel Organization: SEPRO Robotique Message-ID: <39B4AB8B.C37C6325@sepro-robotique.com> Hi, We are working to connect a Compact-Flash card (ATA controller and PCMCIA bus) with an evaluation board called MBX821 (based on MPC 821). The BSP seems to be buggy (bad SIO support, PCMCIA not tested, etc...). Does anyone own a debugged BSP for such board with PCMCIA support ???? Thanks a lot. Emmanuel. --------------------------- Newsgroups: comp.os.vxworks Subject: CodeTest or (ScopeProfile+MemScope) Date: Tue, 05 Sep 2000 16:41:56 +0800 From: Tony Cheung Organization: IMS Netvigator Message-ID: <39B4B1D3.36174246@asiayeah.com> Reply-To: ytcheung@technologist.com Hi, I am evaluating VxWorks for one of my networking projects. I come across the following tools, 1) CodeTest/Coverage + CodeTest/Memory 2) PerformacePak (ScopeProfile+MemScope) Does anyone has any real-life experiences with the above products? They sound very smilar to me, is option 1 or 2 better? Thanks a lot. Tony Cheung --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Green Hills Software Date: Tue, 05 Sep 2000 10:27:46 GMT From: "M Campbell" Organization: @Home Network Message-ID: References: <8ogv5q$rpv$1@nnrp1.deja.com> <39b0638f$2$zpzbeena$mr2ice@news.visi.net> We also used Greenhills with vxWorks on a large program w/powerPCs. We found that Greenhills support for for thier Fortran compiler is starting to be non-existant. Their fortran debugger is practically useless beyond about a ~50K source line limit. When porting ancient fortran code be aware that Greenhills assumes that INTEGER*1 varibles are signed, where older machines (read Gould) make no such distinction. On the Ada side, we had very few problems with developing there. After a certian point, the debugger also became useless. Also found that Greenhills method for implementing multitasking a bit awkward. Would love to see a competior come into the picture on the Fortran side, but don't realistically don't see the enough demand for a company to make the effort to reach the maturity level of the Greenhills product & be competitive. As far as I can tell, GNU compilers / debuggers (fortran, ada) won't work for vxWorks. If anyone has actually used them in a vxWorks environment, i'd love to hear about it, as GNU tools "should" integrate seamlessly into the vxWorks way of doing things. Like Peter, I dislike their Multi environment & we used it only sparingly in the beginning to access the debugger. Once to debugger got unweildly (extremely long compile / link times, then pay the time penalty all over when starting the the debugger), we stopped using that too, resorting the calls to C routines to print statements & stop processing so we could dump memory. About the only part of the product we are currently using is their fortran compiler (we use the vxWorks "ldppc" linker to link our code). From a Greenhills perspective, the upgrade to Tornado II from Tornado I was no problem (esp. since we were only using the compiler). Believe those using the "other" Greeenhills tools will also have little problem. "Peter McMorran" wrote in message news:39b0638f$2$zpzbeena$mr2ice@news.visi.net... > In <8ogv5q$rpv$1@nnrp1.deja.com>, on 08/29/00 > at 06:25 PM, gpbeck@my-deja.com said: > > >Is anyone using the Green Hills software to develop/debug their > >VxWorks applications? > > We're using AdaMulti to port several Ada/Fortran/C Unix programs > to embedded PowerPC. Naturally, we have to make some adaptations > for VxWorks, but none related to the compiler system. > > We were tripped by one thing related to Fortran list-directed > I/O. Most Unix compilers allow strings in a list-directed input > file to be enclosed in double quotes, and we were in the habit of > using these. AdaMulti sticks to the letter of the Fortran > standard and requires single quotes. Since we have routines that > painstakingly write these quotes, we had to do a little work to > get this working. > > >If so, what is your opinion of the their software? Is it > >reliable/stabile? How is their support? > > We found a bug in the Ada runtime which they fixed in a few > weeks. They are now looking at a bug in the Fortran runtime which > produces a memory leak. Support seems good, but these fundamental > and easily reproduced bugs are surprising in a mature product. > Aside from these two bugs, our programs run correctly, and they > are over 100 K lines of code. So the compiler does seem to be > reliable. > > The debugger is pretty neat, letting you grab multiple tasks at > once. I personally dislike the AdaMulti GUI, using it only to > access the debugger, or to find out the code for an option. We > run the build program directly from make, creating the build > files by cloning from existing ones. With a master makefile in > the top directory and makefiles in each subdirectory, we can > update everything with a single make command. The builder keeps > track of all source dependencies, recompiling all affected > programs when a C header, or Fortran include file, or Ada package > spec is changed. > > Cheers, > Peter > > -- > ----------------------------------------------------------- > mcmorran@visi.net (Peter McMorran) > ----------------------------------------------------------- > --------------------------- End of New-News digest ********************** From vxwexplo-errs@csg.lbl.gov Wed Sep 6 04:03:08 2000 From: Vxworks Exploder Date: Wed Sep 6 04:03:11 PDT 2000 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Wed Sep 6 04:03:05 PDT 2000 Subject: re: TFFS on MV260x Subject: @@ TELECOM - VxWorks S/W Developers Needed - Ottawa Subject: use of cascaded macros Subject: Device doesn't show up as an interface Subject: Second serial port on MBX860 Subject: Re: use of cascaded macros Subject: Re: use of cascaded macros Subject: PCI and ISR Subject: Route table corruption Subject: Re: CodeTest or (ScopeProfile+MemScope) Subject: Re: Load Error "Relocation value does not fit in 24 bits" Subject: Re: Driver Questions Subject: Re: Problems with WindView under Windows 2000. Subject: gcc asm / ppc Subject: DHCP service testing Subject: Re: Problems with WindView under Windows 2000. Subject: Re: CodeTest or (ScopeProfile+MemScope) Subject: Re: TrueFFS Subject: Re: gcc asm / ppc Subject: Re: gcc asm / ppc Subject: Re: gcc asm / ppc Subject: Re: gcc asm / ppc Subject: ppc asm: loading 32 bit value into CTR reg Subject: Partitioning and Formatting Subject: Re: gcc asm / ppc Subject: Re: use of cascaded macros Subject: Re: ppc asm: loading 32 bit value into CTR reg Subject: scsi write corruption with large buffers Subject: Re: ppc asm: loading 32 bit value into CTR reg Subject: Re: ppc asm: loading 32 bit value into CTR reg Subject: Re: Spawning C++ task Subject: TELECOM - VxWorks S/W Developers Needed - Ottawa Subject: Re: Problems with PCI ISR Subject: Re: Green Hills Software Subject: Bus Bandwidth and CPU load for VxWorks RTOS and Network stack question. Subject: Santa Barbara Opportunities fo VxWorks Developers Subject: Re: ppc asm: loading 32 bit value into CTR reg Subject: rpclib - errno 0xd0003 Subject: Which IRQ for PCMCIA Card? Subject: Re: ppc asm: loading 32 bit value into CTR reg Subject: Re: rpclib - errno 0xd0003 Subject: dosFs2 chkdsk slow to run from bootrom Subject: Re: Second serial port on MBX860 Subject: Re: use of cascaded macros Subject: Re: Second serial port on MBX860 Subject: Re: Second serial port on MBX860 Subject: Re: How the target Shell handles the console device driver ? Subject: Vxworks wtih MPC8260ads Subject: Re: Device doesn't show up as an interface ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: re: TFFS on MV260x Date: Tue, 05 Sep 2000 11:11:07 GMT From: bob_42690@my-deja.com Organization: Deja.com - Before you buy. Message-ID: <8p2kc6$8du$1@nnrp1.deja.com> References: <8or5h7$t1n$1@overload.lbl.gov> I integrated a non-standard flash (nand type) flash with TFFS. Talk to WRS to see if they have driver source code that is similar to what your requirements are. They may let you adapt their code. When I did this a while back while TFFS was in beta (ugh), I found that the example code did not apear to closely match the structure indicated in the documentation. This situation may have improved. Good luck! In article <8or5h7$t1n$1@overload.lbl.gov>, "Tim Allen" wrote: > >>has anybody already integrated TFFS to on board flash of a MV230x, > >>MV240x, MV260x or MV270x ? > >>Could you share your code changes or help me to get on the road ? > >>I think vxWorks documentation is as simple as usually. > > > >Nobody answered :-(( > > Since I haven't noticed any responses, this should stop you feeling lonely, > but I don't think it will help! > > >Is there really nobody who has done this job ? > >Has anybody succeeded in integrating any other non-standard flash ? > > I have integrated TFFS with a customer's flash programming library, which > presented a non-standard interface. It wasn't as straightforward as the > documentation would suggest. As well as some inconsistent function/structure > definitions, I noticed that the Translation Layer was accessing the flash > hardware directly and not going through the MTD layer as I expected. > > I did get it going though. Unfortunately, I don't think I know enough about > the Mot boards/flash to help you specifically. > > Regards, > Tim. > -------------------- > Tim Allen Ltd > Real Time Embedded Software Consultancy > > Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks,comp.realtime Subject: @@ TELECOM - VxWorks S/W Developers Needed - Ottawa Date: Thu, 31 Aug 2000 16:31:52 -0400 From: "PROCOM" Organization: IGS - Information Gateway Services Message-ID: <8omf4l$htu$1@news.igs.net> Title: C++ / OOD VxWorks Software Developers Description: Our client, one of the world's largest telecommunications companies, currently requires 10 Junior to Senior Software Specialists to design, develop, implement and support Wireless products and applications. Minimum Requirements: All candidates should possess a minimum of 1 year experience developing real-time C/C++, OOD S/W on a VxWorks platform. Desired skills would include any previous Wireless development experience. Location: Ottawa, Ontario, Canada Type: Full-Time Placement Compensation: Open PROCOM: Established in 1978, Professional Computer Consultants Group Ltd. (Procom) is a national leader in the provision of Computer personnel on a contract and full-time basis. Our clients are comprised of the largest national and international corporations that utilize technical resources extensively across a wide range of disciplines. In the Financial Post (March 1999) Procom was ranked as the 6th largest professional Services Company in Canada. In November of 1999, Procom was named a Regional finalist in Canada' s 50 Best Managed Private companies. Our track record is proven with more than 180 consultants servicing Ottawa's high tech community and more than 1600 Procom consultants currently on assignment throughout. North America. For further information on this and other opportunities please visit our web site at www.procom.ca. Interested candidates are invited to forward their resumes or questions in confidence to: Derek Weber Technical Recruiter 300 March Road Kanata, Ontario K2K 2E2 phone: (613) 270-9339 ext 231 e-mail: derekw@procom.ca --------------------------- Newsgroups: comp.os.vxworks Subject: use of cascaded macros Date: Tue, 5 Sep 2000 13:36:18 +0200 From: Gunter Ritter Organization: Tenovis GmbH u. Co. KG Message-ID: <8p2lqt$3g9$1@proxy2.fe.internet.bosch.de> Hello everybody, I have the problem that I get errors "macro recursion too deep" when compiling a module that uses cascaded macros. The macros are defined as follows: #define K_MACRO_1 0 #define K_MACRO_2 K_MACRO_1+1 #define K_MACRO_3 K_MACRO_2+1 #define K_MACRO_4 K_MACRO_3+1 #define K_MACRO_5 K_MACRO_4+1 : : What can I do to compile my code. It is not possible to change the macro definition, becausethey are used in more than one project. I didn´t find a hint in the documentation. Thanks. Gunter --------------------------- Newsgroups: comp.os.vxworks Subject: Device doesn't show up as an interface Date: Tue, 05 Sep 2000 13:52:16 +0200 From: Mario Wolframm Organization: ONELINE AG Message-ID: <39B4DE70.2B5E87AD@oneline.de> Hi, I'm implementing an END which is supposed to be loaded several times to correspond to a dynamic amount of devices. Unfortunately it is not visible in ifShow(), not resolvable via ifunit() etc. It is based on the template that came with Tornado which had to be modified to a certain extend to actually work at all. I call muxDevLoad and get back a reasonable address. I call muxDevStart and get OK I see the driver registered via muxShow however: neither ipAttach nor ifAddrSet work with that new interface! Did I forget something important here? I tried via configNet.h as well but same result! Any help would bring me back to normal work :) ciao Mario --------------------------- Newsgroups: comp.os.vxworks Subject: Second serial port on MBX860 Date: Tue, 05 Sep 2000 12:35:14 GMT From: "Øyvind Holmeide" Organization: Enitel Internet Public Access Message-ID: <6M5t5.7434$Dxe.201282560@news.telia.no> I am trying to configure VxWorks for a second serial port (RS232) on the MBX860 board from Motorola. The BSP is only prepared for one serial port (the console). Thus, an update of sysSerial.c for e.g. SMC2 is needed. I have tried without success. Anyone been through this? regards Øyvind Holmeide --------------------------- Newsgroups: comp.os.vxworks Subject: Re: use of cascaded macros Date: Tue, 05 Sep 2000 08:29:33 -0500 From: Greg Willden Organization: Posted via Supernews, http://www.supernews.com Message-ID: <39B4F533.F56D7B9B@swri.edu> References: <8p2lqt$3g9$1@proxy2.fe.internet.bosch.de> This is a multi-part message in MIME format. - --------------15FD0D22356396C137D54613 Content-Type: text/plain; charset=iso-8859-1 Content-Transfer-Encoding: 8bit Gunter Ritter wrote: > Hello everybody, > > I have the problem that I get errors "macro recursion too deep" when > compiling a module that uses cascaded macros. > > The macros are defined as follows: > > #define K_MACRO_1 0 > #define K_MACRO_2 K_MACRO_1+1 > #define K_MACRO_3 K_MACRO_2+1 > #define K_MACRO_4 K_MACRO_3+1 > #define K_MACRO_5 K_MACRO_4+1 > : > : > What can I do to compile my code. It is not possible to change the macro > definition, becausethey are used in more than one project. > I didn´t find a hint in the documentation. I don't understand why you would use a macro to count in this manner. It appears to be totally unnecessary. The macros just clutter up the code when a simple value of 0, 1, 2, 3, 4 or 5 would do the job. But let me give you a general tip: These macros are just asking for trouble. Macros using +-*/ need to have parenthesis to protect them from order of operations issues. Like this: #define K_MACRO_2 (K_MACRO_1+1) #define K_MACRO_3 (K_MACRO_2+1) : : This probably won't fix the error that you are getting but it will help you in other situations. Cheers - --------------15FD0D22356396C137D54613 Content-Type: text/x-vcard; charset=us-ascii; name="gwillden.vcf" Content-Transfer-Encoding: 7bit Content-Description: Card for Greg Willden Content-Disposition: attachment; filename="gwillden.vcf" begin:vcard n:Willden;Greg tel;fax:(210) 522-2709 tel;work:(210) 522-6241 x-mozilla-html:FALSE org:Southwest Research Institute;www.swri.edu version:2.1 email;internet:gwillden@swri.edu title:Engineer adr;quoted-printable:;;6220 Culebra Rd=0D=0A=0D=0A;San Antonio, TX 78238;;; fn:Greg Willden end:vcard - --------------15FD0D22356396C137D54613-- --------------------------- Newsgroups: comp.os.vxworks Subject: Re: use of cascaded macros Date: Tue, 5 Sep 2000 16:29:36 +0200 From: Gunter Ritter Organization: Tenovis GmbH u. Co. KG Message-ID: <8p2vvs$hmq$1@proxy2.fe.internet.bosch.de> References: <8p2lqt$3g9$1@proxy2.fe.internet.bosch.de> <39B4F533.F56D7B9B@swri.edu> Thanks for your help. I will try this tomorrow. Well, why do we use macros to count in this manner. These macros define constants to access large stringtables. It is easier to access strings in this table by using a macro with names like "K_MY_TEXTSTRING" than using an numerical index. If a string will be deleted in this table, you just have to correkt the macros and you can access the strings after the deleted string with the same constant. Otherwise you have to learn the new numbers of the following strings ;-) Gunter --------------------------- Newsgroups: comp.os.vxworks Subject: PCI and ISR Date: Tue, 5 Sep 2000 16:33:06 +0100 From: "Kai Mattern" Organization: Heidelberger Druckmaschinen AG Message-ID: <8p305d$7sk$1@news.hdpp.de> Hi, I'm having some problems with my ISR. My card use INT A on PCI side and i figured out, that INT A is 0x0A with an added offset of 0x10 for external Interrupt. So i assigend my ISR as follow intConnect(0x19, (VOIDFUNCPTR)clc_int_service_routine, NULL); intEnable(0x19); INUM_TO_IVEC just returns the value so i wrote the Interrupt number direct. I then got no Interrupt at all. So i did a litte try and error and with intConnect(0x10, (VOIDFUNCPTR)clc_int_service_routine, NULL); intEnable(0x10); i got my interrupts....My problem is, i cant disable the interrupt over the interrupt mask register on my PCI Card nor can i disable it over the 9080 chip on the PCI Card. So i think there is something totaly wrong. What i want is...connect the INT A from PCI device to my interrupt service rountine i wrote. how can i figure out the right interrupt number to connect to ? Or did i something else wrong ? is there something more to do to connect a ISR to the interrupt ? thanks for reading btw. i have a M603 board. If more information is needed to solve my problem e-mail me please. Kai Kai_Mattern@web.de --------------------------- Newsgroups: comp.os.vxworks Subject: Route table corruption Date: 5 Sep 2000 16:45:21 +0200 From: jochen@siemens.com.br Organization: Mailgate.ORG Server - http://www.Mailgate.ORG Message-ID: <39B50627.AFB0EBEF@siemens.com.br> Hello vxworkers, When a PPP connection is disrupted, it seems that in some occasions the interface name corresponding to that PPP connection in the route table gets garbage (corrupted). For what I know, there is a known bug concerning this problem when several calls to slipDelete are made. Is it the same for pppDelete? Has anybody experienced this problem?Is there a workaround for this? Thanks, Jochen - -- - ---------------------------------------------------------- Jochen Mielke de Lima - ICN TR ES - Software Development Siemens Information and Communication Networks - ---------------------------------------------------------- - -- Posted from [200.230.253.34] via Mailgate.ORG Server - http://www.Mailgate.ORG --------------------------- Newsgroups: comp.os.vxworks Subject: Re: CodeTest or (ScopeProfile+MemScope) Date: Tue, 05 Sep 2000 16:58:46 +0200 From: Senthil Velmurugan Organization: Ericsson Eurolab Deutschland GmbH Message-ID: <39B50A26.ABB1BE32@eed.ericsson.se> References: <39B4B1D3.36174246@asiayeah.com> Hi, > I am evaluating VxWorks for one of my networking projects. I come across > the following tools, > > 1) CodeTest/Coverage + CodeTest/Memory > 2) PerformacePak (ScopeProfile+MemScope) > > Does anyone has any real-life experiences with the above products? They > sound very smilar to me, is option 1 or 2 better? As far as I know, the major technical differences/similarities I have observed are as follows: 1. Code Instrumentation: ======================= CodeTest: Code instrumentation is necessary. You have to compile with a flag on module basis and it will add intrumentation code automatically by a sort of pre-processor before the code goes into compiler. The average instrumented code size is about 20% to 30% more than your 'clean' code. (The amount of branches in your code like 'if', 'switch' cases and calling memory routines makes the size of instrumentation code size bigger!) PerformancePak: To use both tools, no need to instrument your existing code. With a target-server connection, you can start & work with these tools on the target. 2. CodeTest(Coverage) vs ScopeProfile ===================================== CodeTest will be mostly useful when you are in the development cycle to improve your code and test cases. With this tool you can have an idea in a particular instace, which part of a function executed more often than the other block of your code. On the other hand, ScopeProfile is useful in measuring the performance of the target system. You can get an idea about each function how much time it takes to finish that function. For example, you can have a call stack at any point and from that point you are able to analyse fuction by fuction about execution time and CPU usage. ScopeProfile works by sampling the target for a (user-defined) period of time and makes the statistics out of that gathered information. You can customize the amount of time to sample and how often to sample. 3. CodeTest(Memory) vs. MemScope ================================ As per the functionality, both are almost the same. With these tools you can monitor locate memory allocation/de-allocation and memory related errors, currently used memory blocks and so on. With CodeTest you can locate the actual source code, but with MemScope you will be able to see only the call stack. CodeTest gathers information about memory usage by instrumented code. MemScope gathers by installing wrappers at run-time around allocation/de-allocation routines (malloc(), free(), realloc(), memPartCreate()...). I think you can decide which fit to your target and application! Regards, Senthil. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Load Error "Relocation value does not fit in 24 bits" Date: Tue, 05 Sep 2000 10:58:09 -0400 From: Al Johnston Organization: MindSpring Enterprises Message-ID: <39B50A01.15E8E9B4@mindspring.com> References: <6KBr5.11372$K4.483872@newsread1.prod.itd.earthlink.net> <8on6hv$81e$1@nnrp1.deja.com> <39B0846F.E5FCCF83@mindspring.com> <8oqgb5$vru$1@nnrp1.deja.com> > The destination address is sign extended, and simply takes the place Thanks... I guess the doc writers thought this was too obvious to document. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Driver Questions Date: Tue, 05 Sep 2000 09:56:21 -0500 From: Eric Lesch Organization: Lockheed Martin Missiles and Firing Control - Dallas Message-ID: <39B50995.900FC01C@lmco.com> References: <8p0pl3$g9j$1@news.news.kodak.com> I would suggest taking a look at the VxWorks Programmer's Guide section 3.9.  Also, take a look at the drivers that come with the BSP kit, if you have them.

For your device you should create a new structure with the device header as the first member.

typedef struct myDevHdr
   {
   /* first member must be device header */
   /* so that pointer to this header will
   /*also be pointer to device header */
   DEV_HDR devHdr;

   /* here you can keep track of whether */
   /* it was opened or not */
   BOOL bOpen;

   /* you can store other options here */
   int nMode;
   } MY_DEV_HDR;
 

Your code to add the device will look like this

/* create the device as a global */
MY_DEV_HDR myDev;

/* initialize your device structure here */
myDev.bOpen = FALSE;
myDev.nMode = 0;

/* add the device to the I/O system */
iosDevAdd(&myDev, "my device", drvNum);
 

Now when xxxOpen is called the pointer to your structure is passed as the first argument.

MY_DEV_HDR *xxxOpen(MY_DEV_HDR *pMyDevHdr,
                    char *pRem, int nMode)
   {
   /* check if device has already been opened */
   if(pMyDevHdr->bOpen == TRUE)
      return((MY_DEV_HDR *)ERROR);

   /* other stuff */

   /* set opened boolean to true */
   pMyDevHdr->bOpen = TRUE;

   /* save the mode if you want */
   pMyDevHdr->nMode = nMode;

   /* this will cause the I/O system to pass */
   /* the pointer to your device header to the */
   /* other routines in your driver */
   return(pMyDevHdr);
   }
 

Since open returned the pointer to the structure it will be passed as the first argument to xxxRead, xxxWrite, xxxIoctl, and xxxClose.

STATUS xxxClose(MY_DEV_HDR *pMyDevHdr)
   {
   /* set open boolean to false */
   pMyDevHdr->bOpen = FALSE;

   return(OK);
   }
 
 

Tim Stanka wrote:

I just designed my first VXWORKS driver. My designed was OKed but I have not
coded it up yet. I have a couple of questions:

1) How could I implement marking a device as "open for reading" or "open for
writing" so a device
    could only be open for reading with one fd.

    I wanted to use a section of the device descriptor to store information
about how a device was opened.
    (RD, WR, RDWR etc). This was easy to do in my xxOpen() function but when
I started to design
    xxClose() I realized I needed an association between an fd and the open
mode in order to correctly
    unmark the device for RD, WR etc.

    I want to look at the flags in xxOpen() and mark the device accordingly.
ie, if the flag is O_RDONLY
    I would mark the device for reading. In xxClose() I want to be able to
"unmark" the device but
    xxOpen and xxClose do not have access to the fd table. Also, an fd is
not fed to these functions.
    Without the fd to open mode association I do not know how to properly
"unmark" the device

2) The VXWORKS device descriptor (DEV_HDR) contains three fields: name,
driver number and dl. The dl
    field is a type DL_NODE. This is a self-referencing structure with two
fields:
         typedef struct dl_node
            {
            dl_node *prev;
            dl_node * next;
           } DL_NODE;

   I'm not sure why they did this. I would have a "void *data;" data field
either in DEV_HDR or DL_NODE
   so a user could connect a custom data structure to the device descriptor.

   Has anyone used the device descriptor to store device unique information?

Thanks,
Tim Stanka

--------------------------- Newsgroups: comp.os.vxworks Subject: Re: Problems with WindView under Windows 2000. Date: Tue, 5 Sep 2000 08:08:33 -0700 From: "Dan Pinson" Message-ID: <8p329j$c320f$1@ID-48686.news.cis.dfn.de> References: Hello Simon, Yes, I have the same problem. WRS's response is "Tornado is not supported on Windows 2000". While I haven't found a way to bypass the error message, Tornado seems to run fine. Dan Pinson "Simon DiMaio" wrote in message news:Pine.SUN.3.95.1000901211732.23944G-100000@salcudean.ece.ubc.ca... > Hi > > I have done a full install of Tornado 2.0 under Windows 2000. Upon running > Tornado, I receive the following message: > > couldn't read file "TRY=localhost\host\resource\tcl\WindView.win32.tcl": > no such file or directory > > Tornado runs fine, but WindView is not available. The base directory for > my Tornado installation is C:\Tornado, so I don't know where the > "TRY=localhost" has come from. > > Has anybody experienced a similar error message? > > Any ideas for a remedy would be greatly appreciated. > > Thanks in advance, > Simon DiMaio > > ------------------------------------------------------------------------ > Simon DiMaio Email: simond@ece.ubc.ca > Robotics and Control Group > Dept. of Electrical and Computer Engineering Phone: (604)822-9215 > University of British Columbia Fax : (604)822-5949 > --------------------------- Newsgroups: comp.os.vxworks Subject: gcc asm / ppc Date: Tue, 05 Sep 2000 11:16:58 -0400 From: Al Johnston Organization: MindSpring Enterprises Message-ID: <39B50E6A.294016D2@mindspring.com> I have two easy questions: 1) How do I get asm("addi r1,r1,0x0"); line to compile? Anytime I try to use an immediate operand the compiler chokes with a "Error: unresolved expression that must be resolved". 2) What is the encoding of the register fields for the power pc? The ppc programming environment book uses 5 bit fields titled rA, rD, rS, etc, but never bothers (that I can find) to tell you what values to use for which registers.. is it something as simple as 0x00000 => r0, 0x00001 => r1, 0x00010 => r2, etc? I am still hoping someone will tell me where the "asm" keyword is documented. The gcc doc only covers the use of the ":" fields, not the basics on the commands. --------------------------- Newsgroups: comp.os.vxworks Subject: DHCP service testing Date: Tue, 05 Sep 2000 11:58:43 -0400 From: Eva Qin Organization: Motorola Message-ID: <39B51833.5F491F72@motorola.com> Hi, I need to test DHCP service (without relay agent) in VxWorks, but got several problems and here are some questions I want to ask or clearify: 1. Which table in which file we have to modify to create the pool of configuration parameter sets for DHCP server? In "VxWorks 5.4 Network Programmer's Guide", it says "to control the contents of these memory-resident structures, you must modify the source code that defines these structures". Which structures in which source code need to be modified? I figured it is the dhcpsLeaseTbl[] in usrNetDhcpsCfg.c, is that right? 2. After creating the pool, is there any must-follow procedures need to follow to set up the DHCP server and client in order that they can communication to each other? Right now I have 2 750 targets, they are connected via a hub. I create two images: one for DHCP server and one for DHCP client, each of them used on 1 target. 3. When I create the VxWorks images, except I need to include "DHCP server" and "DHCPv4 runtime client" componenets, is there any other components needed to include for them to function well? By the way, I always got configuration conflicts between "DHCP client timestamp removal" and "DHCP client timestamp setup". Which one of these two I have to include into the image in order for DHCP client function correctly, or it doesn't matter? 4. When I run the test, I called serveral client functions, they are (in order): dhcpcInit( ), dhcpcOptionSet( ), dhcpcBind( ), dhcpcServerGet( ), dhcpcParamGet( ). The return for the first three calls are ok, but failed the last two. The error for the dhcpcServerGet is "not-bound". I assume it's because the client didn't bind to DHCP server correctly. 5. Is there anybody have any write-downs for DHCP server and client test procedures or test programs? or if you can show me a sample client application code? I'd like to take a look if you have and it's ok to show me. My reply address may has a problem, so please send replys to "yihua.qin@motorola.com". Thanks a lot for any help in advance. Yihua --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Problems with WindView under Windows 2000. Date: Tue, 05 Sep 2000 09:30:10 -0700 From: Lori Fraleigh Organization: Real-Time Innovations Message-ID: <39B51F92.E55B3ED4@rti.com> References: <8p329j$c320f$1@ID-48686.news.cis.dfn.de> Simon, Dan - Go to the Tools menu in Tornado and choose Options, then pick 'Tornado Registry' all the way over at the right. Change the Tornado registry from local machine to remote machine and type in the IP address of your Windows 2000 machine. This should make the error go away. The "localhost" is coming from the WIND_REGISTRY setting. Regards, Lori - -- Lori Fraleigh Real-Time Innovations Senior Software Engineer ScopeTools Manager Phone: 408-734-4200 x121 Fax: 408-734-5009 http://www.rti.com Please send/cc all support related email to support@rti.com Dan Pinson wrote: > > Hello Simon, > > Yes, I have the same problem. WRS's response is "Tornado is not supported on > Windows 2000". While I haven't found a way to bypass the error message, > Tornado seems to run fine. > > Dan Pinson > > "Simon DiMaio" wrote in message > news:Pine.SUN.3.95.1000901211732.23944G-100000@salcudean.ece.ubc.ca... > > Hi > > > > I have done a full install of Tornado 2.0 under Windows 2000. Upon running > > Tornado, I receive the following message: > > > > couldn't read file "TRY=localhost\host\resource\tcl\WindView.win32.tcl": > > no such file or directory > > > > Tornado runs fine, but WindView is not available. The base directory for > > my Tornado installation is C:\Tornado, so I don't know where the > > "TRY=localhost" has come from. > > > > Has anybody experienced a similar error message? > > > > Any ideas for a remedy would be greatly appreciated. > > > > Thanks in advance, > > Simon DiMaio > > > > ------------------------------------------------------------------------ > > Simon DiMaio Email: simond@ece.ubc.ca > > Robotics and Control Group > > Dept. of Electrical and Computer Engineering Phone: (604)822-9215 > > University of British Columbia Fax : (604)822-5949 > > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: CodeTest or (ScopeProfile+MemScope) Date: Tue, 05 Sep 2000 09:29:29 -0700 From: Lori Fraleigh Organization: Real-Time Innovations Message-ID: <39B51F69.78F9E005@rti.com> References: <39B4B1D3.36174246@asiayeah.com> <39B50A26.ABB1BE32@eed.ericsson.se> Senthil Velmurugan wrote: > > Hi, > > > I am evaluating VxWorks for one of my networking projects. I come across > > the following tools, > > > > 1) CodeTest/Coverage + CodeTest/Memory > > 2) PerformacePak (ScopeProfile+MemScope) > > > > Does anyone has any real-life experiences with the above products? They > > sound very smilar to me, is option 1 or 2 better? > > As far as I know, the major technical differences/similarities I have > observed are as follows: > > 1. Code Instrumentation: > ======================= > CodeTest: Code instrumentation is necessary. You have to compile with > a flag on module basis and it will add intrumentation code > automatically by a sort of pre-processor before the code goes into > compiler. The average instrumented code size is about 20% to 30% more > than your 'clean' code. (The amount of branches in your code like > 'if', 'switch' cases and calling memory routines makes the size of > instrumentation code size bigger!) > > PerformancePak: To use both tools, no need to instrument your existing > code. With a target-server connection, you can start & work with these > tools on the target. The other major difference not covered in the analysis below is the visibility provided by ScopeProfile and MemScope since these tools require no instrumentation. ScopeProfile can provide CPU utilization for any code in your system - even code you don't have the source to. This can help you track down calls to expensive kernel or third-party library functions in addition to identifying hot spots in your own code. CodeTest provides coverage information for only your own code. These two tools actually solve different problems. MemScope will track all memory allocations in the system including memory allocated by the kernel and third party libraries. CodeTest can only tell you about memory allocated in code which you have the source to and can recompile. It can't see any memory allocated on your behalf by the kernel or third-party libraries - this is where many memory leaks occur. The patching technology used by MemScope is also much less intrusive than instrumentation of code. Regards, Lori > 2. CodeTest(Coverage) vs ScopeProfile > ===================================== > CodeTest will be mostly useful when you are in the development cycle > to improve your code and test cases. With this tool you can have an > idea in a particular instace, which part of a function executed more > often than the other block of your code. > On the other hand, ScopeProfile is useful in measuring the > performance of the target system. You can get an idea about each > function how much time it takes to finish that function. For example, > you can have a call stack at any point and from that point you are > able to analyse fuction by fuction about execution time and CPU usage. > ScopeProfile works by sampling the target for a (user-defined) period > of time and makes the statistics out of that gathered information. You > can customize the amount of time to sample and how often to sample. > > 3. CodeTest(Memory) vs. MemScope > ================================ > As per the functionality, both are almost the same. With these tools > you can monitor locate memory allocation/de-allocation and memory > related errors, currently used memory blocks and so on. With CodeTest > you can locate the actual source code, but with MemScope you will be > able to see only the call stack. CodeTest gathers information about > memory usage by instrumented code. MemScope gathers by installing > wrappers at run-time around allocation/de-allocation routines > (malloc(), free(), realloc(), memPartCreate()...). > > I think you can decide which fit to your target and application! > > Regards, > Senthil. - -- Lori Fraleigh Real-Time Innovations Senior Software Engineer ScopeTools Manager Phone: 408-734-4200 x121 Fax: 408-734-5009 http://www.rti.com Please send/cc all support related email to support@rti.com --------------------------- Newsgroups: comp.os.vxworks Subject: Re: TrueFFS Date: Tue, 5 Sep 2000 17:31:58 +0100 From: "Dave Korn" Organization: (Posted via) U-NET Internet Ltd. Message-ID: References: <39a4022a@news.barak.net.il> Mark Bentley wrote in message ... > >3) Just so you know it is a b---- to debug! They do things like any error >code that you try to return, will be changed to the ERROR define. That's part of the stdio stuff: when a stdio function has an error to report, whether from a lower level such as your driver, or internal to the stdio function itself, the error return goes into errno, and the function itself returns ERROR. So what's (probably) happening is that when your driver returns an error, the stdio code says to itself "Oh dear, an error - the code must be in errno but I'll just return ERROR to the application and let it interpret errno itself". Alas, your driver isn't putting the error in errno, it's trying to return it, and that doesn't work. You are expected to write your driver according to the stdio convention, so that when you want to return an error it's up to your driver to set errno, and your driver ought to be returning ERROR to the calling code in the ioLib. Either way, what they're doing there is *correct* and *you* are using an incorrect method to attempt to return your error. hth, DaveK - -- They laughed at Galileo. They laughed at Copernicus. They laughed at Columbus. But remember, they also laughed at Bozo the Clown. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: gcc asm / ppc Date: Tue, 05 Sep 2000 13:20:24 -0400 From: Al Johnston Organization: MindSpring Enterprises Message-ID: <39B52B58.DD77B490@mindspring.com> References: <39B50E6A.294016D2@mindspring.com> Its worse that I thought. I can not get the compiler to accept: asm("mtctr r1"); either. Why does this not work? I also tried asm("mtctr r1" : : "r1"); which according to the book should also work; but it doesn't. I obviously don't understand the basics of asm, which is not surprizing since I cant find the basics documented anywhere. In case you are wondering, I am intentionally not tying the asm code to c variables. I am just trying to get the raw numbers for these instructions, not actually running the code. Someone give me a clue please --------------------------- Newsgroups: comp.os.vxworks Subject: Re: gcc asm / ppc Date: Tue, 05 Sep 2000 13:44:08 -0400 From: Martin Beaucage Message-ID: <39B530E8.9448D267@tech-m4.com> References: <39B50E6A.294016D2@mindspring.com> The asm synthax is described in the GNU toolKit user guide section 3.32. For the assembler instructions description you can get the following PDF documents on Motorola's web site. (your cpu) User Manual (your cpu) Programming environnement manual Yes it is as simple as that: The prototype for addi is addi rD,rA,SIMM so rD means a Global Purpose Register for the destination rA means a Global Purpose Register for the source or the destination SIMM is a 16 bits constant. asm("addi 1,1,0"); means GPR1 = GPR1 + EXTS(SIMM) (r1 = r1 + SIMM) where EXTS means extending on the left with sign bits. Al Johnston wrote: > I have two easy questions: > > 1) How do I get > asm("addi r1,r1,0x0"); > line to compile? Anytime I try to use an immediate operand the > compiler chokes with a "Error: unresolved expression that must > be resolved". > > 2) What is the encoding of the register fields for the power pc? > The ppc programming environment book uses 5 bit fields titled > rA, rD, rS, etc, but never bothers (that I can find) to tell you > what values to use for which registers.. is it something as > simple as 0x00000 => r0, 0x00001 => r1, 0x00010 => r2, etc? > > I am still hoping someone will tell me where the "asm" keyword > is documented. The gcc doc only covers the use of the ":" fields, > not the basics on the commands. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: gcc asm / ppc Date: Tue, 05 Sep 2000 14:51:13 -0400 From: Al Johnston Organization: MindSpring Enterprises Message-ID: <39B540A1.AFEFA0A6@mindspring.com> References: <39B50E6A.294016D2@mindspring.com> <39B530E8.9448D267@tech-m4.com> > The asm synthax is described in the GNU toolKit user guide section 3.32. I wish it was. That section starts out "...you can now specify the operands of the instruction using c expressions." It is documenting the use of the ":" fields as an addition to the simpler case that does not use that feature. The basic use of asm is not covered. I wish it was. > asm("addi 1,1,0"); Your kidding... no your not! it works just as you said. I tried %r1, r1, p1, %p1, gpr1,... but never just "1"... arrg.... THANKS --------------------------- Newsgroups: comp.os.vxworks Subject: Re: gcc asm / ppc Date: Tue, 05 Sep 2000 15:14:46 -0400 From: "J. Barry Wandell" Message-ID: <39B54626.BC1D44D6@ellacoya.com> References: <39B50E6A.294016D2@mindspring.com> <39B52B58.DD77B490@mindspring.com> "r1" is an undefined expression, the register is just "1" so drop the "r"'s and you should be all set for the asm() call. I think your suspicions are correct regarding the registers - 5 bits for 32 registers makes sense, right?! Never really looked to verify it. --------------------------- Newsgroups: comp.sys.powerpc.tech,comp.os.vxworks Subject: ppc asm: loading 32 bit value into CTR reg Date: Tue, 05 Sep 2000 15:22:48 -0400 From: Al Johnston Organization: MindSpring Enterprises Message-ID: <39B54808.B227A591@mindspring.com> I am trying to do something very simple. Loading a 32bit unsigned number into the CTR (SPR 9) register with minimal side effect. I have now spent 4 hrs trying to do this... I even broke down and tried to find it int the doc and faq... What is the proper way of doing this? using AND has sideffects on the CR. Using addi seemed promising, and was suggested by someone, so I tried lis r0,HI(value) li r0,LO(value) which didn't work because the add's ignore r0 as a "S" fields (use 0 rather than [rS]); so I tried lis r11,HI(value) li r11,LO(value) and that screws up if the hi bit of LO(value) is set (it sign extends the sign bit trashing the value set in the previous lis operation! I thought about swapping the order, but that wont work either because the "addi" would treat the LO(value) loaded into the low 16 bits of CTR as a negative number... This should not be this hard... what's the magic answer guys? thanks. --------------------------- Newsgroups: comp.os.vxworks Subject: Partitioning and Formatting Date: Tue, 05 Sep 2000 19:22:04 GMT From: Ted Dennison Organization: Deja.com - Before you buy. Message-ID: <8p3h4c$7b5$1@nnrp1.deja.com> Is there any way to partition and format a fresh ATA drive from vxWorks? I know we can do it from DOS with a boot floppy. But I'm writing a procedure for installing our software, and our customer will not nessecarily have access to DOS unless we buy it for them (and add that to the cost of the product). - -- T.E.D. http://www.telepath.com/~dennison/Ted/TED.html Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: gcc asm / ppc Date: Tue, 05 Sep 2000 15:35:23 -0400 From: Al Johnston Organization: MindSpring Enterprises Message-ID: <39B54AFB.264F519A@mindspring.com> References: <39B50E6A.294016D2@mindspring.com> <39B52B58.DD77B490@mindspring.com> <39B54626.BC1D44D6@ellacoya.com> > "r1" is an undefined expression, the register is just "1" so drop the > "r"'s and you should be all set for the asm() call. Man I feel dumb... > > I think your suspicions are correct regarding the registers - 5 bits for Seems be working for me... I am sure its in the pe32 somewhere... but it is not in an obvious place. thanks for the help. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: use of cascaded macros Date: Tue, 5 Sep 2000 13:48:00 -0600 From: "Richard" Message-ID: <39b54d5d.0@news> References: <8p2lqt$3g9$1@proxy2.fe.internet.bosch.de> <39B4F533.F56D7B9B@swri.edu> <8p2vvs$hmq$1@proxy2.fe.internet.bosch.de> try the following typedef enum { K_STRING_1, K_STRING_2, K_STRING_3, K_STRING_4, ...., NUMBER_K_STRINGS } K_STRING_DEF_TYPE; you can remove entries from the definition, and the following ones will be renumbered. I have not looked it up in ANSI C, but in all of the compilers I have used, the first enumberated value is 0, and each successive one is one larger than the previous one. These values are usable anywhere an int is. As a bonus, NUMBER_K_STRINGS can be used as the upper limit of the valid indices (all valid index values are less than NUMBER_K_STRINGS), and can be used to allocate your string table. Richard "Gunter Ritter" wrote in message news:8p2vvs$hmq$1@proxy2.fe.internet.bosch.de... > Thanks for your help. I will try this tomorrow. > > Well, why do we use macros to count in this manner. These macros define > constants to access large stringtables. > It is easier to access strings in this table by using a macro with names > like "K_MY_TEXTSTRING" than using an numerical index. > If a string will be deleted in this table, you just have to correkt the > macros and you can access the strings after the deleted string with the same > constant. > Otherwise you have to learn the new numbers of the following strings ;-) > > Gunter > > --------------------------- Newsgroups: comp.sys.powerpc.tech,comp.os.vxworks Subject: Re: ppc asm: loading 32 bit value into CTR reg Date: Tue, 05 Sep 2000 16:40:53 -0400 From: Martin Beaucage Message-ID: <39B55A55.29356DC1@tech-m4.com> References: <39B54808.B227A591@mindspring.com> Al Johnston wrote: > I am trying to do something very simple. Loading a 32bit unsigned > number into the CTR (SPR 9) register with minimal side effect. > You can use the instructions mtctr rS (move to control register) mfctr rD (move from control register) or use mtspr 9,rS (move to special purpose register) mfspr rD,9 (move from special purpose register) Some thing like this should do the trick: asm( "mtctr %0" : : "m" (value) ); --------------------------- Newsgroups: comp.os.vxworks Subject: scsi write corruption with large buffers Date: Tue, 05 Sep 2000 20:31:11 GMT From: kealk@my-deja.com Organization: Deja.com - Before you buy. Message-ID: <8p3l6f$c4f$1@nnrp1.deja.com> Hi Everyone, It seems like when I do a write() to a SCSI device with a large (1-8 Meg) buffer the data is not written correctly. I am using the DOS file system on a Power PC 603 and 604. I still see it even if I do a FIOFLUSH ioctl() after each write(). I do NOT see the memory write corruption() if I allocate the large buffer then break it into pieces before sending it to the write() command. This leads me to believe the problem is actually associated with passing the large buffer to the write() command. Breaking it up causes a performance hit which is why I want to use one single larger buffer. Has anyone else seen problems like this or know of a solution. Any help appreciated. Regards, Kerry Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.sys.powerpc.tech,comp.os.vxworks Subject: Re: ppc asm: loading 32 bit value into CTR reg Date: Tue, 05 Sep 2000 17:12:24 -0400 From: Martin Beaucage Message-ID: <39B561B8.ECC297A1@tech-m4.com> References: <39B54808.B227A591@mindspring.com> <39B55A55.29356DC1@tech-m4.com> Excuse me: mtctr rS means: move to count register, not control register mfctr rD means: move from count register, not control register Martin Beaucage wrote: > Al Johnston wrote: > > > I am trying to do something very simple. Loading a 32bit unsigned > > number into the CTR (SPR 9) register with minimal side effect. > > > > You can use the instructions > > mtctr rS (move to control register) > mfctr rD (move from control register) > > or use > > mtspr 9,rS (move to special purpose register) > mfspr rD,9 (move from special purpose register) > > Some thing like this should do the trick: > asm( "mtctr %0" : : "m" (value) ); --------------------------- Newsgroups: comp.sys.powerpc.tech,comp.os.vxworks Subject: Re: ppc asm: loading 32 bit value into CTR reg Date: 5 Sep 2000 14:28:17 -0700 From: andrew@user2.teleport.com (Andrew Klossner) Message-ID: <8p3ohh$64h$1@user2.teleport.com> References: <39B54808.B227A591@mindspring.com> > I am trying to do something very simple. Loading a 32bit unsigned > number into the CTR (SPR 9) register with minimal side effect. addis 0,0,HI(value) ori 0,0,LO(value) mtctr 0 > using AND has sideffects on the CR. AND has no such side effect, but AND. (opcode ends in a period) will change the CR. -=- Andrew Klossner (andrew@teleport.com) --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Spawning C++ task Date: Tue, 5 Sep 2000 15:57:37 -0600 From: "George Malone" Organization: Ball Corporation Message-ID: <968190933.274975@newsfeed> References: "G Kraemer" wrote in message news:CkSq5.35744$P4.268715@news1.rdc1.az.home.com... > VxWorks experts: > > I have a MVME-2700 using vxWorks 5.4 (T2) and I have added a C function, > usrpgms() called from usrApplInit, which installs some of my own routines > (file system date/time stamping, printer driver install, clock > initializations, symFindByName(), and taskSpawn()). > > I have been delivered several C++ modules which are compiled with the Green > Hills 1.8.9 C++ compiler. This executable (65M) resides on an exported > network drive which is nfs mounted to the MVME-2700. > > Here's my problem: usrpgms() can load the C++ executable into memory but > when I attempt to spawn it, the C++ executable does not function correctly. > When I let usrpgms() load the C++ executable and taskSpawn from the target > shell it functions correctly. > > Target shell execution: > taskSpawn("niu_main",140,0x08,165000,(FUNCTPTR)symaddr) > > Can anyone provide me with sample code on how to spawn the C++ executable > from my C function? > > Thanks in advance > > Garry > > > int niu_main() // does not appear to be within a class > { > // C++ code .... > } > > Garry, indeed, this is a very mysterious symptom with an easy solution. What's missing when you spawn the C++ function is its object pointer, i.e. the "this" pointer. Every C++ function assumes that the first argument on the stack is the "this" pointer. Simply add the object pointer as an argument after (FUNCTPTR)symaddr and your symaddr method will be happy. George --------------------------- Newsgroups: comp.os.vxworks,comp.realtime Subject: TELECOM - VxWorks S/W Developers Needed - Ottawa Date: Tue, 5 Sep 2000 18:51:55 -0400 From: "PROCOM" Organization: IGS - Information Gateway Services Message-ID: <8p3t74$1t6$1@news.igs.net> Title: C++ / OOD VxWorks Software Developers Description: Our client, one of the world's largest telecommunications companies, currently requires 10 Junior to Senior Software Specialists to design, develop, implement and support Wireless products and applications. Minimum Requirements: All candidates should possess a minimum of 1 year experience developing real-time C/C++, OOD S/W on a VxWorks platform. Desired skills would include any previous Wireless development experience. Location: Ottawa, Ontario, Canada Type: Full-Time Placement Compensation: Open PROCOM: Established in 1978, Professional Computer Consultants Group Ltd. (Procom) is a national leader in the provision of Computer personnel on a contract and full-time basis. Our clients are comprised of the largest national and international corporations that utilize technical resources extensively across a wide range of disciplines. In the Financial Post (March 1999) Procom was ranked as the 6th largest professional Services Company in Canada. In November of 1999, Procom was named a Regional finalist in Canada' s 50 Best Managed Private companies. Our track record is proven with more than 180 consultants servicing Ottawa's high tech community and more than 1600 Procom consultants currently on assignment throughout. North America. For further information on this and other opportunities please visit our web site at www.procom.ca. Interested candidates are invited to forward their resumes or questions in confidence to: Derek Weber Technical Recruiter 300 March Road Kanata, Ontario K2K 2E2 phone: (613) 270-9339 ext 231 e-mail: derekw@procom.ca --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Problems with PCI ISR Date: 5 Sep 2000 22:57:21 GMT From: tsikes@netcom.com (Terry Sikes) Organization: NETCOM / MindSpring Enterprises, Inc. Message-ID: <8p3toh$idf$1@slb6.atl.mindspring.net> References: <8op2eq$t6l$1@slb6.atl.mindspring.net> <8ovfmr$1ps$1@nnrp1.deja.com> In article <8ovfmr$1ps$1@nnrp1.deja.com>, Thomas Rahn wrote: >Hi Terry, > >from your code snippets I get the impression you do not >enable the interrupts (something like sysIntEnablePic(IRQ_LVL)) > >Maybe this helps > >Bye > >Thomas Rahn Hi Thomas. Thanks for responding. Actually, I do have that call in my code, I just forget to include it: - ---revised snippet 1--- int myInterrupt = 10; if (intConnect( INUM_TO_IVEC(INT_VEC_GET(myInterrupt)), (VOIDFUNCPTR) myIsr, (int) myIntSem) == ERROR) { logError("tm: intConnect failed"); } if (sysIntEnablePIC(myInterrupt) == ERROR) { logError("tm: failed to enable interrupt"); } - ------ What documentation I've been able to find on sysIntEnablePic has been pretty sparse, so I hope I'm doing (mostly) the right things. Now, if I can just figure out what's actually going on... (I don't see any of my error messages.) Terry - -- tsikes@netcom.com --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Green Hills Software Date: Tue, 5 Sep 2000 19:46:47 -0400 From: "Jeff Creem" Organization: MV Communications, Inc. Message-ID: <8p40db$36f$1@pyrite.mv.net> References: <8ogv5q$rpv$1@nnrp1.deja.com> <39b0638f$2$zpzbeena$mr2ice@news.visi.net> "M Campbell" wrote in message news:CU3t5.124591$eS6.1263211@news1.rdc1.md.home.com... > On the Ada side, we had very few problems with developing there. After a > certian point, the debugger also became useless. Also found that Greenhills > method for implementing multitasking a bit awkward. > > Would love to see a competior come into the picture on the Fortran side, but > don't realistically don't see the enough demand for a company to make the > effort to reach the maturity level of the Greenhills product & be > competitive. As far as I can tell, GNU compilers / debuggers (fortran, ada) > won't work for vxWorks. If anyone has actually used them in a vxWorks > environment, i'd love to hear about it, as GNU tools "should" integrate > seamlessly into the vxWorks way of doing things. > Hmm..vxWorks is a supported platform from ACT (www.gnat.com) for GNAT (Ada 95). I've used gcc for BSP stuff on vxworks for years..Only question left is FORTRAN (might be a little tougher to find a vendor to provide support). --------------------------- Newsgroups: comp.os.vxworks Subject: Bus Bandwidth and CPU load for VxWorks RTOS and Network stack question. Date: Tue, 5 Sep 2000 18:26:12 -0700 From: "Greg FitzJohn" Organization: Posted via Supernews, http://www.supernews.com Message-ID: We are planning on using the MPC855T processor at 50MHz (maybe 80MHz) for one of our products. Is there a place where I can find out what percentage of bus bandwidth and CPU load that the VxWorks RTOS and the network stack typically consume. Thanks, Greg FitzJohn gregf@vertical.com Vertical Networks --------------------------- Newsgroups: comp.os.vxworks Subject: Santa Barbara Opportunities fo VxWorks Developers Date: Tue, 5 Sep 2000 18:46:32 -0700 From: "Recruiting" Organization: Connected Systems Message-ID: Reply-To: "Recruiting" Connected Systems is a high tech startup in Santa Barbara, California nearing launch of our Internet messaging & voice appliance. Our products enable service providers to enhance their Broadband/Internet offerings with our new breed of high-value applications such as messaging, voice mail and beyond. We are growing rapidly and currently seeking experience in voice, CTI, messaging, networking, real-time systems for the following positions: · Software Engineers · Sr. Software Engineers Design and develop integrated applications, drivers on our IP-telephony platforms. C/C++ and networking experience required. Senior positions require RTOS and 5+ years of industry experience. VxWorks experience and familiarity with Unified Messaging, VoIP, LDAP, IMAP protocols desired. We offer competitive salaries, benefits, SEP program, and pre-IPO stock options. Please send your resumes to: jobs@connsys.com or Fax: (775) 205-6564 --------------------------- Newsgroups: comp.sys.powerpc.tech,comp.os.vxworks Subject: Re: ppc asm: loading 32 bit value into CTR reg Date: Tue, 05 Sep 2000 23:07:06 -0400 From: Al Johnston Organization: MindSpring Enterprises Message-ID: <39B5B4DA.ECC176A3@mindspring.com> References: <39B54808.B227A591@mindspring.com> <8p3ohh$64h$1@user2.teleport.com> Thanks to all who answered. > addis 0,0,HI(value) > ori 0,0,LO(value) > mtctr 0 sigh... and thanks, will try this tomorrow. > AND has no such side effect, but AND. (opcode ends in a period) will I know I checked this; Ah, there is a "if ." on the side effects... thanks for straightening me out.ch... - -al --------------------------- Newsgroups: comp.os.vxworks Subject: rpclib - errno 0xd0003 Date: Tue, 05 Sep 2000 23:09:23 -0400 From: Al Johnston Organization: MindSpring Enterprises Message-ID: <39B5B563.D0E3CB51@mindspring.com> Anybody know why the rpcLib routines seem to like to set errno? I know errno is not valid except when an error has been detected... still, its annoying! Anybody know? none of the other lib's seem to "miss" behave like this. - -al --------------------------- Newsgroups: comp.os.vxworks Subject: Which IRQ for PCMCIA Card? Date: Tue, 5 Sep 2000 20:38:01 -0700 From: "Dan Pinson" Message-ID: <8p4e7a$bu99h$1@ID-48686.news.cis.dfn.de> I should remember this but don't. Can someone help? Platform: x86 (Pentium) PCMCIA controller on PCI bus Question: How is the IRQ determined? The card/slot is configured for INT#A which is a PCI interrupt. How do I determine what IRQ is raised? For background the controller is a Cirrus Logic PD6729. I have a 3COM Ethernet III card installed. Networking doesn't seem to be working although there are no error messages during boot. I cannot ping from/to the target nor can I get WDB to work via the network. By peeking in RAM I can see that the card is there and correctly identified. I have PCIC_INT_LVL set to 0x0a. Does that cause the interrupt to map IRQ10 or is that telling VxWorks that it is already mapped to that IRQ? Thanks, Dan Pinson --------------------------- Newsgroups: comp.sys.powerpc.tech,comp.os.vxworks Subject: Re: ppc asm: loading 32 bit value into CTR reg Date: Wed, 06 Sep 2000 04:59:36 GMT From: "Martin Usher" Organization: EarthLink Inc. -- http://www.EarthLink.net Message-ID: References: <39B54808.B227A591@mindspring.com> Reply-To: "Martin Usher" > lis r11,HI(value) > li r11,LO(value) > and that screws up if the hi bit of LO(value) is set (it > sign extends the sign bit trashing the value set in the > previous lis operation! The problem isn't the mtspr order, its getting that number into the register..... > This should not be this hard... what's the magic answer guys? > The answer is to use ori as the second order...... addis reg,0,HI(value) (lis is a simplified mneomic version of addis) ori reg,reg,LO(value) That's the good news. The bad news is that the versions of gcc that I have use the construct addis/add (lis/ls in simplified format) so get the constant wrong half the time. Getting a CTR value wrong is one thing, sending the processor to completely the wrong address is really bad news! --------------------------- Newsgroups: comp.os.vxworks Subject: Re: rpclib - errno 0xd0003 Date: Wed, 06 Sep 2000 11:10:27 +0530 From: Vijay Kumar Peshkar Organization: Ye 'Ol Disorganized NNTPCache groupie Message-ID: <39B5D8CB.C2BFFE7F@cisco.com> References: <39B5B563.D0E3CB51@mindspring.com> Reply-To: vpeshkar@cisco.com This is a multi-part message in MIME format. - --------------E73B9B9B1B91B06DFB6ED5B2 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Al, The value you are seeing is a common residual errno left behind during various phases of rpc... If this is disconcerting, yes you can safely clear errno yourself. Infact the behavior occurs across all libs in vxworks. I had observed one while I was trying to create my own serial device driver...and yes I too did not like the idea ;0) In a nutshell, errno is only guaranteed to contain meaningful information if a system call returns an error. regards, Vijay Al Johnston wrote: > Anybody know why the rpcLib routines seem to like to > set errno? I know errno is not valid except when an > error has been detected... still, its annoying! > > Anybody know? none of the other lib's seem to "miss" > behave like this. > > -al - --------------E73B9B9B1B91B06DFB6ED5B2 Content-Type: text/x-vcard; charset=us-ascii; name="vpeshkar.vcf" Content-Transfer-Encoding: 7bit Content-Description: Card for Vijay Kumar Peshkar Content-Disposition: attachment; filename="vpeshkar.vcf" begin:vcard n:Peshkar;Vijay x-mozilla-html:FALSE org:Wipro Technologies,K2, Bangalore;Wishing you blue skies, a couple of daisies, some fresh air and a smile! adr:;;;;;; version:2.1 email;internet:vpeshkar@cisco.com title:Senior Software Engineer fn:Vijay Kumar Peshkar end:vcard - --------------E73B9B9B1B91B06DFB6ED5B2-- --------------------------- Newsgroups: comp.os.vxworks Subject: dosFs2 chkdsk slow to run from bootrom Date: Wed, 06 Sep 2000 06:40:24 GMT From: paulvroberts@my-deja.com Organization: Deja.com - Before you buy. Message-ID: <8p4osn$ja8$1@nnrp1.deja.com> Hi there, Does anyone know why chkdsk would be much slower when run from the bootrom, then when run after VxWorks is loaded? We are using FAT32 (singly dos partitioned) 9GB SCSI disks and it takes 65 seconds in the bootrom, whereas after VxWorks is running, it takes just 3-4 seconds! What I did was to "enhance" the bootConfig.c file to allow booting using dosFs2. I use a 128K cache, and call chkdsk via the icotl FIOCHKDSK. I am using the MVME2700 BSP with the bootrom_uncmp target, so presumably that's not rom-resident. Any other ideas? Thanks, Paul Roberts. Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Second serial port on MBX860 Date: Wed, 06 Sep 2000 07:46:58 GMT From: lourens@mecalc.co.za (Lourens Geldenhuys) Organization: Mecalc (Pty) Ltd Message-ID: <39b5f508.90937170@modem.mecalc> References: <6M5t5.7434$Dxe.201282560@news.telia.no> Reply-To: lourens@mecalc.co.za Øyvind The ads860 BSP supports SMC2 as the second serial port. This sysSerial.c might be similar to the mbx860. Also, see SPR: #10005 titled "Second SMC initialization in sysSerial.c for ads860" on WindSurf. These changes have already been incorporated in the ads860 BSP for T2. Regards, Lourens Geldenhuys On Tue, 05 Sep 2000 12:35:14 GMT, "Øyvind Holmeide" wrote: > >I am trying to configure VxWorks for a second serial port (RS232) on the >MBX860 board from Motorola. The BSP is only prepared for one serial port >(the console). Thus, an update of sysSerial.c for e.g. SMC2 is needed. I >have tried without success. Anyone been through this? > >regards > >Øyvind Holmeide > > ================================================================ Lourens Geldenhuys Tel: +27 (0)12 665 1480 Project Engineer Fax: +27 (0)12 665 1495 Mecalc (Pty) Ltd e-mail: lourens@mecalc.co.za 86 Oak Avenue, Highveld Technopark, Centurion, South Africa ================================================================ --------------------------- Newsgroups: comp.os.vxworks Subject: Re: use of cascaded macros Date: Wed, 6 Sep 2000 09:51:11 +0200 From: Gunter Ritter Organization: Tenovis GmbH u. Co. KG Message-ID: <8p4t0n$6am$1@proxy2.fe.internet.bosch.de> References: <8p2lqt$3g9$1@proxy2.fe.internet.bosch.de> <39B4F533.F56D7B9B@swri.edu> <8p2vvs$hmq$1@proxy2.fe.internet.bosch.de> Gunter Ritter schrieb in im Newsbeitrag: 8p2vvs$hmq$1@proxy2.fe.internet.bosch.de... > Thanks for your help. I will try this tomorrow. > > Well, why do we use macros to count in this manner. These macros define > constants to access large stringtables. > It is easier to access strings in this table by using a macro with names > like "K_MY_TEXTSTRING" than using an numerical index. > If a string will be deleted in this table, you just have to correkt the > macros and you can access the strings after the deleted string with the same > constant. > Otherwise you have to learn the new numbers of the following strings ;-) > > Gunter > > Okay, i tried it..... The brackets didn´t change anything in the behaviour. Still the same error. The compiler of the other project has no problems with this contruction. Gunter --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Second serial port on MBX860 Date: Wed, 06 Sep 2000 09:25:27 GMT From: "Øyvind Holmeide" Organization: Enitel Internet Public Access Message-ID: References: <6M5t5.7434$Dxe.201282560@news.telia.no> <39b5f508.90937170@modem.mecalc> I have used the ads860 BSP as a starting point. However, these two BSP's do differ (e.g. dual port RAM use). I have change the sysSerial.c according to the ads BSP and changed NUM_TTY from 1 to 2 without success. Do you know if there is any other places where changes are required. Lourens Geldenhuys wrote in message news:39b5f508.90937170@modem.mecalc... > Øyvind > > The ads860 BSP supports SMC2 as the second serial port. This > sysSerial.c might be similar to the mbx860. Also, see SPR: #10005 > titled "Second SMC initialization in sysSerial.c for ads860" on > WindSurf. These changes have already been incorporated in the ads860 > BSP for T2. > > Regards, > Lourens Geldenhuys > > On Tue, 05 Sep 2000 12:35:14 GMT, "Øyvind Holmeide" > wrote: > > > > >I am trying to configure VxWorks for a second serial port (RS232) on the > >MBX860 board from Motorola. The BSP is only prepared for one serial port > >(the console). Thus, an update of sysSerial.c for e.g. SMC2 is needed. I > >have tried without success. Anyone been through this? > > > >regards > > > >Øyvind Holmeide > > > > > > ================================================================ > Lourens Geldenhuys Tel: +27 (0)12 665 1480 > Project Engineer Fax: +27 (0)12 665 1495 > Mecalc (Pty) Ltd e-mail: lourens@mecalc.co.za > 86 Oak Avenue, Highveld Technopark, Centurion, South Africa > ================================================================ --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Second serial port on MBX860 Date: Wed, 06 Sep 2000 09:37:42 GMT From: lourens@mecalc.co.za (Lourens Geldenhuys) Organization: Mecalc (Pty) Ltd Message-ID: <39b60e53.97412772@modem.mecalc> References: <6M5t5.7434$Dxe.201282560@news.telia.no> <39b5f508.90937170@modem.mecalc> Reply-To: lourens@mecalc.co.za Øyvind I am unfortunately not too familiar with the mbx860 BSP. I am currently modifying the ads860 BSP for a custom board and I am hoping that both serial ports will work 'out-of-the-box'. Maybe SPR# 10005 can help? Regards, Lourens Geldenhuys On Wed, 06 Sep 2000 09:25:27 GMT, "Øyvind Holmeide" wrote: >I have used the ads860 BSP as a starting point. However, these two BSP's do >differ (e.g. dual port RAM use). I have change the sysSerial.c according to >the ads BSP and changed NUM_TTY from 1 to 2 without success. Do you know if >there is any other places where changes are required. > > >Lourens Geldenhuys wrote in message >news:39b5f508.90937170@modem.mecalc... >> Øyvind >> >> The ads860 BSP supports SMC2 as the second serial port. This >> sysSerial.c might be similar to the mbx860. Also, see SPR: #10005 >> titled "Second SMC initialization in sysSerial.c for ads860" on >> WindSurf. These changes have already been incorporated in the ads860 >> BSP for T2. >> >> Regards, >> Lourens Geldenhuys >> >> On Tue, 05 Sep 2000 12:35:14 GMT, "Øyvind Holmeide" >> wrote: >> >> > >> >I am trying to configure VxWorks for a second serial port (RS232) on the >> >MBX860 board from Motorola. The BSP is only prepared for one serial port >> >(the console). Thus, an update of sysSerial.c for e.g. SMC2 is needed. >I >> >have tried without success. Anyone been through this? >> > >> >regards >> > >> >Øyvind Holmeide >> > >> > >> >> ================================================================ >> Lourens Geldenhuys Tel: +27 (0)12 665 1480 >> Project Engineer Fax: +27 (0)12 665 1495 >> Mecalc (Pty) Ltd e-mail: lourens@mecalc.co.za >> 86 Oak Avenue, Highveld Technopark, Centurion, South Africa >> ================================================================ > > ================================================================ Lourens Geldenhuys Tel: +27 (0)12 665 1480 Project Engineer Fax: +27 (0)12 665 1495 Mecalc (Pty) Ltd e-mail: lourens@mecalc.co.za 86 Oak Avenue, Highveld Technopark, Centurion, South Africa ================================================================ --------------------------- Newsgroups: comp.os.vxworks Subject: Re: How the target Shell handles the console device driver ? Date: Wed, 06 Sep 2000 09:45:10 GMT From: "Øyvind Holmeide" Organization: Enitel Internet Public Access Message-ID: References: read() and write() from STD_IN and STD_OUT should work. However, it is important that your task has higher priority than the target shell if your task is spawned from e.g. usrAppInit.c. The target shell will receive the input from STD_IN if the target shell has higher priority. Note also that you might run into problems when switching from RS232 to telnet if your task is hanging on a read() when the target shell is re-directed from e.g. RS232 to telnet. Thus, use select() in order to avoid this problem. regards Øyvind Holmeide JiHye, Sohn wrote in message news:qIMr5.5636$jM4.28353@news2.bora.net... > Hello > I'd like to know how the VxWorks target shell handles the console I/O. > > When a user try to communicate the target using RS-232C serial port or > telent, > he can get the shell prompt(->) in telnet terminal or serial comm. S/W like > hyterterminal. > I'm trying to develop my own monitor program like this scheme. > > First, I tried to create a file to control the ethernet device driver like > the consoleFd. > (- the File Descriptor to control the RS-232C serial port), but failed. > I don't know how can I receive the monitoring request and send the response > data > by ethernet device driver or RS-232C serial port respectively. > > If you have any diffferent idea to implement this function, please let me > know. > Thanks for your help. > > --------------------------- Newsgroups: comp.os.vxworks Subject: Vxworks wtih MPC8260ads Date: Wed, 6 Sep 2000 17:49:48 +0800 From: "Tony Kwok" Organization: IMS Netvigator Message-ID: <8p53rd$rqn18@imsp212.netvigator.com> Hi, I need to do a project with the Motorola MPC8260ADS board, TCOM board and a FPGA self-develop board. I am planning to use the vxworks as the system OS. I have the following questions. Thanks a lot for your help. The FPGA will be connected to the 60x bus. It will be acting as a new interface between MPC8260 and another system. I know that there is a BSP for MPC8260ADS. Do I need to self-develop new BSP for the TCOM board and the FPGA board? What is the difference between the driver and the BSP of the vxworks? I am just a beginner in the embedded system. Pls correct me if my concept is wrong. Thanks a lot. Regards, Tony Kwok --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Device doesn't show up as an interface Date: Wed, 06 Sep 2000 12:06:31 +0200 From: Mario Wolframm Organization: ONELINE AG Message-ID: <39B61727.C13AD5D9@oneline.de> References: <39B4DE70.2B5E87AD@oneline.de> <39B612E0.694B2150@atos-group.com> "J. Grimbert" wrote: > > > Did I forget something important here? > > Yes, check the value of IP_MAX_UNITS for your image. > You where right, the value was 1! Set it to 20 I rebuilt everything but it's not done yet, still the same behaviour I will keep trying but if somebody else has an idea? Anyway thanks a lot! Mario --------------------------- End of New-News digest ********************** From vxwexplo-errs@csg.lbl.gov Wed Sep 6 04:15:21 2000 From: "David Anderson @ Data Track" Date: Wed Sep 6 04:15:23 PDT 2000 Subject: Why are Negative block sizes displayed by memShow & memFreeMax vxWorks 5.3.1 & Tornado 1.0.1 / Motorola processor. Dear vxWorkers, We have a problem with memory management which someone must have seen before ;-). memFreeMax() is returning a negative number of around -22KBytes. To make the situation even stranger, memShow 1 from the host shell shows four FREE BLOCKS these are shown as: block 1 small positive value 100 Bytes block 2 larger positive value 500 Bytes block 3 large positive value 26K block 4 large negative number. -22K If we then dump the memory block header for block 4 it shows the size as FFFC 9899 or similar. If this is a signed long then this corresponds to a negative -0x36766 (not correcting for two's complement). ------------------------------------------------------------------------ ---------------------------------------- Q1/ Does this mean that the system has run out of malloc memory space ??. ------------------------------------------------------------------------ --------- Q2/ What is the significance of the negative sized block. Resulting Problems ---------------------------- This completely destroys code that looks like this : if ( blocksize-memFreeMax() > required_later ) proceed_with_function() where memFreeMax() is assumed to be in the range 0-large positive value. David Anderson. mailto:danderson@dtrack.demon.co.uk ----------------------------------------------------- Who said "There's no life here in outer space." ----------------------------------------------------- Disclaimer ---------- This email and any files transmitted with it are confidential and intended solely for the use of the individual or entity to whom they are addressed. If you have received this email in error please destroy and notify Data Track Technology Plc +44 1425 271900. ------------------------------------------------------- From vxwexplo-errs@csg.lbl.gov Wed Sep 6 05:08:46 2000 From: "Ramachandramurthy, Bharat (Bharat)** CTR **" Date: Wed Sep 6 05:08:48 PDT 2000 Subject: Clarification Can this question be answered:- 1. Does vxWorks support opening multiple instances of the same file. if yes then is there any provisions for LOCKING the file. If LOCKING is not supported then how do we prevent the file from being used by the other users when one user is manipulating it. if not does that mean no other user can operate on the file until and unless the current user finishes. If so could u explain how this is done . Can mail me back to bharathramchandra@yahoo.com/br41@lucent.com regards bharath From vxwexplo-errs@csg.lbl.gov Wed Sep 6 05:57:01 2000 From: mmenge@dspt.com Date: Wed Sep 6 05:57:03 PDT 2000 Subject: Re: Why are Negative block sizes displayed by memShow & memFreeMax I have seen this happen when the memory has been corrupted. Somewhere someone overwrote the memory block header. tornado vxworks From vxwexplo-errs@csg.lbl.gov Wed Sep 6 08:58:37 2000 From: David Laight Date: Wed Sep 6 08:58:40 PDT 2000 Subject: re: Device doesn't show up as an interface > I call muxDevLoad and get back a reasonable address. > I call muxDevStart and get OK > I see the driver registered via muxShow > however: > neither ipAttach nor ifAddrSet work with that new interface! > Did I forget something important here? Does 'endFindByName( "xyz", 0 ) work? The one in vxWorks 5.3.1 doesn't support multiple drivers! If it fails, patch in the following: END_OBJ * fixed_endFindByName( char *name, int unit ) { END_TBL_ROW *row; END_OBJ *end; extern int *muxDevExists; for (row = (void *)lstFirst( (void *)*(&muxDevExists-1) ); row; row = (void *)lstNext( &row->node )) { if (strcmp( name, row->name )) continue; for (end = (void *)lstFirst( &row->units ); end; end = (void *)lstNext( &end->node )) if (end->devObject.unit == unit) return end; } return 0; } Compile the above somewhere and then execute something like: extern int endFindByName; endFindByName = 0xea000000 | (((int *)&fixed_endFindByName - &endFindByName - 2) & 0xffffff); to patch a branch instruction over the first instruction of the old routine. (NB the above is an ARM branch, you will need to modify the constants for a different architecture.) David ---------------------------------------------------------------- David Laight email: dsl@tadpole.co.uk Tadpole Technology plc phone: +44 1223 428 232 Cambridge, UK fax: +44 1223 428 201 From vxwexplo-errs@csg.lbl.gov Thu Sep 7 00:10:07 2000 From: rndmail Date: Thu Sep 7 00:10:09 PDT 2000 Subject: Registering Newsgroup Hi, I would like to register myself to Vxworks newsgroup. I got your info from one of my friends who is a member of your news group. My email id is: shah_jignesh_s@yahoo.com I hope you will do the needful in this matter. Lookinf forward to your reply, With regards, Jignesh Shah From vxwexplo-errs@csg.lbl.gov Thu Sep 7 03:44:56 2000 From: "David Anderson @ Data Track" Date: Thu Sep 7 03:44:58 PDT 2000 Subject: Negative returns from memFindMax() : Solution. vxWorks Tornado Thanks to those who provided hints and tips. BSP+Linker problem as explained below. PROBLEM & SOLUTION FOR THE RECORD: >memFreeMax() is returning a negative number of around -22KBytes. > >To make the situation even stranger, memShow 1 from the host shell shows four FREE BLOCKS >these are shown as: > >block 1 small positive value 100 Bytes >block 2 larger positive value 500 Bytes >block 3 large positive value 26K >block 4 large negative number. -22K > >If we then dump the memory block header for block 4 it shows the size as FFFC 9899 or similar. If this is a signed >long then this corresponds to a negative -0x36766 (not correcting for two's complement). System is working in all other respects. INVESTIGATION & SOLUTION: Investigation using advice from this group provided the following: 1) Negative block size is actually a very large positive value. 2) Using memFreeMax() immediately after initialising the memory pool also indicates a large negative number. 3) The pool size passed into the initialisation routine was found to be a very large positive value, so large that it is displayed as a negative number by the shell ( which treats it as an int !). 4) The pool size is passed to the usrRoot task from KernelInit(...) which takes a start and end address for the pool... 5) KernelInit does a (start_address - end_address) to calculate the pool size for usrRoot(..) WITHOUT any checking. 6) The global variable space in the system had increased so that the first free ram address was now after sysMemTop() !. Hence calculated pool size goes to a negative int value ( == large positive unsigned ). (but well below sysPhysMemTop() ) SOLUTION... 7) Increasing LOCAL_MEM_SIZE so that sysMemTop() is increased above the first free memory address cures the problem. Therefore if you ever see a negative block size with memFindMax() or memShow() check your kernelInit memory pool parameters are valid. One ( or more ) of these are assigned by the linker..... . BSP MODIFICATIONS I've now added explicit checking of the memory pool length in usrRoot() which range check the memory pool size parameter. /* Initialize the memory pool before initializing any other package. * The memory associated with the root task will be reclaimed at the * completion of its activities. */ if ( (long)memPoolSize < 0 ) /* checking for overflow into sign bit */ { errnoSet(EDOM); // Out of range. taskSuspend(0); // can be seen with debug tools. } memInit (pMemPoolStart, memPoolSize ); /* initialize memory pool */ David Anderson. mailto:danderson@dtrack.demon.co.uk ----------------------------------------------------- Who said "There's no life here in outer space." ----------------------------------------------------- Disclaimer ---------- This email and any files transmitted with it are confidential and intended solely for the use of the individual or entity to whom they are addressed. If you have received this email in error please destroy and notify Data Track Technology Plc +44 1425 271900. ------------------------------------------------------- From vxwexplo-errs@csg.lbl.gov Thu Sep 7 04:03:10 2000 From: Vxworks Exploder Date: Thu Sep 7 04:03:12 PDT 2000 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Thu Sep 7 04:03:05 PDT 2000 Subject: Why are Negative block sizes displayed by memShow & memFreeMax Subject: tcp/ip Subject: Subject: ANN: software engineering for real-time systems discussion group Subject: Clarification Subject: How to add properly new memory space to MMU Subject: re: Why are Negative block sizes displayed by memShow & memFreeMax Subject: Re: RTOS Comparison Subject: How to make a boot hard disk? Subject: Re: ppc asm: loading 32 bit value into CTR reg Subject: Re: How to work with PCI ? Subject: interrupt on pci bus Subject: Question on WindView 1.0 Subject: Linking archives Subject: messing up of nvram Subject: Re: Linking archives Subject: Re: How to make a boot hard disk? Subject: Re: Subject: Re: ppc asm: loading 32 bit value into CTR reg Subject: re: Device doesn't show up as an interface Subject: Re: How to add properly new memory space to MMU Subject: Re: use of cascaded macros Subject: Re: Problems with WindView under Windows 2000. Subject: how to use configure VxSim Subject: Re: PCMCIA (ATA card) and MBX821 Subject: Re: ppc asm: loading 32 bit value into CTR reg Subject: Re: ppc asm: loading 32 bit value into CTR reg Subject: gcc 2.95.2 Subject: Re: Partitioning and Formatting Subject: Re: Spawning C++ task Subject: Re: ppc asm: loading 32 bit value into CTR reg Subject: Re: ppc asm: loading 32 bit value into CTR reg Subject: Synch Driver for 8530 Subject: HIGH TECH - VxWorks S/W Developers Needed - Ottawa Subject: Re: dosFs2 chkdsk slow to run from bootrom Subject: Flash File system Subject: 64x vme chassis vendor? Subject: Re: ppc asm: loading 32 bit value into CTR reg Subject: Re: use of cascaded macros Subject: Re: dosFs2 chkdsk slow to run from bootrom Subject: Re: Linking archives Subject: Re: use of cascaded macros Subject: Re: use of cascaded macros Subject: Re: How to make a boot hard disk? Subject: Re: tcp/ip Subject: Re: Linking archives Subject: Registering Newsgroup Subject: Re: tcp/ip Subject: Migrating from Tornado1 project to Tornado2 Subject: Re: ppc asm: loading 32 bit value into CTR reg Subject: Re: ppc asm: loading 32 bit value into CTR reg Subject: bootrom/ftp w/cache failure Subject: how to serial? Subject: Re: how to serial? Subject: Spawning routine more than once? ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Why are Negative block sizes displayed by memShow & memFreeMax Date: Wed, 6 Sep 2000 12:07:07 +0100 From: "David Anderson @ Data Track" Organization: Lawrence Berkeley National Laboratory Message-ID: <8p58v7$c5m$1@overload.lbl.gov> vxWorks 5.3.1 & Tornado 1.0.1 / Motorola processor. Dear vxWorkers, We have a problem with memory management which someone must have seen before ;-). memFreeMax() is returning a negative number of around -22KBytes. To make the situation even stranger, memShow 1 from the host shell shows four FREE BLOCKS these are shown as: block 1 small positive value 100 Bytes block 2 larger positive value 500 Bytes block 3 large positive value 26K block 4 large negative number. -22K If we then dump the memory block header for block 4 it shows the size as FFFC 9899 or similar. If this is a signed long then this corresponds to a negative -0x36766 (not correcting for two's complement). - ------------------------------------------------------------------------ - ---------------------------------------- Q1/ Does this mean that the system has run out of malloc memory space ??. - ------------------------------------------------------------------------ - --------- Q2/ What is the significance of the negative sized block. Resulting Problems - ---------------------------- This completely destroys code that looks like this : if ( blocksize-memFreeMax() > required_later ) proceed_with_function() where memFreeMax() is assumed to be in the range 0-large positive value. David Anderson. mailto:danderson@dtrack.demon.co.uk - ----------------------------------------------------- Who said "There's no life here in outer space." - ----------------------------------------------------- Disclaimer - ---------- This email and any files transmitted with it are confidential and intended solely for the use of the individual or entity to whom they are addressed. If you have received this email in error please destroy and notify Data Track Technology Plc +44 1425 271900. - ------------------------------------------------------- --------------------------- Newsgroups: comp.os.vxworks Subject: tcp/ip Date: 6 Sep 2000 13:35:20 +0200 From: skrow@rediffmail.com ("shameek roy") Organization: Mailgate.ORG Server - http://www.Mailgate.ORG Message-ID: <20000906083316.5382.qmail@mailweb5.rediffmail.com> hai, iam just learning the basics... i haven't yet understood why would a realtime OS like VxWorks ever need to support tcp/ip stack. what would be the kind of applications that run on VxWorks and that require tcp/ip support? when VxWorks is an RTOS and as it also requires a host OS for prog. development, why does it ever need to support tcp/ip which doesn't provide any guarantees? please do provide some info or do direct me to appropriate sources. thanx in advance, skrow. //////////////////////////////////////////////////////// Heroes have an infinite capacity for stupidity. Thus are legends born! --Anonymous \\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\ _________________________________________________ Get Your Free Email At, http://www.rediffmail.com Partcipate in crazy Re.1 auctions at http://www.rediff.com/auctions - -- Posted from IDENT:qmailr@[202.54.124.150] via Mailgate.ORG Server - http://www.Mailgate.ORG --------------------------- Newsgroups: comp.os.vxworks Subject: Date: 6 Sep 2000 13:40:08 +0200 From: skrow@rediffmail.com ("shameek roy") Organization: Mailgate.ORG Server - http://www.Mailgate.ORG Message-ID: <20000906083415.27333.qmail@mailweb6.rediffmail.com> hai, iam just learning the basics... i haven't yet understood why would a realtime OS like VxWorks ever need to support tcp/ip stack. what would be the kind of applications that run on VxWorks and that require tcp/ip support? when VxWorks is an RTOS and as it also requires a host OS for prog. development, why does it ever need to support tcp/ip which doesn't provide any guarantees? please do provide some info or do direct me to appropriate sources. thanx in advance, skrow. //////////////////////////////////////////////////////// Heroes have an infinite capacity for stupidity. Thus are legends born! --Anonymous \\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\ _________________________________________________ Get Your Free Email At, http://www.rediffmail.com Partcipate in crazy Re.1 auctions at http://www.rediff.com/auctions - -- Posted from IDENT:qmailr@[202.54.124.151] via Mailgate.ORG Server - http://www.Mailgate.ORG --------------------------- Newsgroups: comp.os.vxworks Subject: ANN: software engineering for real-time systems discussion group Date: Wed, 06 Sep 2000 11:41:02 GMT From: Mark T Organization: Deja.com - Before you buy. Message-ID: <8p5agc$6hr$1@nnrp1.deja.com> Discussion group for all aspects of software engineering for real-time embedded, non-embedded and partially embedded systems. Group name: real-time software development (rtswdev) Main Page URL: http://www.egroups.com/group/RTSWDEV Discussion of requirements, analysis, and design should be from an implementation language independent view point. Discussion of mapping language independent design models (such as from UML) into an implementation language are most welcome. Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Clarification Date: Wed, 6 Sep 2000 17:38:36 +0530 From: "Ramachandramurthy, Bharat (Bharat)** CTR **" Organization: Lawrence Berkeley National Laboratory Message-ID: <8p5cfm$gtf$1@overload.lbl.gov> Can this question be answered:- 1. Does vxWorks support opening multiple instances of the same file. if yes then is there any provisions for LOCKING the file. If LOCKING is not supported then how do we prevent the file from being used by the other users when one user is manipulating it. if not does that mean no other user can operate on the file until and unless the current user finishes. If so could u explain how this is done . Can mail me back to bharathramchandra@yahoo.com/br41@lucent.com regards bharath --------------------------- Newsgroups: comp.os.vxworks Subject: How to add properly new memory space to MMU Date: Wed, 6 Sep 2000 15:23:46 +0300 From: "Ezra Koper" Organization: NetVision Israel Message-ID: <8p5cv9$s8v$1@news.netvision.net.il> Our target hardware is based 90% on regular PC and use Pentium2-333Mhz. In order to add a PCI device I had to do two steps: 1. Add new PCI device and configure its relevant PCI configuration registers by using VxWorks pciDevConfig() & pciConfigOutWord() functions. 2. Command the memory unit to open additional memory space for the new PCI memory window space. Failing to do so will cause to exception each time the software will try to access this memory area. Since I didn't find any MMU VxWorks function to add new memory window, I had to do it in the hard way. a. Add new element to MMU DB structure by calling sysMmuMapAdd() - it just add it to the structure sysPhysMemDesc[], nothing more. b. call vmBaseGlobalMapInit( sysPhysMemDesc,sysPhysMemDescNumEnt,TRUE); which initialize MMU from beginning. Since the source code for sysMmuMapAdd() is available I noticed that it use var sysPhysMemDescNumEnt to keep track of how many memory elements are valid inside sysPhysMemDesc[] structure array. After re initialization of the MMU I was able to access the PCI memory space of the device I have added. Question - As you can see my solution is far from elegant or correct, so I would like to know what is the correct way to add a new memory space to the memory management unit? Note - only basic MMU is activated ( INCLUDE_MMU_BASIC). --------------------------- Newsgroups: comp.os.vxworks Subject: re: Why are Negative block sizes displayed by memShow & memFreeMax Date: Wed, 6 Sep 2000 08:56:02 -0400 From: mmenge@dspt.com Organization: Lawrence Berkeley National Laboratory Message-ID: <8p5g07$i4e$1@overload.lbl.gov> I have seen this happen when the memory has been corrupted. Somewhere someone overwrote the memory block header. tornado vxworks --------------------------- Newsgroups: comp.os.vxworks Subject: Re: RTOS Comparison Date: Wed, 06 Sep 2000 18:43:26 +0530 From: shankar Organization: SSO-IT, Hewlett-Packard Co. Message-ID: <39B642F6.D1E822A8@india.hp.com> References: hi casey check the march 99 online version of ESP at www.embedded.com shankar Casey Ballentine wrote: > > Hello, > > I'm looking for an article or URL that compares several RTOS products. I've > found some by searching industry rags, but the articles have all been > industry specific (i.e. RTOSes in the Automotive Industry). What I'd like > is a more general comparison of RTOSes using several benchmarks (ROM and RAM > use, CPU usage, interrupt latency, etc.) Does anyone have a good reference? > > Thanks --------------------------- Newsgroups: comp.os.vxworks Subject: How to make a boot hard disk? Date: Wed, 6 Sep 2000 15:53:50 +0200 From: "zbody" Organization: Slovenija OnLine - SiOL Message-ID: Hi! I made a bootROM from the shell editing config.h first like "ata=0,0(0,0)..." and it's OK it goes to search for the image on the hard drive, renamed bootrom_uncmp to bootrom.dat. Then I tried with "mkboot c: bootrom.dat" from the floppy with VxWorks utils and it says at first : "Boot sector installed OK." and right after it "This program cannot be run in DOS mode. Failed to copy bootrom.dat to c:bootrom.sys" and freezes... If somebody knows how to do a hard disk bootable, I would be grateful ! Thanks! Darko --------------------------- Newsgroups: comp.sys.powerpc.tech,comp.os.vxworks Subject: Re: ppc asm: loading 32 bit value into CTR reg Date: 6 Sep 2000 07:15:32 -0700 From: andrew@user2.teleport.com (Andrew Klossner) Message-ID: <8p5ji4$irm$1@user2.teleport.com> References: <39B54808.B227A591@mindspring.com> > The bad news is that the versions of gcc that I have use the > construct addis/add (lis/ls in simplified format) so get the > constant wrong half the time. Getting a CTR value wrong is one > thing, sending the processor to completely the wrong address is > really bad news! The sequence addis r9, 0,value@ha addi r9,r9,value@l will do the right thing for any value because it uses "@ha" (highword adjusted) instead of "@h". Perhaps that's what your compiler is doing? "ls" is not a standard simplified mnemonic (per appendix F of the green book.) Where do you find it? -=- Andrew Klossner (andrew@teleport.com) --------------------------- Newsgroups: comp.os.vxworks Subject: Re: How to work with PCI ? Date: Wed, 6 Sep 2000 15:59:21 +0100 From: "Kai Mattern" Organization: Heidelberger Druckmaschinen AG Message-ID: <8p5ilv$dd7$1@news.hdpp.de> References: <8oif6g$5li$1@news.netvision.net.il> i dunno if this helps.....thats what i did to get my card running... first there seams to be a bug with the show routines. Go to sysLib.c and disable the ifdef like this /*#ifdef INCLUDE_SHOW_ROUTINES*/ # include "pci/pciConfigShow.c" /* display of PCI config space */ /*#endif*/ after this is done and your bsp is newly build all show routines should work (worked for me) the adressing its easy to do. The IO and MEM maped adresses are configured in ravenpci.c but thats only for me because i have a M603 board. But it should work the same for you just in another file. IO is starting at 0x80000000 but there is an offset of 0x01000000 so the real startadress is 0x81000000 (same for mem...real start is C1000000). Theses offsets are configured in mv2600.h (like before thats for my board and bsp). So i just added my configuration via pciConfigOut Long in sysLib.c. If you write now to 0x81000000 you will be on PCI IO adress 0x00 and if you write to adress 0xC1000000 you will be on PCI MEM adress 0x00. if you figure out how to configure the interrupt let me know..im working on that now for 3 days and cant get him running. "Ezra Koper" wrote in message news:8oif6g$5li$1@news.netvision.net.il... > Hello > > 1. > We are using a Pentium2 333Mhz PC platform. I need to be able to open memory > space (and later I/O space) with a PCI device connected to primary PCI Slot. > I have no experience with PCI & VxWorks, and would like to know how it can > be done. > > For example address 0x800000 on my target has to be mapped to address > 0x0000000 on primary PCI device. > So writing to address 0x800010 will actually generate writing to address > 0x10 on the primary PCI device. > > There is almost no ducementation on how to use PCI with VxWorks. Any idea > were i can find more information? > > 2. > I can't activate commands as sysPciDevShow. This command paper in Widriver > WindSurf WEB site for Technical Solution 130785.5 > so what might be the problem? > > Note - I have included the PCI support component at VxWorks. > > > --------------------------- Newsgroups: comp.os.vxworks Subject: interrupt on pci bus Date: Wed, 6 Sep 2000 16:07:30 +0100 From: "Kai Mattern" Organization: Heidelberger Druckmaschinen AG Message-ID: <8p5j57$ddl$1@news.hdpp.de> Hi, anybody know what to do to connect a routine to an PCI interrupt INT A on a mtx603 board ? i did intConnect(INUM_TO_IVEC(0x19), (VOIDFUNCPTR)my_isr, NULL); intEnable(0x19); the 0x19 is out of mv2600.h #else /* (MTX603/604) */ # define PMC_INT_LVL1 ( 0x09 + EXT_INTERRUPT_BASE ) where EXT_INT is 0x10. I thought if my card now do an interrupt my isr will answer...but it dont do anything... i tried numbers from 0x00 to 0x20. At 0x10 there is a lot of traffic. Seems LAN or SCSI is on that Interrupt. Anybody know what to do ? thanks a lot Kai --------------------------- Newsgroups: comp.os.vxworks Subject: Question on WindView 1.0 Date: Wed, 06 Sep 2000 16:25:58 +0200 From: Carsten Boeckmann Organization: Ericsson Eurolab Deutschland GmbH Message-ID: <39B653F6.E89E8AFF@eed.ericsson.se> Reply-To: Carsten.Boeckmann@eed.ericsson.se Hi VxWorkers, I am inspecting some tasks on our VxWorks system using WindView 1.0. Since I would like to inspect the events and interactions of just a very small set of the tasks running on our system, I want to switch off the other tasks in WindView's graphical presentation. I have in mind that this is already possible with WindView 2.0, but I didn't find a way on how to do this using WindView 1.0. Does somebody of you know if and how this can be done ? Thanks in advance, Carsten - -- ____________________________________________ -= Carsten.Boeckmann@eed.ericsson.se =- --------------------------- Newsgroups: comp.os.vxworks Subject: Linking archives Date: Wed, 6 Sep 2000 16:27:08 +0200 From: Gunter Ritter Organization: Tenovis GmbH u. Co. KG Message-ID: <8p5k76$90e$1@proxy2.fe.internet.bosch.de> Hello NG, and again a problem using Tornado. It might be a simple problem, but I can´t locate an error. Maybe the error is again sittung in front of the keyboard ;-). I have a bootable project and several downloadable projects. The output of the downloadable projects are archives like project1.a and project2.a. Functions in project2 are using functions in project1 (lets say there are library functions in project 1 for all other bootable projects like project2. The compilation of each project works fine. These archives have to be linked against the bootable VxWorks-Project. When linking the bootable project, the linker tells me, that there undefined references. These references are some functions in project1 that will be used by functions in project2. Other functions in project1 that are used by project2 don´t make any trouble. Moving the modules from project1 to project2 solves the problem, but this is not a solution for me.. Putting all files in one project is no alternative. This makes the project unreadable. What´s wrong now? Thanks for any help. GUnter --------------------------- Newsgroups: comp.os.vxworks Subject: messing up of nvram Date: Wed, 06 Sep 2000 14:30:52 GMT From: ddinker@my-deja.com Organization: Deja.com - Before you buy. Message-ID: <8p5kes$id7$1@nnrp1.deja.com> i have a very peculiar problem here...we have nvram on our board...when i take th bsp for the board , build it and download it...i am able to use it properly... but when i include by application it is unable to write to nvram as some chip selcts get messed up....i am not doing anything with the chip selects in the application... any idea what's happenning? any lead wil be appreciated... dinker Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Linking archives Date: 6 Sep 2000 15:04:08 GMT From: johan@borksoft.xs4all.nl Organization: Borksoft Message-ID: <8p5md8$40e$1@news1.xs4all.nl> References: <8p5k76$90e$1@proxy2.fe.internet.bosch.de> Gunter Ritter wrote: : Hello NG, : I have a bootable project and several downloadable projects. The output of : the downloadable projects are archives like project1.a and project2.a. : Functions in project2 are using functions in project1 (lets say there are : library functions in project 1 for all other bootable projects like : project2. : The compilation of each project works fine. : These archives have to be linked against the bootable VxWorks-Project. : When linking the bootable project, the linker tells me, that there undefined : references. These references are some functions in project1 that will be : used by functions in project2. Other functions in project1 that are used by : project2 don't make any trouble. The linking is done at the moment you build your complete image, so your bootable application. When you add yout project1.a to the LIBS in your bootable project they will be linked. You can remove the project1.a from your other projects. Groeten, Johan - -- o o o o o o o . . . ___________________________________ o _____ || Johan Borkhuis | .][__n_n_|DD[ ====_____ | johan@borksoft.xs4all.nl | >(________|__|_[_________]_|________________________________| _/oo OOOOO oo` ooo ooo 'o!o!o o!o!o` === VxWorks FAQ: http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html === --------------------------- Newsgroups: comp.os.vxworks Subject: Re: How to make a boot hard disk? Date: Wed, 6 Sep 2000 08:05:08 -0700 From: "Dan Pinson" Message-ID: <8p5mfb$c4tq3$1@ID-48686.news.cis.dfn.de> References: mkboot.bat invokes vxcopy.exe which insists on running in a DOS window. Here's what I did: On Windows workstation: 'vxcopy bootrom_uncmp a:bootrom.sys' On target booted to DOS (boot from a floppy if neccessary): 'copy /b a:bootrom.sys c:' Then you should run 'chkdsk c:\bootrom.sys' to ensure that it is contiguous. Regards, Dan "zbody" wrote in message news:r0st5.468$yb7.4751@news.siol.net... > Hi! > > I made a bootROM from the shell editing config.h first like > "ata=0,0(0,0)..." and it's OK it goes to search for the image on the hard > drive, renamed bootrom_uncmp to bootrom.dat. Then I tried with "mkboot c: > bootrom.dat" from the floppy with VxWorks utils and it says at first : "Boot > sector installed OK." and right after it "This program cannot be run in DOS > mode. > Failed to copy bootrom.dat to c:bootrom.sys" and freezes... > > If somebody knows how to do a hard disk bootable, I would be grateful ! > > Thanks! > > Darko > > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Date: Wed, 06 Sep 2000 11:02:59 -0700 From: Joe Durusau Organization: Lockheed Martin Corporation Message-ID: <39B686D3.127A4793@lmco.com> References: <20000906083415.27333.qmail@mailweb6.rediffmail.com> Real-time systems exist that do remote data logging or some such. While the data logging need to occur, it does not have to be fast, if you have big buffers or and afford to lose data from the log in some situations. TCP does provide (more-or-less) guaranteed delivery of messages, just not in any particular time, with using UDP does not guarantee delivery, but is faster. Some folks even have such things as web servers on vxWorks, to create easy operator interfaces, while the critical stuff goes on 'under the table' as it were. HTH, Joe Durusau shameek roy wrote: > > hai, > iam just learning the basics... > i haven't yet understood why would a realtime OS like VxWorks ever need to support tcp/ip stack. > what would be the kind of applications that run on VxWorks and that require tcp/ip support? > when VxWorks is an RTOS and as it also requires a host OS for prog. development, why does it ever need to support tcp/ip which doesn't provide any guarantees? > please do provide some info or do direct me to appropriate sources. > thanx in advance, > skrow. > > //////////////////////////////////////////////////////// > > Heroes have an infinite capacity for stupidity. Thus are legends born! > --Anonymous > > \\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\ > > _________________________________________________ > Get Your Free Email At, http://www.rediffmail.com > > Partcipate in crazy Re.1 auctions at http://www.rediff.com/auctions > > -- > Posted from IDENT:qmailr@[202.54.124.151] > via Mailgate.ORG Server - http://www.Mailgate.ORG --------------------------- Newsgroups: comp.sys.powerpc.tech,comp.os.vxworks Subject: Re: ppc asm: loading 32 bit value into CTR reg Date: Wed, 6 Sep 2000 17:27:19 +0200 From: laire@t-online.de (Ralph Schmidt) Organization: T-Online Message-ID: <8p5non$ukk$11$1@news.t-online.com> References: <39B54808.B227A591@mindspring.com> <8p5ji4$irm$1@user2.teleport.com> On 09/06/00, Andrew Klossner wrote: >> The bad news is that the versions of gcc that I have use the >> construct addis/add (lis/ls in simplified format) so get the >> constant wrong half the time. Getting a CTR value wrong is one >> thing, sending the processor to completely the wrong address is >> really bad news! > >The sequence > > addis r9, 0,value@ha > addi r9,r9,value@l > >will do the right thing for any value because it uses "@ha" (highword >adjusted) instead of "@h". Perhaps that's what your compiler is >doing? > >"ls" is not a standard simplified mnemonic (per appendix F of the >green book.) Where do you find it? > "lis" is a standard macro for addis Rn,0,value defined on the addis manual page. "li" is the same for addi Rn,0,value - -- Ralph Schmidt,laire@t-online.de(private),NextMail welcome --------------------------- Newsgroups: comp.os.vxworks Subject: re: Device doesn't show up as an interface Date: Wed, 6 Sep 2000 16:58:51 +0100 (BST) From: David Laight Organization: Lawrence Berkeley National Laboratory Message-ID: <200009061558.QAA07784@dsl-2.tadpole.co.uk> > I call muxDevLoad and get back a reasonable address. > I call muxDevStart and get OK > I see the driver registered via muxShow > however: > neither ipAttach nor ifAddrSet work with that new interface! > Did I forget something important here? Does 'endFindByName( "xyz", 0 ) work? The one in vxWorks 5.3.1 doesn't support multiple drivers! If it fails, patch in the following: END_OBJ * fixed_endFindByName( char *name, int unit ) { END_TBL_ROW *row; END_OBJ *end; extern int *muxDevExists; for (row = (void *)lstFirst( (void *)*(&muxDevExists-1) ); row; row = (void *)lstNext( &row->node )) { if (strcmp( name, row->name )) continue; for (end = (void *)lstFirst( &row->units ); end; end = (void *)lstNext( &end->node )) if (end->devObject.unit == unit) return end; } return 0; } Compile the above somewhere and then execute something like: extern int endFindByName; endFindByName = 0xea000000 | (((int *)&fixed_endFindByName - &endFindByName - 2) & 0xffffff); to patch a branch instruction over the first instruction of the old routine. (NB the above is an ARM branch, you will need to modify the constants for a different architecture.) David - ---------------------------------------------------------------- David Laight email: dsl@tadpole.co.uk Tadpole Technology plc phone: +44 1223 428 232 Cambridge, UK fax: +44 1223 428 201 --------------------------- Newsgroups: comp.os.vxworks Subject: Re: How to add properly new memory space to MMU Date: Wed, 06 Sep 2000 18:29:20 +0200 From: Gilbert Cabillic Organization: Irisa - Campus de Beaulieu - RENNES Message-ID: <39B670E0.CAA354D0@irisa.fr> References: <8p5cv9$s8v$1@news.netvision.net.il> Ezra Koper wrote: > > Question - As you can see my solution is far from elegant or correct, so I > would like to know what is the correct way to add a new memory space to the > memory management unit? > Note - only basic MMU is activated ( INCLUDE_MMU_BASIC). Well, it works. Your goal is to extend a one to one virtual memory mapping (equivalence for physical and virtual memory) for the memory of your device. To do that, you need to enable the access to the memory (that meens to create one or more valid page directory entries and the corresponding page tables). I don't know exactly how works the sysMmuMapAdd, but if this call doesn't manage the holes of empty memory, you have the risk to create a lot of non usable page directory entries and page tables for the paging management. So it can cost a lot of extra memory. The option is to do it manually. I give you some details, but it's a little bit far from my mind. See memory management chapter of the Intel documentation. First way : create your page tables to describe your device memory (use linear address). Take CR3 which point on the page directory. Add your entries using the page tables pointers you have created. Be careful to create a non cachable page entry (for consistency of the datas) and to create a page align pointer for pages. Second way : just disable the paging (clear the paging flag in CR0). Hope it helps, Gilbert. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: use of cascaded macros Date: Wed, 06 Sep 2000 16:26:52 GMT From: lvwhitebir@my-deja.com Organization: Deja.com - Before you buy. Message-ID: <8p5r7g$r5b$1@nnrp1.deja.com> References: <8p2lqt$3g9$1@proxy2.fe.internet.bosch.de> In article <8p2lqt$3g9$1@proxy2.fe.internet.bosch.de>, Gunter Ritter wrote: > Hello everybody, > > I have the problem that I get errors "macro recursion too deep" when > compiling a module that uses cascaded macros. > > The macros are defined as follows: > > #define K_MACRO_1 0 > #define K_MACRO_2 K_MACRO_1+1 > #define K_MACRO_3 K_MACRO_2+1 > #define K_MACRO_4 K_MACRO_3+1 > #define K_MACRO_5 K_MACRO_4+1 > : > : > What can I do to compile my code. It is not possible to change the macro > definition, becausethey are used in more than one project. > I didn´t find a hint in the documentation. > > Thanks. > > Gunter > > In my projects, we typically have a base and then an increment philosophy such that #define K_MACRO_1 0 #define K_MACRO_2 (K_MACRO_1 + 1) #define K_MACRO_3 (K_MACRO_1 + 2) ... Not as much flexibility when you need to delete entries, but most often, we don't change the numbers once they're assigned. We just stop using one or another, skipping it. This will help your current compiler problem, however, since there is very little recursion of the macros. HTH, Larry Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Problems with WindView under Windows 2000. Date: Wed, 6 Sep 2000 11:44:50 -0500 From: "fuzzy_duck" Organization: SkyPoint Communications, Inc. Message-ID: <8p5scb$2p06$1@shadow.skypoint.net> References: Install the windows 2000 support tools from the win2k CD. Run the application compatiblity tool and set the file tornado.exe to run as a Windows NT SP5 program. This appears to work, although I have not verified it in detail. You must have adminstrator privleges for it to stick. At least it gets rid of the messages. If you install Win2k SP1, another bunch of windows pop up saying RPCExit, extra call. Don't know what this is but unless you need SP1, I would recommend waiting until someone figures out how to fix that. Tools seem to run if you dismiss all the info boxes. Hey Windriver guys--lets get this fixed. "Simon DiMaio" wrote in message news:Pine.SUN.3.95.1000901211732.23944G-100000@salcudean.ece.ubc.ca... > Hi > > I have done a full install of Tornado 2.0 under Windows 2000. Upon running > Tornado, I receive the following message: > > couldn't read file "TRY=localhost\host\resource\tcl\WindView.win32.tcl": > no such file or directory > > Tornado runs fine, but WindView is not available. The base directory for > my Tornado installation is C:\Tornado, so I don't know where the > "TRY=localhost" has come from. > > Has anybody experienced a similar error message? > > Any ideas for a remedy would be greatly appreciated. > > Thanks in advance, > Simon DiMaio > > ------------------------------------------------------------------------ > Simon DiMaio Email: simond@ece.ubc.ca > Robotics and Control Group > Dept. of Electrical and Computer Engineering Phone: (604)822-9215 > University of British Columbia Fax : (604)822-5949 > --------------------------- Newsgroups: comp.os.vxworks Subject: how to use configure VxSim Date: Wed, 06 Sep 2000 17:34:27 GMT From: v_mathi_vanan@my-deja.com Organization: Deja.com - Before you buy. Message-ID: <8p5v72$93$1@nnrp1.deja.com> Hi, I have an application (Client server programe) which I want to compile and test it in VxWorks Simulator. I installed Tornado 2.0 and I did the following steps 1. Created a workspace and project. 2. Added all required files, and compiled them. No error. 3. Launched a simulator and didn't start default target server. 4. Before that I installed ULIP and configured the IP address as 90.0.0.245. My routing table is updated automatically. 5. I configured a target server with backend as wdppipe and target ip address as 90.0.0.245. 6. I launched the target server and the communcation between the target server and target agent (simulator) was ok. 7. Started shell 9. Downloaded my project image. No error. 10. I started my application from the shell. 11. Now the simulator has my server program. 12 I started my client program in NT dos prompt, but if I send a request from the client it is not reaching my server which is running in simulator. can any body tell me the reason why my request is not reached in my server program ? Do I need to configure anything more ? Your help is appreciated. Thanks Mathi. Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: PCMCIA (ATA card) and MBX821 Date: Wed, 06 Sep 2000 18:00:45 GMT From: ed.moore@snellwilcox.com Organization: Deja.com - Before you buy. Message-ID: <8p60nv$22f$1@nnrp1.deja.com> References: <39B4AB8B.C37C6325@sepro-robotique.com> You should be aware that the MBX821 board and the MPC821 processor have been End Of Line'd, ie obsoleted by Motorola. Ed. In article <39B4AB8B.C37C6325@sepro-robotique.com>, Emmanuel wrote: > Hi, > > We are working to connect a Compact-Flash card > (ATA controller and PCMCIA bus) with an evaluation > board called MBX821 (based on MPC 821). > > The BSP seems to be buggy (bad SIO support, PCMCIA > not tested, etc...). Does anyone own a debugged > BSP for such board with PCMCIA support ???? > > Thanks a lot. > Emmanuel. > Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.sys.powerpc.tech,comp.os.vxworks Subject: Re: ppc asm: loading 32 bit value into CTR reg Date: Wed, 06 Sep 2000 12:07:51 -0600 From: "Christopher S. Kush" Organization: Storage Technology Corporation Message-ID: <39B687F7.127D0666@ib.stortek.com> References: <39B54808.B227A591@mindspring.com> <8p5ji4$irm$1@user2.teleport.com> <8p5non$ukk$11$1@news.t-online.com> Ralph Schmidt wrote: > > On 09/06/00, Andrew Klossner wrote: > >Sometime, somebody else wrote: > > > The bad news is that the versions of gcc that I have use the > > > construct addis/add (lis/ls in simplified format) so get the > > > [snip] > > > >"ls" is not a standard simplified mnemonic (per appendix F of the > >green book.) Where do you find it? > > > > "lis" > is a standard macro for addis Rn,0,value defined on the addis manual > page. > > "li" > is the same for addi Rn,0,value You didn't answer his question. Since the original poster is a PowerPC neophyte, I expect they meant to type "addis/addi (lis/li ...". - -- Christopher S. Kush (303) 661 7509 kushcs@ib.stortek.com --------------------------- Newsgroups: comp.sys.powerpc.tech,comp.os.vxworks Subject: Re: ppc asm: loading 32 bit value into CTR reg Date: Wed, 06 Sep 2000 12:11:40 -0600 From: "Christopher S. Kush" Organization: Storage Technology Corporation Message-ID: <39B688DC.A8CD1507@ib.stortek.com> References: <39B54808.B227A591@mindspring.com> <8p5ji4$irm$1@user2.teleport.com> <8p5non$ukk$11$1@news.t-online.com> <39B687F7.127D0666@ib.stortek.com> "Christopher S. Kush" wrote: > > Ralph Schmidt wrote: > > > > On 09/06/00, Andrew Klossner wrote: > > >Sometime, somebody else wrote: > > > > The bad news is that the versions of gcc that I have use the > > > > construct addis/add (lis/ls in simplified format) so get the > > > > [snip] > > > > > >"ls" is not a standard simplified mnemonic (per appendix F of the > > >green book.) Where do you find it? > > > > > > > "lis" > > is a standard macro for addis Rn,0,value defined on the addis manual > > page. > > > > "li" > > is the same for addi Rn,0,value > > You didn't answer his question. > > Since the original poster is a PowerPC neophyte, I expect they meant > to type "addis/addi (lis/li ...". On second thought, probably "addis/add (lis/add..." To the original poster: You realize that there is no way to load an arbitrary 32 bit value into a register with one instruction. You have to load the lower sixteen and then add in (or in, whichever) the upper sixteen. The PowerPC Compiler Writer's Guide (get yourself a copy from IBM's web site) recommends using adds instead of ors, because future PowerPC implementations may have three-input adders. - -- Christopher S. Kush (303) 661 7509 kushcs@ib.stortek.com --------------------------- Newsgroups: comp.os.vxworks Subject: gcc 2.95.2 Date: Wed, 6 Sep 2000 11:53:32 -0700 From: "Ryan Tanner" Organization: Posted via Supernews, http://www.supernews.com Message-ID: Can I use the standard gcc release to with Tornado I? Do I really need to get the WindRiver's released compiler? Thanks --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Partitioning and Formatting Date: Wed, 06 Sep 2000 19:26:25 GMT From: Ted Dennison Organization: Deja.com - Before you buy. Message-ID: <8p65o8$8lm$1@nnrp1.deja.com> References: <8p3h4c$7b5$1@nnrp1.deja.com> In article <8p3h4c$7b5$1@nnrp1.deja.com>, Ted Dennison wrote: > Is there any way to partition and format a fresh ATA drive from > vxWorks? > > I know we can do it from DOS with a boot floppy. But I'm writing a > procedure for installing our software, and our customer will not > nessecarily have access to DOS unless we buy it for them (and add that > to the cost of the product). Well, I did find one gratis solution. There's a bootable disk image of GNU parted available at http://www.gnu.org/software/parted/ . If you download that and a GPL version of rawrite.exe that runs on NT (I found one at http://www.ualberta.ca/FTP/OpenBSD/2.5/tools/ ), you can supply your customers with a bootable disk partitioning tool. However, apparently we have already bought a copy of Win95 to cover this very eventuality. We'll probably just use fdisk from there, as its a lot more user-friendly. - -- T.E.D. http://www.telepath.com/~dennison/Ted/TED.html Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Spawning C++ task Date: Wed, 06 Sep 2000 19:47:14 +0100 From: molochai Message-ID: References: <968190933.274975@newsfeed> In message <968190933.274975@newsfeed> "George Malone" wrote: > > "G Kraemer" wrote in message > news:CkSq5.35744$P4.268715@news1.rdc1.az.home.com... > > VxWorks experts: > > > > I have a MVME-2700 using vxWorks 5.4 (T2) and I have added a C function, > > usrpgms() called from usrApplInit, which installs some of my own routines > > (file system date/time stamping, printer driver install, clock > > initializations, symFindByName(), and taskSpawn()). > > > > I have been delivered several C++ modules which are compiled with the > Green > > Hills 1.8.9 C++ compiler. This executable (65M) resides on an exported > > network drive which is nfs mounted to the MVME-2700. > > > > Here's my problem: usrpgms() can load the C++ executable into memory but > > when I attempt to spawn it, the C++ executable does not function > correctly. > > When I let usrpgms() load the C++ executable and taskSpawn from the > > target shell it functions correctly. > > > > Target shell execution: > > taskSpawn("niu_main",140,0x08,165000,(FUNCTPTR)symaddr) > > > > Can anyone provide me with sample code on how to spawn the C++ executable > > from my C function? > > > > Thanks in advance > > > > Garry > > > > > > int niu_main() // does not appear to be within a class { // C++ code > > .... } > > > > > Garry, indeed, this is a very mysterious symptom with an easy solution. > What's missing when you spawn the C++ function is its object pointer, i.e. > the "this" pointer. Absolutely correct about this. > Every C++ function assumes that the first argument on > the stack is the "this" pointer. Simply add the object pointer as an > argument after (FUNCTPTR)symaddr and your symaddr method will be happy. > But the above is definately *not* a reliable or portable was of achieving this - as has been pointed out on countless occasions myself and others. The method above *assumes* that every C++ implementation does indeed use the first argument 'slot' as the object pointer at a function call (note I didn't use the term 'stack location' as many processors pass arguments in registers wherever possible). It just so happens that this is the case on most architectures, but it needn't be so. You'll also find that GCC and most other decent compilers give a warning or error if you attempt to cast an implicit C++ member function pointer into a C function pointer - as they should. The correct and reliable way pick up an object pointer is with a small 'wrapper' function which is used as the task entry, and casts a parameter into an object pointer, upon which you can safely call class methods. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: ppc asm: loading 32 bit value into CTR reg Date: Wed, 06 Sep 2000 16:02:18 -0400 From: Al Johnston Organization: MindSpring Enterprises Message-ID: <39B6A2CA.6407B6B9@mindspring.com> References: <39B54808.B227A591@mindspring.com> <8p5ji4$irm$1@user2.teleport.com> <8p5non$ukk$11$1@news.t-online.com> <39B687F7.127D0666@ib.stortek.com> <39B688DC.A8CD1507@ib.stortek.com> > To the original poster: > > You realize that there is no way to load an arbitrary 32 bit value into a Yes, I am a ppc neophyte (g), but not quite that bad... re Compiler Writers Guide... yet another book!!! You think there would be an end to the list of books... but then again, its probably not a good idea to gripe about having TOO much doc... Thanks to all for the "help the dumb newbie" responses. --------------------------- Newsgroups: comp.sys.powerpc.tech,comp.os.vxworks Subject: Re: ppc asm: loading 32 bit value into CTR reg Date: Wed, 6 Sep 2000 22:18:56 +0200 From: laire@t-online.de (Ralph Schmidt) Organization: T-Online Message-ID: <8p68rg$6d3$11$1@news.t-online.com> References: <39B54808.B227A591@mindspring.com> <8p5ji4$irm$1@user2.teleport.com> <8p5non$ukk$11$1@news.t-online.com> <39B687F7.127D0666@ib.stortek.com> <39B688DC.A8CD1507@ib.stortek.com> On 09/06/00, "Christopher S. Kush" wrote: >"Christopher S. Kush" wrote: >> >> Ralph Schmidt wrote: >> > >> > On 09/06/00, Andrew Klossner wrote: >> > >"ls" is not a standard simplified mnemonic (per appendix F of the >> > >green book.) Where do you find it? >> > > >> > >> > "lis" >> > is a standard macro for addis Rn,0,value defined on the addis manual >> > page. >> > >> > "li" >> > is the same for addi Rn,0,value >> >> You didn't answer his question. >> >> Since the original poster is a PowerPC neophyte, I expect they meant >> to type "addis/addi (lis/li ...". > >On second thought, probably "addis/add (lis/add..." > Mhh..where wasn't i clear with my answer where lis is defined ? Confused... - -- Ralph Schmidt,laire@t-online.de(private),NextMail welcome --------------------------- Newsgroups: comp.os.vxworks Subject: Synch Driver for 8530 Date: Wed, 6 Sep 2000 13:20:31 -0700 From: "AG" Organization: SBC Internet Services Message-ID: I'm looking for a synch driver for the 8530 SCC. Any clues would be greatly appreciated. --------------------------- Newsgroups: comp.os.vxworks,comp.realtime Subject: HIGH TECH - VxWorks S/W Developers Needed - Ottawa Date: Wed, 6 Sep 2000 17:08:47 -0400 From: "PROCOM" Organization: IGS - Information Gateway Services Message-ID: <8p6bhn$cco$1@news.igs.net> Title: C++ / OOD VxWorks Software Developers Description: Our client, one of the world's largest telecommunications companies, currently requires 10 Junior to Senior Software Specialists to design, develop, implement and support Wireless products and applications. Minimum Requirements: All candidates should possess a minimum of 1 year experience developing real-time C/C++, OOD S/W on a VxWorks platform. Desired skills would include any previous Wireless development experience. Location: Ottawa, Ontario, Canada Type: Full-Time Placement Compensation: Open PROCOM: Established in 1978, Professional Computer Consultants Group Ltd. (Procom) is a national leader in the provision of Computer personnel on a contract and full-time basis. Our clients are comprised of the largest national and international corporations that utilize technical resources extensively across a wide range of disciplines. In the Financial Post (March 1999) Procom was ranked as the 6th largest professional Services Company in Canada. In November of 1999, Procom was named a Regional finalist in Canada' s 50 Best Managed Private companies. Our track record is proven with more than 180 consultants servicing Ottawa's high tech community and more than 1600 Procom consultants currently on assignment throughout. North America. For further information on this and other opportunities please visit our web site at www.procom.ca. Interested candidates are invited to forward their resumes or questions in confidence to: Derek Weber Technical Recruiter 300 March Road Kanata, Ontario K2K 2E2 phone: (613) 270-9339 ext 231 e-mail: derekw@procom.ca --------------------------- Newsgroups: comp.os.vxworks Subject: Re: dosFs2 chkdsk slow to run from bootrom Date: Wed, 06 Sep 2000 23:20:41 GMT From: andreyk@home.com (Andray Kaganovsky) Organization: @Home Message-ID: <8FA7C192Bandreykhomecom@24.2.9.61> References: <8p4osn$ja8$1@nnrp1.deja.com> paulvroberts@my-deja.com wrote in <8p4osn$ja8$1@nnrp1.deja.com>: >Hi there, > >Does anyone know why chkdsk would be much slower when run from the >bootrom, then when run after VxWorks is loaded? > >We are using FAT32 (singly dos partitioned) 9GB SCSI disks and it takes >65 seconds in the bootrom, whereas after VxWorks is running, it takes >just 3-4 seconds! > >What I did was to "enhance" the bootConfig.c file to allow booting using >dosFs2. I use a 128K cache, and call chkdsk via the icotl FIOCHKDSK. > >I am using the MVME2700 BSP with the bootrom_uncmp target, so presumably >that's not rom-resident. > >Any other ideas? > >Thanks, > >Paul Roberts. > > >Sent via Deja.com http://www.deja.com/ >Before you buy. > Could it be that CPU cache is disabled in the bootrom region ? --------------------------- Newsgroups: comp.os.vxworks Subject: Flash File system Date: Wed, 6 Sep 2000 16:36:09 -0700 From: "Randy Schafer" Message-ID: Can anyone recommend a Flash File System (other than TrueFFS) that will work with vxWorks? TFFS does not work properly with our configuration. Thanks - -- Randy Schafer Sr.Software Engineer Cinta Corportaion randy.schafer@cinta-corp.com --------------------------- Newsgroups: comp.os.vxworks Subject: 64x vme chassis vendor? Date: Wed, 06 Sep 2000 19:57:29 -0400 From: Al Johnston Organization: MindSpring Enterprises Message-ID: <39B6D9E9.48A82C15@mindspring.com> Anyone have any recommendations / good experiences with a vme64x chassis vendor? We have pretty much had it with our current vendor. A Johnston SOF EISE Tech Lead. - -------------------------------------------------------------------------- Lifetime warranty - A description of how long it will take to get it right. --------------------------- Newsgroups: comp.sys.powerpc.tech,comp.os.vxworks Subject: Re: ppc asm: loading 32 bit value into CTR reg Date: Thu, 07 Sep 2000 02:57:22 GMT From: "Martin Usher" Organization: EarthLink Inc. -- http://www.EarthLink.net Message-ID: References: <39B54808.B227A591@mindspring.com> <8p5ji4$irm$1@user2.teleport.com> <8p5non$ukk$11$1@news.t-online.com> <39B687F7.127D0666@ib.stortek.com> <39B688DC.A8CD1507@ib.stortek.com> Reply-To: "Martin Usher" > > > >"ls" is not a standard simplified mnemonic (per appendix F of the > > > >green book.) Where do you find it? I think this comes under the category 'typo'. I know what the original poster meant, Green Book or no Green Book. I don't bother with the simplified codes. Its not only more things to learn (and get wrong) but also the debuggers I have only show the standard forms when disassembling. > > You realize that there is no way to load an arbitrary 32 bit value into a > register with one instruction. You have to load the lower sixteen and > then add in (or in, whichever) the upper sixteen. > This goes for all RISCs that I know of except for one. Its just that commonsense would dictate that you have some simple and reliable mechanism to load constants to a register - maybe a load/loadhigh instruction pair? (Like the 29K) (The exception is the Hitachi processors. They're 32/64 bit units that are not dissimilar to the PowerPC, they've got 16 bit instructions and no load literal instructions at all.) > The PowerPC Compiler Writer's Guide (get yourself a copy from IBM's > web site) recommends using adds instead of ors, because future PowerPC > implementations may have three-input adders. > That's very nice of them but to mention this but I don't see the relevance at all to loading a literal. I suspect what's happening here is a bit of "old company obsurfurcation" (if you've ever had any significant experience of Token Ring networking (but not just TR networks) then you will know what I mean....). Speaking personally - and this is definitely "IHMO", no flames please - there's not going to be any future for the PowerPC processor in my world. I build embedded systems and I can't wait to put as much distance between this architecture and my products as possible. There are far, far, easier ways to achieve product. (I will grant everyone that once the thing is working it does go very fast for its clock speed, but in my world performance is not just a matter of how fast a processor executes numeric algorithms.) --------------------------- Newsgroups: comp.os.vxworks Subject: Re: use of cascaded macros Date: 06 Sep 2000 16:53:33 -1000 From: Jim Thomas Organization: Canada France Hawai`i Telescope Message-ID: References: <8p2lqt$3g9$1@proxy2.fe.internet.bosch.de> <39B4F533.F56D7B9B@swri.edu> <8p2vvs$hmq$1@proxy2.fe.internet.bosch.de> <39b54d5d.0@news> >>>>> "Richard" == Richard writes: Richard> try the following Richard> typedef enum Richard> { Richard> K_STRING_1, Richard> K_STRING_2, Richard> K_STRING_3, Richard> K_STRING_4, Richard> ...., Richard> NUMBER_K_STRINGS Richard> } K_STRING_DEF_TYPE; Richard> you can remove entries from the definition, and the following Richard> ones will be renumbered. I have not looked it up in ANSI C, but Richard> in all of the compilers I have used, the first enumberated value Richard> is 0, and each successive one is one larger than the previous Richard> one. These values are usable anywhere an int is. As a bonus, Richard> NUMBER_K_STRINGS can be used as the upper limit of the valid Richard> indices (all valid index values are less than NUMBER_K_STRINGS), Richard> and can be used to allocate your string table. But there is no longer any relationship between the values and the indices. I use exactly his macros very happily (well, with parens :-). char foo[] = { #define bar 0 "bar string", #define mumble (bar + 1) "mumble string", #define last (mumble + 1) (char *) NULL } .... assert((last + 1) == (sizeof foo / sizeof (char *))); Also to the poster who said "just use 0 1 2 3 4 ..." - baloney. He, I think from following postings, used numeric names in the example but uses real names in the actual code. The reason for symbols is exactly to eliminate meaningless numbers. As for the original problem, how many levels do you have? I don't see any way of controlling memory allocation with GNU cpp - maybe you need more swap space? --------------------------- Newsgroups: comp.os.vxworks Subject: Re: dosFs2 chkdsk slow to run from bootrom Date: Thu, 07 Sep 2000 04:54:13 GMT From: paulvroberts@my-deja.com Organization: Deja.com - Before you buy. Message-ID: <8p771e$fag$1@nnrp1.deja.com> References: <8p4osn$ja8$1@nnrp1.deja.com> <8FA7C192Bandreykhomecom@24.2.9.61> In article <8FA7C192Bandreykhomecom@24.2.9.61>, andreyk@home.com (Andray Kaganovsky) wrote: > paulvroberts@my-deja.com wrote in <8p4osn$ja8$1@nnrp1.deja.com>: > > >Hi there, > > > >Does anyone know why chkdsk would be much slower when run from the > >bootrom, then when run after VxWorks is loaded? > > >> > > Could it be that CPU cache is disabled in the bootrom region ? > Thanks for the reply Andray. Good idea, however, I added a cacheEnable(DATA_CACHE) to the bootConfig.c and no difference. Paul. Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Linking archives Date: Thu, 7 Sep 2000 07:28:48 +0200 From: Gunter Ritter Organization: Tenovis GmbH u. Co. KG Message-ID: <8p791k$gss$1@proxy2.fe.internet.bosch.de> References: <8p5k76$90e$1@proxy2.fe.internet.bosch.de> <8p5md8$40e$1@news1.xs4all.nl> Thanks for your reply, but thats excatly what I have done. I added all downloadable projects to LIBS of the bootable project to link them together. But the linker will not find all functions. I´ve been looking for a switch, but I could not find one. So I´ve put one module with some missing functions to the other project and linking has been successful. Gunter --------------------------- Newsgroups: comp.os.vxworks Subject: Re: use of cascaded macros Date: Thu, 7 Sep 2000 07:35:57 +0200 From: Gunter Ritter Organization: Tenovis GmbH u. Co. KG Message-ID: <8p79f2$h59$1@proxy2.fe.internet.bosch.de> References: <8p2lqt$3g9$1@proxy2.fe.internet.bosch.de> <8p5r7g$r5b$1@nnrp1.deja.com> Well, we used numbers in the project before and we changed it a new project. The compiler of this project (Mitsubishi-Compiler) has no problems with these cascaed macros. The problem is, that stopping using the table entries increases the usage of memory. And you never have enough memory in an embedded system. Now we use the code of the new project for another project with VxWorks, and the compiler doesn´t like these macros. Gunter --------------------------- Newsgroups: comp.os.vxworks Subject: Re: use of cascaded macros Date: Thu, 7 Sep 2000 07:37:47 +0200 From: Gunter Ritter Organization: Tenovis GmbH u. Co. KG Message-ID: <8p79ig$h6q$1@proxy2.fe.internet.bosch.de> References: <8p2lqt$3g9$1@proxy2.fe.internet.bosch.de> <39B4F533.F56D7B9B@swri.edu> <8p2vvs$hmq$1@proxy2.fe.internet.bosch.de> <39b54d5d.0@news> The problem begins at level ~400 of 800 levels. Gunter. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: How to make a boot hard disk? Date: Thu, 7 Sep 2000 08:18:57 +0300 From: "Ezra Koper" Organization: NetVision Israel Message-ID: <8p78ei$lcg$1@news.netvision.net.il> References: Hello . Hope this will help!! Creating Bootable Floppy from Boot-ROM · Open DOS window. · From DOS window execute tornado\host\x86-win32\bin\torvars (it will fix environment vars). · From the location of your boot-rom file (tornado\target\config\ ItoneCPU2_5\bootrom_uncmp) type mkboot a: bootrom_uncmp. The mkboot utility will create a bootable floppy disk Creating Bootable Hard Disk from Boot-ROM 1st stage will be to prepare a Hard Disk that will be suitable for VxWorks environment. Prepare a Floppy Disk with DOS6.2, FDISK & FORMAT files on it. Turn off your PC, remove your HD and place VxWorks HD instead. Boot your PC from DOS 6.2 Floppy Disk. Run FDISK Create FAT16 2Gbyte partition and set it to be the active partition. Exit from FDISK, reboot from the Floppy Disk and format the HD by using the FORMAT command from the Floppy Disk. Turn Off your PC, reconnect your original HD, and add VxWorks HD as a secondary HD (for example as disk D). Turn ON your PC (Windows NT should be activated). · Open DOS window. · From DOS window execute tornado\host\x86-win32\bin\torvars (it will fix environment vars). · From the location of your boot-rom file (tornado\target\config\ ItoneCPU2_5\bootrom_uncmp) copy bootrom_uncmp to file name bootrom.dat. · Type vxsys d: (assuming VxWorks HD is mapped as disk D). The vxsys.com installs a bootstrap loader that will be used to load bootrom.sys file. · Type vxcopy bootrom.dat d:bootrom.sys . vxcopy copy’s & converts the bootrom_uncmp file to the required format. zbody wrote in message news:r0st5.468$yb7.4751@news.siol.net... > Hi! > > I made a bootROM from the shell editing config.h first like > "ata=0,0(0,0)..." and it's OK it goes to search for the image on the hard > drive, renamed bootrom_uncmp to bootrom.dat. Then I tried with "mkboot c: > bootrom.dat" from the floppy with VxWorks utils and it says at first : "Boot > sector installed OK." and right after it "This program cannot be run in DOS > mode. > Failed to copy bootrom.dat to c:bootrom.sys" and freezes... > > If somebody knows how to do a hard disk bootable, I would be grateful ! > > Thanks! > > Darko > > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: tcp/ip Date: Thu, 07 Sep 2000 08:51:38 +0200 From: Emmanuel Organization: SEPRO Robotique Message-ID: <39B73AFA.E3FD6E58@sepro-robotique.com> References: <20000906083316.5382.qmail@mailweb5.rediffmail.com> shameek roy wrote: > i haven't yet understood why would a realtime OS like VxWorks ever need > to support tcp/ip stack. Hi, Many new embedded computers (robotics, etc...) use an ethernet adapter to transfer datas & programs to/from a PC (portable, or desktop). TCP/IP is a standard for such communication (remote management throught the Internet or a local network, for example). Althought, this new systems usually use file system (FAT16, FAT32, ...) over flash memory to store their datas (instead of raw EEPROM memory, for example). The TCP/IP stack includes FTP and NFS protocol for file sharing. Nowaday, it's possible to integrate a WEB Server in an embedded networked device : customers can manage their devices from a standard browser, without any other specifics softwares... Regards Emmanuel. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Linking archives Date: 7 Sep 2000 07:12:39 GMT From: johan@borksoft.xs4all.nl Organization: Borksoft Message-ID: <8p7f57$enc$1@news1.xs4all.nl> References: <8p5k76$90e$1@proxy2.fe.internet.bosch.de> <8p5md8$40e$1@news1.xs4all.nl> <8p791k$gss$1@proxy2.fe.internet.bosch.de> Gunter Ritter wrote: : Thanks for your reply, but thats excatly what I have done. I added all : downloadable projects to LIBS of the bootable project to link them together. : But the linker will not find all functions. I've been looking for a switch, : but I could not find one. So I've put one module with some missing functions : to the other project and linking has been successful. Have you checked the command line of the link command? All your libraries should be in there. You could also try to change the order of the libraries, put your global library at the end, just before the VxWorks libraries. Groeten, Johan - -- o o o o o o o . . . ___________________________________ o _____ || Johan Borkhuis | .][__n_n_|DD[ ====_____ | johan@borksoft.xs4all.nl | >(________|__|_[_________]_|________________________________| _/oo OOOOO oo` ooo ooo 'o!o!o o!o!o` === VxWorks FAQ: http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html === --------------------------- Newsgroups: comp.os.vxworks Subject: Registering Newsgroup Date: Thu, 07 Sep 2000 12:38:35 +0530 From: rndmail Organization: Lawrence Berkeley National Laboratory Message-ID: <8p7f95$1ti$1@overload.lbl.gov> Hi, I would like to register myself to Vxworks newsgroup. I got your info from one of my friends who is a member of your news group. My email id is: shah_jignesh_s@yahoo.com I hope you will do the needful in this matter. Lookinf forward to your reply, With regards, Jignesh Shah --------------------------- Newsgroups: comp.os.vxworks Subject: Re: tcp/ip Date: Thu, 07 Sep 2000 10:17:23 +0200 From: someone@siemens.de Organization: Siemens AG,(Hofmannstr) Munich-Germany-Europe. Message-ID: <39B74F13.D95A87AE@siemens.de> References: <20000906083316.5382.qmail@mailweb5.rediffmail.com> Hi! We at Siemens are developing toasters with an TCP/IP interface so you can see via HTML if the toasts are ready to eat ;-) Yes, we use it for networking purpose! We do not really need the "real time", but VxWorks is a small nice OS for network devices! shameek roy wrote: > hai, > iam just learning the basics... > i haven't yet understood why would a realtime OS like VxWorks ever need to support tcp/ip stack. > what would be the kind of applications that run on VxWorks and that require tcp/ip support? > when VxWorks is an RTOS and as it also requires a host OS for prog. development, why does it ever need to support tcp/ip which doesn't provide any guarantees? > please do provide some info or do direct me to appropriate sources. > thanx in advance, > skrow. > > //////////////////////////////////////////////////////// > > Heroes have an infinite capacity for stupidity. Thus are legends born! > --Anonymous > > \\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\ > > _________________________________________________ > Get Your Free Email At, http://www.rediffmail.com > > Partcipate in crazy Re.1 auctions at http://www.rediff.com/auctions > > -- > Posted from IDENT:qmailr@[202.54.124.150] > via Mailgate.ORG Server - http://www.Mailgate.ORG --------------------------- Newsgroups: comp.os.vxworks Subject: Migrating from Tornado1 project to Tornado2 Date: Thu, 07 Sep 2000 08:14:41 GMT From: Gaël Trinquart Organization: Deja.com - Before you buy. Message-ID: <8p7ip4$sqn$1@nnrp1.deja.com> Is there a way in Tornado2 to import Tornado1's projects based on makefiles ? Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.sys.powerpc.tech,comp.os.vxworks Subject: Re: ppc asm: loading 32 bit value into CTR reg Date: Thu, 7 Sep 2000 11:03:16 +0200 From: laire@t-online.de (Ralph Schmidt) Organization: T-Online Message-ID: <8p7lkk$kr7$13$1@news.t-online.com> References: <39B54808.B227A591@mindspring.com> <8p5ji4$irm$1@user2.teleport.com> <8p5non$ukk$11$1@news.t-online.com> <39B687F7.127D0666@ib.stortek.com> <39B688DC.A8CD1507@ib.stortek.com> On 09/07/00, "Martin Usher" wrote: >Speaking personally - and this is definitely "IHMO", no flames please - >there's not going to be any future for the PowerPC processor in my world. I >build embedded systems and I can't wait to put as much distance between this >architecture and my products as possible. There are far, far, easier ways to >achieve product. (I will grant everyone that once the thing is working it >does go very fast for its clock speed, but in my world performance is not >just a matter of how fast a processor executes numeric algorithms.) > Oh well...for every cpu you have to jump over a new learning border and getting a personal crisis about this simple addis,addi,addis,ori, @ha,@l reloc modes is ridiculous. Personally i prefer the value loading over any TOC/Smalldata mess for performance and cleanness reasons. And as you asked in your private email why the ppc has no non signed extended add you obviously missed other reloc16 combinations like this addis r3,0,Symbol@ha lwz r4,Symbol@l(r3) addis r3,0,Symbol@ha stw r4,Symbol@l(r3) would you want non sign extended loadstore offsets just to avoid the hidden handling of the sign bit ? - -- Ralph Schmidt,laire@t-online.de(private),NextMail welcome --------------------------- Newsgroups: comp.sys.powerpc.tech,comp.os.vxworks Subject: Re: ppc asm: loading 32 bit value into CTR reg Date: 07 Sep 2000 11:12:55 +0200 From: Holger Bettag Organization: Universitaet Bremen, Germany. Message-ID: <8xits8d0q0.fsf@s63.informatik.uni-bremen.de> References: <39B54808.B227A591@mindspring.com> <8p5ji4$irm$1@user2.teleport.com> <8p5non$ukk$11$1@news.t-online.com> <39B687F7.127D0666@ib.stortek.com> <39B688DC.A8CD1507@ib.stortek.com> "Martin Usher" writes: [someone else wrote:] > > The PowerPC Compiler Writer's Guide (get yourself a copy from IBM's > > web site) recommends using adds instead of ors, because future PowerPC > > implementations may have three-input adders. > > > > That's very nice of them but to mention this but I don't see the relevance > at all to loading a literal. I suspect what's happening here is a bit of > "old company obsurfurcation" (if you've ever had any significant experience > of Token Ring networking (but not just TR networks) then you will know what > I mean....). > Using adds can also be beneficial on the 603e and related cores, because some types of integer addition can be executed in the "system unit" in parallel with some other integer instruction in the integer unit. The newer PPC cores (and the good old 604 line) have two full-featured integer units, so it makes no difference there. Holger --------------------------- Newsgroups: comp.os.vxworks Subject: bootrom/ftp w/cache failure Date: Thu, 07 Sep 2000 09:46:03 GMT From: Michael Fainstein Organization: Deja.com - Before you buy. Message-ID: <8p7o4o$2pm$1@nnrp1.deja.com> Hi all! I am using Tornado2 with mpc860 target. I am using bootrom_res. This bootrom works fine if I build it without any cache, but if I turn on instruction cache it doesn't load application file using FTP: It is stuck in bootLoadModule, and on PC's FTP server I have error message about socket data errors. BTW, in my application I work with instruction cache enabled, and I don't have any problems. Any ideas what can be wrong? Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: how to serial? Date: Thu, 7 Sep 2000 19:21:06 +0900 From: "moony" Organization: System Engineering Research Institute (SERI) Message-ID: <8p7q8t$n6u$1@green.kreonet.re.kr> Thank you for reading this. I am programming application using serial communication. But, I don't know how to control serial. I succeed the job such as open(), read(), write(), and so on. However, I want to know how to set the flow control, parity bit, stop bit, etc. I need anyone's help. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: how to serial? Date: 7 Sep 2000 10:33:49 GMT From: borkhuis@lucent.com Organization: Borksoft Message-ID: <8p7qud$6i3$1@news1.xs4all.nl> References: <8p7q8t$n6u$1@green.kreonet.re.kr> moony wrote: : Thank you for reading this. : I am programming application using serial communication. : But, I don't know how to control serial. : I succeed the job such as open(), read(), write(), and so on. : However, I want to know how to set the flow control, parity bit, stop bit, : etc. You should use ioctl calls for that. Groeten, Johan - -- o o o o o o o . . . ___________________________________ o _____ || Johan Borkhuis | .][__n_n_|DD[ ====_____ | johan@borksoft.xs4all.nl | >(________|__|_[_________]_|________________________________| _/oo OOOOO oo` ooo ooo 'o!o!o o!o!o` === VxWorks FAQ: http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html === --------------------------- Newsgroups: comp.os.vxworks Subject: Spawning routine more than once? Date: Thu, 07 Sep 2000 12:46:34 +0200 From: Albert Wijnja Organization: MECO Equipment Engineers BV Message-ID: <39B7720A.43A4720D@meco.nl> Reply-To: Albert.Wijnja@meco.nl Hello, Currently we are doing a project with ACE/TAO on VxWorks. I found something strange. It concerns a server (notifier, with my_notifier as entry point) and multiple clients (consumer with my_consumer as entry point). The programs are started from the target shell via the spa-command. This command is a wrapper around taskSpawn to set the priority and stacksize. When I start it as follows (command result and output skipped): ld < notifier spa my_notifier ld < consumer spa my_consumer, "1" spa my_consumer, "2" spa my_consumer, "3" The second consumer seems to be running, however the ERRNO for this task is set to 0xC (ENOMEM) (via th i-command). The third consumer exits with an error (something about a socket connection). I found that when the consumer is loaded everytime before a new start, it works fine. When the commands are entered as follows: ld < notifier spa my_notifier ld < consumer spa my_consumer, "1" ld < consumer spa my_consumer, "2" ld < consumer spa my_consumer, "3" everything runs smoothley. Does anyone have an explanation for this behaviour? Albert --------------------------- End of New-News digest ********************** From vxwexplo-errs@csg.lbl.gov Fri Sep 8 04:03:10 2000 From: Vxworks Exploder Date: Fri Sep 8 04:03:13 PDT 2000 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Fri Sep 8 04:03:05 PDT 2000 Subject: Re: Spawning routine more than once? Subject: Negative returns from memFindMax() : Solution. Subject: Re: Linking archives Subject: Re: gcc asm / ppc Subject: telnet and tftp Subject: Re: Linking archives Subject: Re: how to use configure VxSim Subject: 1553 bus Subject: Re: use of cascaded macros Subject: Re: Linking archives Subject: Re: use of cascaded macros Subject: Re: use of cascaded macros Subject: Re: Problems with WindView under Windows 2000. Subject: Re: use of cascaded macros Subject: Re: rpclib - errno 0xd0003 Subject: Question about TFTP Subject: Re: PCMCIA (ATA card) and MBX821 Subject: Re: dosFs2 chkdsk slow to run from bootrom Subject: Re: 1553 bus Subject: Re: Linking archives Subject: WDB_COM_TYPE Subject: How to run a C++ object method from the Shell? Subject: Re: how to use configure VxSim Subject: configuring PC target??? Subject: Re: configuring PC target??? Subject: Re: use of cascaded macros Subject: signals on VxSim Subject: Re: tcp/ip Subject: Re: NFS woes Subject: Re: ppc asm: loading 32 bit value into CTR reg Subject: Re: use of cascaded macros Subject: Re: use of cascaded macros Subject: Re: use of cascaded macros Subject: Re: ppc asm: loading 32 bit value into CTR reg Subject: Re: tcp/ip Subject: How to download a modem driver to vxworks image Subject: Problem with VxWorks Ncr810 SCSI Driver Subject: Does Windows2000 support Tornado 2.0? Subject: Re: configuring PC target??? Subject: [Q]: DHCPC Subject: Re: Does Windows2000 support Tornado 2.0? Subject: HOT HOT - VxWorks S/W Developers Needed - Ottawa Subject: Re: use of cascaded macros Subject: Re: use of cascaded macros Subject: GUI Query Subject: Re: ricky Subject: Re: 1553 bus Subject: Re: configuring PC target??? Subject: Re: GUI Query Subject: Re: How to download a modem driver to vxworks image Subject: socket Subject: Re: Linking archives Subject: Re: use of cascaded macros Subject: Re: Device doesn't show up as an interface Subject: MPC8260 Subject: Re: socket Subject: Re: Device doesn't show up as an interface Subject: Re: 1553 bus ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: Spawning routine more than once? Date: 7 Sep 2000 11:05:58 GMT From: borkhuis@lucent.com Organization: Lucent Technologies, Nieuwegein NL Message-ID: <8p7sqm$99u$1@news1.xs4all.nl> References: <39B7720A.43A4720D@meco.nl> Albert Wijnja wrote: : Hello, : Currently we are doing a project with ACE/TAO on VxWorks. I found : something strange. It concerns a server (notifier, with my_notifier as : entry point) and multiple clients (consumer with my_consumer as entry : point). The programs are started from the target shell via the : spa-command. This command is a wrapper around taskSpawn to set the : priority and stacksize. When I start it as follows (command result and : output skipped): : ld < notifier : spa my_notifier : ld < consumer : spa my_consumer, "1" : spa my_consumer, "2" : spa my_consumer, "3" : The second consumer seems to be running, however the ERRNO for this task : is set to 0xC (ENOMEM) (via th i-command). The third consumer exits with : an error (something about a socket connection). : I found that when the consumer is loaded everytime before a new start, : it works fine. When the commands are entered as follows: : ld < notifier : spa my_notifier : ld < consumer : spa my_consumer, "1" : ld < consumer : spa my_consumer, "2" : ld < consumer : spa my_consumer, "3" : everything runs smoothley. This means that your consumer task is using some static/global data. In the first example you are executing my_consumer in the same memory space, so overwriting the static/global data. In the second example each instance of my_consumer has its own memory space, as they are loaded and linked seperately. How can you solve this? Make your task reentrant. You are using global or static data. Instead you should be using normal variables or malloc-ed memory areas. Groeten, Johan - -- o o o o o o o . . . ______________________________ o _____ || Johan Borkhuis | .][__n_n_|DD[ ====_____ | borkhuis@lucent.com | >(________|__|_[_________]_|___________________________| _/oo OOOOO oo` ooo ooo 'o!o!o o!o!o` === VxWorks FAQ: http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html === --------------------------- Newsgroups: comp.os.vxworks Subject: Negative returns from memFindMax() : Solution. Date: Thu, 7 Sep 2000 11:42:36 +0100 From: "David Anderson @ Data Track" Organization: Lawrence Berkeley National Laboratory Message-ID: <8p7tb6$7is$1@overload.lbl.gov> vxWorks Tornado Thanks to those who provided hints and tips. BSP+Linker problem as explained below. PROBLEM & SOLUTION FOR THE RECORD: >memFreeMax() is returning a negative number of around -22KBytes. > >To make the situation even stranger, memShow 1 from the host shell shows four FREE BLOCKS >these are shown as: > >block 1 small positive value 100 Bytes >block 2 larger positive value 500 Bytes >block 3 large positive value 26K >block 4 large negative number. -22K > >If we then dump the memory block header for block 4 it shows the size as FFFC 9899 or similar. If this is a signed >long then this corresponds to a negative -0x36766 (not correcting for two's complement). System is working in all other respects. INVESTIGATION & SOLUTION: Investigation using advice from this group provided the following: 1) Negative block size is actually a very large positive value. 2) Using memFreeMax() immediately after initialising the memory pool also indicates a large negative number. 3) The pool size passed into the initialisation routine was found to be a very large positive value, so large that it is displayed as a negative number by the shell ( which treats it as an int !). 4) The pool size is passed to the usrRoot task from KernelInit(...) which takes a start and end address for the pool... 5) KernelInit does a (start_address - end_address) to calculate the pool size for usrRoot(..) WITHOUT any checking. 6) The global variable space in the system had increased so that the first free ram address was now after sysMemTop() !. Hence calculated pool size goes to a negative int value ( == large positive unsigned ). (but well below sysPhysMemTop() ) SOLUTION... 7) Increasing LOCAL_MEM_SIZE so that sysMemTop() is increased above the first free memory address cures the problem. Therefore if you ever see a negative block size with memFindMax() or memShow() check your kernelInit memory pool parameters are valid. One ( or more ) of these are assigned by the linker..... . BSP MODIFICATIONS I've now added explicit checking of the memory pool length in usrRoot() which range check the memory pool size parameter. /* Initialize the memory pool before initializing any other package. * The memory associated with the root task will be reclaimed at the * completion of its activities. */ if ( (long)memPoolSize < 0 ) /* checking for overflow into sign bit */ { errnoSet(EDOM); // Out of range. taskSuspend(0); // can be seen with debug tools. } memInit (pMemPoolStart, memPoolSize ); /* initialize memory pool */ David Anderson. mailto:danderson@dtrack.demon.co.uk - ----------------------------------------------------- Who said "There's no life here in outer space." - ----------------------------------------------------- Disclaimer - ---------- This email and any files transmitted with it are confidential and intended solely for the use of the individual or entity to whom they are addressed. If you have received this email in error please destroy and notify Data Track Technology Plc +44 1425 271900. - ------------------------------------------------------- --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Linking archives Date: Thu, 7 Sep 2000 14:01:46 +0200 From: Gunter Ritter Organization: Tenovis GmbH u. Co. KG Message-ID: <8p802g$gjl$1@proxy2.fe.internet.bosch.de> References: <8p5k76$90e$1@proxy2.fe.internet.bosch.de> <8p5md8$40e$1@news1.xs4all.nl> <8p791k$gss$1@proxy2.fe.internet.bosch.de> <8p7f57$enc$1@news1.xs4all.nl> Thanks for your help. VxWorks and Tornado are quite new for me. Okay, the command line of the link command seems to be ok, all archives are present. They are linked to partialimage.o I changed the order of the archives from project1.a project2.a to project2.a project1.a. Well, the problem has changed a little bit. Now some functions in project 2 that are referenced in project1 are marked as unreferenced by the linker. What do I have to do to make it independent of the link order? Gunter --------------------------- Newsgroups: comp.os.vxworks Subject: Re: gcc asm / ppc Date: Thu, 7 Sep 2000 12:55:51 +0100 From: "Dave Korn" Organization: (Posted via) U-NET Internet Ltd. Message-ID: References: <39B50E6A.294016D2@mindspring.com> <39B530E8.9448D267@tech-m4.com> <39B540A1.AFEFA0A6@mindspring.com> Al Johnston wrote in message <39B540A1.AFEFA0A6@mindspring.com>... > >> The asm synthax is described in the GNU toolKit user guide section 3.32. > >I wish it was. That section starts out "...you can now specify the >operands of the instruction using c expressions." It is documenting >the use of the ":" fields as an addition to the simpler case that does >not use that feature. The basic use of asm is not covered. I wish it >was. > >> asm("addi 1,1,0"); > >Your kidding... no your not! it works just as you said. I tried %r1, >r1, >p1, %p1, gpr1,... but never just "1"... arrg.... THANKS There's a gcc option, -mregnames, that is *supposed* to make the assembler accept the 'r' in front of the register numbers. I say supposed, because it didn't work when I tried it, but YMMV. It may be needed as a parameter to the assembler, so if -mregnames doesn't work, perhaps - -Wa,a,-mregnames (or however the syntax goes) would do the trick. DaveK - -- They laughed at Galileo. They laughed at Copernicus. They laughed at Columbus. But remember, they also laughed at Bozo the Clown. --------------------------- Newsgroups: comp.os.vxworks Subject: telnet and tftp Date: Thu, 07 Sep 2000 15:11:32 +0300 From: Ofer Goren Organization: Bay Networks, Inc. Message-ID: <39B785F4.AD196CC8@nortelnetworks.com> Hi. I have a problem, and I can use some help. I'm using Tor 1.0.1 for PPC740. Here is the problem: 1.I'm connecting to my target with a serial connection. 2. I'm starting a tftp session from a remote host to my local target. 3. while the tftp session is in progress (!!!), I'm opening a telnet session from a remote host to my target. 4. The "serial" connection is halted (like it should be). 5. I'm exiting my telnet session to return to my serial connection. 6. When I'm using "i" to view my tasks, I see a tftp task in the task list. 7. If I'm doing the same over and over again, number of tftp tasks are listed in my task list, and they never end. I can protect my tftp sessions using shellLock(), so no telnet session will interrupt them, But I don't like this solution. I also interesting in understanding the cause of such behaviour. Any clue? Thanks, - -- Embedded SW Team Leader Email : ogoren@nortelnetworks.com Phone number: +9723-6456023 Fax number : +9723-6479579 ESN : 826-6023 http://www.nortelnetworks.com - ---------------------------------------------------- If you lend someone $20, and never see that person again, it was probably worth the investment. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Linking archives Date: Thu, 07 Sep 2000 13:21:02 +0100 From: Hugh Duggan Organization: Lucent Technologies, Chippenham, United Kingdom Message-ID: <39B7882E.BBFB0BF5@lucent.com> References: <8p5k76$90e$1@proxy2.fe.internet.bosch.de> <8p5md8$40e$1@news1.xs4all.nl> <8p791k$gss$1@proxy2.fe.internet.bosch.de> <8p7f57$enc$1@news1.xs4all.nl> <8p802g$gjl$1@proxy2.fe.internet.bosch.de> Gunter Ritter wrote: > > Thanks for your help. VxWorks and Tornado are quite new for me. > > Okay, the command line of the link command seems to be ok, all archives are > present. They are linked to partialimage.o > > I changed the order of the archives from project1.a project2.a to project2.a > project1.a. > Well, the problem has changed a little bit. Now some functions in project 2 > that are referenced in project1 are marked as unreferenced by the linker. > > What do I have to do to make it independent of the link order? > > Gunter It sounds to me as if you have both archives dependent on each other. If I am correct, then this will _never_ work [no matter what operating system and compiler you use]. You will have to do one of the following: a] Combine project1.a and project2.a into a single archive. or b] Redesign your code so that the two modules are not mutually dependent. I have found the book "Large Scale C++ Software Design" by John Lakos to be very helpful in eliminating dependencies... Hugh - ----------------------------------------------------------- Hugh Duggan | Phone : +44 1666 832118 Lucent Technologies | FAX : +44 1666 832925 Chippenham, UNITED KINGDOM | E-mail : hduggan@lucent.com --------------------------- Newsgroups: comp.os.vxworks Subject: Re: how to use configure VxSim Date: Thu, 07 Sep 2000 12:27:07 GMT From: hruesing@my-deja.com Organization: Deja.com - Before you buy. Message-ID: <8p81i9$cr8$1@nnrp1.deja.com> References: <8p5v72$93$1@nnrp1.deja.com> In article <8p5v72$93$1@nnrp1.deja.com>, v_mathi_vanan@my-deja.com wrote: > Hi, > > I have an application (Client server programe) which I want to compile > and test it in VxWorks Simulator. I installed Tornado 2.0 and I did the > following steps > > 1. Created a workspace and project. > 2. Added all required files, and compiled them. No error. > 3. Launched a simulator and didn't start default target server. > 4. Before that I installed ULIP and configured the IP address as > 90.0.0.245. My routing table is updated automatically. > 5. I configured a target server with backend as wdppipe and target ip > address as 90.0.0.245. > 6. I launched the target server and the communcation between the target > server and target agent (simulator) was ok. > 7. Started shell > 9. Downloaded my project image. No error. > 10. I started my application from the shell. > 11. Now the simulator has my server program. > 12 I started my client program in NT dos prompt, but if I send a request > from the client it is not reaching my server which is running in > simulator. > > can any body tell me the reason why my request is not reached in my > server program ? > > Do I need to configure anything more ? > > Your help is appreciated. > > Thanks > Mathi. > > Sent via Deja.com http://www.deja.com/ > Before you buy. > If you are speaking from the "full" simulator (not included in standart T2) you have to config your target server as wdbrpc. Hope this helps. Harud Rüsing Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: 1553 bus Date: Thu, 07 Sep 2000 12:27:23 GMT From: S Austin Organization: Deja.com - Before you buy. Message-ID: <8p81io$crb$1@nnrp1.deja.com> Is it possible to comparison shop for 1553 support for vxWorks or is Compware Corp. the only game in town? Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: use of cascaded macros Date: Thu, 07 Sep 2000 12:28:51 GMT From: lvwhitebir@my-deja.com Organization: Deja.com - Before you buy. Message-ID: <8p81lg$cs3$1@nnrp1.deja.com> References: <8p2lqt$3g9$1@proxy2.fe.internet.bosch.de> <8p5r7g$r5b$1@nnrp1.deja.com> <8p79f2$h59$1@proxy2.fe.internet.bosch.de> In article <8p79f2$h59$1@proxy2.fe.internet.bosch.de>, Gunter Ritter wrote: > Well, we used numbers in the project before and we changed it a new project. > The compiler of this project (Mitsubishi-Compiler) has no problems with > these cascaed macros. > The problem is, that stopping using the table entries increases the usage of > memory. And you never have enough memory in an embedded system. > > Now we use the code of the new project for another project with VxWorks, and > the compiler doesn´t like these macros. > > Gunter > > I'm not sure I understand your problem that well, but you seem to have two solutions. For the first, you can rewrite the code so that the macros aren't nested so deep (if you post a longer explanation of why you do it this way someone might have a more portable work-around). My misunderstanding is from the statement "The problem is, that stopping using the table entries increases the usage of memory." Second, you can use a different compiler. You might ask in a NG specific to your compiler if there is some sort of switch that you can use to increase the nesting. Going to a different compiler, however, is a short-term solution. Without modifying the code, you'll probably experience this kind of pain again in the future. HTH, Larry Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Linking archives Date: Thu, 7 Sep 2000 14:45:04 +0200 From: Gunter Ritter Organization: Tenovis GmbH u. Co. KG Message-ID: <8p82jm$jgs$1@proxy2.fe.internet.bosch.de> References: <8p5k76$90e$1@proxy2.fe.internet.bosch.de> <8p5md8$40e$1@news1.xs4all.nl> <8p791k$gss$1@proxy2.fe.internet.bosch.de> <8p7f57$enc$1@news1.xs4all.nl> <8p802g$gjl$1@proxy2.fe.internet.bosch.de> <39B7882E.BBFB0BF5@lucent.com> You are right. These archives depend on each other. The code I have to use can not be redesigned. I want to structure the source files in several projects and link them together to a bootable project. Solution b) is not possible, because the code is used in another project, too. How can I combine the archives as proposed in solution a)? Is there a different solution to link the object files (or any ohter kind of output) of several projects to a bootable project? Gunter --------------------------- Newsgroups: comp.os.vxworks Subject: Re: use of cascaded macros Date: Thu, 7 Sep 2000 15:20:28 +0200 From: Gunter Ritter Organization: Tenovis GmbH u. Co. KG Message-ID: <8p84m2$m2g$1@proxy2.fe.internet.bosch.de> References: <8p2lqt$3g9$1@proxy2.fe.internet.bosch.de> <8p5r7g$r5b$1@nnrp1.deja.com> <8p79f2$h59$1@proxy2.fe.internet.bosch.de> <8p81lg$cs3$1@nnrp1.deja.com> I will try it again. The code we have shall be used for several software packages for an embedded system. For the first package we define the text strings for the display in a table (about 800 strings). To access the strings in the table we use the constants (see my first posting). The next package is based on the first package. But some text strings are changed (no problem), inserted or deleted. If we define constants like #define K_MACRO_1 0 #define K_MACRO_2 1 #define K_MACRO_3 2 #define K_MACRO_4 3 we have to change many of the defines. The string table needs more and more memory if we leave entries in the table that are no more used. Now we have started a new project using a new hardware platform with the ARM9 processor using VxWorks, Tornado 2.0.1and the gnu-compiler delivered by WindRiver. The code shall not be redesigned and be used as it is (as far as possible). The code can be compiled without errors with a different compiler (Mitsubishi). So it should be possible to compile the code for ARM9. Maybe I just have to set a switch for the gnu-compiler. Gunter --------------------------- Newsgroups: comp.os.vxworks Subject: Re: use of cascaded macros Date: Thu, 07 Sep 2000 08:34:55 -0500 From: Greg Willden Organization: Posted via Supernews, http://www.supernews.com Message-ID: <39B7997E.6F7DE630@swri.edu> References: <8p2lqt$3g9$1@proxy2.fe.internet.bosch.de> <39B4F533.F56D7B9B@swri.edu> <8p2vvs$hmq$1@proxy2.fe.internet.bosch.de> <39b54d5d.0@news> This is a multi-part message in MIME format. - --------------70218FDEEC3F3DAFD88B559A Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Jim Thomas wrote: > >>>>> "Richard" == Richard writes: > > Richard> try the following > Richard> typedef enum > Richard> { > Richard> K_STRING_1, > Richard> K_STRING_2, > Richard> K_STRING_3, > Richard> K_STRING_4, > Richard> ...., > Richard> NUMBER_K_STRINGS > Richard> } K_STRING_DEF_TYPE; > > Richard> you can remove entries from the definition, and the following > Richard> ones will be renumbered. I have not looked it up in ANSI C, but > Richard> in all of the compilers I have used, the first enumberated value > Richard> is 0, and each successive one is one larger than the previous > Richard> one. These values are usable anywhere an int is. As a bonus, > Richard> NUMBER_K_STRINGS can be used as the upper limit of the valid > Richard> indices (all valid index values are less than NUMBER_K_STRINGS), > Richard> and can be used to allocate your string table. > > But there is no longer any relationship between the values and the > indices. I use exactly his macros very happily (well, with parens :-). I don't understand what you mean about losing the relationship between values and the indices. Using the enumerated type K_STRING_1 will have the value of 0, K_STRING_2 will have the value of 1, etc... Just like the original macros. > Also to the poster who said "just use 0 1 2 3 4 ..." - baloney. He, I > think from following postings, used numeric names in the example but uses > real names in the actual code. The reason for symbols is exactly to > eliminate meaningless numbers. Yes that was baloney. I retract my previous statement. :) I say the best, most portable way of doing this is to use an enumerated type as has been mentioned. - --------------70218FDEEC3F3DAFD88B559A Content-Type: text/x-vcard; charset=us-ascii; name="gwillden.vcf" Content-Transfer-Encoding: 7bit Content-Description: Card for Greg Willden Content-Disposition: attachment; filename="gwillden.vcf" begin:vcard n:Willden;Greg tel;fax:(210) 522-2709 tel;work:(210) 522-6241 x-mozilla-html:FALSE org:Southwest Research Institute;www.swri.edu version:2.1 email;internet:gwillden@swri.edu title:Engineer adr;quoted-printable:;;6220 Culebra Rd=0D=0A=0D=0A;San Antonio, TX 78238;;; fn:Greg Willden end:vcard - --------------70218FDEEC3F3DAFD88B559A-- --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Problems with WindView under Windows 2000. Date: Thu, 7 Sep 2000 07:08:17 -0700 From: "Charles Krinke" Message-ID: <3eNt5.1$592.71@news.connectnet.com> References: I have found a similar problem just trying to go "cd\torrnado\host\x86-win32" with w2k. It appears to be an issue with DOS windows. "cd\tornado\host\x86-wi~1" works, but that is just exercising microsoft's hack. Apparently, gates&co crippled DOS windows a bit more then in previous versions of their software. "Simon DiMaio" wrote in message news:Pine.SUN.3.95.1000901211732.23944G-100000@salcudean.ece.ubc.ca... > Hi > > I have done a full install of Tornado 2.0 under Windows 2000. Upon running > Tornado, I receive the following message: > > couldn't read file "TRY=localhost\host\resource\tcl\WindView.win32.tcl": > no such file or directory > > Tornado runs fine, but WindView is not available. The base directory for > my Tornado installation is C:\Tornado, so I don't know where the > "TRY=localhost" has come from. > > Has anybody experienced a similar error message? > > Any ideas for a remedy would be greatly appreciated. > > Thanks in advance, > Simon DiMaio > > ------------------------------------------------------------------------ > Simon DiMaio Email: simond@ece.ubc.ca > Robotics and Control Group > Dept. of Electrical and Computer Engineering Phone: (604)822-9215 > University of British Columbia Fax : (604)822-5949 > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: use of cascaded macros Date: Thu, 7 Sep 2000 08:16:43 -0600 From: "Richard" Message-ID: <39b7a2b4.0@news> References: <8p2lqt$3g9$1@proxy2.fe.internet.bosch.de> <39B4F533.F56D7B9B@swri.edu> <8p2vvs$hmq$1@proxy2.fe.internet.bosch.de> <39b54d5d.0@news> "Jim Thomas" wrote in message news:wwn1hknc9e.fsf@atlas.cfht.hawaii.edu... > >>>>> "Richard" == Richard writes: > > Richard> try the following > Richard> typedef enum > Richard> { > Richard> K_STRING_1, > Richard> K_STRING_2, > Richard> K_STRING_3, > Richard> K_STRING_4, > Richard> ...., > Richard> NUMBER_K_STRINGS > Richard> } K_STRING_DEF_TYPE; > > Richard> you can remove entries from the definition, and the following > Richard> ones will be renumbered. I have not looked it up in ANSI C, but > Richard> in all of the compilers I have used, the first enumberated value > Richard> is 0, and each successive one is one larger than the previous > Richard> one. These values are usable anywhere an int is. As a bonus, > Richard> NUMBER_K_STRINGS can be used as the upper limit of the valid > Richard> indices (all valid index values are less than NUMBER_K_STRINGS), > Richard> and can be used to allocate your string table. > > But there is no longer any relationship between the values and the > indices. I use exactly his macros very happily (well, with parens :-). But they do retain that relationship, The K_STRING_1 has a value of 0, K_STRING_2 = 1, and so on. This is what allows one to be deleted from both the table, and the enumerated type, and still retain the same relationship. This way of defining the values results in the exact same values he originally had without the problem of nesting defines, and without the need to change the relationship of the defines when a string is removed/added. As with the defines, it does require that the enumerated type be updated anytime the table is changed. - -- rrapier sieversinst To reply to me, replace the space with @, and add the .com on the end. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: rpclib - errno 0xd0003 Date: Thu, 7 Sep 2000 07:29:32 -0700 From: "Charles Krinke" Message-ID: References: <39B5B563.D0E3CB51@mindspring.com> Dear Al: There is a header file that decodes these cryptic numbers. I believe errno.h may help. I have seen the infamous 0xd00003 error many times and basically it occurs whenever the network stack cannot get a file it thinks it needs. It can occur from something as simple as unplugging the ethernet cable. I have seen it as I got my ethernet driver functioning for the AMD 973 over the past couple of months. I posted viertually your same question 3 months ago and someone pointed me to errno.h then. "Al Johnston" wrote in message news:39B5B563.D0E3CB51@mindspring.com... > Anybody know why the rpcLib routines seem to like to > set errno? I know errno is not valid except when an > error has been detected... still, its annoying! > > Anybody know? none of the other lib's seem to "miss" > behave like this. > > -al --------------------------- Newsgroups: comp.os.vxworks Subject: Question about TFTP Date: Thu, 07 Sep 2000 11:27:27 -0400 From: JB Organization: none really Message-ID: <39B7B3DE.1AB470DE@my-deja.com> Hi, We have the VxWorks 5.4 running on a MV3604 card. We are transferring a few dozen files via TFTP during code initialization. After a few files have been transferred, the socket counter in the WFTP32.EXE window jumps to three and the next file transfer is delayed until the socket counter goes back to 1. The point in the code initialization that this problem happens is very consistent. Is this a VxWorks problem? Or is there any FTP parameters that need to be tweaked for this to work? Thanks in advance, JB --------------------------- Newsgroups: comp.os.vxworks Subject: Re: PCMCIA (ATA card) and MBX821 Date: Thu, 7 Sep 2000 09:44:23 +0100 From: "Jeremy Pye" Organization: Customer of Energis Squared Message-ID: References: <39B4AB8B.C37C6325@sepro-robotique.com> We use Intoto PCMCIA stack and client drivers (ethernet, ata, modem). Works well. We dumped our 821 because of cache not working. Changed to 860 C.1 + instead. Emmanuel wrote in message news:39B4AB8B.C37C6325@sepro-robotique.com... > Hi, > > We are working to connect a Compact-Flash card > (ATA controller and PCMCIA bus) with an evaluation > board called MBX821 (based on MPC 821). > > The BSP seems to be buggy (bad SIO support, PCMCIA > not tested, etc...). Does anyone own a debugged > BSP for such board with PCMCIA support ???? > > Thanks a lot. > Emmanuel. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: dosFs2 chkdsk slow to run from bootrom Date: Thu, 07 Sep 2000 15:46:33 GMT From: andreyk@home.com (Andray Kaganovsky) Organization: @Home Message-ID: <8FA8743BCandreykhomecom@24.2.9.61> References: <8p4osn$ja8$1@nnrp1.deja.com> <8FA7C192Bandreykhomecom@24.2.9.61> <8p771e$fag$1@nnrp1.deja.com> paulvroberts@my-deja.com wrote in <8p771e$fag$1@nnrp1.deja.com>: >In article <8FA7C192Bandreykhomecom@24.2.9.61>, > andreyk@home.com (Andray Kaganovsky) wrote: >> paulvroberts@my-deja.com wrote in <8p4osn$ja8$1@nnrp1.deja.com>: >> >> >Hi there, >> > >> >Does anyone know why chkdsk would be much slower when run from the >> >bootrom, then when run after VxWorks is loaded? >> > >>> > >> >> Could it be that CPU cache is disabled in the bootrom region ? >> > >Thanks for the reply Andray. Good idea, however, I added a >cacheEnable(DATA_CACHE) to the bootConfig.c and no difference. > How about proving that the CPU cache is enabled on the *code* section indeed ? Say, by measuring time taken by the following loop: volatile unsigned long s = 0; void foo() { register int i; for (i = 0; i < 10000000; i++ ) { /* Since 's' is volatile var which is only written to and never from, and 'i' (hopefully) resides in the CPU register, the fact that cache is enabled on the *data* segment should have no effect. */ s = (i | (i / 2)) & (~(i + 1234)); } } In my opinion, this loop should take much less time when CPU cache is enabled on *code* section (i.e. the bootrom itself). regards, Andray --------------------------- Newsgroups: comp.os.vxworks Subject: Re: 1553 bus Date: Thu, 07 Sep 2000 11:50:17 -0700 From: Joe Durusau Organization: Lockheed Martin Corporation Message-ID: <39B7E369.D972DA92@lmco.com> References: <8p81io$crb$1@nnrp1.deja.com> If you happen to be talking about the right device, you can get sample code from the vendor. The last time I looked, the Greenspring device came with a driver and samples of how to use it. It still will take some effort to make it work with your sytem, however. BTW, I don't work for GS. Speaking only for myself, Joe Durusau S Austin wrote: > > Is it possible to comparison shop for 1553 support for vxWorks or is > Compware Corp. the only game in town? > > Sent via Deja.com http://www.deja.com/ > Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Linking archives Date: 7 Sep 2000 11:32:45 -0500 From: Bob Westbrook Message-ID: References: <8p5k76$90e$1@proxy2.fe.internet.bosch.de> <8p5md8$40e$1@news1.xs4all.nl> <8p791k$gss$1@proxy2.fe.internet.bosch.de> <8p7f57$enc$1@news1.xs4all.nl> <8p802g$gjl$1@proxy2.fe.internet.bosch.de> On Thu, 7 Sep 2000 14:01:46 +0200, Gunter Ritter wrote: >Thanks for your help. VxWorks and Tornado are quite new for me. > >Okay, the command line of the link command seems to be ok, all archives are >present. They are linked to partialimage.o > >I changed the order of the archives from project1.a project2.a to project2.a >project1.a. >Well, the problem has changed a little bit. Now some functions in project 2 >that are referenced in project1 are marked as unreferenced by the linker. > >What do I have to do to make it independent of the link order? > >Gunter > I had a similar problem with cyclic dependencies. Try changing the LIBS macro to project1.a project2.a project1.a or project2.a project1.a project2.a Hope this helps --------------------------- Newsgroups: comp.os.vxworks Subject: WDB_COM_TYPE Date: Thu, 07 Sep 2000 16:40:37 GMT From: rohitkapur@my-deja.com Organization: Deja.com - Before you buy. Message-ID: <8p8ge1$v6b$1@nnrp1.deja.com> hi: I am currently writing an ethernet driver for vxWorks using Tornado 1.0.1. We are trying to implement the 4.3/4.4 BSD stack. We are not creating an Enhanced Network driver. Does anyone know what should be entered for the 'WDB_COMM_TYPE" field in the config.h file for a Non-enhance ethernet driver. Can we define 'WDB_COMM_TYPE' as 'WDB_COMM_NETWORK', and 'WDB_MODE' as 'WDB_TASK'. I know if we are creating an enhanced network driver(END), then we need to set WDB_COMM_TYPE equal to WDB_COM_END, and WDB_MODE equal to WDB_DUAL. If anyone can assist me with this question, i would appreciate it a lot. Thanks, Rohit Kapur Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: How to run a C++ object method from the Shell? Date: Thu, 07 Sep 2000 07:32:09 -0500 From: Kenneth Arakaki Organization: Nortel Networks Message-ID: <39B78AC9.6A39982D@nortelnetworks.com> VxWorks Group: I know that we can load a load module ld < foo.o .... and run a function ->foo( ) .... in the VxWorks shell, but how to run an C++ object method in the shell???? We can obviously load the C++ module the same way... but how to call the method? Do we have to create a C wrapper? .... void main () { .... } Please, an example.... Thanks in advance....!!! Ken --------------------------- Newsgroups: comp.os.vxworks Subject: Re: how to use configure VxSim Date: Thu, 07 Sep 2000 17:58:58 GMT From: v_mathi_vanan@my-deja.com Organization: Deja.com - Before you buy. Message-ID: <8p8l0m$521$1@nnrp1.deja.com> References: <8p5v72$93$1@nnrp1.deja.com> <8p81i9$cr8$1@nnrp1.deja.com> Hi, Thanks for your informatin. I advanced little ahead. I built vxworks image by cofiguring rps, ping client, udp, tcp, igmp, icmp, etc. Now I am able to start simulator (vxsim) and connect target server using the ip 90.0.0.1. then I started shell and typed the following commands and got the below results. - -> ifShow lo (unit number 0): Flags: (0x8069) UP LOOPBACK RUNNING ARP MULTICAST Type: SOFTWARE_LOOPBACK Internet address: 127.0.0.1 Netmask 0xff000000 Subnetmask 0xff000000 Metric is 0 Maximum Transfer Unit size is 32768 0 packets received; 0 packets sent 0 multicast packets received 0 multicast packets sent 0 input errors; 0 output errors 0 collisions; 0 dropped value = 0 = 0x0 interface 90.0.0.254 is missing. - -> hostShow hostname inet address aliases - -------- ------------ ------- localhost 127.0.0.1 value = 0 = 0x0 - -> target address is missing. Any idea ? Thanks Mathi. In article <8p81i9$cr8$1@nnrp1.deja.com>, hruesing@my-deja.com wrote: > In article <8p5v72$93$1@nnrp1.deja.com>, > v_mathi_vanan@my-deja.com wrote: > > Hi, > > > > I have an application (Client server programe) which I want to compile > > and test it in VxWorks Simulator. I installed Tornado 2.0 and I did > the > > following steps > > > > 1. Created a workspace and project. > > 2. Added all required files, and compiled them. No error. > > 3. Launched a simulator and didn't start default target server. > > 4. Before that I installed ULIP and configured the IP address as > > 90.0.0.245. My routing table is updated automatically. > > 5. I configured a target server with backend as wdppipe and target ip > > address as 90.0.0.245. > > 6. I launched the target server and the communcation between the > target > > server and target agent (simulator) was ok. > > 7. Started shell > > 9. Downloaded my project image. No error. > > 10. I started my application from the shell. > > 11. Now the simulator has my server program. > > 12 I started my client program in NT dos prompt, but if I send a > request > > from the client it is not reaching my server which is running in > > simulator. > > > > can any body tell me the reason why my request is not reached in my > > server program ? > > > > Do I need to configure anything more ? > > > > Your help is appreciated. > > > > Thanks > > Mathi. > > > > Sent via Deja.com http://www.deja.com/ > > Before you buy. > > > > If you are speaking from the "full" simulator (not included in standart > T2) you have to config your target server as wdbrpc. > > Hope this helps. > > Harud Rüsing > > Sent via Deja.com http://www.deja.com/ > Before you buy. > Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: configuring PC target??? Date: Thu, 07 Sep 2000 13:54:49 -0400 From: Bruce Chrustie Organization: Nortel Message-ID: <39B7D669.50B20513@root.com> Hi gang, Host is running Windowz and my code works fine under VxSim but now it is time to download it to a 'real' processor for execution. Hooked up the target PC to the LAN and booted up the PC via the floppy and default boot config. But I cannot get the target to communicate via the LAN, nor the terminal emulator (serial). The tornado manual does not seem to explain this very well. Any insight from those who have done this before???? Bruce, --------------------------- Newsgroups: comp.os.vxworks Subject: Re: configuring PC target??? Date: Thu, 07 Sep 2000 13:55:56 -0400 From: Bruce Chrustie Organization: Nortel Message-ID: <39B7D6AC.8343E52F@root.com> References: <39B7D669.50B20513@root.com> one more thing....I can see the target booting via the VGA monitor on the PC itself and am able to control the booting via the keyboard on the target. Bruce Chrustie wrote: > > Hi gang, > > Host is running Windowz and my code works fine under VxSim but now it is > time to download it to a 'real' processor for execution. Hooked up the > target PC to the LAN and booted up the PC via the floppy and default > boot config. But I cannot get the target to communicate via the LAN, nor > the terminal emulator (serial). > > The tornado manual does not seem to explain this very well. Any insight > from those who have done this before???? > > Bruce, --------------------------- Newsgroups: comp.os.vxworks Subject: Re: use of cascaded macros Date: Thu, 07 Sep 2000 15:02:05 -0400 From: Douglas Fraser Organization: Lucent Technologies Message-ID: <39B7E62D.68DA292D@lucent.com> References: <8p2lqt$3g9$1@proxy2.fe.internet.bosch.de> <8p5r7g$r5b$1@nnrp1.deja.com> <8p79f2$h59$1@proxy2.fe.internet.bosch.de> <8p81lg$cs3$1@nnrp1.deja.com> <8p84m2$m2g$1@proxy2.fe.internet.bosch.de> Gunter Ritter wrote: > > I will try it again. > The code we have shall be used for several software packages for an embedded > system. > For the first package we define the text strings for the display in a table > (about 800 strings). > The code can be compiled without errors with a different compiler > (Mitsubishi). So it should be possible to compile the code for ARM9. Maybe I > just have to set a switch for the gnu-compiler. > > Gunter I agree with Richard, you should really try to use enums in this case. They were designed to solve this very problem. Adding or deleting an element in the enumeration renumbers all the elements subsequent to that element, just as you want it to. It is also 'type safe' as an array index since it has type equivalency to an integer. Give them a try, you will probably find it to be a far more natural expression of what you are doing. The best part is, converting your current header file to the appropriate enum representation using emacs, vi, or Perl is easy. If you are using Windows, I am not sure how you would do it, but I assume it can be done. Give it a try. Only one file needs to change in your whole system, and it will be far simpler to read and maintain. - -- Douglas Fraser --------------------------- Newsgroups: comp.os.vxworks Subject: signals on VxSim Date: Thu, 07 Sep 2000 18:57:19 GMT From: edabobojr@yahoo.com Organization: Deja.com - Before you buy. Message-ID: <8p8oe3$98r$1@nnrp1.deja.com> Could somebody offer me some advice on signal handling on vxsim. I can install the handler fine, and it gets called when the signal occurs. What is the appropriate way to continue normal execution? I've tried using setjmp/longjmp and also modifying the pc through the sigcontext passed to the handler. These both result in General Protection Faults. Can anybody help? Thanks Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: tcp/ip Date: Thu, 07 Sep 2000 19:14:14 GMT From: djhystad@msn.com Organization: Deja.com - Before you buy. Message-ID: <8p8pdg$afu$1@nnrp1.deja.com> References: <20000906083316.5382.qmail@mailweb5.rediffmail.com> We use VxWorks primarily for the TCP/IP stack. The target shell is nice for debugging during development, and not having to rewrite all our code when changing SBC's is great. The scheduler is inadequate for our needs. To each his own. <20000906083316.5382.qmail@mailweb5.rediffmail.com>, skrow@rediffmail.com ("shameek roy") wrote: > hai, > iam just learning the basics... > i haven't yet understood why would a realtime OS like VxWorks ever need to support tcp/ip stack. > what would be the kind of applications that run on VxWorks and that require tcp/ip support? > when VxWorks is an RTOS and as it also requires a host OS for prog. development, why does it ever need to support tcp/ip which doesn't provide any guarantees? > please do provide some info or do direct me to appropriate sources. > thanx in advance, > skrow. > > //////////////////////////////////////////////////////// > > Heroes have an infinite capacity for stupidity. Thus are legends born! > --Anonymous > > \\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\ > > _________________________________________________ > Get Your Free Email At, http://www.rediffmail.com > > Partcipate in crazy Re.1 auctions at http://www.rediff.com/auctions > > -- > Posted from IDENT:qmailr@[202.54.124.150] > via Mailgate.ORG Server - http://www.Mailgate.ORG > Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: NFS woes Date: Thu, 07 Sep 2000 19:21:54 GMT From: waqarahmad@my-deja.com Organization: Deja.com - Before you buy. Message-ID: <8p8pro$b5r$1@nnrp1.deja.com> References: <4perqs4m4ob1mt7hr2b4sjvh4mni4rdvgf@4ax.com> Mark, >The Windows NFS client software > will usually see the VxWorks machine, but either timeout trying to > read the disk or not see all the files. The file system on the vxWorks side might really be slow and the time out on the client side might be kicking in. You may wanna look into increasing the timeout on the NFS client side. > NFS appears to be working in > general, because we also open files on the Wintel box from VxWorks (so > WinNT is the server and VxWorks the client), and this works with no > problems. One can not say that NFS appears to be working by simply reversing the usage (i.e client and server). Because if there is a problem in the NFS server on the vxWorks side you won't notice a thing because in this case the vxWorks server is not being used. I think in your case you might simply have WinNT setup issue. and adjusting the time out might just do the trick. Waqar In article <4perqs4m4ob1mt7hr2b4sjvh4mni4rdvgf@4ax.com>, Hosehead Jones 2nd-1/2 wrote: > All righty, then - the background. We're running VxWorks 5.4 (Tornado > II) with dosFs 2.0 on a MV2700 board; our kernel includes both NFS > client and server. The MV2700 has a SCSI hard drive attached to it > that's been divided up into four partitions. The host platform is > Windows NT. > > What we want to do is use NFS to export the local partitions in such a > way that the Windows box will be able to see the local drive and copy > files back and forth between the two systems. Unfortunately, we've > only had marginal success doing this. The Windows NFS client software > will usually see the VxWorks machine, but either timeout trying to > read the disk or not see all the files. > > Since we can sometimes see the VxWorks partitions, I know we're > exporting the volumes correctly (I can't recall the exact call to do > this - something like nfsExport). NFS appears to be working in > general, because we also open files on the Wintel box from VxWorks (so > WinNT is the server and VxWorks the client), and this works with no > problems. > > At this point, I don't know if the problem lies with WRS's NFS server > software, or with the WinNT setup. (We've tried a few different > Windows NFS packages - FrontierTech, Omni Enterprise, Hummingbird - > with no improvement.) Has anyone else gotten this to work reliably? > Is NFS even the answer? > > ----------------------------- > Mark Morrison > Coherent Technologies, Inc. > markm@ctilidar.com > > Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.sys.powerpc.tech,comp.os.vxworks Subject: Re: ppc asm: loading 32 bit value into CTR reg Date: Thu, 07 Sep 2000 15:34:45 -0400 From: Al Johnston Organization: MindSpring Enterprises Message-ID: <39B7EDD5.8C791003@mindspring.com> References: <39B54808.B227A591@mindspring.com> <8p5ji4$irm$1@user2.teleport.com> <8p5non$ukk$11$1@news.t-online.com> <39B687F7.127D0666@ib.stortek.com> <39B688DC.A8CD1507@ib.stortek.com> > The PowerPC Compiler Writer's Guide (get yourself a copy from IBM's > web site) recommends using adds instead of ors, because future PowerPC I have had no luck finding this... can you give me a site? thanks, - -al --------------------------- Newsgroups: comp.os.vxworks Subject: Re: use of cascaded macros Date: 7 Sep 2000 19:52:47 GMT From: mathar@mpia-hd.mpg.de (Richard Mathar) Organization: Max-Planck-Institut fuer Astronomie Keywords: cccp, VxWorks Message-ID: <8p8rmf$s6n$1@news.urz.uni-heidelberg.de> References: <8p2lqt$3g9$1@proxy2.fe.internet.bosch.de> Reply-To: mathar@mpia-hd.mpg.de Gunter Ritter writes: |> |> I have the problem that I get errors "macro recursion too deep" when |> compiling a module that uses cascaded macros. |> |> The macros are defined as follows: |> |> #define K_MACRO_1 0 |> #define K_MACRO_2 K_MACRO_1+1 |> #define K_MACRO_3 K_MACRO_2+1 |> #define K_MACRO_4 K_MACRO_3+1 |> #define K_MACRO_5 K_MACRO_4+1 |> : |> : If your're using gcc-2.95 there is indeed a maximum of 400 set to the input stack. See the lines #define INPUT_STACK_MAX 400 and #define CHECK_DEPTH(code) \ if (indepth >= (INPUT_STACK_MAX - 1)) \ { \ error_with_line (line_for_error (instack[indepth].lineno), \ "macro or `#include' recursion too deep"); \ code; \ } in the file gcc-2.95/gcc/cccp.c (after gunzip and tar xf of the source code). Increasing this number and recompilation of the compiler should help. mathar@mpia-hd.mpg.de http://www.mpia-hd.mpg.de/homes/mathar --------------------------- Newsgroups: comp.os.vxworks Subject: Re: use of cascaded macros Date: Thu, 07 Sep 2000 19:50:52 GMT From: djhystad@msn.com Organization: Deja.com - Before you buy. Message-ID: <8p8rii$df6$1@nnrp1.deja.com> References: <8p2lqt$3g9$1@proxy2.fe.internet.bosch.de> <39B4F533.F56D7B9B@swri.edu> <8p2vvs$hmq$1@proxy2.fe.internet.bosch.de> <39b54d5d.0@news> In article , Jim Thomas wrote: > >>>>> "Richard" == Richard writes: > > Richard> try the following > Richard> typedef enum > Richard> { > Richard> K_STRING_1, > Richard> K_STRING_2, > Richard> K_STRING_3, > Richard> K_STRING_4, > Richard> ...., > Richard> NUMBER_K_STRINGS > Richard> } K_STRING_DEF_TYPE; > > Richard> you can remove entries from the definition, and the following > Richard> ones will be renumbered. I have not looked it up in ANSI C, but > Richard> in all of the compilers I have used, the first enumberated value > Richard> is 0, and each successive one is one larger than the previous > Richard> one. These values are usable anywhere an int is. As a bonus, > Richard> NUMBER_K_STRINGS can be used as the upper limit of the valid > Richard> indices (all valid index values are less than NUMBER_K_STRINGS), > Richard> and can be used to allocate your string table. > > But there is no longer any relationship between the values and the > indices. I use exactly his macros very happily (well, with parens :- ). > > char foo[] = { > #define bar 0 > "bar string", > #define mumble (bar + 1) > "mumble string", > #define last (mumble + 1) > (char *) NULL > } > .... > assert((last + 1) == (sizeof foo / sizeof (char *))); > > Also to the poster who said "just use 0 1 2 3 4 ..." - baloney. He, I > think from following postings, used numeric names in the example but uses > real names in the actual code. The reason for symbols is exactly to > eliminate meaningless numbers. > > As for the original problem, how many levels do you have? I don't see any > way of controlling memory allocation with GNU cpp - maybe you need more > swap space? > How about rewriting your example as: enum {foo, bar, mumble, last} char *strings[last]; void initializeStrings() { strings[foo] = "foo string"; strings[bar] = "bar string"; strings[mumble] = "mumble string"; } I know it requires an extra function, but it is easier to read. Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: use of cascaded macros Date: Thu, 07 Sep 2000 19:50:24 GMT From: djhystad@msn.com Organization: Deja.com - Before you buy. Message-ID: <8p8rhm$deq$1@nnrp1.deja.com> References: <8p2lqt$3g9$1@proxy2.fe.internet.bosch.de> <39B4F533.F56D7B9B@swri.edu> <8p2vvs$hmq$1@proxy2.fe.internet.bosch.de> <39b54d5d.0@news> <39b7a2b4.0@news> In article <39b7a2b4.0@news>, "Richard" wrote: > "Jim Thomas" wrote in message > news:wwn1hknc9e.fsf@atlas.cfht.hawaii.edu... > > >>>>> "Richard" == Richard writes: > > > > Richard> try the following > > Richard> typedef enum > > Richard> { > > Richard> K_STRING_1, > > Richard> K_STRING_2, > > Richard> K_STRING_3, > > Richard> K_STRING_4, > > Richard> ...., > > Richard> NUMBER_K_STRINGS > > Richard> } K_STRING_DEF_TYPE; > > > > Richard> you can remove entries from the definition, and the following > > Richard> ones will be renumbered. I have not looked it up in ANSI C, but > > Richard> in all of the compilers I have used, the first enumberated value > > Richard> is 0, and each successive one is one larger than the previous > > Richard> one. These values are usable anywhere an int is. As a bonus, > > Richard> NUMBER_K_STRINGS can be used as the upper limit of the valid > > Richard> indices (all valid index values are less than NUMBER_K_STRINGS), > > Richard> and can be used to allocate your string table. > > > > But there is no longer any relationship between the values and the > > indices. I use exactly his macros very happily (well, with parens :-). > > But they do retain that relationship, The K_STRING_1 has a value of 0, > K_STRING_2 = 1, > and so on. This is what allows one to be deleted from both the table, and > the enumerated type, and still retain the same relationship. This way of > defining the values results in the exact same values he originally had > without the problem of nesting defines, and without the need to change the > relationship of the defines when a string is removed/added. As with the > defines, it does require that the enumerated type be updated anytime the > table is changed. > > > -- > rrapier sieversinst > To reply to me, replace the space with @, and add the .com on the end. > > I think his point is that there is no longer an obvious relationship between the enumeration and the values stored in the array. The enumeration is defined in one place, and the array at another. To remove an entry you would either have to determine the index of element in the array (either by counting or by "grokking" the nameing convention). This is somewhat error prone. I think this looks like a good place to use automatic code generation. Save the index/string definitions in a file using a format like; index1, String 1 index2, String 2 index3, String 3 Writing a parser to convert this to: enum { index1, index2, index3, maxIndex}; char *strings[maxIndex] = { "String 1", "String 2", "String 3" }; Would be pretty easy. Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.sys.powerpc.tech,comp.os.vxworks Subject: Re: ppc asm: loading 32 bit value into CTR reg Date: 7 Sep 2000 13:37:24 -0700 From: andrew@user2.teleport.com (Andrew Klossner) Message-ID: <8p8ua4$fkr$1@user2.teleport.com> References: <39B54808.B227A591@mindspring.com> <39B687F7.127D0666@ib.stortek.com> <39B688DC.A8CD1507@ib.stortek.com> <39B7EDD5.8C791003@mindspring.com> >> The PowerPC Compiler Writer's Guide (get yourself a copy from IBM's >> web site) recommends using adds instead of ors, because future PowerPC > I have had no luck finding this... can you give me a site? deja.com, where you can search prior postings to comp.sys.powerpc.tech for information such as this. The Guide itself is at http://www.chips.ibm.com/techlib/products/powerpc/manuals/compiler/cwg.pdf ... at least it was three weeks ago, the last time this question was asked. -=- Andrew Klossner (andrew@teleport.com) --------------------------- Newsgroups: comp.os.vxworks Subject: Re: tcp/ip Date: Thu, 07 Sep 2000 15:53:34 -0500 From: "M. Kotiaho" Organization: LMMFC - D Message-ID: <39B8004E.3C254526@nospam.lmco.com> References: <20000906083316.5382.qmail@mailweb5.rediffmail.com> In embedded/distributed systems, the different subsystems need to communicate with each other, and quite often also with the outside world. In many defense projects, the subsystems will communicate with each other based on a message protocol that is unique to the project, but still built on top of TCP/IP (or UDP/IP). This is much easier than trying to develop a messaging protocol from scratch. MK shameek roy wrote: > hai, > iam just learning the basics... > i haven't yet understood why would a realtime OS like VxWorks ever need to support tcp/ip stack. > what would be the kind of applications that run on VxWorks and that require tcp/ip support? > when VxWorks is an RTOS and as it also requires a host OS for prog. development, why does it ever need to support tcp/ip which doesn't provide any guarantees? > please do provide some info or do direct me to appropriate sources. > thanx in advance, > skrow. --------------------------- Newsgroups: comp.os.vxworks Subject: How to download a modem driver to vxworks image Date: Thu, 07 Sep 2000 21:15:54 GMT From: bryan Organization: Deja.com - Before you buy. Message-ID: <8p90hk$jo3$1@nnrp1.deja.com> Hi everyone, This is my first time to send a question in this newsgroup. I am developing a modem driver for VxWorks. I got a target hardware from the other company and the image of VxWorks is inside the IDE. That's mean I don't have any source code of BSP for this target. So I need to download my modem driver as a downloadable application. Is it possible to do that?! If so, how can I do it? Best Regards Bryan Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Problem with VxWorks Ncr810 SCSI Driver Date: Thu, 07 Sep 2000 21:43:53 GMT From: syee@connex.com Organization: Deja.com - Before you buy. Message-ID: <8p926j$ljj$1@nnrp1.deja.com> I'm trying to get the ncr810 driver to work on the 810 controller (and eventually the 895 controller). I'm using Tornado 1.0.1 and VxWorks 5.3.1. on a Pentium machine. The system is crashing everytime I try to initialize the SCSI controller. Below are the functions calls I make that lead up to the crash: ncr810CtrlCreate() Creates controller structure intConnect() Connects ISR to interrupt vector sysIntEnablePIC() enables interrupt SetHwRegisters() sets hardware-dependent registers ncr810CtrlInit() initializes controller The system crashes and reboots at the controller init function. I'm assuming that there is some conflict between the interrupt and the Interrupt Service Routine (ncr810Intr). If I don't include the sysIntEnablePIC() function, the system doesn't crash and I'm able to do a controller init. However, when I try a scsiPhysDevCreate or an scsiAutoConfig, the system hangs. I'm positive the base address and interrupts I use are correct because I verified it in the PCI configuration space. What's even stranger is that I got it to work once but then all of a sudden it crashed and I didn't have a backup at the time. Has anyone experienced a similar problem? Any suggestions on how to fix this? Please e-mail me at syee@connex.com. Thanks in advance. Stanley Yee Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Does Windows2000 support Tornado 2.0? Date: Thu, 7 Sep 2000 13:18:27 +0800 From: "ricky" Organization: utstar Message-ID: <8p78fi$1ek4$1@news.cz.js.cn> Reply-To: "ricky" Hello,everyone under Windows 2000: Yestoday I install Tornado 2.0 into my PC with Windows2000 successfully. When I run Tornado, it said couldn`t read file "TRY=localhost\host\resource\tcl\WindView.win32.tcl", then I ignored it and entered main window of Tornado, I can`t find WindView in Tools menu. Maybe WindView.dll is initialized by Tornado unsuccessfully without WindView.win32.tcl. But why Tornado can`t read file WindView.win32.tcl? Is there a path mistake or other reasons? How should I do? Yours, ricky --------------------------- Newsgroups: comp.os.vxworks Subject: Re: configuring PC target??? Date: Thu, 07 Sep 2000 16:36:07 -0500 From: Bo Yu Organization: Motorola CIG Message-ID: <39B80A47.33A010DB@email.mot.com> References: <39B7D669.50B20513@root.com> <39B7D6AC.8343E52F@root.com> - --------------92BD3F017BCA83438FE9A7B7 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Bruce Chrustie wrote: > one more thing....I can see the target booting via the VGA monitor on > the PC itself and am able to control the booting via the keyboard on the > target. > > Bruce Chrustie wrote: > > > > Hi gang, > > > > Host is running Windowz and my code works fine under VxSim but now it is > > time to download it to a 'real' processor for execution. Hooked up the > > target PC to the LAN and booted up the PC via the floppy and default > > boot config. But I cannot get the target to communicate via the LAN, nor > > the terminal emulator (serial). > > > > The tornado manual does not seem to explain this very well. Any insight > > from those who have done this before???? > > > > Bruce, Did you rebuild the VxWorks using cross compiler ? - -- Bo Yu - --------------92BD3F017BCA83438FE9A7B7 Content-Type: text/html; charset=us-ascii Content-Transfer-Encoding: 7bit Bruce Chrustie wrote:
one more thing....I can see the target booting via the VGA monitor on
the PC itself and am able to control the booting via the keyboard on the
target.

Bruce Chrustie wrote:
>
> Hi gang,
>
> Host is running Windowz and my code works fine under VxSim but now it is
> time to download it to a 'real' processor for execution. Hooked up the
> target PC to the LAN and booted up the PC via the floppy and default
> boot config. But I cannot get the target to communicate via the LAN, nor
> the terminal emulator (serial).
>
> The tornado manual does not seem to explain this very well. Any insight
> from those who have done this before????
>
> Bruce,

Did you rebuild the VxWorks using cross compiler ? 
-- 
Bo Yu
  - --------------92BD3F017BCA83438FE9A7B7-- --------------------------- Newsgroups: comp.os.vxworks Subject: [Q]: DHCPC Date: Thu, 07 Sep 2000 22:27:44 GMT From: Stephen Organization: Deja.com - Before you buy. Message-ID: <8p94o6$opq$1@nnrp1.deja.com> I'm having some trouble with DHCPC. I'm trying to use DHCPC after VxWorks is up to get an IP address from a DHCP server that I'm running on LINUX (DHCPD). I'm not trying to configure an interface; I'm only trying to get an address for my application. I see that the DHCP server is getting the DHCP discovery messages, and it is responding with an address to my interface's MAC address. However, on the VxWorks side, the client simply repeats the WAIT_OFFER state until it eventually times out. It looks like the timeout message is the only event it ever sees; DHCPC is not seeing the response from the DHCP server at all. ifShow on the interface shows that it *is* receiving multicast packets. I have the Network Protocol Toolkit installed, and I wonder if that is an issue. (It was on a separate occasion.) Anyway, here's my fairly straightforward code. Anybody got any ideas? #include "stdio.h" #include "dhcpcLib.h" void dhcpc(int init) { int cookie; char addr[16]; STATUS result; struct dhcp_param params; cookie = dhcpcInit(ifunit("RX2"), 0); if (!cookie) { printf("dhcpInit failed\n"); return; } else printf("cookie is %d\n", cookie); result = dhcpcBind(cookie, TRUE); <-- it fails here, of course.. if (result != OK) { printf("dhcpcBind failed: %d.\n", errnoGet()); return; } result = dhcpcParamsGet(cookie, ¶ms); if (result != OK) { printf("dhcpcParamsGet failed: %d.\n", errnoGet()); return; } inet_ntoa_b(params.yiaddr, addr); printf("DHCP returned %s.\n", addr); } Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Does Windows2000 support Tornado 2.0? Date: Thu, 07 Sep 2000 16:02:28 -0700 From: Lori Fraleigh Organization: Real-Time Innovations Message-ID: <39B81E84.4C005492@rti.com> References: <8p78fi$1ek4$1@news.cz.js.cn> Ricky - Go to the Tools menu in Tornado and choose Options, then pick 'Tornado Registry' all the way over at the right. Change the Tornado registry from local machine to remote machine and type in the IP address of your Windows 2000 machine. This should make the error go away. The "localhost" is coming from the WIND_REGISTRY setting. Regards, Lori ricky wrote: > > Hello,everyone under Windows 2000: > > Yestoday I install Tornado 2.0 into my PC with Windows2000 successfully. > When I run Tornado, it said couldn`t read file > "TRY=localhost\host\resource\tcl\WindView.win32.tcl", > then I ignored it and entered main window of Tornado, > I can`t find WindView in Tools menu. > > Maybe WindView.dll is initialized by Tornado unsuccessfully without > WindView.win32.tcl. > But why Tornado can`t read file WindView.win32.tcl? > Is there a path mistake or other reasons? > > How should I do? > > Yours, > ricky - -- Lori Fraleigh Real-Time Innovations Senior Software Engineer ScopeTools Manager Phone: 408-734-4200 x121 Fax: 408-734-5009 http://www.rti.com Please send/cc all support related email to support@rti.com --------------------------- Newsgroups: comp.os.vxworks,comp.realtime Subject: HOT HOT - VxWorks S/W Developers Needed - Ottawa Date: Thu, 7 Sep 2000 19:39:54 -0400 From: "PROCOM" Organization: IGS - Information Gateway Services Message-ID: <8p98p4$nc$1@news.igs.net> Title: C++ / OOD VxWorks Software Developers Description: Our client, one of the world's largest telecommunications companies, currently requires 10 Junior to Senior Software Specialists to design, develop, implement and support Wireless products and applications. Minimum Requirements: All candidates should possess a minimum of 1 year experience developing real-time C/C++, OOD S/W on a VxWorks platform. Desired skills would include any previous Wireless development experience. Location: Ottawa, Ontario, Canada Type: Full-Time Placement Compensation: Open PROCOM: Established in 1978, Professional Computer Consultants Group Ltd. (Procom) is a national leader in the provision of Computer personnel on a contract and full-time basis. Our clients are comprised of the largest national and international corporations that utilize technical resources extensively across a wide range of disciplines. In the Financial Post (March 1999) Procom was ranked as the 6th largest professional Services Company in Canada. In November of 1999, Procom was named a Regional finalist in Canada' s 50 Best Managed Private companies. Our track record is proven with more than 180 consultants servicing Ottawa's high tech community and more than 1600 Procom consultants currently on assignment throughout. North America. For further information on this and other opportunities please visit our web site at www.procom.ca. Interested candidates are invited to forward their resumes or questions in confidence to: Derek Weber Technical Recruiter 300 March Road Kanata, Ontario K2K 2E2 phone: (613) 270-9339 ext 231 e-mail: derekw@procom.ca --------------------------- Newsgroups: comp.os.vxworks Subject: Re: use of cascaded macros Date: 07 Sep 2000 14:53:10 -1000 From: Jim Thomas Organization: Canada France Hawai`i Telescope Message-ID: References: <8p2lqt$3g9$1@proxy2.fe.internet.bosch.de> <8p8rmf$s6n$1@news.urz.uni-heidelberg.de> >>>>> "Richard" == Richard Mathar writes: Richard> If your're using gcc-2.95 there is indeed a maximum of 400 set to Richard> the input stack. See the lines Richard> #define INPUT_STACK_MAX 400 .... Richard> in the file gcc-2.95/gcc/cccp.c (after gunzip and tar xf of the Richard> source code). Arrgh! What ever happened to the 0,1,many rule? :-( --------------------------- Newsgroups: comp.os.vxworks Subject: Re: use of cascaded macros Date: 07 Sep 2000 15:03:10 -1000 From: Jim Thomas Organization: Canada France Hawai`i Telescope Message-ID: References: <8p2lqt$3g9$1@proxy2.fe.internet.bosch.de> <39B4F533.F56D7B9B@swri.edu> <8p2vvs$hmq$1@proxy2.fe.internet.bosch.de> <39b54d5d.0@news> <8p8rii$df6$1@nnrp1.deja.com> >>>>> "djhystad" == djhystad writes: djhystad> In article , djhystad> Jim Thomas wrote: >> char foo[] = { >> #define bar 0 >> "bar string", >> #define mumble (bar + 1) >> "mumble string", >> #define last (mumble + 1) >> (char *) NULL >> } djhystad> How about rewriting your example as: djhystad> enum {foo, bar, mumble, last} djhystad> char *strings[last]; djhystad> void initializeStrings() djhystad> { djhystad> strings[foo] = "foo string"; djhystad> strings[bar] = "bar string"; djhystad> strings[mumble] = "mumble string"; djhystad> } djhystad> I know it requires an extra function, but it is easier to read. First, your interpretation of my meaning was correct. The proposed enum removes the relationship between the indices and the values. Second, your function above is nice, but it's not quite as easy to tell if all strings are initialized. --------------------------- Newsgroups: comp.os.vxworks Subject: GUI Query Date: Thu, 07 Sep 2000 18:12:28 -0700 From: Robert Sharp Organization: SBC Internet Services Message-ID: <39B83CFC.AE96C651@hotmail.com> Hi all, I'm at the beginning of a project, researching GUI issues, and I was wondering what, besides C/C++ GUI apps, others are developing/have developed their GUIs with and what tools/vendors you would recommend (or not recommend). I'm dealing with a 133 MIPS processor, with an IR remote control input and a TV (NTSC resolution) for output. I was originally considering using Wind River's HTMLWorks so I (and others) could create a GUI by HTML authoring, but they have dropped that product. I'm interested to see if others have implemented HTML-based GUIs and are happy with them. I assume a fair amount of Java GUI apps are being done. Anyone using XML technology? Others? Thanks in advance! Bob --------------------------- Newsgroups: comp.os.vxworks Subject: Re: ricky Date: Fri, 8 Sep 2000 10:09:17 +0800 From: "ricky" Organization: utstar Message-ID: <8p9hot$2ouf$1@news.cz.js.cn> Reply-To: "ricky" ·¢¼þÈË: Lori Fraleigh ÊÕ¼þÈË: Ö÷Ìâ: Re: Does Windows2000 support Tornado 2.0? ÈÕÆÚ: 2000Äê9ÔÂ8ÈÕ 7:04 Ricky - Go to the Tools menu in Tornado and choose Options, then pick 'Tornado Registry' all the way over at the right. Change the Tornado registry from local machine to remote machine and type in the IP address of your Windows 2000 machine. This should make the error go away. The "localhost" is coming from the WIND_REGISTRY setting. Regards, Lori ricky wrote: > > Hello,everyone under Windows 2000: > > Yestoday I install Tornado 2.0 into my PC with Windows2000 successfully. > When I run Tornado, it said couldn`t read file > "TRY=localhost\host\resource\tcl\WindView.win32.tcl", > then I ignored it and entered main window of Tornado, > I can`t find WindView in Tools menu. > > Maybe WindView.dll is initialized by Tornado unsuccessfully without > WindView.win32.tcl. > But why Tornado can`t read file WindView.win32.tcl? > Is there a path mistake or other reasons? > > How should I do? > > Yours, > ricky - -- Lori Fraleigh Real-Time Innovations Senior Software Engineer ScopeTools Manager Phone: 408-734-4200 x121 Fax: 408-734-5009 http://www.rti.com Please send/cc all support related email to support@rti.com --------------------------- Newsgroups: comp.os.vxworks Subject: Re: 1553 bus Date: Fri, 08 Sep 2000 09:01:24 +0530 From: Vijay Kumar Peshkar Organization: Ye 'Ol Disorganized NNTPCache groupie Message-ID: <39B85D8C.6BDB687A@cisco.com> References: <8p81io$crb$1@nnrp1.deja.com> Reply-To: vpeshkar@cisco.com This is a multi-part message in MIME format. - --------------50C97A34EA14E5AF58F0071E Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Austin, This is not a reply to your query. I am also interested in knowing more about the mil-std-1553b support. Can you please update me if you receive any info..? Thanks and Regards, Vijay S Austin wrote: > Is it possible to comparison shop for 1553 support for vxWorks or is > Compware Corp. the only game in town? > > Sent via Deja.com http://www.deja.com/ > Before you buy. - --------------50C97A34EA14E5AF58F0071E Content-Type: text/x-vcard; charset=us-ascii; name="vpeshkar.vcf" Content-Transfer-Encoding: 7bit Content-Description: Card for Vijay Kumar Peshkar Content-Disposition: attachment; filename="vpeshkar.vcf" begin:vcard n:Peshkar;Vijay x-mozilla-html:FALSE org:Wipro Technologies,K2, Bangalore;Wishing you blue skies, a couple of daisies, some fresh air and a smile! adr:;;;;;; version:2.1 email;internet:vpeshkar@cisco.com title:Senior Software Engineer fn:Vijay Kumar Peshkar end:vcard - --------------50C97A34EA14E5AF58F0071E-- --------------------------- Newsgroups: comp.os.vxworks Subject: Re: configuring PC target??? Date: Fri, 08 Sep 2000 09:09:53 +0530 From: Vijay Kumar Peshkar Organization: Ye 'Ol Disorganized NNTPCache groupie Message-ID: <39B85F89.D2FA9DD3@cisco.com> References: <39B7D669.50B20513@root.com> <39B7D6AC.8343E52F@root.com> Reply-To: vpeshkar@cisco.com This is a multi-part message in MIME format. - --------------0D6C3602ACC7104B523B9FF8 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Bruce, Have you verified the host ip address to be correct?. Have you specified the path of the .sym file residing in Host correctly. Need more info about how you are trying to communicate. Also I remember building my first PC target using the manuals ;0) Just browse thru the prog guide once more...sorry no more details...it was long ago... Do reply. have fun, Vijay Bruce Chrustie wrote: > one more thing....I can see the target booting via the VGA monitor on > the PC itself and am able to control the booting via the keyboard on the > target. > > Bruce Chrustie wrote: > > > > Hi gang, > > > > Host is running Windowz and my code works fine under VxSim but now it is > > time to download it to a 'real' processor for execution. Hooked up the > > target PC to the LAN and booted up the PC via the floppy and default > > boot config. But I cannot get the target to communicate via the LAN, nor > > the terminal emulator (serial). > > > > The tornado manual does not seem to explain this very well. Any insight > > from those who have done this before???? > > > > Bruce, - --------------0D6C3602ACC7104B523B9FF8 Content-Type: text/x-vcard; charset=us-ascii; name="vpeshkar.vcf" Content-Transfer-Encoding: 7bit Content-Description: Card for Vijay Kumar Peshkar Content-Disposition: attachment; filename="vpeshkar.vcf" begin:vcard n:Peshkar;Vijay x-mozilla-html:FALSE org:Wipro Technologies,K2, Bangalore;Wishing you blue skies, a couple of daisies, some fresh air and a smile! adr:;;;;;; version:2.1 email;internet:vpeshkar@cisco.com title:Senior Software Engineer fn:Vijay Kumar Peshkar end:vcard - --------------0D6C3602ACC7104B523B9FF8-- --------------------------- Newsgroups: comp.os.vxworks Subject: Re: GUI Query Date: Fri, 08 Sep 2000 09:16:38 +0530 From: Vijay Kumar Peshkar Organization: Ye 'Ol Disorganized NNTPCache groupie Message-ID: <39B8611E.1808ABCE@cisco.com> References: <39B83CFC.AE96C651@hotmail.com> Reply-To: vpeshkar@cisco.com This is a multi-part message in MIME format. - --------------4F704F397F5D56078733BB05 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Robert, I had done a web brower pad using UGL last year. Think its OK. cheers, Vijay Robert Sharp wrote: > Hi all, > > I'm at the beginning of a project, researching GUI issues, and I was > wondering what, besides C/C++ GUI apps, others are developing/have > developed their GUIs with and what tools/vendors you would recommend (or > not recommend). I'm dealing with a 133 MIPS processor, with an IR remote > control input and a TV (NTSC resolution) for output. > > I was originally considering using Wind River's HTMLWorks so I (and > others) could create a GUI by HTML authoring, but they have dropped that > product. I'm interested to see if others have implemented HTML-based > GUIs and are happy with them. > > I assume a fair amount of Java GUI apps are being done. Anyone using XML > technology? Others? > > Thanks in advance! > > Bob - --------------4F704F397F5D56078733BB05 Content-Type: text/x-vcard; charset=us-ascii; name="vpeshkar.vcf" Content-Transfer-Encoding: 7bit Content-Description: Card for Vijay Kumar Peshkar Content-Disposition: attachment; filename="vpeshkar.vcf" begin:vcard n:Peshkar;Vijay x-mozilla-html:FALSE org:Wipro Technologies,K2, Bangalore;Wishing you blue skies, a couple of daisies, some fresh air and a smile! adr:;;;;;; version:2.1 email;internet:vpeshkar@cisco.com title:Senior Software Engineer fn:Vijay Kumar Peshkar end:vcard - --------------4F704F397F5D56078733BB05-- --------------------------- Newsgroups: comp.os.vxworks Subject: Re: How to download a modem driver to vxworks image Date: Fri, 08 Sep 2000 09:24:52 +0530 From: Vijay Kumar Peshkar Organization: Ye 'Ol Disorganized NNTPCache groupie Message-ID: <39B8630C.2B9BB20C@cisco.com> References: <8p90hk$jo3$1@nnrp1.deja.com> Reply-To: vpeshkar@cisco.com This is a multi-part message in MIME format. - --------------44BE564D72038D037FDD0D9F Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Hi Bryan, If the wdb component is enabled in the target image and if you are able to launch wdb from your host PC, you should be able to download the .obj files to the target anytime. Hope this helps. regards, Vijay bryan wrote: > Hi everyone, > > This is my first time to send a question in this newsgroup. > > I am developing a modem driver for VxWorks. I got a target hardware > from the other company and the image of VxWorks is inside the IDE. > That's mean I don't have any source code of BSP for this target. > So I need to download my modem driver as a downloadable application. > Is it possible to do that?! If so, how can I do it? > > Best Regards > Bryan > > Sent via Deja.com http://www.deja.com/ > Before you buy. - --------------44BE564D72038D037FDD0D9F Content-Type: text/x-vcard; charset=us-ascii; name="vpeshkar.vcf" Content-Transfer-Encoding: 7bit Content-Description: Card for Vijay Kumar Peshkar Content-Disposition: attachment; filename="vpeshkar.vcf" begin:vcard n:Peshkar;Vijay x-mozilla-html:FALSE org:Wipro Technologies,K2, Bangalore;Wishing you blue skies, a couple of daisies, some fresh air and a smile! adr:;;;;;; version:2.1 email;internet:vpeshkar@cisco.com title:Senior Software Engineer fn:Vijay Kumar Peshkar end:vcard - --------------44BE564D72038D037FDD0D9F-- --------------------------- Newsgroups: comp.os.vxworks Subject: socket Date: Fri, 08 Sep 2000 05:45:03 GMT From: ba_ajay@my-deja.com Organization: Deja.com - Before you buy. Message-ID: <8p9ucs$llg$1@nnrp1.deja.com> Hi, I am learning basics of Vxworks. I want to use socket programme with a tcp/ip stack. Right now, i am using a bootable application. No compilation errors. Configured according to the procedures in one of the vxworks faq sites. VxWorks / Tornado II Running multiple simulators on NT URL :http://www.xs4all.nl/~borkhuis/vxworks/multsimnt.html problem : a. ping localhost is not pinging. Can anyone help me know how to make a client-server communication application possible on vxworks. configured according to the procedure given in the site. 1. Which symbol table to use(downloadable or built-in) ? 2. is built-in symbol table the right one for a bootable appln ? 3. if use a downloadable application , i get no errors in compilation. while downloading, error message is received as undefined symbols socket,bind etc.. What do i need to include ? 4. IS this approach the right way of testing client-server on different processors. Client running on Shell linked to target1 processor 0 (ip-addrress 90.0.0.1) server running on Shell linked to target2 processor 1 (ip-addrress 90.0.0.2) Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Linking archives Date: Fri, 8 Sep 2000 07:56:37 +0200 From: Gunter Ritter Organization: Tenovis GmbH u. Co. KG Message-ID: <8p9v1n$4nh$1@proxy2.fe.internet.bosch.de> References: <8p5k76$90e$1@proxy2.fe.internet.bosch.de> <8p5md8$40e$1@news1.xs4all.nl> <8p791k$gss$1@proxy2.fe.internet.bosch.de> <8p7f57$enc$1@news1.xs4all.nl> <8p802g$gjl$1@proxy2.fe.internet.bosch.de> I thought about this, too. But I didn´t try it yet. I find it a little bit strange to do it this way. Well, I try it now. Gunter --------------------------- Newsgroups: comp.os.vxworks Subject: Re: use of cascaded macros Date: Fri, 8 Sep 2000 08:03:02 +0200 From: Gunter Ritter Organization: Tenovis GmbH u. Co. KG Message-ID: <8p9vdo$57d$1@proxy2.fe.internet.bosch.de> References: <8p2lqt$3g9$1@proxy2.fe.internet.bosch.de> <8p8rmf$s6n$1@news.urz.uni-heidelberg.de> I think theres a limit in the compiler. But the compiler is delivered by WindRiver as binary. I try now the variant with the enumeration. Gunter --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Device doesn't show up as an interface Date: Fri, 08 Sep 2000 11:50:00 +0530 From: Yusuf Motiwala Organization: Lucent Technologies, Columbus, Ohio Message-ID: <39B88510.92280A0E@lucent.com> References: <39B4DE70.2B5E87AD@oneline.de> You need to assign address for it to get shoed up in ifShow etc. Since, ipAttach does not work with your interface, if will not be shhown yup in ifShow etc. Yusuf Mario Wolframm wrote: > I call muxDevLoad and get back a reasonable address. > I call muxDevStart and get OK > I see the driver registered via muxShow > however: > neither ipAttach nor ifAddrSet work with that new interface! --------------------------- Newsgroups: comp.os.vxworks Subject: MPC8260 Date: Fri, 08 Sep 2000 10:19:19 +0300 From: Assaf Hoffman Organization: Galileo Technology Message-ID: <39B892F7.9DC0315B@galileo.co.il> Hi VxWorkers !! I'm trying to run the MPC8260 based on the PPCEC603 tool chain. I'm experiencing heavy stability problem with my BSP: 1) Debug printing in boot process (bootrom.hex) hangs the boot. 2) Once in 5-6 boots the VxWorks image hangs ('ping' is functional). Does anybody have any experience with this CPU ?? - -- The only true freedom is freedom from your heart's desires.... ======================================================================== _/_/_/_/_/ _/_/_/_/_/_/_/ _/_/_/_/_/_/_/_/ _/_/_/_/ _/_/_/ _/_/_/ _/_/_/_/ _/_/_/ _/_/_/ _/_/_/ _/_/_/ Assaf Hoffman _/_/_/_/_/_/_/_/ Galileo Technology Ltd. _/_/_/_/_/_/ E-mail - hoffman@galileo.co.il _/_/_/_/ Tel - 972-4-9999555 ext. 203 Check our Web site: http://www.galileoT.com ========================================================================= --------------------------- Newsgroups: comp.os.vxworks Subject: Re: socket Date: 8 Sep 2000 08:06:06 GMT From: borkhuis@lucent.com Organization: Lucent Technologies, Nieuwegein NL Message-ID: <8pa6le$ao$1@news1.xs4all.nl> References: <8p9ucs$llg$1@nnrp1.deja.com> ba_ajay@my-deja.com wrote: : 1. Which symbol table to use(downloadable or built-in) ? : 2. is built-in symbol table the right one for a bootable appln ? Yes, that is the right one. : 3. if use a downloadable application , i get no errors in compilation. : while downloading, error message is received as undefined symbols : socket,bind etc.. What do i need to include ? In your bootable application you should include the network components, including the Ping client. In the simulator the first lines should be like this: Attached TCP/IP interface to nt unit 0 Attaching interface lo0...done Adding 3069 symbols for standalone. Groeten, Johan - -- o o o o o o o . . . ______________________________ o _____ || Johan Borkhuis | .][__n_n_|DD[ ====_____ | borkhuis@lucent.com | >(________|__|_[_________]_|___________________________| _/oo OOOOO oo` ooo ooo 'o!o!o o!o!o` === VxWorks FAQ: http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html === --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Device doesn't show up as an interface Date: Fri, 08 Sep 2000 10:29:32 +0200 From: Mario Wolframm Organization: ONELINE AG Message-ID: <39B8A36C.EAD5C65B@oneline.de> References: <39B4DE70.2B5E87AD@oneline.de> <39B88510.92280A0E@lucent.com> Yusuf Motiwala wrote: > > You need to assign address for it to get shoed up in ifShow etc. > Since, ipAttach does not work with your interface, if will not be shown > up in ifShow etc. > > Yusuf Hi fellows, the following approach worked now: set IP_MAX_UNITS to something big enough and then 1. pCookie = muxDevLoad ( unit, /* 0 */ loadFunc, /* driverLoad */ loadString, /* "a:b:c:... */ loan, /* unused: 0 */ pBSP /* NULL */ ); 2. muxDevStart (pCookie); 3. ipAttach (unit /* 0 */, pDevice /* "ln" */); 4. ifMaskSet (pDeviceName /* "ln0" */, ifMask /* 0xffffff00 */); ifBroadcastSet (pDeviceName /* "ln0" */, bcastAddr /* "10.2.3.255" */); ifAddrSet (pDeviceName /* "ln0" */, ifAddr /* "10.2.3.4" */); By that I managed to assign several interfaces based on that driver now Thanks to all who answered (or considered to) Mario --------------------------- Newsgroups: comp.os.vxworks Subject: Re: 1553 bus Date: Fri, 08 Sep 2000 08:46:45 GMT From: David Briggs Organization: Deja.com - Before you buy. Message-ID: <8pa91h$1q5$1@nnrp1.deja.com> References: <8p81io$crb$1@nnrp1.deja.com> In article <8p81io$crb$1@nnrp1.deja.com>, S Austin wrote: > Is it possible to comparison shop for 1553 support for vxWorks or is > Compware Corp. the only game in town? Not that I know. There are a wide number of 1553 vendors - including my employers Radstone Technology. http://www.radstone.com HTH - -- - -- David Briggs, Principal Software Engineer Radstone Technology Tel: (+44/0) 1327 359444 Water Lane, Towcester Fax: (+44/0) 1327 358113 Northants NN12 6JN E-mail: briggs@radstone.co.uk UK WWW: http://www.radstone.co.uk/ Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- End of New-News digest ********************** From vxwexplo-errs@csg.lbl.gov Fri Sep 8 05:12:28 2000 From: f.pertin@staubli.com Date: Fri Sep 8 05:12:30 PDT 2000 Subject: debbugging a source code loaded from the target loader...... Hello VXWORKS and TORNADO gurus, I have a problem to debug my application. I loaded objects from the target loader, my symbols tables (host and target) are synchronized. Does any body know how to debug source code correponding to this object file. Everything is going like if the gdb does not know the debug informations of this object file. best regards Francois From vxwexplo-errs@csg.lbl.gov Fri Sep 8 08:43:24 2000 From: David Laight Date: Fri Sep 8 08:43:27 PDT 2000 Subject: re: DHCPC > I'm having some trouble with DHCPC. I'm trying to > use DHCPC after VxWorks is up to get an IP address > from a DHCP server that I'm running on LINUX > (DHCPD). I'm not trying to configure an interface; > I'm only trying to get an address for my application. > I see that the DHCP server is getting the DHCP > discovery messages, and it is responding with an > address to my interface's MAC address. > However, on the VxWorks side, the client simply repeats > the WAIT_OFFER state until it eventually times out. It > looks like the timeout message is the only event it ever > sees; DHCPC is not seeing the response from the DHCP server > at all. > ifShow on the interface shows that it *is* receiving > multicast packets. Ah yes - you will (almost certainly) find that the response from you dhcp server terminates after the last option byte. The vxworks client code expects the responses to be exactly the size specified in the RFC. Unfortunately the RFC has been superseded by one that allows an unrestricted options length. The lack of backward compatibility between the two RFCs was almost certainly not intenetional. It is probably that vxWorks is the only client code that performs this length check (it doesn't do the more usefule check that the TLV option don't fall of the end of the receive data). There are two checks for the packet length in the dhcpc code, a little hacking and they can both be patched out. However you will find that there are rather a lot of bugs in this code. I ended up raising so many TSRs the support group got overloaded:-) Anyway I fixed a significant number of bugs then gave WRS the fixed code back. David ---------------------------------------------------------------- David Laight email: dsl@tadpole.co.uk Tadpole Technology plc phone: +44 1223 428 232 Cambridge, UK fax: +44 1223 428 201 From vxwexplo-errs@csg.lbl.gov Fri Sep 8 09:31:59 2000 From: Charles Johnson Date: Fri Sep 8 09:32:02 PDT 2000 Subject: Alternate File Systems for VxWorks Does anyone know if there are any alternatives in terms of file systems (commercial or freeware) for VxWorks other than the dosFs2.0 that comes with VxWorks ?? A journaling file system with support for preallocation, contiguous allocation and load leveling (for flash disks) would be great. I've heard of a company called CrossStor (http://www.crosstor.com/home/index.html ) which appears to have something. Thanks. Charles Johnson Oresis Communications, Inc. cjohnson@oresis.com From vxwexplo-errs@csg.lbl.gov Sat Sep 9 04:03:08 2000 From: Vxworks Exploder Date: Sat Sep 9 04:03:10 PDT 2000 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Sat Sep 9 04:03:04 PDT 2000 Subject: debbugging a source code loaded from the target loader...... Subject: Porting a BSP from Tornado 1.0.1 to Tornado 2.0 Subject: Re: use of cascaded macros Subject: Re: How to download a modem driver to vxworks image Subject: Re: How to download a modem driver to vxworks image Subject: Re: Problem with VxWorks Ncr810 SCSI Driver Subject: re: DHCPC Subject: Alternate File Systems for VxWorks Subject: Re: Problem with VxWorks Ncr810 SCSI Driver Subject: Starting off Subject: Exceptions inside ISRs Subject: Re: Starting off Subject: Are these different: __asm vs __asm__ Subject: printing 64 bit unsigned decimal value Subject: gethostbyname() etc.. Subject: Re: printing 64 bit unsigned decimal value Subject: Re: gethostbyname() etc.. Subject: printf of "long long" Subject: Re: rpclib - errno 0xd0003 Subject: Re: gethostbyname() etc.. Subject: Re: Multiple instances of same (modern) NIC in an x86 target? Subject: Re: BSD4.4 Required Driver Entry point and structures. Subject: Everyone please atleast read this message over..... ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: debbugging a source code loaded from the target loader...... Date: Fri, 8 Sep 2000 14:12:10 +0200 From: f.pertin@staubli.com Organization: Lawrence Berkeley National Laboratory Message-ID: <8pal7m$p6m$1@overload.lbl.gov> Hello VXWORKS and TORNADO gurus, I have a problem to debug my application. I loaded objects from the target loader, my symbols tables (host and target) are synchronized. Does any body know how to debug source code correponding to this object file. Everything is going like if the gdb does not know the debug informations of this object file. best regards Francois --------------------------- Newsgroups: comp.os.vxworks Subject: Porting a BSP from Tornado 1.0.1 to Tornado 2.0 Date: Fri, 08 Sep 2000 14:58:03 +0200 From: Thomas Scholz Organization: Ericsson Eurolab Germany Message-ID: <39B8E25B.7550135B@eed.ericsson.se> Hi vxW wizards, 1) I would like to port a BSP from T101 to T2 using the T2 project tools. I am having a lot of problems with undefined and doubly defined macros because many definitions have been made in non-standard places (configAll.h instead of config.h, etc.) Can somebody provide me with information about what the project tools do when they scan my source files, how they build a BSP, etc. I hope to be able to fix many problems if I get a deeper insight into what happens. 2) There is one specific bug (maybe related to the above problems) that occurs here that is not our fault (I hope so, at least). I get the error: .../Tornado_2.0/target/h/private/cplusLibP.h:38: syntax error before `0' The code from cplusLibP.h is: typedef enum { OFF = 0, TERSE = 1, COMPLETE = 2 } CPLUS_DEMANGLER_MODES; It comes from the third line of the enumeration above, because there is an ugly macro definition in /Tornado_2.0/target/h/tffs/flbase.h #ifndef OFF #define OFF 0 #endif that makes "OFF" unusable for the enumeration. What is the best fix for this problem. Thanks, Th. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: use of cascaded macros Date: Fri, 8 Sep 2000 08:15:36 -0600 From: "Richard" Message-ID: <39b8f3ee.0@news> References: <8p2lqt$3g9$1@proxy2.fe.internet.bosch.de> <39B4F533.F56D7B9B@swri.edu> <8p2vvs$hmq$1@proxy2.fe.internet.bosch.de> <39b54d5d.0@news> "Jim Thomas" wrote in message news:wwn1hknc9e.fsf@atlas.cfht.hawaii.edu... > >>>>> "Richard" == Richard writes: > > Richard> try the following > Richard> typedef enum > Richard> { > Richard> K_STRING_1, > Richard> K_STRING_2, > Richard> K_STRING_3, > Richard> K_STRING_4, > Richard> ...., > Richard> NUMBER_K_STRINGS > Richard> } K_STRING_DEF_TYPE; > > Richard> you can remove entries from the definition, and the following > Richard> ones will be renumbered. I have not looked it up in ANSI C, but > Richard> in all of the compilers I have used, the first enumberated value > Richard> is 0, and each successive one is one larger than the previous > Richard> one. These values are usable anywhere an int is. As a bonus, > Richard> NUMBER_K_STRINGS can be used as the upper limit of the valid > Richard> indices (all valid index values are less than NUMBER_K_STRINGS), > Richard> and can be used to allocate your string table. > > But there is no longer any relationship between the values and the > indices. I use exactly his macros very happily (well, with parens :-). There is no less of a relationship between the enumeration and the indices and the defines and the indices. Just as with the defines each enumeration value is 1+previous. With the defined constants embedded in the string table, there is a difference, but that is not what was being described originally. In addition, at least with the compiler he is using, your proposed solution is not feasible, so he would be reduced to defining each one as a numeric value, not dependant on the previous ones which would cause huge headaches when it came time to remove/add entries from anywhere but the end of the table. If the process of generating the defines were automated, that problem could be resolved. > > char foo[] = { > #define bar 0 > "bar string", > #define mumble (bar + 1) > "mumble string", > #define last (mumble + 1) > (char *) NULL > } > .... > assert((last + 1) == (sizeof foo / sizeof (char *))); > Another solution that could resolve both problems would be something along the lines of: typedef enum { K_STRING_1, #define K_STRING_1_VALUE "String 1" K_STRING_2, #define K_STRING_2_VALUE "String 2" . . . NUMBER_K_STRINGS }; char StringTable[] = { K_STRING_1_VALUE, K_STRING_2_VALUE, . . . }; assert(NUMBER_K_STRINGS == (sizeof (StringTable) / sizeof (char *))); This would still assure that the table was at least the corect size, and solve the nested macros problem, and have a reasonable assurance that the strings and indices are matched. > Also to the poster who said "just use 0 1 2 3 4 ..." - baloney. He, I > think from following postings, used numeric names in the example but uses > real names in the actual code. The reason for symbols is exactly to > eliminate meaningless numbers. > > As for the original problem, how many levels do you have? I don't see any > way of controlling memory allocation with GNU cpp - maybe you need more > swap space? - -- rrapier sieversinst To reply to me, replace the space with @, and add the .com on the end. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: How to download a modem driver to vxworks image Date: Fri, 08 Sep 2000 14:29:51 GMT From: bryan Organization: Deja.com - Before you buy. Message-ID: <8pat41$ob6$1@nnrp1.deja.com> References: <8p90hk$jo3$1@nnrp1.deja.com> <39B8630C.2B9BB20C@cisco.com> Thank you for your help. But Actually I want to know that the driver will work or not if I add it as a .obj files. For my understand of VxWorks, when I install a driver, I need to add some functions in usrConfig.c. Is it correct? As I read from the User's Guide, I can use the windsh(shell) with startup(like a autoexec.bat) to creating additional devices. Do you know how to do it? It will help me a lot. Thanks again!!! Bryan > Hi Bryan, > > If the wdb component is enabled in the target image and if you are able > to > launch wdb from your host PC, you should be able to download the .obj > files > to the target anytime. > Hope this helps. > > regards, > Vijay > > bryan wrote: > > > Hi everyone, > > > > This is my first time to send a question in this newsgroup. > > > > I am developing a modem driver for VxWorks. I got a target hardware > > from the other company and the image of VxWorks is inside the IDE. > > That's mean I don't have any source code of BSP for this target. > > So I need to download my modem driver as a downloadable application. > > Is it possible to do that?! If so, how can I do it? > > > > Best Regards > > Bryan > > > > Sent via Deja.com http://www.deja.com/ > > Before you buy. > > --------------44BE564D72038D037FDD0D9F > Content-Type: text/x-vcard; charset=us-ascii; > name="vpeshkar.vcf" > Content-Transfer-Encoding: 7bit > Content-Description: Card for Vijay Kumar Peshkar > Content-Disposition: attachment; > filename="vpeshkar.vcf" > > begin:vcard > n:Peshkar;Vijay > x-mozilla-html:FALSE > org:Wipro Technologies,K2, Bangalore;Wishing you blue skies, a couple of daisies, some fresh air and a smile! > adr:;;;;;; > version:2.1 > email;internet:vpeshkar@cisco.com > title:Senior Software Engineer > fn:Vijay Kumar Peshkar > end:vcard > > --------------44BE564D72038D037FDD0D9F-- > > Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: How to download a modem driver to vxworks image Date: 8 Sep 2000 15:16:16 GMT From: borkhuis@lucent.com Organization: Lucent Technologies, Nieuwegein NL Message-ID: <8pavs0$grn$1@news1.xs4all.nl> References: <8p90hk$jo3$1@nnrp1.deja.com> <39B8630C.2B9BB20C@cisco.com> <8pat41$ob6$1@nnrp1.deja.com> bryan wrote: : Thank you for your help. But Actually I want to know that the : driver will work or not if I add it as a .obj files. For my : understand of VxWorks, when I install a driver, I need to add : some functions in usrConfig.c. Is it correct? No, usrConfig.c is not used for normal projects anymore. If you want to include drivers in the kernel you have to add these to sysLib.c or you have to include these in the project facility using CDF files. : As I read from the User's Guide, I can use the windsh(shell) with : startup(like a autoexec.bat) to creating additional devices. Do you : know how to do it? It will help me a lot. Thanks again!!! Yes, you can do that. Just enable the Startup scripting in your project and define a startup script in the boor parameters and write a startup script. In your case if you load the object code of your driver and call the start function of this driver the driver is enabled. You can start drivers all the time, as these are just some pieces of code, as long as you don't need some special initialisation code before interrupts are enabled. Groeten, Johan - -- o o o o o o o . . . ______________________________ o _____ || Johan Borkhuis | .][__n_n_|DD[ ====_____ | borkhuis@lucent.com | >(________|__|_[_________]_|___________________________| _/oo OOOOO oo` ooo ooo 'o!o!o o!o!o` === VxWorks FAQ: http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html === --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Problem with VxWorks Ncr810 SCSI Driver Date: Fri, 08 Sep 2000 17:38:07 +0200 From: Sebastien Fauris Organization: Club-Internet (France) Message-ID: <39B907DF.A508EE6F@club-internet.fr> References: <8p926j$ljj$1@nnrp1.deja.com> Hi, I don't know if this is your problem, but the driver of NCR810 I use from Wind River (Tornado 1.0.1) does not support shared PCI interrupt. I am waiting from them to correct this bug or hope if they don't, they would give me the source so I can correct it myself. Sebastien. --------------------------- Newsgroups: comp.os.vxworks Subject: re: DHCPC Date: Fri, 8 Sep 2000 16:43:38 +0100 (BST) From: David Laight Organization: Lawrence Berkeley National Laboratory Message-ID: <200009081543.QAA11506@dsl-2.tadpole.co.uk> > I'm having some trouble with DHCPC. I'm trying to > use DHCPC after VxWorks is up to get an IP address > from a DHCP server that I'm running on LINUX > (DHCPD). I'm not trying to configure an interface; > I'm only trying to get an address for my application. > I see that the DHCP server is getting the DHCP > discovery messages, and it is responding with an > address to my interface's MAC address. > However, on the VxWorks side, the client simply repeats > the WAIT_OFFER state until it eventually times out. It > looks like the timeout message is the only event it ever > sees; DHCPC is not seeing the response from the DHCP server > at all. > ifShow on the interface shows that it *is* receiving > multicast packets. Ah yes - you will (almost certainly) find that the response from you dhcp server terminates after the last option byte. The vxworks client code expects the responses to be exactly the size specified in the RFC. Unfortunately the RFC has been superseded by one that allows an unrestricted options length. The lack of backward compatibility between the two RFCs was almost certainly not intenetional. It is probably that vxWorks is the only client code that performs this length check (it doesn't do the more usefule check that the TLV option don't fall of the end of the receive data). There are two checks for the packet length in the dhcpc code, a little hacking and they can both be patched out. However you will find that there are rather a lot of bugs in this code. I ended up raising so many TSRs the support group got overloaded:-) Anyway I fixed a significant number of bugs then gave WRS the fixed code back. David - ---------------------------------------------------------------- David Laight email: dsl@tadpole.co.uk Tadpole Technology plc phone: +44 1223 428 232 Cambridge, UK fax: +44 1223 428 201 --------------------------- Newsgroups: comp.os.vxworks Subject: Alternate File Systems for VxWorks Date: Fri, 8 Sep 2000 09:31:46 -0700 From: Charles Johnson Organization: Lawrence Berkeley National Laboratory Message-ID: <8pb6q6$bsj$1@overload.lbl.gov> Does anyone know if there are any alternatives in terms of file systems (commercial or freeware) for VxWorks other than the dosFs2.0 that comes with VxWorks ?? A journaling file system with support for preallocation, contiguous allocation and load leveling (for flash disks) would be great. I've heard of a company called CrossStor (http://www.crosstor.com/home/index.html ) which appears to have something. Thanks. Charles Johnson Oresis Communications, Inc. cjohnson@oresis.com --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Problem with VxWorks Ncr810 SCSI Driver Date: Fri, 08 Sep 2000 17:21:51 GMT From: yeemoney@my-deja.com Organization: Deja.com - Before you buy. Message-ID: <8pb76m$56v$1@nnrp1.deja.com> References: <8p926j$ljj$1@nnrp1.deja.com> I think I may have gotten it to work. I set the ncr810PciMemOffset (located in ncr810Lib.c) to 0 and it stopped crashing. I still need to see how stable it is. Only me, Stanley Yee In article <8p926j$ljj$1@nnrp1.deja.com>, syee@connex.com wrote: > I'm trying to get the ncr810 driver to work on the 810 controller (and > eventually the 895 controller). I'm using Tornado 1.0.1 and VxWorks > 5.3.1. on a Pentium machine. The system is crashing everytime I try to > initialize the SCSI controller. Below are the functions calls I make > that lead up to the crash: > > ncr810CtrlCreate() Creates controller structure > intConnect() Connects ISR to interrupt vector > sysIntEnablePIC() enables interrupt > SetHwRegisters() sets hardware-dependent registers > ncr810CtrlInit() initializes controller > > The system crashes and reboots at the controller init function. I'm > assuming that there is some conflict between the interrupt and the > Interrupt Service Routine (ncr810Intr). If I don't include the > sysIntEnablePIC() function, the system doesn't crash and I'm able to do > a controller init. However, when I try a scsiPhysDevCreate or an > scsiAutoConfig, the system hangs. > > I'm positive the base address and interrupts I use are correct because > I verified it in the PCI configuration space. What's even stranger is > that I got it to work once but then all of a sudden it crashed and I > didn't have a backup at the time. Has anyone experienced a similar > problem? Any suggestions on how to fix this? Please e-mail me at > syee@connex.com. Thanks in advance. > > Stanley Yee > > Sent via Deja.com http://www.deja.com/ > Before you buy. > Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Starting off Date: Fri, 8 Sep 2000 22:08:01 +0200 From: "Pavius" Organization: Bezeq International Ltd. Message-ID: <39b938ac@news.bezeqint.net> Hi. I have been programming for quite some time, and have gathered enough knowledge in C and C++, and have a fairly good working knowledge in assembly. I'm about to start developing in a RTOS environment. As i read in the VxWorks programmers guide (which is about the only guide i found on the net), everything evolves around multitasking, event handling and dispatching. Thing is, i have no clue on where to actually begin ... I don't assume i can get my hands on some vxworks libs, just to "get the feeling of things", and since i can't find any books on the topics (and i don't mean "Real time programming in 24 hours), i can't seem to grasp how i can start writing programs that will eventually reside on DSPs and such. So my questions in a nutshell are : - - Where can i find tutorials about RTOS programming ? (and is this the best place to start) - - Are there any non-commercial licenses for a (maybe crippled) version of the vxworks kernel ( i am using VC6)? TIA omg, i'm a newbie ;) --------------------------- Newsgroups: comp.os.vxworks Subject: Exceptions inside ISRs Date: Fri, 08 Sep 2000 14:57:37 -0500 From: Michael White Organization: Motorola Message-ID: <39B944B0.3E0E633B@ftw.mot.com> Hi all, Does anyone out there know the behavior of VxWorks 5.4 on a MPC8260 if an exception (such as an instruction translation miss) occurs inside of an external interrupt handler? Also, is there any way to get text to the TTY output under this situation? The default VxWorks TTY drivers are being used. Thanks! - --- Michael White, qa1402@ftw.mot.com Cellular Infrastructure Group Ft. Worth, Texas (817)245-6418 --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Starting off Date: Fri, 8 Sep 2000 23:40:34 +0200 From: "Martin Raabe" Organization: T-Online Message-ID: <8pbmek$uff$18$1@news.t-online.com> References: <39b938ac@news.bezeqint.net> Hello Pavius, go and try Tornado Prototyper. Maybe this is what you want? http://www.WindRiver.com Ciao - -- Martin Raabe --------------------------- Newsgroups: comp.os.vxworks Subject: Are these different: __asm vs __asm__ Date: Fri, 08 Sep 2000 23:33:48 GMT From: "Jim" Organization: Magma Communications Ltd. Message-ID: Hello all, I have a real quick question for anyone who knows ... I am writing some assembly in VxWorks for an MPC860 and I was wondering if there is a difference between ... __asm ("........"); and __asm__ ("........"); Cheers and thanks in advance, Jim --------------------------- Newsgroups: comp.os.vxworks,cisco.eng.su-dev-vxworks,cisco.eng.cr.su-dev-vxworks Subject: printing 64 bit unsigned decimal value Date: Fri, 08 Sep 2000 17:49:10 -0700 From: Rajesh Goudar Organization: Cisco Systems, Inc. Message-ID: <39B98906.67FD69A2@cisco.com> Hi, How do I print a 64 bit (unsigned long long) variable using printf without getting any compiler warnings ? eg: main() { usigned long long time_cnt; ................ ................. printf(" the time tick = %??? \n", time_cnt); } i have used combinations of %d, %u, %ld %gu etc... all results in compiler warning. Any tips or help will be greatly appreciated. Thanks in advance, Rajesh --------------------------- Newsgroups: comp.os.vxworks Subject: gethostbyname() etc.. Date: Sat, 09 Sep 2000 01:33:02 GMT From: nunya Organization: Nunya Business Message-ID: <39B9934E.9C26A7A0@nunya.com> Hello all; I'm looking for the gethostbyname() stuff for VxWorks. I just need simple IP address resolution, nothing extensive. Is there any code out there for this already? Thanks! JLM jlm@nospam.dessertfirst.com --------------------------- Newsgroups: comp.os.vxworks,cisco.eng.su-dev-vxworks,cisco.eng.cr.su-dev-vxworks Subject: Re: printing 64 bit unsigned decimal value Date: Sat, 09 Sep 2000 01:36:33 GMT From: johnstona@avionics2.robins.af.mil Organization: Deja.com - Before you buy. Message-ID: <8pc46v$7to$1@nnrp1.deja.com> References: <39B98906.67FD69A2@cisco.com> > How do I print a 64 bit (unsigned long long) variable using printf > without getting any compiler warnings ? Amazing... I was just going to ask a related question. But first the answer to your query. Somewhere in the prog. guide it says that vxworks does not support %L (long long). So as far as wrs software, we are out of luck. This has been this way for as long as I can remember. I am going to ask if anyone has a "fix" for this defect/lack. It has been bugging me for quiet awhile now. - -al (sofeise) Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: gethostbyname() etc.. Date: Fri, 08 Sep 2000 21:48:06 -0400 From: Al Johnston Organization: MindSpring Enterprises Message-ID: <39B996D6.698E8F9B@mindspring.com> References: <39B9934E.9C26A7A0@nunya.com> > I'm looking for the gethostbyname() stuff for VxWorks. I just need > simple > IP address resolution, nothing extensive. Depends on how you want to resolve it... if dns, then vxworks already supports it. If you want yp (nis) then for some STRANGE reason WRS says your out of luck. We have a kludge that hooks into the vxworks into yp client, but it is hardware dependent. Have versions for sun2ce, mv2700, and hkv30 (maybe) if you are interested. I warn you though, the way it is implemented is via "self modifying code". But we have been using it here for 8 years now... - -al --------------------------- Newsgroups: comp.os.vxworks Subject: printf of "long long" Date: Fri, 08 Sep 2000 21:55:20 -0400 From: Al Johnston Organization: MindSpring Enterprises Message-ID: <39B99888.6A31EEC7@mindspring.com> the lack of support for printing a 64 bit integer has long annoyed me. My fingers just keep typing the %L, and the compiler has to remind me its not there... this happens every year or so... their memory is not to good.... Does someone have a "real" fix for this? (not a divide by ten and map to char in an array approach).... thanks --------------------------- Newsgroups: comp.os.vxworks Subject: Re: rpclib - errno 0xd0003 Date: Fri, 08 Sep 2000 21:59:15 -0400 From: Al Johnston Organization: MindSpring Enterprises Message-ID: <39B99973.96C9F9AA@mindspring.com> References: <39B5B563.D0E3CB51@mindspring.com> I sure wish they would not do that... I am never sure if I missed a "check for error return" somewhere... or if vxw is just being ill natured... thanks guys. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: gethostbyname() etc.. Date: Sat, 09 Sep 2000 04:38:51 GMT From: nunya Organization: Nunya Business Message-ID: <39B9BED9.34431EB2@nunya.com> References: <39B9934E.9C26A7A0@nunya.com> <39B996D6.698E8F9B@mindspring.com> Yes, a simple DNS resolver is all I need, but...I don't see any man pages or prototypes for the gethostbyname() stuff in a Tornado 1.0.1, non-SENS, x86 target directory. I do see some resolvxxx() routines in a SENS PPC target directory that look like the items I need there. Of course, the code I'm working on runs on multiple targets so I need something I can use on (at least) both targets. I don't mind writing some pseudo-api that front-ends resolvxxx(PPC) or gethostbyname()(x86) but I want to try to avoid writing the gethostbyname() etc..., again. Thanks! JLM Al Johnston wrote: > > I'm looking for the gethostbyname() stuff for VxWorks. I just need > > simple > > IP address resolution, nothing extensive. > > Depends on how you want to resolve it... if dns, then vxworks already > supports it. If you want yp (nis) then for some STRANGE reason > WRS says your out of luck. > > We have a kludge that hooks into the vxworks into yp client, but > it is hardware dependent. Have versions for sun2ce, mv2700, and > hkv30 (maybe) if you are interested. I warn you though, the > way it is implemented is via "self modifying code". > > But we have been using it here for 8 years now... > > -al --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Multiple instances of same (modern) NIC in an x86 target? Date: Sat, 09 Sep 2000 05:42:37 GMT From: Marcelo & Keyly Mourier Organization: AJAX Productions Message-ID: <39B9CED9.C608B01C@earthlink.com> References: <8nb8ng$ghd$1@overload.lbl.gov> Reply-To: jukema@earthlink.net The Intel Pro 100B is a widely available PCI 10/100Tx NIC that uses the i82559 MAC chip. There's a vxWorks driver for this board (if_fei.c) that we modified to support multiple units (NICs.) Most of the work was done in the driver's "attach" routine, where we had to assign a different block of PCI I/O space to each of the NICs. We also shared a single interrupt among all the NICs, so we had to modify the driver's interrupt handler to poll each card to find out the one(s) that needed to be serviced. Overall, it was a few days worth of work. Henrik Bergstrom wrote: > Dear VxWorks-users, > > I am having trouble finding a modern, commercially available, NIC which > works with several instances of the same card in an x86 target. Some > drivers are prepared for multiple instances, but don't work in > practice. > > I am in need of a END (MUX) driver and the only one my local > vxw-representative knew of was Intel's 82559, but I don't have access > to that NIC. > > What would make me really happy are END/MUX-drivers with support for > multiple instances of the same card for any of the following: > > o 3com 3c905C-TX "Fast EtherLink XL PCI" > (This has "theoretical support" for multiple instances, but not > fully so in practice.) > o SMC 9432TX "EtherPower II 10/100" > o any other modern, commercially available, card > > Any help, tips or recommendations is appreciated. > > Best regards, > > Henrik Bergstrom > > - -- > Henrik Bergstrom +46-8-628 28 28 henrik.bergstrom@intertex.se > Intertex Data AB (fax) 628 64 14 --------------------------- Newsgroups: comp.os.vxworks Subject: Re: BSD4.4 Required Driver Entry point and structures. Date: Sat, 09 Sep 2000 05:44:02 GMT From: Marcelo & Keyly Mourier Organization: AJAX Productions Message-ID: <39B9CF30.5FACD8B9@earthlink.com> References: <3999938A.50D6CDF2@lucent.com> Reply-To: jukema@earthlink.net The book TCP/IP Illustrated (volume 2) is what you're looking for... myhandle@lucent.com wrote: > Hi Group, > I need documentation which specifies the BSD4.4 network interface driver format, > specially the required driver entry points and the details of the various > interface data structures that are involved. If One of U can give me a pointer, > it will be appreciated... > thanx > Jish --------------------------- Newsgroups: comp.os.vxworks Subject: Everyone please atleast read this message over..... Date: Sat, 09 Sep 2000 07:27:02 GMT From: secret@netzerro.net Organization: Serious Cheat Proof PayPal $$$ Fast! Message-ID: <09090003.3307@netzerro.net> Reply-To: secretman@netzer0.com This is a Brand New CHEAT PROOF program involving PayPal, & it really works! Please check this out! Sorry for the Spam…It won't happen again, but this is a great new opportunity! Make $10,000 in 3 weeks using PayPal - CHEAT PROOF! This is a brand new, very serious business opportunity that will bring in thousands through your personal PayPal account, all from the safety of your home, with minimal effort! It WILL work because unlike other programs, this one is CHEAT PROOF! Nobody can cheat so everyone gets paid. $5 (that's it, no more ever!) WILL get you $10,000 in 3 weeks (I've personally made around $6,100, but $10K IS possible)! If you aren't familiar w/PayPal - Accounts are FREE, and you can sign up on my site below. But seriously, everyone should at least check this out if nothing else, you'll be sorry if you don't: http://zap.to/paypalcash --------------------------- End of New-News digest ********************** From vxwexplo-errs@csg.lbl.gov Sun Sep 10 04:03:06 2000 From: Vxworks Exploder Date: Sun Sep 10 04:03:08 PDT 2000 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Sun Sep 10 04:03:03 PDT 2000 Subject: Re: Exceptions inside ISRs Subject: Re: use of cascaded macros Subject: Problem with updating DHCPS entries ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: Exceptions inside ISRs Date: Sat, 09 Sep 2000 18:24:57 GMT From: john_94501@my-deja.com Organization: Deja.com - Before you buy. Message-ID: <8pdv98$5ng$1@nnrp1.deja.com> References: <39B944B0.3E0E633B@ftw.mot.com> Hi Michael, > Does anyone out there know the behavior of VxWorks 5.4 on a MPC8260 > if > an exception (such as an instruction translation miss) occurs inside of > an external interrupt handler? The example you quote is not a good one since that will be handled correctly by the OS (since the MMU and interrupt enable state are restored at the entry to an ISR, the OS needs to be able to handle the misses and load the TLB with the appropriate info from the hash table). Now, other exceptions, including the instruction translation error case, will result in two things. A message is stored in a special location (pointed to by sysExcMsg), and the target is rebooted. The bootrom should then display the exception message. True exceptions (i.e. everything except the MMU miss ones) in interrupt handlers are considered fatal; moral of the story: be very careful in your ISRs! HTH, John... Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: use of cascaded macros Date: 09 Sep 2000 18:59:45 -1000 From: Jim Thomas Organization: Canada France Hawai`i Telescope Message-ID: References: <8p2lqt$3g9$1@proxy2.fe.internet.bosch.de> <39B4F533.F56D7B9B@swri.edu> <8p2vvs$hmq$1@proxy2.fe.internet.bosch.de> <39b54d5d.0@news> <39b8f3ee.0@news> >>>>> "Richard" == Richard writes: Richard> "Jim Thomas" wrote in message Richard> news:wwn1hknc9e.fsf@atlas.cfht.hawaii.edu... >> But there is no longer any relationship between the values and the >> indices. I use exactly his macros very happily (well, with parens :-). Richard> There is no less of a relationship between the enumeration and Richard> the indices and the defines and the indices. As has been made clear, I was referring to physical location in the source code. Since the compiler provides no way of checking these or generating them (oh for a "real" macro processor :-) the visual check is all we have. Richard> Another solution that could resolve both problems would be Richard> something along the lines of: Richard> typedef enum Richard> { Richard> K_STRING_1, Richard> #define K_STRING_1_VALUE "String 1" Richard> K_STRING_2, Richard> #define K_STRING_2_VALUE "String 2" Richard> . Richard> . Richard> . Richard> NUMBER_K_STRINGS Richard> }; Richard> char StringTable[] = Richard> { Richard> K_STRING_1_VALUE, Richard> K_STRING_2_VALUE, Richard> . Richard> . Richard> . Richard> }; Richard> assert(NUMBER_K_STRINGS == (sizeof (StringTable) / sizeof (char *))); Richard> This would still assure that the table was at least the corect Richard> size, and solve the nested macros problem, and have a reasonable Richard> assurance that the strings and indices are matched. Nice inversion :-) Unfortunately when the symbols are not numeric but meaningful it's harder to make sure the table is right. The assert helps, but it's still easy to leave one out and duplicate another :-( Jim --------------------------- Newsgroups: comp.os.vxworks Subject: Problem with updating DHCPS entries Date: Sun, 10 Sep 2000 07:53:44 GMT From: anderberg@my-deja.com Organization: Deja.com - Before you buy. Message-ID: <8pfem4$ou3$1@nnrp1.deja.com> Can anybody suggest how one can update an entry added to DHCPS by dhcpsLeaseEntryAdd call. Or, alternatively, restart DHCP server without board reboot, to use new tables. There are some functions (dhcpsCleanup, dhcpsFreeResource...) which it seems are concerned to this issue, but amazingly, they are not documented. I use VxWorks 5.4 on PPC860 Thanks in advance Alex Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- End of New-News digest ********************** From vxwexplo-errs@csg.lbl.gov Sun Sep 10 21:07:04 2000 From: Neelesh Dwivedi Date: Sun Sep 10 21:07:07 PDT 2000 Subject: Route Change. Hello All, I have been studying VxWorks and Tornado for the past one month. I had a query regarding the Routing Table maintained by VxWorks. 1. Does it confirm to any MIB ? and 2. In case of any change in the routing table, does it generate any route change notification/ trap for other parts of the application? Thanks. Neelesh Dwivedi. From vxwexplo-errs@csg.lbl.gov Mon Sep 11 02:41:58 2000 From: "Balaji Ramachandran" Date: Mon Sep 11 02:42:00 PDT 2000 Subject: North / South Bridge in PCI? Tornado, VxWorks Hi VxWorks veterans, Can any one give clear picture about, what is the difference between South Bridge / North Bridge in PCI?, how to identify in a system, whether the bridge is north / south?. Thanks & Regards, R.Balaji. From vxwexplo-errs@csg.lbl.gov Mon Sep 11 04:03:07 2000 From: Vxworks Exploder Date: Mon Sep 11 04:03:09 PDT 2000 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Mon Sep 11 04:03:04 PDT 2000 Subject: Re: dosFs2 chkdsk slow to run from bootrom Subject: Re: HOT HOT - VxWorks S/W Developers Needed - Ottawa Subject: Route Change. Subject: Booting the pcPentium target ? Subject: Re: debbugging a source code loaded from the target loader...... Subject: Re: socket Subject: Re: Targetting COM ports 3 and 4 - need help Subject: Re: Linking archives Subject: Re: gethostbyname() etc.. Subject: North / South Bridge in PCI? ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: dosFs2 chkdsk slow to run from bootrom Date: Sun, 10 Sep 2000 13:26:15 +0200 From: Leonid Rosenboim Organization: BitBand Technologies Ltd. http://www.bitband.com Message-ID: <39BB6FD7.76855B10@bitband.com> References: <8p4osn$ja8$1@nnrp1.deja.com> Reply-To: leonid@bitband.com Actually, this is very simple - the CHKDSK utility in DosFs2 would take advantage of available memory to hasten the process by keeping the entire FAT in RAM, if this however is impossible, it will use the "second" FAT copy on the disk as its working copy of FAT. The "working copy" is where the software keeps track of cluster chains traversed etc. So what you need is to define a few defines to make bootCOnfig.c allocate more memory, for PC target these are: :#define INCLUDE_ADD_BOOTMEM /* Add upper memory to low memory bootrom*/ :#define ADDED_BOOTMEM_SIZE 0x00200000 /* 2MB */ Dont know how much memory will be sufficient for you, your milage will vary ... Leonid paulvroberts@my-deja.com wrote: > Hi there, > > Does anyone know why chkdsk would be much slower when run from the > bootrom, then when run after VxWorks is loaded? > > We are using FAT32 (singly dos partitioned) 9GB SCSI disks and it takes > 65 seconds in the bootrom, whereas after VxWorks is running, it takes > just 3-4 seconds! > > What I did was to "enhance" the bootConfig.c file to allow booting using > dosFs2. I use a 128K cache, and call chkdsk via the icotl FIOCHKDSK. > > I am using the MVME2700 BSP with the bootrom_uncmp target, so presumably > that's not rom-resident. > > Any other ideas? > > Thanks, > > Paul Roberts. > > Sent via Deja.com http://www.deja.com/ > Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: HOT HOT - VxWorks S/W Developers Needed - Ottawa Date: Sun, 10 Sep 2000 12:50:38 +0100 From: molochai Message-ID: <9055d3fb49%molochai@vapour-trail.demon.co.uk> References: <8p98p4$nc$1@news.igs.net> In message <8p98p4$nc$1@news.igs.net> "PROCOM" wrote: I'm sure that nobody objects to an occasional job advert on this NG, but sending the same advert several times a week is excessive IMHO. > Title: C++ / OOD VxWorks Software Developers > [snip rest of advert] --------------------------- Newsgroups: comp.os.vxworks Subject: Route Change. Date: Mon, 11 Sep 2000 09:38:23 +0530 From: Neelesh Dwivedi Organization: Lawrence Berkeley National Laboratory Message-ID: <8phm7k$rqk$1@overload.lbl.gov> Hello All, I have been studying VxWorks and Tornado for the past one month. I had a query regarding the Routing Table maintained by VxWorks. 1. Does it confirm to any MIB ? and 2. In case of any change in the routing table, does it generate any route change notification/ trap for other parts of the application? Thanks. Neelesh Dwivedi. --------------------------- Newsgroups: comp.os.vxworks Subject: Booting the pcPentium target ? Date: Mon, 11 Sep 2000 04:49:08 GMT From: mishra.r@dcmtech.co.in Organization: Deja.com - Before you buy. Message-ID: <8pho7v$73m$1@nnrp1.deja.com> Hi all, I need to boot my pcPentium target over the Ethernet.(ie. by downloading the vxWorks image). I have a NE2000 compatible ethernet card, have set the base address and IRQ for that in the config.h. So, after when the image is created the loader is unable to load that image from the host. Moreover, theres no message on the ftp server running on the host. regards Rajesh. Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: debbugging a source code loaded from the target loader...... Date: Mon, 11 Sep 2000 08:42:41 +0200 From: "Markus MItterer" Organization: KPNQwest customer news service Message-ID: <96%u5.6$AP1.615@nreader1.kpnqwest.net> References: <8pal7m$p6m$1@overload.lbl.gov> Hi Francois, I found out two possibilities to solve the problem: - - start your debugger and open the command-line window (Debug -> Debug Window -> DebugCommandLine). Then you have to type: add-symbol-file and your debugger reads all symbols from the object file. - - start your debugger and download the modul you want to debug with Tornado2. Of course, the object-files must be the same as already loaded. hope that helps Markus --------------------------- Newsgroups: comp.os.vxworks Subject: Re: socket Date: Mon, 11 Sep 2000 06:29:21 GMT From: ba_ajay@my-deja.com Organization: Deja.com - Before you buy. Message-ID: <8phu3l$d6t$1@nnrp1.deja.com> References: <8p9ucs$llg$1@nnrp1.deja.com> <8pa6le$ao$1@news1.xs4all.nl> In article <8pa6le$ao$1@news1.xs4all.nl>, borkhuis@lucent.com wrote: > ba_ajay@my-deja.com wrote: > : 1. Which symbol table to use(downloadable or built-in) ? > > : 2. is built-in symbol table the right one for a bootable appln ? > > Yes, that is the right one. > > : 3. if use a downloadable application , i get no errors in compilation. > : while downloading, error message is received as undefined symbols > : socket,bind etc.. What do i need to include ? > > In your bootable application you should include the network components, > including the Ping client. In the simulator the first lines should be like > this: > > Attached TCP/IP interface to nt unit 0 > Attaching interface lo0...done > > Adding 3069 symbols for standalone. > > Groeten, > Johan > Hi, Thanks for ur guidance. Simulator has the lines as above. I have included the ping client in the network components, but still i am not able to ping ip address 90.0.0.1 & 90.0.0.2 on NT(dos prompt) . 1. In NT ping localhost works whereas in the VxWorks Shell prompt sometimes doesnt work(sometimes does work). I have changed true to false in Config.h . Do i need to run any of the makefiles for the change come into effect ? Question : Can VxWorks be able to see 2 NT hosts as part of networking feature.i.e communication b/n VxWorks host machines. Can u help me on this ? ajay > -- > o o o o o o o . . . ______________________________ > o _____ || Johan Borkhuis | > .][__n_n_|DD[ ====_____ | borkhuis@lucent.com | > >(________|__|_[_________]_|___________________________| > _/oo OOOOO oo` ooo ooo 'o!o!o o!o!o` > === VxWorks FAQ: http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html === > Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Targetting COM ports 3 and 4 - need help Date: Mon, 11 Sep 2000 08:39:29 +0200 From: "Michael Lawnick" Organization: Deja.com - Before you buy Message-ID: <200009110638.IAA32669@net.softec.de> References: <967826453.9627.0.nnrp-14.c1ed3722@news.demon.co.uk> Hi Peter, verify that you have assigned a fixed port/interrupt pair to your COMs in BIOS. I had a similar problem on an other PC-board were this parameters were set to automatic. After setting up fixed, the problem had gone. HTH > Hi > > I am running my application on an Arcom Media GX Board which uses a 233Mhz > Cyriz chip. > > This board works with the Pentium BSP and contains 4 COM ports. I want to > read and write to these ports but can only manage to access ports COM1 and > COM2. I have added the relevant code to sysSerial.c and config.h but it > makes no difference. WindRiver themselves tell me that it is probably more > difficult than I think and I will probably have to allow COM1 and 3 to share > one Interrupt Leven and COM2 and COM4 to share another and then do the > detection of which COM port is relevant within ISR code. > > I do have enough interrupts for 4 ports but I realise I may not be allowed > to handle them separately. > > Has anyone done interrupt sharing before ?? > > Many thanks to replies in advance > > Peter Deed > Infocell Telecom > ================== > > > > > > > _____________________________________________________________ > Deja.com: Before you buy. > http://www.deja.com/ > * To modify or remove your subscription, go to > http://www.deja.com/edit_sub.xp?group=comp.os.vxworks > * Read this thread at > http://www.deja.com/thread/%3C967826453.9627.0.nnrp-14.c1ed3722%40news.demon.co.uk%3E > MfG Lawnick, SOFTEC GmbH ============================================== SOFTEC GmbH Tel +49-731-96600-0 Promenade 17 Fax +49-731-96600-23 D-89073 Ulm Michael Lawnick Germany lawnick@softec.de ============================================== Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Linking archives Date: 11 Sep 2000 04:04:03 -0500 From: Bob Westbrook Message-ID: References: <8p5k76$90e$1@proxy2.fe.internet.bosch.de> <8p5md8$40e$1@news1.xs4all.nl> <8p791k$gss$1@proxy2.fe.internet.bosch.de> <8p7f57$enc$1@news1.xs4all.nl> <8p802g$gjl$1@proxy2.fe.internet.bosch.de> <8p9v1n$4nh$1@proxy2.fe.internet.bosch.de> On Fri, 8 Sep 2000 07:56:37 +0200, Gunter Ritter wrote: > >I thought about this, too. But I didn´t try it yet. I find it a little bit >strange to do it this way. >Well, I try it now. > >Gunter > Did it work ? --------------------------- Newsgroups: comp.os.vxworks Subject: Re: gethostbyname() etc.. Date: Mon, 11 Sep 2000 11:42:50 +0200 From: Christian Olsson Organization: Ericsson Microwave Systems AB Message-ID: <39BCA91A.FBF8AFFE@emw.ericsson.se> References: <39B9934E.9C26A7A0@nunya.com> Maybe "vxman hostGetByName" will do it for you? /Christian Olsson nunya wrote: > > Hello all; > > I'm looking for the gethostbyname() stuff for VxWorks. I just need > simple > IP address resolution, nothing extensive. > > Is there any code out there for this already? > > Thanks! > JLM > > jlm@nospam.dessertfirst.com --------------------------- Newsgroups: comp.os.vxworks Subject: North / South Bridge in PCI? Date: Mon, 11 Sep 2000 15:22:22 +0530 From: "Balaji Ramachandran" Organization: Lawrence Berkeley National Laboratory Message-ID: <8pibak$7jf$1@overload.lbl.gov> Tornado, VxWorks Hi VxWorks veterans, Can any one give clear picture about, what is the difference between South Bridge / North Bridge in PCI?, how to identify in a system, whether the bridge is north / south?. Thanks & Regards, R.Balaji. --------------------------- End of New-News digest ********************** From vxwexplo-errs@csg.lbl.gov Mon Sep 11 04:50:20 2000 From: David Laight Date: Mon Sep 11 04:50:22 PDT 2000 Subject: Re: North / South Bridge in PCI? > Hi VxWorks veterans, > > Can any one give clear picture about, what is the difference between > South Bridge / North Bridge in PCI?, how to identify in a system, > whether the bridge is north / south?. > The North (or host) bridge converts the CPU bus to a PCI bus. This also tends to contain the memory controller (it simplifies the bus arbitration if this is all done in one place). The South bridge is a PCI device that support all the standard (PC) IO devices (serial ports, disk controller, ISA bus etc). Some of the newest PC PCI chipsets are putting the disk controller into the north bridge. This gives the disk controlles a faster path into memory. Also look out for systems that use a private bus to pass data between the north and south bridges - again to get faster throughput and/or lower latencies. David ---------------------------------------------------------------- David Laight email: dsl@tadpole.co.uk Tadpole Technology plc phone: +44 1223 428 232 Cambridge, UK fax: +44 1223 428 201 From vxwexplo-errs@csg.lbl.gov Tue Sep 12 03:05:33 2000 From: "Krishna Kishore Reddy Sureddy" Date: Tue Sep 12 03:05:35 PDT 2000 Subject: vxWorks Task Hi, Can a task in vxWorks get suspended other than the conditions presented below.Correct me if i am Wrong. 1. vxWorks calls a. ts() b. taskSuspend() c. sigsuspend() d. pause() e. nanosleep() f. aio_suspend g. sigtimedwait() h. sigwaitinfo() 2. Task level exception 3. Breakpoint in task 4. Attaching task to Debugger 5. CS_MSEC_DELAY(msec), a vxWorks Macro 6. Error in Allocating/Freeing Memory 7. When a pipe is full and task attempts to write to a pipe 8. One task stack overflows and corrupts other tasks stack 9. ioctl() call causes non-recoverable error 10. If an invalid device descriptor in the valid range of device descriptors numbers is passed to an I/O function 11. Some errors caused in the system even though the task itself may not have caused that error such as a. Device driver doing a hardware access that generates an error b. Bus error caused by a DMA cycle which was set up by a totally different task c. Bus errors and access violations due to task level code, driver code, interrupt handlers Thanks and Regards, S.K.K.Reddy. From vxwexplo-errs@csg.lbl.gov Tue Sep 12 04:03:07 2000 From: Vxworks Exploder Date: Tue Sep 12 04:03:10 PDT 2000 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Tue Sep 12 04:03:04 PDT 2000 Subject: Re: Are these different: __asm vs __asm__ Subject: re: North / South Bridge in PCI? Subject: Intel PRO/10+ ISA (82595) network driver? Subject: second serial port on the MBX860 Subject: how to reverse DEFLATE Subject: Re: second serial port on the MBX860 Subject: ATA CD-ROM access Subject: shared memory Subject: Simulator Fonts Subject: Re: Linking archives Subject: Re: printf of "long long" Subject: Re: gethostbyname() etc.. Subject: powerpc stack address in romInit() Subject: Re: powerpc stack address in romInit() Subject: Re: powerpc stack address in romInit() Subject: Re: powerpc stack address in romInit() Subject: Choosing a Tornado 2 Intel x86 BSP Subject: How to get DosFS2 ? Subject: network communication Subject: Re: Choosing a Tornado 2 Intel x86 BSP Subject: How to delay current task for some microseconds? Subject: Re: printf of "long long" Subject: Re: How to delay current task for some microseconds? Subject: vxWorks Task ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: Are these different: __asm vs __asm__ Date: Mon, 11 Sep 2000 12:08:26 -0500 From: "Keith McHugh" Organization: Esat Net Customer Message-ID: <39bcbc59.0@seat.s3two.ie> References: Reply-To: "Keith McHugh" I don't know but __asm__ works fine to me. Did you try both of them on your board? I'm afraid __asm won't work. "Jim" wrote in message news:wHeu5.1254$gi1.22001@news.magma.ca... > Hello all, > I have a real quick question for anyone who knows ... I am writing some > assembly in VxWorks for an MPC860 and I was wondering if there is a > difference between ... > > __asm ("........"); > and > __asm__ ("........"); > > Cheers and thanks in advance, > Jim > > > --------------------------- Newsgroups: comp.os.vxworks Subject: re: North / South Bridge in PCI? Date: Mon, 11 Sep 2000 12:50:34 +0100 (BST) From: David Laight Organization: Lawrence Berkeley National Laboratory Message-ID: <200009111150.MAA17403@dsl-2.tadpole.co.uk> > Hi VxWorks veterans, > > Can any one give clear picture about, what is the difference between > South Bridge / North Bridge in PCI?, how to identify in a system, > whether the bridge is north / south?. > The North (or host) bridge converts the CPU bus to a PCI bus. This also tends to contain the memory controller (it simplifies the bus arbitration if this is all done in one place). The South bridge is a PCI device that support all the standard (PC) IO devices (serial ports, disk controller, ISA bus etc). Some of the newest PC PCI chipsets are putting the disk controller into the north bridge. This gives the disk controlles a faster path into memory. Also look out for systems that use a private bus to pass data between the north and south bridges - again to get faster throughput and/or lower latencies. David - ---------------------------------------------------------------- David Laight email: dsl@tadpole.co.uk Tadpole Technology plc phone: +44 1223 428 232 Cambridge, UK fax: +44 1223 428 201 --------------------------- Newsgroups: comp.os.vxworks Subject: Intel PRO/10+ ISA (82595) network driver? Date: Mon, 11 Sep 2000 14:43:31 +0200 From: Albert Wijnja Organization: MECO Equipment Engineers BV Message-ID: <39BCD373.E63BC046@meco.nl> Reply-To: Albert.Wijnja@meco.nl Hello, Does anyone know if there is a driver for the Intel PRO/10+ ISA network adapter available? This card has a 82595 chipset. Albert Wijnja --------------------------- Newsgroups: comp.os.vxworks Subject: second serial port on the MBX860 Date: Mon, 11 Sep 2000 13:31:44 GMT From: "Øyvind Holmeide" Organization: Enitel Internet Public Access Message-ID: <495v5.8872$Dxe.201235968@news.telia.no> I am trying to configure VxWorks for a second serial port (RS232) on the MBX860 board from Motorola. The BSP is only prepared for one serial port (the console). Thus, an update of sysSerial.c for e.g. SMC2 is needed. I have tried without success based on the corresponding BSP for the ads860 board. Anyone been through this? regards Øyvind Holmeide --------------------------- Newsgroups: comp.os.vxworks Subject: how to reverse DEFLATE Date: Mon, 11 Sep 2000 16:35:06 +0200 From: "Ton den Hartog" Organization: UUNET-NL (http://www.nl.uu.net) Message-ID: <8piqfh$gu8$1@porthos.nl.uu.net> Hi, I have an old program in a bootrom.bin image that cannot easily be re-generated and that I would like to patch a little bit (only change a string with an IP-addres in) Apparantly this image is compressed with the Tornado defalte tool. Is there a Tornado-deflate compatible tool available to uncompress the file so that I can make my patch ? We use Tornado 2 on Windows NT but a HP-UNIX tool is fine also. Ton den Hartog --------------------------- Newsgroups: comp.os.vxworks Subject: Re: second serial port on the MBX860 Date: 11 Sep 2000 14:35:57 GMT From: borkhuis@lucent.com Organization: Lucent Technologies, Nieuwegein NL Message-ID: <8piqkd$46v$1@news1.xs4all.nl> References: <495v5.8872$Dxe.201235968@news.telia.no> oeyvind Holmeide wrote: : I am trying to configure VxWorks for a second serial port (RS232) on the : MBX860 board from Motorola. The BSP is only prepared for one serial port : (the console). Thus, an update of sysSerial.c for e.g. SMC2 is needed. I : have tried without success based on the corresponding BSP for the ads860 : board. : Anyone been through this? There is a package available from WindRiver with support for the 3 SCC's. I got this package from our FAE, so ask your contact at WRS for this. Groeten, Johan - -- o o o o o o o . . . ______________________________ o _____ || Johan Borkhuis | .][__n_n_|DD[ ====_____ | borkhuis@lucent.com | >(________|__|_[_________]_|___________________________| _/oo OOOOO oo` ooo ooo 'o!o!o o!o!o` === VxWorks FAQ: http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html === --------------------------- Newsgroups: comp.os.vxworks Subject: ATA CD-ROM access Date: Mon, 11 Sep 2000 14:43:09 GMT From: newcastle1@my-deja.com Organization: Deja.com - Before you buy. Message-ID: <8pir1n$e6a$1@nnrp1.deja.com> Anyone know how to acess a ATA/IDE CDROM? I have used a similar route as mounting a hard disk, using ataDevCreate (). The CDROM is on the second device on the first controller. The device itself becomes visible, as though it is mounted. But then I cannot access any of the files using the cdromFS functions. any help appreciated cheers Richard Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: shared memory Date: Mon, 11 Sep 2000 16:56:47 +0200 From: Marc Waber Organization: Swisscom IP+ (post doesn't reflect views of Swisscom) Message-ID: <39BCF2AF.F80C8C03@stest.ch> Dear all! I am trying to use shared memory with a CPV5350 as a master and a MCPN765 as a slave. The master sets up the anchor correctly, but the slave can't find it, and hangs itself with the procedure "smIsAlive:..." with a wrong ready-value and no heartbeat. Any hints or ideas? Thanks, Marc - -- Marc Waber E-mail: marc.waber@stest.ch STS Switching Test Solutions AG Phone: +41-76-355-6713 Foerrlibuckstr. 62 / Postfach 74 Fax: +41-1-355-6605 CH-8037 Zurich Homepage: www.stest.com --------------------------- Newsgroups: comp.os.vxworks Subject: Simulator Fonts Date: Mon, 11 Sep 2000 17:24:25 +0200 From: Lars Herczeg Organization: EPFL Message-ID: Hi folks! I'm working on a Windoz NT machine and the fonts of the VXWorks Simulator window are just so tiny you can't read'em. Where can I change the size of the VXWorks Simulator font? Any help is apreciated. Cheers, Lars - ----------------------------------------------------------------- Lars Herczeg Swiss Federal Institute of Technology Lausanne (EPFL) Electrical Engineering Division Ch. des Triaudes 9/344 CH-1024 Ecublens Switzerland E-mail: Lars.Herczeg@epfl.ch - ----------------------------------------------------------------- --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Linking archives Date: Mon, 11 Sep 2000 16:12:15 GMT From: werner_engeln@my-deja.com Organization: Deja.com - Before you buy. Message-ID: <8pj08d$kv9$1@nnrp1.deja.com> References: <8p5k76$90e$1@proxy2.fe.internet.bosch.de> <8p5md8$40e$1@news1.xs4all.nl> <8p791k$gss$1@proxy2.fe.internet.bosch.de> <8p7f57$enc$1@news1.xs4all.nl> <8p802g$gjl$1@proxy2.fe.internet.bosch.de> <39B7882E.BBFB0BF5@lucent.com> <8p82jm$jgs$1@proxy2.fe.internet.bosch.de> If you cannot redesign nor can combine these libraries into one, add a library twice: project1.a project2.a project1.a The linker works sequentially and takes all unresolved symbols from project1.a, afterwards takes all unresolved symbols from project2.a and again takes all yet unresolved symbols from project1.a. good luck, Werner In article <8p82jm$jgs$1@proxy2.fe.internet.bosch.de>, Gunter Ritter wrote: > You are right. These archives depend on each other. > > The code I have to use can not be redesigned. > I want to structure the source files in several projects and link them > together to a bootable project. > > Solution b) is not possible, because the code is used in another project, > too. > > How can I combine the archives as proposed in solution a)? > Is there a different solution to link the object files (or any ohter kind of > output) of several projects to a bootable project? > > Gunter > > Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: printf of "long long" Date: Mon, 11 Sep 2000 18:01:37 GMT From: John Fusco Organization: Deja.com - Before you buy. Message-ID: <8pj6lg$t2f$1@nnrp1.deja.com> References: <39B99888.6A31EEC7@mindspring.com> The problem is that without a prototype argument, the compiler assumes that it must put an int on the stack. Given the printf prototype: printf(char *, ... ); The compiler knows nothing about what arguments the function expects, so it has to make assumptions. This is what the old K&R compilers did before there were prototypes. That is, all integer arguments are converted to "int", all floating point arguments are converted to "double" and all structures, arrays, etc. become pointers. I don't suppose a C++ iostream would have this problem. In article <39B99888.6A31EEC7@mindspring.com>, Al Johnston wrote: > the lack of support for printing a 64 bit integer has long annoyed me. > > My fingers just keep typing the %L, and the compiler has to remind > me its not there... this happens every year or so... their memory > is not to good.... > > Does someone have a "real" fix for this? (not a divide by ten > and map to char in an array approach).... > > thanks > Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: gethostbyname() etc.. Date: Mon, 11 Sep 2000 18:28:06 GMT From: nunya Organization: Nunya Business Message-ID: <39BD242C.455C79E0@nunya.com> References: <39B9934E.9C26A7A0@nunya.com> <39BCA91A.FBF8AFFE@emw.ericsson.se> Unfortunately, that only returns entries in the host table that have been added (by you) using hostAdd(). I need something like the resolvLib stuff that is in a SENS install, but I need it to work for older versions of vxWorks, too.. Given the responses I've gotten, it doesn't sound like there is anything available. I haven't talked to WRS yet, but, based on past experience with them, they won't be able to help either. :-( Thanks! JLM Christian Olsson wrote: > Maybe "vxman hostGetByName" will do it for you? > > /Christian Olsson > > nunya wrote: > > > > Hello all; > > > > I'm looking for the gethostbyname() stuff for VxWorks. I just need > > simple > > IP address resolution, nothing extensive. > > > > Is there any code out there for this already? > > > > Thanks! > > JLM > > > > jlm@nospam.dessertfirst.com --------------------------- Newsgroups: comp.os.vxworks Subject: powerpc stack address in romInit() Date: Mon, 11 Sep 2000 13:46:15 -0500 From: Bo Yu Organization: Motorola CIG Message-ID: <39BD2877.E4A02565@email.mot.com> - --------------AD0D435A75C7423A84ED8ED6 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Hi, All PowerPC BSPs have the following statement in romInit() to load the initial the stack pointer. I cannot find the definition for STACK_ADRS in romInit(). /* initialize the stack pointer */ lis sp, HIADJ(STACK_ADRS) addi sp, sp, LO(STACK_ADRS) Could anyone tell me what the value for STACK_ADRS is? Thanks. - -- Bo Yu - --------------AD0D435A75C7423A84ED8ED6 Content-Type: text/html; charset=us-ascii Content-Transfer-Encoding: 7bit Hi,

All PowerPC BSPs have the following statement in romInit() to load
the initial the stack pointer. I cannot find the definition for
STACK_ADRS in romInit().

  /* initialize the stack pointer */

  lis      sp, HIADJ(STACK_ADRS)
  addi   sp, sp, LO(STACK_ADRS)

Could anyone tell me what the value for STACK_ADRS is? Thanks. 

-- 
Bo Yu
  - --------------AD0D435A75C7423A84ED8ED6-- --------------------------- Newsgroups: comp.os.vxworks Subject: Re: powerpc stack address in romInit() Date: Mon, 11 Sep 2000 20:37:27 GMT From: john_94501@my-deja.com Organization: Deja.com - Before you buy. Message-ID: <8pjfq4$8vr$1@nnrp1.deja.com> References: <39BD2877.E4A02565@email.mot.com> Hi, > All PowerPC BSPs have the following statement in romInit() to load > the initial the stack pointer. I cannot find the definition for > STACK_ADRS in romInit(). It should be in a header file either in the BSP or in the 'all' directory. I think that it will be the address that the kernel image loads to (i.e. the address that the bootrom says after "Starting at"). This works since the stack grows down and the image up from that point. You should be able to verify this by searching for the header file, or by simply disassembling the vxWorks image and seeing what value is loaded! HTH, John... > /* initialize the stack pointer */ > > lis sp, HIADJ(STACK_ADRS) > addi sp, sp, LO(STACK_ADRS) > > Could anyone tell me what the value for STACK_ADRS is? Thanks. > > -- > Bo Yu > > --------------AD0D435A75C7423A84ED8ED6 > Content-Type: text/html; charset=us-ascii > Content-Transfer-Encoding: 7bit > > > > Hi, >

All PowerPC BSPs have the following statement in romInit() to load >
the initial the stack pointer. I cannot find the definition for >
STACK_ADRS in romInit(). >

  /* initialize the stack pointer */ >

  lis      sp, HIADJ(STACK_ADRS) >
  addi   sp, sp, LO(STACK_ADRS) >

Could anyone tell me what the value for STACK_ADRS is? Thanks. > 

-- 
> Bo Yu
>   > > --------------AD0D435A75C7423A84ED8ED6-- > > Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: powerpc stack address in romInit() Date: Tue, 12 Sep 2000 00:34:04 GMT From: rjdejoie@kmrmail.kmr.ll.mit.edu (Bob DeJoie) Message-ID: <39bd797a.13461736@news> References: <39BD2877.E4A02565@email.mot.com> On Mon, 11 Sep 2000 13:46:15 -0500, Bo Yu wrote: >Could anyone tell me what the value for STACK_ADRS is? Thanks. From MV2700 BSP target/config/all/configAll.h #if ((CPU_FAMILY == MIPS) || (CPU_FAMILY == PPC)) #define STACK_RESIDENT RAM_DST_ADRS #else #define STACK_RESIDENT _sdata #endif #if (_STACK_DIR == _STACK_GROWS_DOWN) #ifdef ROM_RESIDENT #define STACK_ADRS STACK_RESIDENT #else #define STACK_ADRS _romInit #endif /* ROM_RESIDENT */ #else /* _STACK_DIR == _STACK_GROWS_UP */ #ifdef ROM_RESIDENT #define STACK_ADRS (STACK_RESIDENT-STACK_SAVE) #else #define STACK_ADRS (_romInit-STACK_SAVE) #endif /* ROM_RESIDENT */ #endif /* _STACK_DIR == _STACK_GROWS_UP */ ================================================ Bob DeJoie Kwajalein Missile Range Republic of the Marshall Islands email: rjdejoie@kmrmail.kmr.ll.mit.edu     phone: 805-355-5855 fax: 805-355-3833 Time Zone: GMT +12 hours                      ================================================ --------------------------- Newsgroups: comp.os.vxworks Subject: Re: powerpc stack address in romInit() Date: Mon, 11 Sep 2000 17:30:22 -0700 From: Rajesh Tripathi Organization: Globix Corp. Message-ID: <39BD791E.EE39FA7C@com21.com> References: <39BD2877.E4A02565@email.mot.com> Hi, It is defined in configAll.h file located in "$WIND_BASE/target/config/all"directory and configAll.h is included in config.h file located in your bsp directory. Chears. Rajesh Bo Yu wrote: > Hi, > > All PowerPC BSPs have the following statement in romInit() to load > the initial the stack pointer. I cannot find the definition for > STACK_ADRS in romInit(). > > /* initialize the stack pointer */ > > lis sp, HIADJ(STACK_ADRS) > addi sp, sp, LO(STACK_ADRS) > > Could anyone tell me what the value for STACK_ADRS is? Thanks. > > -- > Bo Yu > > --------------------------- Newsgroups: comp.os.vxworks Subject: Choosing a Tornado 2 Intel x86 BSP Date: Tue, 12 Sep 2000 02:07:37 GMT From: Marcelo & Keyly Mourier Organization: EarthLink Inc. -- http://www.EarthLink.net Message-ID: <39BD90C7.C79FAFCA@earthlink.com> - --------------9F0F68006FC509C22A580040 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Hi there! There seems to be a large selection of Tornado 2.0 BSPs for the Intel x86 architecture. I was wondering if anybody could recommend a BSP that runs out-the-box on a PC/AT-compatible motherboard based on the Celeron, Pentium II, or Pentium III processors. The keyword here is out-the-box; e.g. I don't want to (can't afford to) do any porting to get vxWorks up and runnig. Thanks! - --------------9F0F68006FC509C22A580040 Content-Type: text/html; charset=us-ascii Content-Transfer-Encoding: 7bit Hi there!

There seems to be a large selection of Tornado 2.0 BSPs for the Intel x86 architecture. I was wondering if anybody could recommend a BSP that runs out-the-box on a PC/AT-compatible motherboard based on the Celeron, Pentium II, or Pentium III processors. The keyword here is out-the-box; e.g. I don't want to (can't afford to) do any porting to get vxWorks up and runnig.

Thanks! - --------------9F0F68006FC509C22A580040-- --------------------------- Newsgroups: comp.os.vxworks Subject: How to get DosFS2 ? Date: Tue, 12 Sep 2000 02:47:54 GMT From: ellin_lin@my-deja.com Organization: Deja.com - Before you buy. Message-ID: <8pk5gk$30s$1@nnrp1.deja.com> As a tenderfoot, I'm puzzled with this problem: Where & how? Does DosFS2 support FAT32? ellin Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: network communication Date: Tue, 12 Sep 2000 03:49:43 GMT From: ba_ajay@my-deja.com Organization: Deja.com - Before you buy. Message-ID: <8pk94g$759$1@nnrp1.deja.com> Hi, Can VxWorks be able to see 2 NT hosts as part of networking feature.i.e communication b/n VxWorks host machines. I want to use one NT machine as client and another NT machine as server using (tcp/ip stack)socket programming in VxWorks. Can u help me on this ? ajay Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Choosing a Tornado 2 Intel x86 BSP Date: Tue, 12 Sep 2000 10:15:21 +0530 From: Vijay Kumar Peshkar Organization: Ye 'Ol Disorganized NNTPCache groupie Message-ID: <39BDB4E1.9BC88A38@cisco.com> References: <39BD90C7.C79FAFCA@earthlink.com> Reply-To: vpeshkar@cisco.com This is a multi-part message in MIME format. - --------------8072015462FA31966951C07A Content-Type: multipart/alternative; boundary="------------A073120CA99CCFB21EC3627A" - --------------A073120CA99CCFB21EC3627A Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Hi, The Pc-pentiunmpro is good one. good luck, Vijay Marcelo & Keyly Mourier wrote: > Hi there! > > There seems to be a large selection of Tornado 2.0 BSPs for the Intel > x86 architecture. I was wondering if anybody could recommend a BSP > that runs out-the-box on a PC/AT-compatible motherboard based on the > Celeron, Pentium II, or Pentium III processors. The keyword here is > out-the-box; e.g. I don't want to (can't afford to) do any porting to > get vxWorks up and runnig. > > Thanks! - --------------A073120CA99CCFB21EC3627A Content-Type: text/html; charset=us-ascii Content-Transfer-Encoding: 7bit Hi,

The Pc-pentiunmpro is good one.

good luck,
Vijay

Marcelo & Keyly Mourier wrote:

Hi there!

There seems to be a large selection of Tornado 2.0 BSPs for the Intel x86 architecture. I was wondering if anybody could recommend a BSP that runs out-the-box on a PC/AT-compatible motherboard based on the Celeron, Pentium II, or Pentium III processors. The keyword here is out-the-box; e.g. I don't want to (can't afford to) do any porting to get vxWorks up and runnig.

Thanks!

- --------------A073120CA99CCFB21EC3627A-- - --------------8072015462FA31966951C07A Content-Type: text/x-vcard; charset=us-ascii; name="vpeshkar.vcf" Content-Transfer-Encoding: 7bit Content-Description: Card for Vijay Kumar Peshkar Content-Disposition: attachment; filename="vpeshkar.vcf" begin:vcard n:Peshkar;Vijay x-mozilla-html:FALSE org:Wipro Technologies,K2, Bangalore;Wishing you blue skies, a couple of daisies, some fresh air and a smile! adr:;;;;;; version:2.1 email;internet:vpeshkar@cisco.com title:Senior Software Engineer fn:Vijay Kumar Peshkar end:vcard - --------------8072015462FA31966951C07A-- --------------------------- Newsgroups: comp.os.vxworks Subject: How to delay current task for some microseconds? Date: Tue, 12 Sep 2000 14:17:51 +0800 From: "Li Xin" Organization: Motorola Message-ID: <8pkh2c$bgi$1@schbbs.mot.com> I am writing a device driver under VxWorks and need a means to delay current task for some microseconds. I know taskDelay() can delay current task, but that will delay at least one tick which I think is too long for our driver. Is there any method to delay current task for some microseconds? --------------------------- Newsgroups: comp.os.vxworks Subject: Re: printf of "long long" Date: Tue, 12 Sep 2000 10:10:41 +0300 From: Ofer Goren Organization: Bay Networks, Inc. Message-ID: <39BDD6F0.BDF7C70C@nortelnetworks.com> References: <39B99888.6A31EEC7@mindspring.com> we use a function to do the trick. The function get a UINT64 variable, and prints each 32 bits of it at a time. Al Johnston wrote: > the lack of support for printing a 64 bit integer has long annoyed me. > > My fingers just keep typing the %L, and the compiler has to remind > me its not there... this happens every year or so... their memory > is not to good.... > > Does someone have a "real" fix for this? (not a divide by ten > and map to char in an array approach).... > > thanks - -- Embedded SW Team Leader Email : ogoren@nortelnetworks.com Phone number: +9723-6456023 Fax number : +9723-6479579 ESN : 826-6023 http://www.nortelnetworks.com - ---------------------------------------------------- If you lend someone $20, and never see that person again, it was probably worth the investment. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: How to delay current task for some microseconds? Date: Tue, 12 Sep 2000 09:12:12 +0100 From: Graham Baxter Organization: Graham Baxter (Software) Limited Message-ID: <39BDE55C.48147D16@NOSPAM.bcs.org.uk> References: <8pkh2c$bgi$1@schbbs.mot.com> Has your processor got a spare timer? If so, use it as a nano-second timer. Put in a time wasting loop: volatile int i; for(i=0;i I am writing a device driver under VxWorks and need a means to delay current > task for some microseconds. I know taskDelay() can delay current task, but > that will delay at least one tick which I think is too long for our driver. > Is there any method to delay current task for some microseconds? --------------------------- Newsgroups: comp.os.vxworks Subject: vxWorks Task Date: Tue, 12 Sep 2000 15:38:45 +0530 From: "Krishna Kishore Reddy Sureddy" Organization: Lawrence Berkeley National Laboratory Message-ID: <8pkvmk$r6r$1@overload.lbl.gov> Hi, Can a task in vxWorks get suspended other than the conditions presented below.Correct me if i am Wrong. 1. vxWorks calls a. ts() b. taskSuspend() c. sigsuspend() d. pause() e. nanosleep() f. aio_suspend g. sigtimedwait() h. sigwaitinfo() 2. Task level exception 3. Breakpoint in task 4. Attaching task to Debugger 5. CS_MSEC_DELAY(msec), a vxWorks Macro 6. Error in Allocating/Freeing Memory 7. When a pipe is full and task attempts to write to a pipe 8. One task stack overflows and corrupts other tasks stack 9. ioctl() call causes non-recoverable error 10. If an invalid device descriptor in the valid range of device descriptors numbers is passed to an I/O function 11. Some errors caused in the system even though the task itself may not have caused that error such as a. Device driver doing a hardware access that generates an error b. Bus error caused by a DMA cycle which was set up by a totally different task c. Bus errors and access violations due to task level code, driver code, interrupt handlers Thanks and Regards, S.K.K.Reddy. --------------------------- End of New-News digest ********************** From vxwexplo-errs@csg.lbl.gov Tue Sep 12 12:38:22 2000 From: James Cunningham Date: Tue Sep 12 12:38:25 PDT 2000 Subject: data access exception in net task Is there a known problem with the Ethernet Controller (DEC 21140) that is used on the Motorola 2400 and 2600 series of PowerPCs. I have a situation where I am seeing a data access exception in the net task under very heavy DMA traffic. I originally thought it was a problem with the BSD network driver (VxWorks 5.4) for our card that I am writing. The symptoms that I saw were: Running on PowerPCs (2431 and 2603) sending/receiving data TCP as fast as possible runs fine for about an hour or so and then I get one of the following conditions: 1) Receives of packets stop happening - DMAs into my receive buffers just seem to stop. 2) A data access exception. The printout points to the net task. 3) The machine just locks up. Looking at mbufShow(), memShow(), spy() and my driver status don't show anything that would indicate a problem. Setting up our driver as a character driver and sending data without using the IP stack (direct IOCTL calls to our driver) still would result in a data access exception in the net task. It would seem that maybe this is a problem with the Ethernet driver built into the PPC board? Modifying the PCI Latency timer in our boards configuration space seemed to change the amount of time before the problem occurred but did not eliminate it. Any ideas as to what might be causing this problem or even how to approach tracking it down would be greatly appreciated. __________________________________________________ Do You Yahoo!? Yahoo! Mail - Free email you can access from anywhere! http://mail.yahoo.com/ From vxwexplo-errs@csg.lbl.gov Tue Sep 12 22:29:07 2000 From: "Jagannatha L.V" Date: Tue Sep 12 22:29:10 PDT 2000 Subject: Query about inetLib Hi, all I got some problems while using inetLib routines. The problem is as follows: see the following line of code: printf("\n%x", inet_addr("90.36.40.2")); Output of this in Vxworks is 228245a But.... Output of this in Unix is 5a242802 (reversing of bytes) 1. I want to know why is this different. 2. If i see the manuals of vxwoks(inetLib) in example it says(I am cutting and pasting) The following example returns 0x5a000002: inet_addr ("90.0.0.2"); but... actually it returns 200005a in vxworks. Can any one put some light on it or is it a bug. reply as soon as posible. with regards from jagan.... __________________________________________________ Do You Yahoo!? Yahoo! Mail - Free email you can access from anywhere! http://mail.yahoo.com/ From vxwexplo-errs@csg.lbl.gov Wed Sep 13 01:57:55 2000 From: "Tim Allen" Date: Wed Sep 13 01:57:57 PDT 2000 Subject: Re: How to get DosFS2 ? >As a tenderfoot, I'm puzzled with this problem: >Where & how? >Does DosFS2 support FAT32? Yes, DOSFS 2.0 does support FAT32, and you should be able to get it as a free upgrade from your Wind River rep. -------------------- Tim Allen Ltd Real Time Embedded Software Consultancy VxWorks From vxwexplo-errs@csg.lbl.gov Wed Sep 13 02:18:10 2000 From: "David Anderson @ Data Track" Date: Wed Sep 13 02:18:12 PDT 2000 Subject: RE: vxWorks Task tornado vxworks > Submitted-by: "Krishna Kishore Reddy Sureddy" > > Hi, > Can a task in vxWorks get suspended other than the conditions presented > below.Correct me if i am Wrong. > > 1. vxWorks calls > a. ts() b. taskSuspend() c. sigsuspend() d. pause() e. > nanosleep() > f. aio_suspend g. sigtimedwait() h. sigwaitinfo() > 2. Task level exception > 3. Breakpoint in task > 4. Attaching task to Debugger > 5. CS_MSEC_DELAY(msec), a vxWorks Macro > 6. Error in Allocating/Freeing Memory > 7. When a pipe is full and task attempts to write to a pipe > 8. One task stack overflows and corrupts other tasks stack > 9. ioctl() call causes non-recoverable error > 10. If an invalid device descriptor in the valid range of device > descriptors > numbers is passed to an I/O function > 11. Some errors caused in the system even though the task itself > may not have caused that error such as > a. Device driver doing a hardware access that generates an error > b. Bus error caused by a DMA cycle which was set up by a totally > different task > c. Bus errors and access violations due to task level code, > driver code, interrupt handlers > Thanks and Regards, > S.K.K.Reddy. > How about: 12. Hardware fault. that covers everything not explained by 1..11 above. David Anderson. mailto:danderson@dtrack.demon.co.uk ----------------------------------------------------- Who said "There's no life here in outer space." ----------------------------------------------------- Disclaimer ---------- This email and any files transmitted with it are confidential and intended solely for the use of the individual or entity to whom they are addressed. If you have received this email in error please destroy and notify Data Track Technology Plc +44 1425 271900. ------------------------------------------------------- From vxwexplo-errs@csg.lbl.gov Wed Sep 13 03:03:26 2000 From: Neelesh Dwivedi Date: Wed Sep 13 03:03:28 PDT 2000 Subject: Re: Query about inetLib Hello Mr. Jagannath, I need to know a thing or two before I look for a suitable answer for your query. Was UNIX running on the same Processor as VxWorks? If it is not so, then I think this is an issue of Little-Endian and Big-Endian Notations. An Intel Processor uses the Little-Endian format ( 0228245a for "90.36.40.02"). So if the VxWorks was running on an Intel Processor, the outcome is not unpredictable. If you are using a Motorola Processor or any other , which supports Big-Endian form, then also the result you get is not wrong. I hope that would solve your problem. Regards, Neelesh Dwivedi. Daewoo Telecom Ltd., New Delhi R&D Centre, Gurgaon, India. From vxwexplo-errs@csg.lbl.gov Wed Sep 13 03:28:52 2000 From: "Mike Anderson" Date: Wed Sep 13 03:28:54 PDT 2000 Subject: RE: Query about inetLib VxWorks Greetings! > printf("\n%x", inet_addr("90.36.40.2")); > > Output of this in Vxworks is > 228245a > But.... > Output of this in Unix is > 5a242802 > > (reversing of bytes) > > > Can any one put some light on it or is it a bug. > reply as soon as posible. You don't say what kind of CPU architecture that you're using, but my guess is that it's probably x86, i960 or one of the other little endian processors. The inet_addr routines will simply convert the dotted quad into an integer with the bytes in the same relative location. On a big endian machine, such as a SPARC or PPC, this would result in the output that you anticipated. But, on a little endian machine, the bytes would be reversed. If you changed your printf to look like this: printf("\n%x", htonl(inet_addr("90.35.40.2"))); you would get the big endian result you expected on both your little endian target and the big endian UN*X box. HTH, ================================================================ Mike Anderson The PTR Group, Inc. Embedded Development, Training, Consulting mailto:mike@theptrgroup.com http://www.theptrgroup.com From vxwexplo-errs@csg.lbl.gov Wed Sep 13 04:01:42 2000 From: "David Anderson @ Data Track" Date: Wed Sep 13 04:01:44 PDT 2000 Subject: RE: Query about inetLib vxWorks Tornado > > Hi, > all I got some problems while using inetLib routines. > The problem is as follows: > > see the following line of code: > printf("\n%x", inet_addr("90.36.40.2")); > > Output of this in Vxworks is > 228245a > But.... > Output of this in Unix is > 5a242802 > > (reversing of bytes) > > 1. I want to know why is this different. > 2. If i see the manuals of vxwoks(inetLib) in example > it says(I am cutting and pasting) > > The following example returns 0x5a000002: > > inet_addr ("90.0.0.2"); > but... > actually it returns 200005a in vxworks. > > Can any one put some light on it or is it a bug. > reply as soon as posible. > > with regards from > jagan.... It is correct , the value returned is in network order ( OK so far ) but when you display the value using your output routine ( printf or whatever ) the value is interpreted in target processor order... if these are different then the little endian/big endian byte swap is apparent. Hop that helps. David Anderson. mailto:danderson@dtrack.demon.co.uk ----------------------------------------------------- Who said "There's no life here in outer space." ----------------------------------------------------- Disclaimer ---------- This email and any files transmitted with it are confidential and intended solely for the use of the individual or entity to whom they are addressed. If you have received this email in error please destroy and notify Data Track Technology Plc +44 1425 271900. ------------------------------------------------------- From vxwexplo-errs@csg.lbl.gov Wed Sep 13 04:03:09 2000 From: Vxworks Exploder Date: Wed Sep 13 04:03:11 PDT 2000 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Wed Sep 13 04:03:05 PDT 2000 Subject: Re: Linking archives Subject: C++ Static Class Members and VxWorks Subject: Re: C++ Static Class Members and VxWorks Subject: Blocked signals in VxWorks Subject: Re: C++ Static Class Members and VxWorks Subject: Re: How to delay current task for some microseconds? Subject: C++ Signal Service Routines must be static? Subject: Re: C++ Static Class Members and VxWorks Subject: Re: How to delay current task for some microseconds? Subject: Re: How to work with PCI ? Subject: Tornado for Managed Switches Subject: Re: PCI Configuration help for a novice Subject: Re: C++ Signal Service Routines must be static? Subject: scsiIoctl() is aborting my SCSI commands Subject: Re: How to delay current task for some microseconds? Subject: Re: How to delay current task for some microseconds? Subject: Re: How to delay current task for some microseconds? Subject: 8xx SCC in UART mode Subject: Re: C++ Signal Service Routines must be static? Subject: dosFS vs DOS compatibility !!! Subject: data access exception in net task Subject: how to set up default editor using Tornado Subject: Re: how to set up default editor using Tornado Subject: Re: data access exception in net task Subject: Re: How to delay current task for some microseconds? Subject: Re: configuring PC target??? Subject: Re: Booting the pcPentium target ? Subject: Query about inetLib Subject: switching between multiple BSPs/toolchains Subject: Re: 8xx SCC in UART mode Subject: Re: switching between multiple BSPs/toolchains Subject: NPT driver question Subject: interrupt delayed Subject: Using SDRAM with MBX860 Subject: re: How to get DosFS2 ? Subject: RE: vxWorks Task Subject: re: Query about inetLib Subject: Re: Query about inetLib ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: Linking archives Date: Tue, 12 Sep 2000 14:09:59 +0200 From: Gunter Ritter Organization: Tenovis GmbH u. Co. KG Message-ID: <8pl6dg$9tc$1@proxy2.fe.internet.bosch.de> References: <8p5k76$90e$1@proxy2.fe.internet.bosch.de> <8p5md8$40e$1@news1.xs4all.nl> <8p791k$gss$1@proxy2.fe.internet.bosch.de> <8p7f57$enc$1@news1.xs4all.nl> <8p802g$gjl$1@proxy2.fe.internet.bosch.de> <8p9v1n$4nh$1@proxy2.fe.internet.bosch.de> Bob Westbrook schrieb in im Newsbeitrag: hv3qrsg5b1vsqvalfski0kginp4hqkjo8u@4ax.com... > On Fri, 8 Sep 2000 07:56:37 +0200, Gunter Ritter > wrote: > > > > >I thought about this, too. But I didn´t try it yet. I find it a little bit > >strange to do it this way. > >Well, I try it now. > > > >Gunter > > > Did it work ? Yes, it works. But I still think that it this a strange solution. Gunter --------------------------- Newsgroups: comp.os.vxworks Subject: C++ Static Class Members and VxWorks Date: Tue, 12 Sep 2000 12:22:40 GMT From: epooleNO@UCEnetis.com (RKT Technologies, Inc.) Organization: Software Consultants to Medical Device Manufacturers Message-ID: <39be1e6f.40485874@news.netis.com> I know that when a member of a class in C++ is declared static, the same copy of that class member is available to all objects of that class (i.e. the same copy is shared across all objects). My question is: In VxWorks, if I use taskSpawn to spawn two independent tasks, each of which creates and uses some objects of the same class (i.e. each #include's "myStuff.hpp" and uses the class definition contained in that file), and that class uses static members, are the same copies of those static members shared across both tasks? Or is one copy of the static members shared across all objects in one task, and a different copy shared across all objects in the other task? I would think that the latter would be the case (each task gets its own copy of the static members), but I have been told differently. So, I look to the experts on this forum for guidance. Thanks to all... Eric Poole (e-mail me at epoole-at-scoot-dot-netis-dot-com) --------------------------- Newsgroups: comp.os.vxworks Subject: Re: C++ Static Class Members and VxWorks Date: 12 Sep 2000 13:40:29 GMT From: lied@nwson.lucent.com (Robert Lied) Organization: Lucent Technologies, Naperville, Illinois, USA Message-ID: <8plbod$b0s@ssbunews.ih.lucent.com> References: <39be1e6f.40485874@news.netis.com> Reply-To: lied@lucent.com In article <39be1e6f.40485874@news.netis.com>, RKT Technologies, Inc. wrote: >I would think that the latter would be the case (each task gets its >own copy of the static members), but I have been told differently. >So, I look to the experts on this forum for guidance. You have been told correctly. There's only one and all the tasks better play nice with it. Classes have no concept of tasks. If you make a class variable (static member), it exists for all the instances of the class. Tasks never enter into it, from the C++ point of view. Another way to look at it is that tasks share all the global memory, and a static member is really placed in global memory, but named with class qualification. - -- Bob Lied --------------------------- Newsgroups: comp.os.vxworks Subject: Blocked signals in VxWorks Date: Tue, 12 Sep 2000 12:59:58 GMT From: vtail@hotmail.com (RKT Technologies, Inc.) Organization: Software Consultants to Medical Device Manufacturers Message-ID: <39be2841.43000639@news.netis.com> In VxWorks, if signal is blocked using sigblock() or sigprocmask(), and a signal appears (and is blocked), is the signal lost or does it pend and then get activated when the signal is unblocked? Thanks ... (Above e-mail address is a spam sink ... reply by e-mail, if you wish, to epoole-at-scoot-dot-netis-dot-com,) --------------------------- Newsgroups: comp.os.vxworks Subject: Re: C++ Static Class Members and VxWorks Date: Tue, 12 Sep 2000 13:16:53 GMT From: vtail@hotmail.com (RKT Technologies, Inc.) Organization: Software Consultants to Medical Device Manufacturers Message-ID: <39be2c19.43984389@news.netis.com> References: <39be1e6f.40485874@news.netis.com> <8plbod$b0s@ssbunews.ih.lucent.com> On 12 Sep 2000 13:40:29 GMT, lied@nwson.lucent.com (Robert Lied) wrote: >You have been told correctly. There's only one and all the >tasks better play nice with it. Good morning, Bob. Well, that's a bummer. Is that a behavior that's unique to VxWorks, or would the same behavior be evident on POSIX-compliant Unix systems? Thanks... Eric (Above e-mail address is a spam sink ... reply by e-mail, if you wish, to epoole-at-scoot-dot-netis-dot-com,) --------------------------- Newsgroups: comp.os.vxworks Subject: Re: How to delay current task for some microseconds? Date: Tue, 12 Sep 2000 10:30:36 -0400 From: Charles H. Chapman Organization: Dynacs Engineering Company Message-ID: <8plemm$256@niven.ksc.nasa.gov> References: <8pkh2c$bgi$1@schbbs.mot.com> <39BDE55C.48147D16@NOSPAM.bcs.org.uk> Reply-To: chc@nasa2.ksc.nasa.gov > There is also a function nanosleep() which may do the job. Nah, nanosleep may have a resolution of nanoseconds but it has a granularity of one clock tick. Chuck --------------------------- Newsgroups: comp.os.vxworks Subject: C++ Signal Service Routines must be static? Date: Tue, 12 Sep 2000 14:19:16 GMT From: vtail@hotmail.com (RKT Technologies, Inc.) Organization: Software Consultants to Medical Device Manufacturers Message-ID: <39be3a63.47642957@news.netis.com> I'm told that in C++ on VxWorks. a signal service routine must be static. Is this true? Eric Poole (Above e-mail address is a spam sink ... reply by e-mail, if you wish, to epoole-at-scoot-dot-netis-dot-com,) --------------------------- Newsgroups: comp.os.vxworks Subject: Re: C++ Static Class Members and VxWorks Date: 12 Sep 2000 15:21:31 GMT From: Troy Cauble Organization: Bell Labs, Holmdel, NJ, USA Message-ID: <8plhlr$9tl@nntpb.cb.lucent.com> References: <39be1e6f.40485874@news.netis.com> <8plbod$b0s@ssbunews.ih.lucent.com> <39be2c19.43984389@news.netis.com> RKT Technologies, Inc. wrote: : On 12 Sep 2000 13:40:29 GMT, lied@nwson.lucent.com (Robert Lied) : wrote: :>You have been told correctly. There's only one and all the :>tasks better play nice with it. : Good morning, Bob. : Well, that's a bummer. : Is that a behavior that's unique to VxWorks, or would the same : behavior be evident on POSIX-compliant Unix systems? In a multithreaded Unix process, yes. In separate Unix processes, no. Unix has processes. VxWorks and many other RTOS's have tasks. (Though terminology may vary.) A primary difference between the two is the lack of memory protection in tasks. VxWorks tasks share memory. Besides globals, they can access each others stack space. So a bad pointer problem in one task can screw up another task. It's not really a C++ or even a POSIX issue. The OS's provide different things. - -troy --------------------------- Newsgroups: comp.os.vxworks Subject: Re: How to delay current task for some microseconds? Date: Tue, 12 Sep 2000 15:31:49 GMT From: Ted Dennison Organization: Deja.com - Before you buy. Message-ID: <8pli95$m8g$1@nnrp1.deja.com> References: <8pkh2c$bgi$1@schbbs.mot.com> <39BDE55C.48147D16@NOSPAM.bcs.org.uk> <8plemm$256@niven.ksc.nasa.gov> In article <8plemm$256@niven.ksc.nasa.gov>, chc@nasa2.ksc.nasa.gov wrote: > > There is also a function nanosleep() which may do the job. > > Nah, nanosleep may have a resolution of nanoseconds but it has a > granularity of one clock tick. Unless you have some external clock device with a higher resolution that is capable of generating interrupts for you, I don't see how you can possibly get anything with a higher resoltion than your system clock. I suppose you could try to just burn time with some bogous assembly instructions. But the timing is going to be CPU (and clock speed) dependent, and mighty tough to accurately calculate if you are using a modern pipelined and cached processor. I'd put in a taskDelay (1) and be done with it. If that's too long, jack up your clock frequency. - -- T.E.D. http://www.telepath.com/~dennison/Ted/TED.html Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: How to work with PCI ? Date: Tue, 12 Sep 2000 16:53:32 +0100 From: James Marshall Organization: Agilent Technologies UK Ltd Message-ID: <39BE517C.27094971@agilent.com> References: <8oif6g$5li$1@news.netvision.net.il> I've just broken through into a clearing in the jungle of PCI Configuration & VxWorks on various PC platforms including CompactPCI. Some things I found useful on the journey: 1) There's a brief WindRiver Technical Note, WTN-49, that talks about some of these issues - it's on Windsurf. In particular it covers VME window mapping, if only briefly. 2) PCI System Architecture by Tom Shanley & Don Anderson, part of the MindShare series. 3) Linux source for PCI configuration. I used the source browser at lxr.linux.no to look at (for example) drivers/pci/pci.c . It lets you click on identifiers to uncover where they are defined and used. The VxWorks PCI show routines aren't included from Project - you need to edit config.h in your (locally-copied) BSP and define INCLUDE_SHOW_ROUTINES. Unfortunately, this is very coarse-grained and pulls in several other unrelated show routines. I guess you could always write a .cdf file to add it to Project. In your driver you would use PCI routines from sysLib [pcPentium]: pciFindDevice() to get the config address (bus,device,function) for each instance of your device, then read the appropriate BAR(s) to get their address(es). I guess your master/slave window stuff mentioned in WTN-49 comes in at that point. Note that the PCI Auto Config Library (src/drv/pci/pciAutoConfigLib.c) mentioned in WTN-49 isn't supported on x86 targets because the PCI BIOS (a callable part of the BIOS) is expected to do all PCI configuration before the OS boots. This is why the default setting for PCI_CFG_TYPE on x86 targets is PCI_CFG_NONE. James Marshall --------------------------- Newsgroups: comp.os.vxworks Subject: Tornado for Managed Switches Date: Tue, 12 Sep 2000 09:21:49 -0700 From: "Eric McDaniel" Organization: Posted via Supernews, http://www.supernews.com Message-ID: Does anyone out there have any experience with this package (TMS) from WindRiver? It's actually 3 separate products - basic layer 2, extended layer 2 and layer 3. Any experience with any of them? How well integrated are they with the rest of the OS and the tools? I assume most or all of the components joined the WindRiver product line as the result of acquisitions. Are they worth the money? What are the competing alternatives, and are they any good? Thanks in advance for any help. Eric McDaniel ericm@vertical.com --------------------------- Newsgroups: comp.os.vxworks Subject: Re: PCI Configuration help for a novice Date: Tue, 12 Sep 2000 17:14:06 +0100 From: James Marshall Organization: Agilent Technologies UK Ltd Message-ID: <39BE564E.69470763@agilent.com> References: <39AABFBA.C71B9657@res.raytheon.com> Use the PCI Auto Config Library (src/drv/pci/pciAutoConfigLib.c). Also get WTN-49 from WindSurf under WindRiver Technical Notes. (See my response to Ezra Koper's How to work with PCI). For interpreting PCI Config registers, the PCICFG program might help though you'll have to port it from its current DOS form. See http://www.cs.cmu.edu/~ralf/pub-files/rbpci119.zip. Tom Shanley's PCI System Architecture has a decent description of what PCI configuration requires. You can also get some ideas from the Linux source. James Marshall kostizt wrote: > Folks, > > I must admit that I am new to the PCI bus and I am having a lot of > trouble getting a vendor supplied driver / serial device into my > system. The symptom is that (a) the BARs don't make a lot of sense to > me and (b) the device (serial), doesn't seem to be mapped (i/o simply > posting in memory with no interrupts). > > I have a 2304 BSP. The vendor's documentation says "ensure that the BSP > itself performs configuration of all PCI devices supported by the PCI > HAL. This includes configuring the PCI base address registers (BARs) > and the interrupt line request register in the PCI configuration space > of each device." > > I thought I would find the code that does this in the BSP and go from > there. The problem is that I can't find where any of this happens! > > Does anyone know how the PCI devices get their configuration information > processed by the BSP? > > If I invoke an xpciShow, it looks as follows: > > -> xpciShow > Scanning function 0 of each PCI device on bus 0 of card 0 > bus slot vendor dev IRQ Bar0 Bar1 Bar2 Bar3 Bar4 > Bar5 > 0 00 1057 4801 00 00000001 fc000000 00000000 00000000 00000000 > 00000000 > 0 11 10ad 0565 00 00000000 00000000 00000000 00000000 00000000 > 00000000 > 0 13 10e3 0000 00 fd050000 00000001 00000000 00000000 00000000 > 00000000 > 0 14 1011 0009 00 00020001 00000000 00000000 00000000 00000000 > 00000000 > 0 17 124b 0001 00 00000000 00000000 00000000 00000000 00000000 > 00000000 > value = 1 = 0x1 > > The information for bus/slot/vendor for slot 17 is the interesting > piece. The vendor info is the PCI device of interest. Does a BAR0 of 0 > make sense? > > Any help would be greatly appreciated. > > Thanks in advance! > > Tom Kostizak --------------------------- Newsgroups: comp.os.vxworks Subject: Re: C++ Signal Service Routines must be static? Date: 12 Sep 2000 16:35:04 GMT From: lied@w-lon.lucent.com (Robert Lied) Organization: Lucent Technologies, Naperville, Illinois, USA Message-ID: <8pllvo$c7f@ssbunews.ih.lucent.com> References: <39be3a63.47642957@news.netis.com> Reply-To: lied@lucent.com In article <39be3a63.47642957@news.netis.com>, RKT Technologies, Inc. wrote: >I'm told that in C++ on VxWorks. a signal service routine must be >static. Is this true? If you mean a member function, yes. It's not a VxWorks thing. The reason is that member functions implicitly get called with an extra parameter (the infamous "this" pointer). A handler expecting a single argument will actually get "this" as its first argument, and havoc ensues. Static member functions don't pass the "this" pointer, so less of a problem. Strictly speaking, the types are still wrong, though. Simple solution: don't use member functions. Use an ordinary function as the service routine; maybe make it a friend. There's not much you can safely do in a service routine, anyway, except maybe set up some condition that the object will be checking later. A similar problem comes up in trying to use member functions as the entry points of tasks. The solution is discussed approximately weekly in comp.programming.threads. - -- Bob Lied --------------------------- Newsgroups: comp.os.vxworks Subject: scsiIoctl() is aborting my SCSI commands Date: Tue, 12 Sep 2000 16:54:24 GMT From: yeemoney@my-deja.com Organization: Deja.com - Before you buy. Message-ID: <8pln3j$sg2$1@nnrp1.deja.com> I'm using: - -the ncr810 driver on the sym53C895 SCSI controller - -VxWorks Version 5.3.1 - -Tornado Version 1.0.1 I pretty sure I havethe driver working. I'm able to initialize and do scsiAutoConfig and the information looks right. However, I'm having some issues running SCSI commands with scsiIoctl(). Has anyone successfully used this particular function? The problem I'm having is that scsiIoctl is returning an error when it is executed via the MyScsiCommand routine. For example, I'm trying to do a SCSI read (opcode 0x8) and I create a function called scsiRead based on the MyScsiCommand template. However, when it gets to the scsiIoctl() command, it returns a -1 (ERROR). I then inserted a printErrno() and it shows 0x370012 which corresponds to "S_scsiLib_ABORTED" error message. So my command is being aborted for some reason. I hooked up a scsi bus analyzer and the seek command is not even getting out to the bus. Are you experiencing any similar problems? Any ideas on why the command is being aborted? Thanks in advance. Only me, Stanley Yee Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: How to delay current task for some microseconds? Date: Tue, 12 Sep 2000 17:58:28 +0100 From: James Marshall Organization: Agilent Technologies UK Ltd Message-ID: <39BE60B4.BA40D4AF@agilent.com> References: <8pkh2c$bgi$1@schbbs.mot.com> Use the timestamp driver if it's implemented on your system (see WindView User's Guide, especially appendix G). Find out which timer device driver your system uses then view its source file in target/src/drv/timer/???.c. This should tell you whether timestamp is supported and to what resolution. The PC Pentium Pro/II/III has a 64-bit TimeStamp Counter which increments at the processor's clock rate so you can get very small delays. Be aware that some (most?) timestamp drivers reset the timestamp counter every tick. Sometimes this is because of hardware limitations, like the PowerPC decrementer register acts for both the system tick and timestamp, but sometimes it's because WindRiver made it so, like the i8253Timer.c driver for x86 which resets the Pentium's TSC every tick. James Marshall --------------------------- Newsgroups: comp.os.vxworks Subject: Re: How to delay current task for some microseconds? Date: Tue, 12 Sep 2000 12:36:07 -0700 From: Harvey Taylor Organization: Newsfeeds.com http://www.newsfeeds.com 73,000+ UNCENSORED Newsgroups. Message-ID: <39BE85A7.2692@despam.pangea.ca> References: <8pkh2c$bgi$1@schbbs.mot.com> <39BDE55C.48147D16@NOSPAM.bcs.org.uk> <8plemm$256@niven.ksc.nasa.gov> <8pli95$m8g$1@nnrp1.deja.com> Ted Dennison wrote: >[..] > I suppose you could try to just burn time with some bogous assembly > instructions. But the timing is going to be CPU (and clock speed) > dependent, and mighty tough to accurately calculate if you are using a > modern pipelined and cached processor. > Hi Ted, Several CPUs allow you to count cycles with an RDTSC type instruction which can be utilized to good effect. - -het - -- "Experience is like lead shot in your behind; one who has never been shot doubts that such a thing exists at all." -Hans Lippman Harvey Taylor mailto:het@despam.pangea.ca http://www.pangea.ca/~het --------------------------- Newsgroups: comp.os.vxworks Subject: Re: How to delay current task for some microseconds? Date: Tue, 12 Sep 2000 19:13:50 GMT From: Ted Dennison Organization: Deja.com - Before you buy. Message-ID: <8plv8r$73s$1@nnrp1.deja.com> References: <8pkh2c$bgi$1@schbbs.mot.com> <39BE60B4.BA40D4AF@agilent.com> In article <39BE60B4.BA40D4AF@agilent.com>, James Marshall wrote: > Use the timestamp driver if it's implemented on your system (see > WindView User's Guide, especially appendix G). Find out which timer > device driver your system uses then view its source file in > target/src/drv/timer/???.c. This should tell you whether timestamp is > supported and to what resolution. The PC Pentium Pro/II/III has a > 64-bit TimeStamp Counter which increments at the processor's clock > rate so you can get very small delays. Wow, thanks a lot for the info James. I always wondered how Windview managed to get that kind of resolution, when the best I could manage was a clock tick. Duuuh. I guess I could have read the WindView users' guide myself. I've got a couple of very good immediate uses for that, too. - -- T.E.D. http://www.telepath.com/~dennison/Ted/TED.html Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: 8xx SCC in UART mode Date: Tue, 12 Sep 2000 10:07:53 +0100 From: "Jeremy Pye" Organization: Customer of Energis Squared Message-ID: Because the SMC ports don't offer handshaking we need to connect our modem to one of the SCC port. These drivers are not in either the MBX860, ADS860 BSPs, so it looks as though I am going to have to write one. I don't know where to start. Can anyone help me with a few pointers. regards, Jeremy --------------------------- Newsgroups: comp.os.vxworks Subject: Re: C++ Signal Service Routines must be static? Date: Tue, 12 Sep 2000 18:37:12 GMT From: vtail@hotmail.com (RKT Technologies, Inc.) Organization: Software Consultants to Medical Device Manufacturers Message-ID: <39be773f.63226063@news.netis.com> References: <39be3a63.47642957@news.netis.com> <8pllvo$c7f@ssbunews.ih.lucent.com> On 12 Sep 2000 16:35:04 GMT, lied@w-lon.lucent.com (Robert Lied) wrote: >In article <39be3a63.47642957@news.netis.com>, >RKT Technologies, Inc. wrote: >>I'm told that in C++ on VxWorks. a signal service routine must be >>static. Is this true? > >If you mean a member function, yes. It's not a VxWorks thing. The >reason is that member functions implicitly get called with an extra >parameter (the infamous "this" pointer). > > ,,, > > Simple solution: don't use member >functions. Use an ordinary function as the service routine; maybe >make it a friend. That's what I ended up doing, been working on it a good part of the afternoon. Didn't make it a friend, though. I should look into that. Thanks much... Eric --------------------------- Newsgroups: comp.os.vxworks Subject: dosFS vs DOS compatibility !!! Date: Tue, 12 Sep 2000 21:58:28 +0200 From: "zbody" Organization: Slovenija OnLine - SiOL Message-ID: Hello! I installed dosFs on the pcPentium target running VxWorks as a hard-boot and made a VxW format floppy. When I booted "pure" DOS6.22 from the DOS sys floppy I expected that DOS will recognize VxWorks formated floppy diskette but... it doesn't !!! (Extended error 82 - "The directory or file cannot be created." -MSDN) I put the params in the dosFsConfigInit() (for floppy) exactly as is written in the Programmer's Guide p.187. On the page 176. of Prog. Guide the first sentence says like this :"Diskettes formatted using the dosFs file system are compatible with MS-DOS diskettes up to and including release 6.2" :-)) Does somebody knows something about? Thanks! Regards! Darko --------------------------- Newsgroups: comp.os.vxworks Subject: data access exception in net task Date: Tue, 12 Sep 2000 12:38:20 -0700 (PDT) From: James Cunningham Organization: Lawrence Berkeley National Laboratory Message-ID: <8pm2rj$680$1@overload.lbl.gov> Is there a known problem with the Ethernet Controller (DEC 21140) that is used on the Motorola 2400 and 2600 series of PowerPCs. I have a situation where I am seeing a data access exception in the net task under very heavy DMA traffic. I originally thought it was a problem with the BSD network driver (VxWorks 5.4) for our card that I am writing. The symptoms that I saw were: Running on PowerPCs (2431 and 2603) sending/receiving data TCP as fast as possible runs fine for about an hour or so and then I get one of the following conditions: 1) Receives of packets stop happening - DMAs into my receive buffers just seem to stop. 2) A data access exception. The printout points to the net task. 3) The machine just locks up. Looking at mbufShow(), memShow(), spy() and my driver status don't show anything that would indicate a problem. Setting up our driver as a character driver and sending data without using the IP stack (direct IOCTL calls to our driver) still would result in a data access exception in the net task. It would seem that maybe this is a problem with the Ethernet driver built into the PPC board? Modifying the PCI Latency timer in our boards configuration space seemed to change the amount of time before the problem occurred but did not eliminate it. Any ideas as to what might be causing this problem or even how to approach tracking it down would be greatly appreciated. __________________________________________________ Do You Yahoo!? Yahoo! Mail - Free email you can access from anywhere! http://mail.yahoo.com/ --------------------------- Newsgroups: comp.os.vxworks Subject: how to set up default editor using Tornado Date: Tue, 12 Sep 2000 20:18:00 GMT From: "joe zhao" Organization: Magma Communications Ltd. Message-ID: HI, I would like to set up SlickEditor as the default editor in WindowsNT. How can I do in within Tornado environment? Thanks. Joe Zhao --------------------------- Newsgroups: comp.os.vxworks Subject: Re: how to set up default editor using Tornado Date: Wed, 13 Sep 2000 00:18:21 GMT From: ellin_lin@my-deja.com Organization: Deja.com - Before you buy. Message-ID: <8pmh3r$t67$1@nnrp1.deja.com> References: In article , "joe zhao" wrote: > HI, > I would like to set up SlickEditor as the default editor in WindowsNT. How > can > I do in within Tornado environment? > > Thanks. > > Joe Zhao > > I think you want to set up SlickEditor as the default editor for Tornado, you can do it as following: Tools->Options->External Editor ellin Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: data access exception in net task Date: Wed, 13 Sep 2000 01:55:14 GMT From: Marcelo & Keyly Mourier Organization: EarthLink Inc. -- http://www.EarthLink.net Message-ID: <39BEDF93.C2D18160@earthlink.com> References: <8pm2rj$680$1@overload.lbl.gov> I'm not sure this is the problem you are seeing, but just in case... Running vxWorks 5.3.1 on a PowerPC 403CX would occasionally cause a data access exception in the NetTask. The data access exception was caused by a misaligned memory access somewhere in the TCP/IP stack. Running the same kernel and application software on a PowerPC 603 (which is able to handle misaligned memory operands) never caused the exception. I can't remember exactly where the problem was, but somehow I recall it had to do with the "dc" driver loaning its buffer to build an mbuf cluster... James Cunningham wrote: > Is there a known problem with the Ethernet Controller (DEC 21140) that > is used on the Motorola 2400 and 2600 series of PowerPCs. I have a > situation where I am seeing a data access exception in the net task > under very heavy DMA traffic. > > I originally thought it was a problem with the BSD network driver > (VxWorks 5.4) for our card that I am writing. The symptoms that I saw > were: > > Running on PowerPCs (2431 and 2603) sending/receiving data TCP as fast > as possible runs fine for about an hour or so and then I get one of the > following conditions: > 1) Receives of packets stop happening - DMAs into my receive buffers > just seem to stop. > 2) A data access exception. The printout points to the net task. > 3) The machine just locks up. > > Looking at mbufShow(), memShow(), spy() and my driver status don't show > anything that would indicate a problem. > > Setting up our driver as a character driver and sending data without > using the IP stack (direct IOCTL calls to our driver) still would > result in a data access exception in the net task. It would seem that > maybe this is a problem with the Ethernet driver built into the PPC > board? > > Modifying the PCI Latency timer in our boards configuration space > seemed to change the amount of time before the problem occurred but did > not eliminate it. > > Any ideas as to what might be causing this problem or even how to > approach tracking it down would be greatly appreciated. > > __________________________________________________ > Do You Yahoo!? > Yahoo! Mail - Free email you can access from anywhere! > http://mail.yahoo.com/ --------------------------- Newsgroups: comp.os.vxworks Subject: Re: How to delay current task for some microseconds? Date: Wed, 13 Sep 2000 05:00:14 GMT From: Marcelo & Keyly Mourier Organization: AJAX Productions Message-ID: <39BF0AF9.65E8BD9C@earthlink.com> References: <8pkh2c$bgi$1@schbbs.mot.com> Reply-To: jukema@earthlink.net Many modern CPUs include a 56-bit or 64-bit Time Base counter that runs at the speed of the CPU bus (e.g. 33MHz or 66MHz) You could use in-line assembly to read this register and break out the loop when the specified number of microseconds have elapsed... Li Xin wrote: > I am writing a device driver under VxWorks and need a means to delay current > task for some microseconds. I know taskDelay() can delay current task, but > that will delay at least one tick which I think is too long for our driver. > Is there any method to delay current task for some microseconds? --------------------------- Newsgroups: comp.os.vxworks Subject: Re: configuring PC target??? Date: Tue, 12 Sep 2000 09:20:20 +0200 From: "Dr. Peter Salewsky" Message-ID: <39BDD934.1E5410CB@resolut.de> References: <39B7D669.50B20513@root.com> <39B7D6AC.8343E52F@root.com> Bruce Chrustie schrieb: > one more thing....I can see the target booting via the VGA monitor on > the PC itself and am able to control the booting via the keyboard on > the target. > What messages do you see on the VGA monitor when the target boots? Do you get a prompt on the VGA monitor after booting? Can you enter commands at this prompt via the keyboard? [...] - -- Dr. Peter Salewsky phone +49 3493 743-20 resolut GmbH fax +49 3493 743-25 Antonienstr. 3 dialog@resolut.de D-06749 Bitterfeld, Germany www.resolut.de --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Booting the pcPentium target ? Date: Tue, 12 Sep 2000 09:29:54 +0200 From: "Dr. Peter Salewsky" Message-ID: <39BDDB72.4A0AEDC2@resolut.de> References: <8pho7v$73m$1@nnrp1.deja.com> mishra.r@dcmtech.co.in schrieb: > I need to boot my pcPentium target over the Ethernet.(ie. by > downloading the vxWorks image). I have a NE2000 compatible ethernet > card, have set the base address and IRQ for that in the config.h. So, > after when the image is created the loader is unable to load that image > from the host. Moreover, theres no message on the ftp server running on > the host. Do you have a monitor connected directly to your target? If yes, what messages do you see on this monitor when the target tries to boot? - -- Dr. Peter Salewsky phone +49 3493 743-20 resolut GmbH fax +49 3493 743-25 Antonienstr. 3 dialog@resolut.de D-06749 Bitterfeld, Germany www.resolut.de --------------------------- Newsgroups: comp.os.vxworks Subject: Query about inetLib Date: Tue, 12 Sep 2000 22:28:35 -0700 (PDT) From: "Jagannatha L.V" Organization: Lawrence Berkeley National Laboratory Message-ID: <8pn60k$m6u$1@overload.lbl.gov> Hi, all I got some problems while using inetLib routines. The problem is as follows: see the following line of code: printf("\n%x", inet_addr("90.36.40.2")); Output of this in Vxworks is 228245a But.... Output of this in Unix is 5a242802 (reversing of bytes) 1. I want to know why is this different. 2. If i see the manuals of vxwoks(inetLib) in example it says(I am cutting and pasting) The following example returns 0x5a000002: inet_addr ("90.0.0.2"); but... actually it returns 200005a in vxworks. Can any one put some light on it or is it a bug. reply as soon as posible. with regards from jagan.... __________________________________________________ Do You Yahoo!? Yahoo! Mail - Free email you can access from anywhere! http://mail.yahoo.com/ --------------------------- Newsgroups: comp.os.vxworks Subject: switching between multiple BSPs/toolchains Date: Wed, 13 Sep 2000 08:33:34 +0200 From: "Markus MItterer" Organization: KPNQwest customer news service Message-ID: Hello VxWorkers, I have build all of my modules with Tornado2 for a specific BSP (mv2303) and everything works fine. Now we will get a new BSP (mv240x) and I have to compile the source for this target. Usual you have to set your toolchain for a specific BSP at the creating time of a project. How can I change these settings afterwards? At last I want to try VxSim, another toolchain.... Any ideas? regards - -- Markus Mitterer Software-Development, STK mailto: markus_mitterer@sbu1.storkgroup.com --------------------------- Newsgroups: comp.os.vxworks Subject: Re: 8xx SCC in UART mode Date: 13 Sep 2000 07:02:19 GMT From: borkhuis@lucent.com Organization: Lucent Technologies, Nieuwegein NL Message-ID: <8pn8pr$a3o$1@news1.xs4all.nl> References: Jeremy Pye wrote: : Because the SMC ports don't offer handshaking : we need to connect our modem to one of the SCC : port. These drivers are not in either the MBX860, : ADS860 BSPs, so it looks as though I am going : to have to write one. : I don't know where to start. Can anyone help me : with a few pointers. Ask your support engineer for the drivers for this. They have some unsupported code for the SCC ports. You have to do some to get it working, but the code itself is OK. Groeten, Johan - -- o o o o o o o . . . ______________________________ o _____ || Johan Borkhuis | .][__n_n_|DD[ ====_____ | borkhuis@lucent.com | >(________|__|_[_________]_|___________________________| _/oo OOOOO oo` ooo ooo 'o!o!o o!o!o` === VxWorks FAQ: http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html === --------------------------- Newsgroups: comp.os.vxworks Subject: Re: switching between multiple BSPs/toolchains Date: 13 Sep 2000 07:07:17 GMT From: borkhuis@lucent.com Organization: Lucent Technologies, Nieuwegein NL Message-ID: <8pn935$a3o$2@news1.xs4all.nl> References: Markus MItterer wrote: : Hello VxWorkers, : I have build all of my modules with Tornado2 for a specific BSP (mv2303) and : everything works fine. Now we will get a new BSP (mv240x) and I have to : compile the source for this target. Usual you have to set your toolchain for : a specific BSP at the creating time of a project. How can I change these : settings afterwards? At last I want to try VxSim, another toolchain.... : Any ideas? You can add another build to your project. Go to the builds tab in your workspace and right-click on the project you want another build for. Now select New build and follow the instuctions. Groeten, Johan - -- o o o o o o o . . . ______________________________ o _____ || Johan Borkhuis | .][__n_n_|DD[ ====_____ | borkhuis@lucent.com | >(________|__|_[_________]_|___________________________| _/oo OOOOO oo` ooo ooo 'o!o!o o!o!o` === VxWorks FAQ: http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html === --------------------------- Newsgroups: comp.os.vxworks Subject: NPT driver question Date: Wed, 13 Sep 2000 07:07:15 GMT From: Kans Organization: Deja.com - Before you buy. Message-ID: <8pn92r$nf9$1@nnrp1.deja.com> Hello VxWork-ers, I am developing a NPT driver. The muxTkBind call takes pNetSvcInfo and pNetDrvInfo as parameters. The NPT document does not talk much about these parameters. Has anybody used it? If yes, of what use is it to the drivers? I would like to store some information with each binding, where can I store that ?. Neither NET_PROTOCOL nor MUX_BIND_ENTRY seem to have such a provision. The muxTkSend call has a pSpareData pointer. Is the network service supposed to call muxTkSend with pSpareData set to pNetDrvInfo ?. How much space is the network service supposed to allocate for pNetDrvInfo ?(since pNetDrvInfo is just a pointer, passed to the muxTkBind call). Or is the usage of these parameters is supposed to just be an understanding between the two layers and mux has nothing to do about them?. How does VxWorks IP stack use them? Hope somebody can help me out on this!. TIA, Kans. Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: interrupt delayed Date: Wed, 13 Sep 2000 09:26:16 +0200 From: Anthony Rouaux Organization: freesbee.fr Message-ID: <39BF2C17.DE9D4475@st2e.com> I am working with the Motorola MPC860T. I am using the intConnect() function to connect my routine to CPM timer1 interruption. I find amazing that it takes 20 us for the core to begin to execute my routine after the interruption occurs. Is this due to a hardware or a software problem? Can you please help me to find a solution to decrease this time. Thanks --------------------------- Newsgroups: comp.os.vxworks Subject: Using SDRAM with MBX860 Date: Wed, 13 Sep 2000 08:45:34 GMT From: francescotorre1965@my-deja.com Organization: Deja.com - Before you buy. Message-ID: <8pner9$up0$1@nnrp1.deja.com> Hello, we are using the MBX860 cpu board from Motorola; it is equipped with 4 Mb of EDO DRAM. We need to extend such memory by means of the SIMM extension. Is it possible to use a Syncronous DRAM ? What we should modify and/or configure ? Thanks! Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: re: How to get DosFS2 ? Date: Wed, 13 Sep 2000 09:52:45 +0100 From: "Tim Allen" Organization: Lawrence Berkeley National Laboratory Message-ID: <8pngi3$qor$1@overload.lbl.gov> >As a tenderfoot, I'm puzzled with this problem: >Where & how? >Does DosFS2 support FAT32? Yes, DOSFS 2.0 does support FAT32, and you should be able to get it as a free upgrade from your Wind River rep. - -------------------- Tim Allen Ltd Real Time Embedded Software Consultancy VxWorks --------------------------- Newsgroups: comp.os.vxworks Subject: RE: vxWorks Task Date: Tue, 12 Sep 2000 15:01:45 +0100 From: "David Anderson @ Data Track" Organization: Lawrence Berkeley National Laboratory Message-ID: <8pnk2j$s09$1@overload.lbl.gov> tornado vxworks > Submitted-by: "Krishna Kishore Reddy Sureddy" > > Hi, > Can a task in vxWorks get suspended other than the conditions presented > below.Correct me if i am Wrong. > > 1. vxWorks calls > a. ts() b. taskSuspend() c. sigsuspend() d. pause() e. > nanosleep() > f. aio_suspend g. sigtimedwait() h. sigwaitinfo() > 2. Task level exception > 3. Breakpoint in task > 4. Attaching task to Debugger > 5. CS_MSEC_DELAY(msec), a vxWorks Macro > 6. Error in Allocating/Freeing Memory > 7. When a pipe is full and task attempts to write to a pipe > 8. One task stack overflows and corrupts other tasks stack > 9. ioctl() call causes non-recoverable error > 10. If an invalid device descriptor in the valid range of device > descriptors > numbers is passed to an I/O function > 11. Some errors caused in the system even though the task itself > may not have caused that error such as > a. Device driver doing a hardware access that generates an error > b. Bus error caused by a DMA cycle which was set up by a totally > different task > c. Bus errors and access violations due to task level code, > driver code, interrupt handlers > Thanks and Regards, > S.K.K.Reddy. > How about: 12. Hardware fault. that covers everything not explained by 1..11 above. David Anderson. mailto:danderson@dtrack.demon.co.uk - ----------------------------------------------------- Who said "There's no life here in outer space." - ----------------------------------------------------- Disclaimer - ---------- This email and any files transmitted with it are confidential and intended solely for the use of the individual or entity to whom they are addressed. If you have received this email in error please destroy and notify Data Track Technology Plc +44 1425 271900. - ------------------------------------------------------- --------------------------- Newsgroups: comp.os.vxworks Subject: re: Query about inetLib Date: Wed, 13 Sep 2000 15:34:40 +0530 From: Neelesh Dwivedi Organization: Lawrence Berkeley National Laboratory Message-ID: <8pnk2k$s0a$1@overload.lbl.gov> Hello Mr. Jagannath, I need to know a thing or two before I look for a suitable answer for your query. Was UNIX running on the same Processor as VxWorks? If it is not so, then I think this is an issue of Little-Endian and Big-Endian Notations. An Intel Processor uses the Little-Endian format ( 0228245a for "90.36.40.02"). So if the VxWorks was running on an Intel Processor, the outcome is not unpredictable. If you are using a Motorola Processor or any other , which supports Big-Endian form, then also the result you get is not wrong. I hope that would solve your problem. Regards, Neelesh Dwivedi. Daewoo Telecom Ltd., New Delhi R&D Centre, Gurgaon, India. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Query about inetLib Date: Wed, 13 Sep 2000 10:39:05 GMT From: "James Kuntz" Organization: Verio Message-ID: References: <8pn60k$m6u$1@overload.lbl.gov> Reply-To: "James Kuntz" Hi, The problem appears to a big vs. little endian difference with the two systems. You should use the function ntohl (network to host long) to resolve it. Try printf("\n%x", ntohl(inet_addr("90.36.40.2"))); on both systems and the results should be the same. James Kuntz Jagannatha L.V wrote in message news:8pn60k$m6u$1@overload.lbl.gov... > Hi, > all I got some problems while using inetLib routines. > The problem is as follows: > > see the following line of code: > printf("\n%x", inet_addr("90.36.40.2")); > > Output of this in Vxworks is > 228245a > But.... > Output of this in Unix is > 5a242802 > > (reversing of bytes) > > 1. I want to know why is this different. > 2. If i see the manuals of vxwoks(inetLib) in example > it says(I am cutting and pasting) > > The following example returns 0x5a000002: > > inet_addr ("90.0.0.2"); > but... > actually it returns 200005a in vxworks. > > Can any one put some light on it or is it a bug. > reply as soon as posible. > > with regards from > jagan.... > > > > > > __________________________________________________ > Do You Yahoo!? > Yahoo! Mail - Free email you can access from anywhere! > http://mail.yahoo.com/ --------------------------- End of New-News digest ********************** From vxwexplo-errs@csg.lbl.gov Wed Sep 13 04:52:59 2000 From: "Balaji Ramachandran" Date: Wed Sep 13 04:53:01 PDT 2000 Subject: Source code archive link VxWorks Tornado Hi all, Anyone can suggest where shall I find Some example source code archive for PCI device access on VxWorks. Thanks & Regards, R.Bala From vxwexplo-errs@csg.lbl.gov Wed Sep 13 07:05:55 2000 From: Jim Way Date: Wed Sep 13 07:05:57 PDT 2000 Subject: RE: data access exception in net task James, > Running on PowerPCs (2431 and 2603) sending/receiving data TCP as fast > as possible runs fine for about an hour or so and then I get one of the > following conditions: > 1) Receives of packets stop happening - DMAs into my receive buffers > just seem to stop. > 2) A data access exception. The printout points to the net task. > 3) The machine just locks up. > > Looking at mbufShow(), memShow(), spy() and my driver status don't show > anything that would indicate a problem. I'm using PPC860T with VxWorks 5.4. Earlier this year, I had a similar sounding problem. After "a while" my Ethernet would stop talking. There were no apparently obvious symptoms, just like you're describing. The problem was fixed by hiring a WRS consultant who applied two SPR fixes (and did not apply another seemingly applicable one). Here is a snippet of his email to me: > I have integrated the fixes for the following : > > SPR 22325 - Race condition between interrupt processing of received packets and polled mode processing in END driver > SPR 23528 - motCpmEnd driver corrupts tNetTask in system mode > > I have not integrated the fixes for the following : > SPR 21143 - motCmpEnd END driver improperly filters incomming frames when promiscuous mode is enabled. Now, I'm not saying that these specific SPRs are related to your problem. Just that my experience in this area suggests that browsing WindSurf for SPRs might be helpful. Your local FAE might also be an excellent resource for this problem. I know that mine is a top-notch techical guy. Depending on your expertise with VxWorks (mine was very limited at the time, now I'm an old-hand ) you might want to consider using their consulting services. We paid for three days of support and one on-site trip, and I think my management all believe it was money very well spent. I know that I had other things to work on during that time. HTH, Jim ----------------------------------------- Jim Way, Software Engineer Datum Austin (Austron Inc.) voice: 512.721.4170 fax : 512.990.9712 email: jwayATdatumDOTcom (no spam please) ----------------------------------------- From vxwexplo-errs@csg.lbl.gov Wed Sep 13 09:06:24 2000 From: "Garner, William F (N-SDC)" Date: Wed Sep 13 09:06:27 PDT 2000 Subject: Changing the Network Address VxSim Uses VxWorks/Tornado users- How or can the IP address of VxSim be changed from 90.0.0.n to 10.xx.xx.xx? Changing the ULIP IP address to 10.0.0.254, the vxTarget IP address remains (90.0.0.1). -> hostShow hostname inet address aliases -------- ------------ ------- vxTarget 90.0.0.1 localhost 127.0.0.1 host 90.0.0.254 trapDest 10.10.6.53 Please help, Bill From vxwexplo-errs@csg.lbl.gov Wed Sep 13 19:00:19 2000 From: "Mike Anderson" Date: Wed Sep 13 19:00:22 PDT 2000 Subject: RE: Changing the Network Address VxSim Uses VxWorks Greetings! > > How or can the IP address of VxSim be changed from 90.0.0.n to > 10.xx.xx.xx? > > [snip] This only works with the full copy of the simulator (not the Simulator Lite that comes with Tornado, but the extra cost version that includes networking. Also, I believe that this only works on the Solaris version of the simulator. Start your simulator & target server. Startup a Windsh. at the -> issue a bootChange command. -> bootChange [you'll be prompted for all of the boot parameters] Key in the new inet on enet (10.x.y.z) and boot host IP (10.x.y.n). Finish entering the rest of the boot parameters. When you get back to the -> prompt, there will be a file saved to the disk subsystem at the location in the file system that the simulator booted from. Unfortunately, I don't remember the file name, but it will be obvious from the name. This will have the boot line in it and can be edited to change the IP parameters. HTH, Mike Anderson The PTR Group, Inc. Embedded Development, Training and Consulting Services mailto:mike@theptrgroup.com http://www.theptrgroup.com From vxwexplo-errs@csg.lbl.gov Thu Sep 14 03:16:26 2000 From: keith.wood@epid.eurotherm.co.uk Date: Thu Sep 14 03:16:28 PDT 2000 Subject: IP Link-Local Address space Does any version of VxWorks support use of the Link-Local address space (169.254/16) as an extension to DHCP? If yes - which versions? If no - does anyone know if there are plans to incorporate it in any future versions? Keith Wood ------------------------- e-mail work: keith.wood@epid.eurotherm.co.uk home: genius@the-wood-family.org.uk phone: work: +44 1903 205277 My views are not my own - they are the intellectual property of my employer From vxwexplo-errs@csg.lbl.gov Thu Sep 14 04:03:12 2000 From: Vxworks Exploder Date: Thu Sep 14 04:03:15 PDT 2000 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Thu Sep 14 04:03:08 PDT 2000 Subject: RE: Query about inetLib Subject: RE: Query about inetLib Subject: Using a single IP for 2 interfaces Subject: Re: Using SDRAM with MBX860 Subject: Source code archive link Subject: Re: interrupt delayed Subject: Re: How to delay current task for some microseconds? Subject: Re: How to delay current task for some microseconds? Subject: Re: printf of "long long" Subject: RE: data access exception in net task Subject: problem in converting from elf to bin Subject: FTP and standard out Subject: test Subject: FTP and standard output Subject: Problems Building gcc-2.95 for powerpc and vxworks Subject: Turning off standard output with FTP Subject: Query:VxWorks Socket Control Subject: Changing the Network Address VxSim Uses Subject: "no carrier" on MVME2400 Subject: Re: 1553 bus Subject: hardware to monitor 32 vxworks "consols" Subject: VxWorks 5.4 NFS server problem Subject: Re: printf of "long long" Subject: Re: hardware to monitor 32 vxworks "consols" Subject: Re: PCI Configuration help for a novice Subject: Re: hardware to monitor 32 vxworks "consols" Subject: Ring Buffer Overflow? Subject: Ring Buffer Overflow? Correction Subject: Re: printf of "long long" Subject: How to get cursor position in Zinc? Subject: Re: "no carrier" on MVME2400 Subject: questions wxwork/tornado Subject: RE: Changing the Network Address VxSim Uses Subject: Re: dosFs2 chkdsk slow to run from bootrom Subject: UDP Socket Receive Subject: Re: How to get cursor position in Zinc? Subject: Re: Source code archive link Subject: Elan 520, AMD BSP Subject: Re: printf of "long long" Subject: Re: Query:VxWorks Socket Control Subject: Re: UDP Socket Receive ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: RE: Query about inetLib Date: Wed, 13 Sep 2000 06:34:30 -0400 From: "Mike Anderson" Organization: Lawrence Berkeley National Laboratory Message-ID: <8pnnj3$ks$1@overload.lbl.gov> VxWorks Greetings! > printf("\n%x", inet_addr("90.36.40.2")); > > Output of this in Vxworks is > 228245a > But.... > Output of this in Unix is > 5a242802 > > (reversing of bytes) > > > Can any one put some light on it or is it a bug. > reply as soon as posible. You don't say what kind of CPU architecture that you're using, but my guess is that it's probably x86, i960 or one of the other little endian processors. The inet_addr routines will simply convert the dotted quad into an integer with the bytes in the same relative location. On a big endian machine, such as a SPARC or PPC, this would result in the output that you anticipated. But, on a little endian machine, the bytes would be reversed. If you changed your printf to look like this: printf("\n%x", htonl(inet_addr("90.35.40.2"))); you would get the big endian result you expected on both your little endian target and the big endian UN*X box. HTH, ================================================================ Mike Anderson The PTR Group, Inc. Embedded Development, Training, Consulting mailto:mike@theptrgroup.com http://www.theptrgroup.com --------------------------- Newsgroups: comp.os.vxworks Subject: RE: Query about inetLib Date: Wed, 13 Sep 2000 11:43:43 +0100 From: "David Anderson @ Data Track" Organization: Lawrence Berkeley National Laboratory Message-ID: <8pnnj3$kt$1@overload.lbl.gov> vxWorks Tornado > > Hi, > all I got some problems while using inetLib routines. > The problem is as follows: > > see the following line of code: > printf("\n%x", inet_addr("90.36.40.2")); > > Output of this in Vxworks is > 228245a > But.... > Output of this in Unix is > 5a242802 > > (reversing of bytes) > > 1. I want to know why is this different. > 2. If i see the manuals of vxwoks(inetLib) in example > it says(I am cutting and pasting) > > The following example returns 0x5a000002: > > inet_addr ("90.0.0.2"); > but... > actually it returns 200005a in vxworks. > > Can any one put some light on it or is it a bug. > reply as soon as posible. > > with regards from > jagan.... It is correct , the value returned is in network order ( OK so far ) but when you display the value using your output routine ( printf or whatever ) the value is interpreted in target processor order... if these are different then the little endian/big endian byte swap is apparent. Hop that helps. David Anderson. mailto:danderson@dtrack.demon.co.uk - ----------------------------------------------------- Who said "There's no life here in outer space." - ----------------------------------------------------- Disclaimer - ---------- This email and any files transmitted with it are confidential and intended solely for the use of the individual or entity to whom they are addressed. If you have received this email in error please destroy and notify Data Track Technology Plc +44 1425 271900. - ------------------------------------------------------- --------------------------- Newsgroups: comp.os.vxworks Subject: Using a single IP for 2 interfaces Date: 13 Sep 2000 14:02:40 +0200 From: yehudak@millimetrix.net ("Yehuda Kasirer") Organization: Mailgate.ORG Server - http://www.Mailgate.ORG Message-ID: This is a multi-part message in MIME format. - ------=_NextPart_000_0004_01C01D94.4F2AE370 Content-Type: text/plain; charset="windows-1255" Content-Transfer-Encoding: 7bit Is it possible to configure vxworks to use a single IP for a device with 4 network interfaces ? I know it is not the standard router configuration (when each interface is part of a different network) but I need to have my box connected in a ring like topology where each box may need to forward from one port to a destination port based on routing table. I do not want to use a different IP address per interface. Yehuda Kasirer - ------=_NextPart_000_0004_01C01D94.4F2AE370 Content-Type: text/html; charset="windows-1255" Content-Transfer-Encoding: quoted-printable
Is it = possible to=20 configure vxworks to use a single
IP for = a device with=20 4 network interfaces ?
 
I know = it is not the=20 standard router configuration
(when = each interface=20 is part of a different network)
but I = need to have=20 my box connected in a ring like topology
where = each box may=20 need to forward from one port to a destination
port = based on=20 routing table. I do not want to use a different IP = address
per=20 interface.
 

Yehuda = Kasirer
  

- ------=_NextPart_000_0004_01C01D94.4F2AE370-- - -- Posted from mailgw1.netvision.net.il [194.90.1.14] via Mailgate.ORG Server - http://www.Mailgate.ORG --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Using SDRAM with MBX860 Date: 13 Sep 2000 12:09:46 GMT From: Geir Frode Raanes Organization: UNINETT news service Message-ID: <8pnqqa$e2b$1@snipp.uninett.no> References: <8pner9$up0$1@nnrp1.deja.com> Sender: Geir Frode Raanes francescotorre1965@my-deja.com wrote: : Hello, : we are using the MBX860 cpu board from Motorola; This board is End-Of-Line announced. And good riddance - it has a basically lousy design and it is flaky if exposed to any external hardware and/or hostile environment. : it is equipped with 4 Mb of EDO DRAM. Not the newer ones - those have 16 MByte. : We need to extend such memory by means of the : SIMM extension. : Is it possible to use a Syncronous DRAM ? Nope. The DIMM module share Universal Programmeable Machine A on the CPU with the soldered DRAM. Stupid. This means that in order to work the DIMM has to feature exactly the same DRAM as the soldered DRAM. The soldered DRAM may vary. You are basically f... The DIMM socket is one piece of useless HW option. : What we should modify and/or configure ? You cant modify the I2C EEPROM resident UPM tables as the soldered DRAM will then no longer work. Sorry. - -- ****************************************************** Never ever underestimate the power of human stupidity. -Robert Anson Heinlein GeirFRS@invalid.and.so.forth ****************************************************** --------------------------- Newsgroups: comp.os.vxworks Subject: Source code archive link Date: Wed, 13 Sep 2000 17:33:39 +0530 From: "Balaji Ramachandran" Organization: Lawrence Berkeley National Laboratory Message-ID: <8pnr3j$263$1@overload.lbl.gov> VxWorks Tornado Hi all, Anyone can suggest where shall I find Some example source code archive for PCI device access on VxWorks. Thanks & Regards, R.Bala --------------------------- Newsgroups: comp.os.vxworks Subject: Re: interrupt delayed Date: Wed, 13 Sep 2000 13:47 +0100 (BST) From: rew@cp.cix.co.uk (Richard Wenham) Organization: CIX - Compulink Information eXchange Message-ID: References: <39BF2C17.DE9D4475@st2e.com> Reply-To: rew@cp.cix.co.uk I'm not familiar with the Motorola MPC860T, but 20 us is the average latency I would expect with the x86 BSP (Pentium 233). Also, we get the occasional delay of up to 200 us, caused by network activity. As far as I can make out, execution is in the kernel at this time, with interrupts off. Richard Wenham Cranfield Precision In article <39BF2C17.DE9D4475@st2e.com>, anthony.rouaux@st2e.com (Anthony Rouaux) wrote: > I am working with the Motorola MPC860T. I am using the intConnect() > function to connect my routine to CPM timer1 interruption. I find > amazing that it takes 20 us for the core to begin to execute my routine > after the interruption occurs. > Is this due to a hardware or a software problem? > Can you please help me to find a solution to decrease this time. > Thanks --------------------------- Newsgroups: comp.os.vxworks Subject: Re: How to delay current task for some microseconds? Date: Wed, 13 Sep 2000 07:54:29 -0500 From: "Ben A. Abbott" Organization: Posted via Supernews, http://www.supernews.com Message-ID: <39BF7905.BF917C89@swri.org> References: <8pkh2c$bgi$1@schbbs.mot.com> <39BF0AF9.65E8BD9C@earthlink.com> > I use the following to read the accurate clock on a PPC: static int _accurate_time_fake=0; static unsigned long _accurate_time_now_upper32bits() { __asm__ (" mftbu 3"); __asm__ (" blr "); return _accurate_time_fake++; /* this line added to avoid warnings */} static unsigned long _accurate_time_now_lower32bits() { __asm__ (" mftb 3"); __asm__ (" blr "); return _accurate_time_fake++; /* this line added to avoid warnings */} Ben Abbott babbott@swri.org --------------------------- Newsgroups: comp.os.vxworks Subject: Re: How to delay current task for some microseconds? Date: Wed, 13 Sep 2000 09:22:43 -0400 From: Douglas Fraser Organization: Lucent Technologies Message-ID: <39BF7FA3.BFBEED3@lucent.com> References: <8pkh2c$bgi$1@schbbs.mot.com> Li Xin wrote: > > I am writing a device driver under VxWorks and need a means to delay current > task for some microseconds. I know taskDelay() can delay current task, but > that will delay at least one tick which I think is too long for our driver. > Is there any method to delay current task for some microseconds? Many CPUs have a block of FLASH that is used for non-volatile storage. This FLASH will NOT be cached (to allow FLASH programming algorithms to work) and will usually have a well defined access time for read cycles. For very small delays, you can just use some integer number of reads from this FLASH block to implement a predictable delay. Yes, it is board specific, thus non-portable. But it is very light weight and reliable. On my last product where I needed this, FLASH reads completed in exactly 140nS. The instruction loop, being VERY small, fit entirely in the instruction cache. We could change the clock rate on the processor by a gross amount and the change in the delay time was minimal because the access time on the FLASH swamped the execution time. Hope that helps. - -- Douglas Fraser --------------------------- Newsgroups: comp.os.vxworks Subject: Re: printf of "long long" Date: Wed, 13 Sep 2000 09:53:43 -0400 From: Al Johnston Organization: MindSpring Enterprises Message-ID: <39BF86E7.9327DE2@mindspring.com> References: <39B99888.6A31EEC7@mindspring.com> <8pj6lg$t2f$1@nnrp1.deja.com> > The problem is that without a prototype argument, the compiler assumes > that it must put an int on the stack. Given the printf prototype: > > printf(char *, ... ); > > The compiler knows nothing about what arguments the function expects, so > it has to make assumptions. This is what the old K&R compilers did Are you telling me that a printf() with a correct typecaste for a "long long" will work? What format string? The doc says %L is not supported... thanks, - -al --------------------------- Newsgroups: comp.os.vxworks Subject: RE: data access exception in net task Date: Wed, 13 Sep 2000 07:02:53 -0700 From: Jim Way Organization: Lawrence Berkeley National Laboratory Message-ID: <8po24j$858$1@overload.lbl.gov> James, > Running on PowerPCs (2431 and 2603) sending/receiving data TCP as fast > as possible runs fine for about an hour or so and then I get one of the > following conditions: > 1) Receives of packets stop happening - DMAs into my receive buffers > just seem to stop. > 2) A data access exception. The printout points to the net task. > 3) The machine just locks up. > > Looking at mbufShow(), memShow(), spy() and my driver status don't show > anything that would indicate a problem. I'm using PPC860T with VxWorks 5.4. Earlier this year, I had a similar sounding problem. After "a while" my Ethernet would stop talking. There were no apparently obvious symptoms, just like you're describing. The problem was fixed by hiring a WRS consultant who applied two SPR fixes (and did not apply another seemingly applicable one). Here is a snippet of his email to me: > I have integrated the fixes for the following : > > SPR 22325 - Race condition between interrupt processing of received packets and polled mode processing in END driver > SPR 23528 - motCpmEnd driver corrupts tNetTask in system mode > > I have not integrated the fixes for the following : > SPR 21143 - motCmpEnd END driver improperly filters incomming frames when promiscuous mode is enabled. Now, I'm not saying that these specific SPRs are related to your problem. Just that my experience in this area suggests that browsing WindSurf for SPRs might be helpful. Your local FAE might also be an excellent resource for this problem. I know that mine is a top-notch techical guy. Depending on your expertise with VxWorks (mine was very limited at the time, now I'm an old-hand ) you might want to consider using their consulting services. We paid for three days of support and one on-site trip, and I think my management all believe it was money very well spent. I know that I had other things to work on during that time. HTH, Jim - ----------------------------------------- Jim Way, Software Engineer Datum Austin (Austron Inc.) voice: 512.721.4170 fax : 512.990.9712 email: jwayATdatumDOTcom (no spam please) - ----------------------------------------- --------------------------- Newsgroups: comp.os.vxworks Subject: problem in converting from elf to bin Date: Wed, 13 Sep 2000 14:10:40 GMT From: Balajee Premraj Organization: Deja.com - Before you buy. Message-ID: <8po1se$k4d$1@nnrp1.deja.com> i Have built the Vxworks_rom image but when try to covert it to binary image using elftobin utility it gives me an error in conversion why is it so? - -- Balajee Premraj Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: FTP and standard out Date: Wed, 13 Sep 2000 08:54:29 -0600 From: Aaron Stone Organization: Hewlett Packard Message-ID: Im trying to use ftpXfer to transfer files. When I open the file socket. Login and other terminal information are dumped to the screen. Is there any way to turn this off? --------------------------- Newsgroups: comp.os.vxworks Subject: test Date: Wed, 13 Sep 2000 08:55:36 -0600 From: Aaron Stone Organization: Hewlett Packard Message-ID: <2b5vrs8rf83h6ll7ku2bk3itigp0f20snl@4ax.com> fsdafsdfsadfsad --------------------------- Newsgroups: comp.os.vxworks Subject: FTP and standard output Date: Wed, 13 Sep 2000 08:59:17 -0600 From: Aaron Stone Organization: Hewlett Packard Message-ID: Im trying to use ftpXfer to transfer files. When I open the file socket. Login and other terminal information are dumped to the screen. Is there any way to turn this off? --------------------------- Newsgroups: comp.os.vxworks Subject: Problems Building gcc-2.95 for powerpc and vxworks Date: Wed, 13 Sep 2000 14:42:34 GMT From: gavreid@my-deja.com Organization: Deja.com - Before you buy. Message-ID: <8po3ok$mfj$1@nnrp1.deja.com> I get the following error when I try and build gcc-2.95 (with binutils 2.10) for the powerpc-wrs-vxworks target ..... I am doing something wrong ? I have try's this using Linux and Cygwin and both produce the same error. Are there any patches I need to apply ? ... Is there a FAQ ? ###Begin error code block /build/gcc-2.95/gcc/xgcc -B/build/gcc-2.95/gcc/ - B/usr/local/ppc/powerpc-wrs-vxw orks/bin/ -c -DHAVE_CONFIG_H -g -O2 -I. -I./../../include fdmatch.c In file included from /build/gcc-2.95/gcc/include/types/vxANSI.h:45, from /build/gcc-2.95/gcc/include/types/vxTypes.h:63, from /build/gcc-2.95/gcc/include/sys/types.h:39, from fdmatch.c:52: /build/gcc-2.95/gcc/include/types/vxArch.h:46: warning: `_BIG_ENDIAN' redefined *Initialization*:1: warning: this is the location of the previous definition In file included from fdmatch.c:53: /build/gcc-2.95/gcc/include/sys/stat.h:43: parse error before `ULONG' /build/gcc-2.95/gcc/include/sys/stat.h:43: warning: no semicolon at end of struc t or union /build/gcc-2.95/gcc/include/sys/stat.h:44: warning: data definition has no type or storage class /build/gcc-2.95/gcc/include/sys/stat.h:45: parse error before `st_mode' /build/gcc-2.95/gcc/include/sys/stat.h:45: warning: data definition has no type or storage class /build/gcc-2.95/gcc/include/sys/stat.h:49: parse error before `st_rdev' /build/gcc-2.95/gcc/include/sys/stat.h:49: warning: data definition has no type or storage class /build/gcc-2.95/gcc/include/sys/stat.h:50: parse error before `st_size' /build/gcc-2.95/gcc/include/sys/stat.h:50: warning: data definition has no type or storage class /build/gcc-2.95/gcc/include/sys/stat.h:51: parse error before `st_atime' /build/gcc-2.95/gcc/include/sys/stat.h:51: warning: data definition has no type or storage class /build/gcc-2.95/gcc/include/sys/stat.h:52: parse error before `st_mtime' /build/gcc-2.95/gcc/include/sys/stat.h:52: warning: data definition has no type or storage class /build/gcc-2.95/gcc/include/sys/stat.h:53: parse error before `st_ctime' /build/gcc-2.95/gcc/include/sys/stat.h:53: warning: data definition has no type or storage class /build/gcc-2.95/gcc/include/sys/stat.h:56: parse error before `st_attrib' /build/gcc-2.95/gcc/include/sys/stat.h:56: warning: data definition has no type or storage class /build/gcc-2.95/gcc/include/sys/stat.h:63: parse error before `}' /build/gcc-2.95/gcc/include/sys/stat.h:128: parse error before `mkdir' /build/gcc-2.95/gcc/include/sys/stat.h:128: warning: data definition has no type or storage class /build/gcc-2.95/gcc/include/sys/stat.h:129: parse error before `fstat' /build/gcc-2.95/gcc/include/sys/stat.h:129: warning: data definition has no type or storage class /build/gcc-2.95/gcc/include/sys/stat.h:130: parse error before `stat' /build/gcc-2.95/gcc/include/sys/stat.h:130: warning: data definition has no type or storage class /build/gcc-2.95/gcc/include/sys/stat.h:131: parse error before `fstatfs' /build/gcc-2.95/gcc/include/sys/stat.h:131: warning: data definition has no type or storage class /build/gcc-2.95/gcc/include/sys/stat.h:132: parse error before `statfs' /build/gcc-2.95/gcc/include/sys/stat.h:132: warning: data definition has no type or storage class fdmatch.c: In function `fdmatch': fdmatch.c:59: storage size of `sbuf1' isn't known fdmatch.c:60: storage size of `sbuf2' isn't known make[1]: *** [fdmatch.o] Error 1 make[1]: Leaving directory `/build/gcc-2.95/powerpc-wrs- vxworks/libiberty' make: *** [all-target-libiberty] Error 2 ###END error code block Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Turning off standard output with FTP Date: Wed, 13 Sep 2000 09:07:45 -0600 From: Aaron Stone Organization: Hewlett-Packard Fort Collins Site Message-ID: <39BF9840.B78A9901@am.exch.hp.com> Im trying to use ftpXfer to transfer files. When I open the file socket. Login and other terminal information are dumped to the screen. Is there any way to turn this off? --------------------------- Newsgroups: comp.os.vxworks Subject: Query:VxWorks Socket Control Date: Wed, 13 Sep 2000 11:17:56 -0400 From: Pallavi Singh Organization: Nortel Message-ID: <39BF9AA4.D722CA8A@nortelnetworks.com> I wanted to know: 1. In what scenarios will call to recv() and ioctl() to read a socket would fail or return an error code? 2. The reason why a socket control occasionally fails to be read? Regards, Pallavi --------------------------- Newsgroups: comp.os.vxworks Subject: Changing the Network Address VxSim Uses Date: Wed, 13 Sep 2000 08:54:39 -0700 From: "Garner, William F (N-SDC)" Organization: Lawrence Berkeley National Laboratory Message-ID: <8po95i$bbi$1@overload.lbl.gov> VxWorks/Tornado users- How or can the IP address of VxSim be changed from 90.0.0.n to 10.xx.xx.xx? Changing the ULIP IP address to 10.0.0.254, the vxTarget IP address remains (90.0.0.1). - -> hostShow hostname inet address aliases - -------- ------------ ------- vxTarget 90.0.0.1 localhost 127.0.0.1 host 90.0.0.254 trapDest 10.10.6.53 Please help, Bill --------------------------- Newsgroups: comp.os.vxworks Subject: "no carrier" on MVME2400 Date: 13 Sep 2000 16:26:58 GMT From: Juergen Naumann Organization: Uni Dortmund, Physik Message-ID: <8po9si$2vc$1@nx6.HRZ.Uni-Dortmund.DE> Reply-To: naumann@Physik.Uni-Dortmund.DE Dear all When booting VxWorks I get the message "interrupt: dc0 - no carrier" very frequently. I learned from DejaNews that this is not something to really worry about but it's annoying. I also found as a workaround to change the DEC_LOAD_STRING in configNet.h, which seems to work on MVME2300, 2600 and 2700 boards, but unfortunately I'm using a MVME2400 board where this string is completly empty in the BSP. Putting in the value by hand also didn't solve the problem. Does anyone have a hint? Thanks, - -- Juergen Naumann e----------------------------->H1<---------------------------p --------------------------- Newsgroups: comp.os.vxworks Subject: Re: 1553 bus Date: Wed, 13 Sep 2000 09:23:58 -0700 From: "Paul Anderson" Message-ID: References: <8p81io$crb$1@nnrp1.deja.com> <39B85D8C.6BDB687A@cisco.com> Greetings, Compware offers both BC & RT-MT driver packages for the ILC DDC ACE (BU6158x). We also offer the RT-MT driver package for the UTMC SuMMIT device. The BC & RT-MT drivers for the ACE also support "dynamic transition" between BC & RT when both drivers are installed. These packages are quite mature, about 5 years in the field. The packages are also quite complete in supporting features and capabilities of the ACE. Programmer's Reference manuals are comprehensive and well written, and are available under NDA (non-disclosure agreement) for qualified procurements. Also, we are very interested to respond to questions about particular 1553 system requirements toward qualifying our driver packages. Our license model consists of two parts, the "Development/Source" license, which delivers entire source code including test applications which validate the installation, and, the "Right-To-Use" license which is procured for each instance of the run-time version to be deployed, i.e., for each instance of hardware deployed. The "RTU" licening is steeply discounted with volume. Our web site: http://www.compware.com Please let me know if I may be of assistance. Thank you, Paul Anderson Compware Corp. 800-788-5855 "Vijay Kumar Peshkar" wrote in message news:39B85D8C.6BDB687A@cisco.com... > Austin, > > This is not a reply to your query. > I am also interested in knowing more about the mil-std-1553b support. > Can > you please update me if you receive any info..? > > Thanks and Regards, > Vijay --------------------------- Newsgroups: comp.os.vxworks Subject: hardware to monitor 32 vxworks "consols" Date: Wed, 13 Sep 2000 12:22:04 -0400 From: Al Johnston Organization: MindSpring Enterprises Message-ID: <39BFA9AC.C24754D1@mindspring.com> We would like to put together some hardware that would provide access to up to 32 cpu cards vxworks "console" ports. we are running vxworks on mv2700/ppc cards. The proposed "monitoring hardware" would connect to the "com1" port on up to 32 of these cards. It should be able to display a virtual window for each processor. one way to do this would to put 32 tty cards in a sun box and then have a xterm running on the sun box for each of these. We are looking for a cheaper and more physically portable solution. maybe some kind of terminal server hooked to a pc running linux? anybody doing this already? thanks. --------------------------- Newsgroups: comp.os.vxworks Subject: VxWorks 5.4 NFS server problem Date: Wed, 13 Sep 2000 10:30:51 -0700 From: "Se;ymour" Organization: MindSpring Enterprises Message-ID: <8pocqb$9s3$1@nntp9.atl.mindspring.net> I'm having a problems copying files from a Linux Client to the Vxworks NFS Server. When I copy a large number of files after about 3 or 4 minutes the Linux Client reports an errror _nfs_fhget: inode 0 still buys, i_count=1 Does anyone else have this or a similar problems --------------------------- Newsgroups: comp.os.vxworks Subject: Re: printf of "long long" Date: Wed, 13 Sep 2000 18:07:39 GMT From: John Fusco Organization: Deja.com - Before you buy. Message-ID: <8pofop$65f$1@nnrp1.deja.com> References: <39B99888.6A31EEC7@mindspring.com> <8pj6lg$t2f$1@nnrp1.deja.com> <39BF86E7.9327DE2@mindspring.com> In article <39BF86E7.9327DE2@mindspring.com>, Al Johnston wrote: > > > The problem is that without a prototype argument, the compiler assumes > > that it must put an int on the stack. Given the printf prototype: > > > > printf(char *, ... ); > > > > The compiler knows nothing about what arguments the function expects, so > > it has to make assumptions. This is what the old K&R compilers did > > Are you telling me that a printf() with a correct typecaste for a > "long long" will work? What format string? The doc says %L is > not supported... > Hmmm. I may have to retract that statement. I tried somem experiments and discovered that the following printf formats work with long long... printf("%lld\n",mylonglong) printf("%llx\n",mylonglong) This is a surprise to me since I didn't think that standard C understood what to do with long long integers in the absense of a prototype. But there it is, I tried it on: gcc 2.8.1 Solaris C 5.0 IRIX MIPSpro Compilers: Version 7.2.1 Unfortunately it did not work on my PowerPC gcc from Tornado gcc 2.7.2. So I guess you're out of luck unless you are brave enough to build your own compiler from a newer GCC release. I imagine it would work, but who knows. Good luck. Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: hardware to monitor 32 vxworks "consols" Date: Wed, 13 Sep 2000 13:18:16 -0500 From: Andrew Johnson Organization: APS, Argonne National Laboratory Message-ID: <39BFC4E8.55B26F9F@aps.anl.gov> References: <39BFA9AC.C24754D1@mindspring.com> Al Johnston wrote: > > We would like to put together some hardware that would provide > access to up to 32 cpu cards vxworks "console" ports. ... > one way to do this would to put 32 tty cards in a sun box and > then have a xterm running on the sun box for each of these. Ouch, wasteful... > We are looking for a cheaper and more physically portable > solution. maybe some kind of terminal server hooked to > a pc running linux? We have lots of terminal servers spread around the APS storage ring which are premanently plugged into our console ports, then we just telnet to the relevent server port to talk to the vxWorks system. We allow engineers to connect to systems they need to diagnose from their offices, or even through a dialup line from home (securely of course). - - Andrew - -- Every great idea appears crazy to start with. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: PCI Configuration help for a novice Date: Wed, 13 Sep 2000 18:21:01 GMT From: John Fusco Organization: Deja.com - Before you buy. Message-ID: <8poghk$78k$1@nnrp1.deja.com> References: <39AABFBA.C71B9657@res.raytheon.com> My version of the 2304 BSP does not do PCI autoconfig. I don't think they have updated that BSP to do so. All the devices have drivers with hard-coded base addresses. Updating the BSP to use the autoconfig library is no small task. A zero is what you would expect to see in a BAR that has not been configured or is not implemented. In article <39AABFBA.C71B9657@res.raytheon.com>, kostizt wrote: > Folks, > > I must admit that I am new to the PCI bus and I am having a lot of > trouble getting a vendor supplied driver / serial device into my > system. The symptom is that (a) the BARs don't make a lot of sense to > me and (b) the device (serial), doesn't seem to be mapped (i/o simply > posting in memory with no interrupts). > > I have a 2304 BSP. The vendor's documentation says "ensure that the BSP > itself performs configuration of all PCI devices supported by the PCI > HAL. This includes configuring the PCI base address registers (BARs) > and the interrupt line request register in the PCI configuration space > of each device." > > I thought I would find the code that does this in the BSP and go from > there. The problem is that I can't find where any of this happens! > > Does anyone know how the PCI devices get their configuration information > processed by the BSP? > > If I invoke an xpciShow, it looks as follows: > > -> xpciShow > Scanning function 0 of each PCI device on bus 0 of card 0 > bus slot vendor dev IRQ Bar0 Bar1 Bar2 Bar3 Bar4 > Bar5 > 0 00 1057 4801 00 00000001 fc000000 00000000 00000000 00000000 > 00000000 > 0 11 10ad 0565 00 00000000 00000000 00000000 00000000 00000000 > 00000000 > 0 13 10e3 0000 00 fd050000 00000001 00000000 00000000 00000000 > 00000000 > 0 14 1011 0009 00 00020001 00000000 00000000 00000000 00000000 > 00000000 > 0 17 124b 0001 00 00000000 00000000 00000000 00000000 00000000 > 00000000 > value = 1 = 0x1 > > The information for bus/slot/vendor for slot 17 is the interesting > piece. The vendor info is the PCI device of interest. Does a BAR0 of 0 > make sense? > > Any help would be greatly appreciated. > > Thanks in advance! > > Tom Kostizak > > Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: hardware to monitor 32 vxworks "consols" Date: Wed, 13 Sep 2000 14:41:14 -0400 From: Al Johnston Organization: MindSpring Enterprises Message-ID: <39BFCA4A.957BDD57@mindspring.com> References: <39BFA9AC.C24754D1@mindspring.com> <39BFC4E8.55B26F9F@aps.anl.gov> Hey, glad you are still around Your "blast tas" code is one reason why I wanted to set this up. I had a failure after three days... no msg was written to the screen (a xterm with a rsh to the target)... I think it was a "network timeout" issue rather than a real failure, but I am not sure. we really want this capability anyway, so now is a good time to get it back... we had it via some hardware that was loaded via the vax... but we got ride of the vax a while back [ :-< i miss my vax] and have never re-established the capability. At the moment I am twiddling my thumbs... our !Q$ dawn vme 64x chassis ate a 3rd card two weeks ago... and I have been refusing to use them until the defects are fixed... man am I board... I hate my job when its like this... but i have two many things going at once already.... I don't want to start another one. thanks for your input. --------------------------- Newsgroups: comp.os.vxworks Subject: Ring Buffer Overflow? Date: Wed, 13 Sep 2000 19:22:01 GMT From: Joe Andrews Organization: Vehicle Managememt Systems Message-ID: <39BFD3D9.F191F871@mw.boeing.com> Reply-To: joe.andrews@mw.boeing.com Sender: nntp@news.boeing.com (Boeing NNTP News Access) I'm having a problem running any downloaded code on my mv2604 boards. The code runs fine in the sim. I rebuilt the project for the board and downloaded in to the card. When I try to execute any function under the shell, I immediately get the following error: LogTask: ring buffer overflow After a few of those, it switches to : LogTask 11xx messages lost ring buffer overflow Which scrolls forever. The only way I can stop it is to reset the card. It seems to happen before any of my code gets called: if I set a breakpoint at the first line of the function, it still happens. Any idea what this error means, (yes, I know it means my ring buffer is overflowing) or what's causing it? Or, how can I find out what the messages are in the buffer? The manual states LogTask gets its input from a message queue filled by LogMsg. The default output device has not been changed and is set to ConsoleFD, but the above messages are all I ever see. Interestingly, I originally had the target server set up to use the backplane network and ran the code on card #2, with card #1 being the gateway. When I tried to execute a function on card #2, the above message would appear on card #1's console. --------------------------- Newsgroups: comp.os.vxworks Subject: Ring Buffer Overflow? Correction Date: Wed, 13 Sep 2000 19:46:00 GMT From: Joe Andrews Organization: Vehicle Managememt Systems Message-ID: <39BFD978.9B790E13@mw.boeing.com> Reply-To: joe.andrews@mw.boeing.com Sender: nntp@news.boeing.com (Boeing NNTP News Access) Oops, I mistakenly gave the wrong error before. What really comes out is this: interrupt: panic: netJobAdd: ring buffer overflow The cycle then repeats with the following: LogTask: xxx messages lost interrupt: panic: netJobAdd: ring buffer overflow Where xxx is a number anywhere from 200-1500 --------------------------- Newsgroups: comp.os.vxworks Subject: Re: printf of "long long" Date: Wed, 13 Sep 2000 17:07:24 -0400 From: Al Johnston Organization: MindSpring Enterprises Message-ID: <39BFEC8C.1EFB7288@mindspring.com> References: <39B99888.6A31EEC7@mindspring.com> <8pj6lg$t2f$1@nnrp1.deja.com> <39BF86E7.9327DE2@mindspring.com> <8pofop$65f$1@nnrp1.deja.com> Hi, > Hmmm. I may have to retract that statement. I tried somem experiments > and discovered that the following printf formats work with long long... > > printf("%lld\n",mylonglong) > printf("%llx\n",mylonglong) > > This is a surprise to me since I didn't think that standard C understood > what to do with long long integers in the absense of a prototype. But > there it is, I tried it on: The only place I have found "%ll" documented was in solaris man pages. So I am not sure what "standard" (if any) it is from. it seems to work with gcc 2.95.2-solaris8; it compiles okay with my vxw/ppc gcc compiler (no warnings, etc... even with -Wall or -ansi! > gcc 2.8.1 > Solaris C 5.0 > IRIX MIPSpro Compilers: Version 7.2.1 ah, another IRIX user!!! > Unfortunately it did not work on my PowerPC gcc from Tornado gcc 2.7.2. > So I guess you're out of luck unless you are brave enough to build your > own compiler from a newer GCC release. I imagine it would work, but who Thanks... at the moment I have no hardware... so I couldn't see if it would work on the ppc... the vxw doc says %L is not supported, but does not mention %ll. Thanks for the answer!!! (sure wish it would have worked!) Seems like a strange thing for wrs to break... but then again, %ll may not have worked in gcc 2.7.x to begin with. I am planing to build my own gcc/ppc sometimes in the near future... but I am putting it off at the moment. And hoping some other nice person will post a faq/howto... thanks for the help! - -al --------------------------- Newsgroups: comp.os.vxworks Subject: How to get cursor position in Zinc? Date: Wed, 13 Sep 2000 16:00:13 +0800 From: "zongjun" Organization: The news server of Changzhou TeleCom Message-ID: <8pnc07$pmn$1@news.cz.js.cn> Hi all: I would like to program InputMethod useing Zinc,but I don't know how to get cursor position. I'm waiting for your help. Thanks in advance! Zong --------------------------- Newsgroups: comp.os.vxworks Subject: Re: "no carrier" on MVME2400 Date: Wed, 13 Sep 2000 17:00:36 -0500 From: Greg Willden Organization: Posted via Supernews, http://www.supernews.com Message-ID: <39BFF904.95180ADE@swri.edu> References: <8po9si$2vc$1@nx6.HRZ.Uni-Dortmund.DE> This is a multi-part message in MIME format. - --------------2484C3DA2D03A07DF063CC85 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Juergen Naumann wrote: > Dear all > > When booting VxWorks I get the message "interrupt: dc0 - no carrier" > very frequently. I learned from DejaNews that this is not something to > really worry about but it's annoying. I also found as a workaround to > change the DEC_LOAD_STRING in configNet.h, which seems to work on > MVME2300, 2600 and 2700 boards, but unfortunately I'm using a MVME2400 > board where this string is completly empty in the BSP. Putting in the > value by hand also didn't solve the problem. > > Does anyone have a hint? > You can also try changing the settings on the ethernet cable to half duplex. We had a Cisco router connected to a MVME2307 and forcing the port on the Cisco to half duplex did the trick for us. I think that if you force the port to 10BaseT as opposed to 100 the problem will go away as well. I not sure about the second solution though. - --------------2484C3DA2D03A07DF063CC85 Content-Type: text/x-vcard; charset=us-ascii; name="gwillden.vcf" Content-Transfer-Encoding: 7bit Content-Description: Card for Greg Willden Content-Disposition: attachment; filename="gwillden.vcf" begin:vcard n:Willden;Greg tel;fax:(210) 522-2709 tel;work:(210) 522-6241 x-mozilla-html:FALSE org:Southwest Research Institute;www.swri.edu version:2.1 email;internet:gwillden@swri.edu title:Engineer adr;quoted-printable:;;6220 Culebra Rd=0D=0A=0D=0A;San Antonio, TX 78238;;; fn:Greg Willden end:vcard - --------------2484C3DA2D03A07DF063CC85-- --------------------------- Newsgroups: comp.os.vxworks Subject: questions wxwork/tornado Date: Thu, 14 Sep 2000 00:10:52 GMT From: e.jolly@home.com (Ric) Organization: Excite@Home - The Leader in Broadband Message-ID: <39c01403.7876622@news> Hi everyone, i have questions on wxvorks and tornado. I hope that an expert will be able to answer the following, thanks in advance. The eggshell from vxwork give us access to several debugging commands (l(), lkup(), i(), tt() etc...). I was reading the tornado (tornado 2) user guide, and i found out that the Wind Sh is really similar to the eggshell from vxwork. My questions: is tornado a graphical debugger interface using commands already available trough vxwork eggsehll? Or tornado really has enhanced feature? Also, when using tornado is it possible to set multiple condtions breakpoints? by that i mean making sure that several condition are met before breaking or breaking when one out of several condition are met (and/or multiple condition). If we are using multiple targets is possible to debug simultaniously several target (ex: communication between 2 or more node, monitor objets on both node)? also when breaking we could get the display of the trace either on assembly, c or c++? thanks -eric --------------------------- Newsgroups: comp.os.vxworks Subject: RE: Changing the Network Address VxSim Uses Date: Wed, 13 Sep 2000 22:05:58 -0400 From: "Mike Anderson" Organization: Lawrence Berkeley National Laboratory Message-ID: <8ppcai$7t$1@overload.lbl.gov> VxWorks Greetings! > > How or can the IP address of VxSim be changed from 90.0.0.n to > 10.xx.xx.xx? > > [snip] This only works with the full copy of the simulator (not the Simulator Lite that comes with Tornado, but the extra cost version that includes networking. Also, I believe that this only works on the Solaris version of the simulator. Start your simulator & target server. Startup a Windsh. at the -> issue a bootChange command. - -> bootChange [you'll be prompted for all of the boot parameters] Key in the new inet on enet (10.x.y.z) and boot host IP (10.x.y.n). Finish entering the rest of the boot parameters. When you get back to the -> prompt, there will be a file saved to the disk subsystem at the location in the file system that the simulator booted from. Unfortunately, I don't remember the file name, but it will be obvious from the name. This will have the boot line in it and can be edited to change the IP parameters. HTH, Mike Anderson The PTR Group, Inc. Embedded Development, Training and Consulting Services mailto:mike@theptrgroup.com http://www.theptrgroup.com --------------------------- Newsgroups: comp.os.vxworks Subject: Re: dosFs2 chkdsk slow to run from bootrom Date: Thu, 14 Sep 2000 02:26:51 GMT From: paulvroberts@my-deja.com Organization: Deja.com - Before you buy. Message-ID: <8ppd0q$9lr$1@nnrp1.deja.com> References: <8p4osn$ja8$1@nnrp1.deja.com> <39BB6FD7.76855B10@bitband.com> Leonid and all, Thanks for the reply. Its nice to have an expert on the newsgroup! Actually, after I posted the message, I noticed this in the code for the check routine, so I added a warning message if the FAT was being used. I didn't see my warning message. This may of course be a bug in my patch, or a build problem. I haven't actually got any time right now to go back over and verify it or to try your suggestion (we are using the MVME2700 family here and I could not find INCLUDE_ADD_BOOTMEM being used - I will need to dig deeper). Hopefully I will be back on this in a few weeks. Sorry if this sounds waffly, I just wanted to make sure I said "thank you" and thought you and others might like to know the status. Paul Roberts. In article <39BB6FD7.76855B10@bitband.com>, leonid@bitband.com wrote: > Actually, this is very simple - > > the CHKDSK utility in DosFs2 would take advantage of available memory to > hasten > the process by keeping the entire FAT in RAM, if this however is impossible, > > it will use the "second" FAT copy on the disk as its working copy of FAT. > The "working copy" is where the software keeps track of cluster chains > traversed etc. > > So what you need is to define a few defines to make bootCOnfig.c allocate > more memory, > for PC target these are: > > :#define INCLUDE_ADD_BOOTMEM /* Add upper memory to low memory bootrom*/ > :#define ADDED_BOOTMEM_SIZE 0x00200000 /* 2MB */ > > Dont know how much memory will be sufficient for you, your milage will vary > ... > > Leonid > > paulvroberts@my-deja.com wrote: > > > Hi there, > > > > Does anyone know why chkdsk would be much slower when run from the > > bootrom, then when run after VxWorks is loaded? > > > > We are using FAT32 (singly dos partitioned) 9GB SCSI disks and it takes > > 65 seconds in the bootrom, whereas after VxWorks is running, it takes > > just 3-4 seconds! > > > > What I did was to "enhance" the bootConfig.c file to allow booting using > > dosFs2. I use a 128K cache, and call chkdsk via the icotl FIOCHKDSK. > > > > I am using the MVME2700 BSP with the bootrom_uncmp target, so presumably > > that's not rom-resident. > > > > Any other ideas? > > > > Thanks, > > > > Paul Roberts. > Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: UDP Socket Receive Date: Fri, 15 Sep 2000 03:12:37 +1000 From: Chris Chock Organization: Pacific Internet (Australia) Message-ID: <39C10705.77D4E80A@hunterlink.net.au> I have been sending data from a Radstone PPC4A card to a PC via ethernet, using UDP. When I try to receive using either "recvfrom" or "select" I never seem to get any UDP packets. I know that the PC is transmitting 80 bytes of data at a 10 Hz rate, but my application never gets them. Are there any known issues with receiving UDP packets ? Thanks. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: How to get cursor position in Zinc? Date: Thu, 14 Sep 2000 10:07:48 +0530 From: Vijay Kumar Peshkar Organization: Wipro Technologies Message-ID: <39C0561B.52A20F44@cisco.com> References: <8pnc07$pmn$1@news.cz.js.cn> Reply-To: vpeshkar@cisco.com This is a multi-part message in MIME format. - --------------7619A8EE6200A7E3CCC60D01 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Zong, I am assuming you want the cursor position of the mouse. Here's how to do it. The basic idea is to create input queues and attach them to the device. Then check the queue for a combination of conditions to see if the mouse has moved. 1. Create Input Driver for mouse using either: ugl8250MsMouseDevCreate - assuming you have a MS serial mouse., else use uglPs2MouseDevCreate. Here you will specify the queue and interrupt number related to the mouse and also port address. 2.call uglInputQEventGet to check for any new events...this can be in a low priority task. The elements in the queue will be in a structure of type UGL_INPUT_EVENT. You need to extract the x,y and 'id' from it. More details with examples are in the Zinc manual. Have Fun, Vijay zongjun wrote: > Hi all: > I would like to program InputMethod useing Zinc,but I don't know how to get > cursor position. > I'm waiting for your help. > > Thanks in advance! > > Zong - --------------7619A8EE6200A7E3CCC60D01 Content-Type: text/x-vcard; charset=us-ascii; name="vpeshkar.vcf" Content-Transfer-Encoding: 7bit Content-Description: Card for Vijay Kumar Peshkar Content-Disposition: attachment; filename="vpeshkar.vcf" begin:vcard n:Peshkar;Vijay Kumar x-mozilla-html:FALSE org:Wipro Technologies,K2, Bangalore;Wishing you blue skies, a couple of daisies, some fresh air and a smile! adr:;;;;;; version:2.1 email;internet:vpeshkar@cisco.com title:Senior Software Engineer fn:Vijay Kumar Peshkar end:vcard - --------------7619A8EE6200A7E3CCC60D01-- --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Source code archive link Date: Thu, 14 Sep 2000 07:11:25 GMT From: Balajee Premraj Organization: Deja.com - Before you buy. Message-ID: <8pptmi$ria$1@nnrp1.deja.com> References: <8pnr3j$263$1@overload.lbl.gov> In article <8pnr3j$263$1@overload.lbl.gov>, "Balaji Ramachandran" wrote: > VxWorks Tornado > > Hi all, > > Anyone can suggest where shall I find Some example source code archive > for PCI device access on VxWorks. > > Thanks & Regards, > > R.Bala > > Hai, check out the directory, target\h\drv\pci inyour tornado installation root ,might be useful. - -- Balajee Premraj Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Elan 520, AMD BSP Date: Thu, 14 Sep 2000 09:44:18 +0200 From: Dieter Schmidhuber Organization: Siemens Inc. Message-ID: <39C081D2.5522CCA@dit.siemens.de> Hello, hello, is there anybody who has experience with the biosless BSP from AMD for the Elan 520, or does anybody know an other way to work with the Elan 520 and VxWorks biosless. Thanks --------------------------- Newsgroups: comp.os.vxworks Subject: Re: printf of "long long" Date: Thu, 14 Sep 2000 10:42:49 +0300 From: Ofer Goren Organization: Bay Networks, Inc. Message-ID: <39C08178.C23AF43D@nortelnetworks.com> References: <39B99888.6A31EEC7@mindspring.com> <8pj6lg$t2f$1@nnrp1.deja.com> <39BF86E7.9327DE2@mindspring.com> <8pofop$65f$1@nnrp1.deja.com> <39BFEC8C.1EFB7288@mindspring.com> Al Johnston wrote: > Seems like a strange thing for wrs to break... but then again, %ll may > not > have worked in gcc 2.7.x to begin with. I am planing to build my own > gcc/ppc sometimes in the near future... but I am putting it off at the > moment. And hoping some other nice person will post a faq/howto... > Well, maybe this nice person will be you. If you DO try to build the newer compiler, can you share the steps required, with us? Thanks, > > thanks for the help! > > -al - -- Embedded SW Team Leader Email : ogoren@nortelnetworks.com Phone number: +9723-6456023 Fax number : +9723-6479579 ESN : 826-6023 http://www.nortelnetworks.com - ---------------------------------------------------- If you lend someone $20, and never see that person again, it was probably worth the investment. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Query:VxWorks Socket Control Date: Thu, 14 Sep 2000 11:18:23 +0100 From: Phil Ashby Organization: Internet Designers Limited - a BT company Message-ID: <39C0A5EF.D0C0CDCC@ashbysoft.cix.co.uk> References: <39BF9AA4.D722CA8A@nortelnetworks.com> recv() can fail for a multitude of reasons, but usually the connection being closed, or the local socket being closed. Less likely is that the buffer address or the length is invalid, or the socket is in non-blocking mode and would block (errno = EWOULDBLOCK). Similar comments apply to ioctl(), particularly invalid buffer size / length. Phil. Pallavi Singh wrote: > > I wanted to know: > > 1. In what scenarios will call to recv() and ioctl() to read a socket > would fail or return an error code? > > 2. The reason why a socket control occasionally fails to be read? > > Regards, > Pallavi --------------------------- Newsgroups: comp.os.vxworks Subject: Re: UDP Socket Receive Date: Thu, 14 Sep 2000 11:26:13 +0100 From: Phil Ashby Organization: Internet Designers Limited - a BT company Message-ID: <39C0A7C4.54A3BA53@ashbysoft.cix.co.uk> References: <39C10705.77D4E80A@hunterlink.net.au> How does the PC send the packets? Does it reply to incoming packets from the VxWorks board, or does it use an internally generated 'struct sockaddr_in' in the sendto() call? Are there any routers involved (and is the routing correct on the PC)?. Does the PC receive packets from the VxWorks board OK? Have you tried snooping the network traffic (Netmon is quite useful :) Phil. Chris Chock wrote: > > I have been sending data from a Radstone PPC4A card to a PC via > ethernet, using UDP. When I try to receive using either "recvfrom" or > "select" I never seem to get any UDP packets. I know that the PC is > transmitting 80 bytes of data at a 10 Hz rate, but my application never > gets them. Are there any known issues with receiving UDP packets ? > > Thanks. --------------------------- End of New-News digest ********************** From vxwexplo-errs@csg.lbl.gov Thu Sep 14 14:52:41 2000 From: "Martin Mayer" Date: Thu Sep 14 14:52:43 PDT 2000 Subject: MAC address This message is in MIME format. Since your mail reader does not understand this format, some or all of this message may not be legible. ------_=_NextPart_001_01C01E95.FC4FEA30 Content-Type: text/plain; charset="iso-8859-1" Hi, Where does vxWorks get the MAC address from when displaying it in the ifShow routine? Is it possible to modify this MAC address dynamically? Thanks, Martin Mayer ------_=_NextPart_001_01C01E95.FC4FEA30 Content-Type: text/html; charset="iso-8859-1" MAC address

Hi,

Where does vxWorks get the MAC address from when displaying it in the ifShow routine?

Is it possible to modify this MAC address dynamically?

Thanks,
Martin Mayer

------_=_NextPart_001_01C01E95.FC4FEA30-- From vxwexplo-errs@csg.lbl.gov Thu Sep 14 15:05:32 2000 From: "Martin Mayer" Date: Thu Sep 14 15:05:34 PDT 2000 Subject: Dynamically changing IP addresses This message is in MIME format. Since your mail reader does not understand this format, some or all of this message may not be legible. ------_=_NextPart_001_01C01E95.C7270F00 Content-Type: text/plain; charset="iso-8859-1" Hi, I would like to write a command that will change the IP address that vxWorks uses. Any hints? Thanks, Martin Mayer ------_=_NextPart_001_01C01E95.C7270F00 Content-Type: text/html; charset="iso-8859-1" Dynamically changing IP addresses

Hi,

I would like to write a command that will change the IP address that vxWorks uses.

Any hints?

Thanks,
Martin Mayer



------_=_NextPart_001_01C01E95.C7270F00-- From vxwexplo-errs@csg.lbl.gov Thu Sep 14 16:13:25 2000 From: Carlos Rimola Date: Thu Sep 14 16:13:28 PDT 2000 Subject: Anyone running RIP V2 out there? Is anyone running VxWorks RIP in version 2 mode? I just found a bug in the latest Wind River code that I can't imagine anyone being able to successfully run RIP in V2 mode. The bug is in the ripSplitPacket function and will result in routes that expire after 120 seconds and a possible page fault due to a 0 ifp address being stored in the route entry. E-mail me if you want more details and a fix.. In a related note - if you were to ship a new router product with RIP support, what would you default its operation (supply) to? RIP V1 or V2? Which one is more popular? (I noticed that the RIP that ships with Linux defaults to RIP 1 mode). Carlos Rimola Sr. Software Engineer Praxon, Inc. From vxwexplo-errs@csg.lbl.gov Thu Sep 14 16:54:58 2000 From: "Jay Salerno-nonTRW" Date: Thu Sep 14 16:55:00 PDT 2000 Subject: Posting I have a SBS POWERDENALI board. Has anyone been able to get vxworks to boot from the 2nd 8Mbyte flash part? Or, does anyone know of a TFFS driver for the intel 28F160 flash memory part? Thanks in advance Jay Salerno email: jay.salerno-nonTrw@trrw.com phone: 310-764-6310 From vxwexplo-errs@csg.lbl.gov Thu Sep 14 19:30:03 2000 From: Fred Roeber Date: Thu Sep 14 19:30:05 PDT 2000 Subject: Re: hardware to monitor 32 vxworks "consols" On Thu, 14 Sep 2000, Al Johnston wrote: > We would like to put together some hardware that would provide > access to up to 32 cpu cards vxworks "console" ports. we are > running vxworks on mv2700/ppc cards. The proposed "monitoring > hardware" would connect to the "com1" port on up to 32 of these > cards. It should be able to display a virtual window for > each processor. We use Lantronix terminal servers for this. They are neat little boxes (3" x 7" x 5") that have an ethernet connection and some number of serial port connections (e.g 8, 12 or 16). You hook the console serial lines on target boards to different serial ports on the Lantronix box. To see what is happening on the serial port, you simply telnet to the Lantronix box (over the ethernet connection) passing it a port number. You then see all serial traffic just as if you were connected directly. Different people can telnet to each of the different ports. Works great for us. Far better than rlogin method since you get to see all the bootup/crash type messages. The only down side is the Lantronix boxes aren't cheap (I think they are like $1K). Sorry, I don't buy the things for our labs so don't have the contact info. Should be able to find it on the web though. Fred | Fred J Roeber, BBN Systems & Technologies | | 4 John Clarke Road Middletown, RI 02842-5202 | | froeber@bbn.com 401-848-3548 | | Division Scientist, High Performance Computing | From vxwexplo-errs@csg.lbl.gov Fri Sep 15 04:03:08 2000 From: Vxworks Exploder Date: Fri Sep 15 04:03:10 PDT 2000 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Fri Sep 15 04:03:04 PDT 2000 Subject: IP Link-Local Address space Subject: rlogin or telnet vxworks Client ? Subject: Re: How to delay current task for some microseconds? Subject: ftpXfer question Subject: How to use socket API in pSOS Subject: Telecom 'Start-up' Equity Opportunities / RTOS Subject: Re: How to delay current task for some microseconds? Subject: Use of spares in WIND_TCB Subject: [Fwd: Bootup Times ... how long????] Subject: task states Subject: Re: printf of "long long" Subject: Re: hardware to monitor 32 vxworks "consols" Subject: Re: How to use socket API in pSOS Subject: Re: How to add hdlc to ppp Subject: tgtsvr Subject: Lint and TornadoII Subject: localtime function in vxWorks Subject: MAC address Subject: Dynamically changing IP addresses Subject: Anyone running RIP V2 out there? Subject: Posting Subject: How can I config local IP address in pSOS? Subject: how to allocate a block of physically contiguous memory Subject: synchronizing local and host sym tables Subject: re: hardware to monitor 32 vxworks "consols" Subject: TFFS problem Subject: Re: how to allocate a block of physically contiguous memory Subject: Re: Use of spares in WIND_TCB Subject: Re: task states Subject: Re: How can I config local IP address in pSOS? Subject: Re: How to use socket API in pSOS Subject: Re: how to allocate a block of physically contiguous memory Subject: VGA controller on MBX860 Subject: Re: how to allocate a block of physically contiguous memory Subject: Re: Lint and TornadoII Subject: Re: tgtsvr ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: IP Link-Local Address space Date: Thu, 14 Sep 2000 11:14:11 +0100 From: keith.wood@epid.eurotherm.co.uk Organization: Lawrence Berkeley National Laboratory Message-ID: <200009141020.LAA22061@castor.epid.eurotherm.co.uk> Does any version of VxWorks support use of the Link-Local address space (169.254/16) as an extension to DHCP? If yes - which versions? If no - does anyone know if there are plans to incorporate it in any future versions? Keith Wood - ------------------------- e-mail work: keith.wood@epid.eurotherm.co.uk home: genius@the-wood-family.org.uk phone: work: +44 1903 205277 My views are not my own - they are the intellectual property of my employer --------------------------- Newsgroups: comp.os.vxworks Subject: rlogin or telnet vxworks Client ? Date: Thu, 14 Sep 2000 12:19:09 GMT From: "Phil" Organization: The World @ Software Tool & Die Message-ID: Sender: news@world.std.com (Mr Usenet Himself) My company is looking for a telnet or rlogin client for Tornado 2.0 - VxWorks 5.4. Thanks, Phil --------------------------- Newsgroups: comp.os.vxworks Subject: Re: How to delay current task for some microseconds? Date: Thu, 14 Sep 2000 14:44:35 +0200 From: Leif Thuresson Organization: TSL Message-ID: <39C0C833.30D8C2C9@tsl.uu.se> References: <8pkh2c$bgi$1@schbbs.mot.com> Reply-To: leif.thuresson@tsl.uu.se This is a multi-part message in MIME format. - --------------22C0C1F50E2107BD3FEA4243 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Li Xin wrote: > I am writing a device driver under VxWorks and need a means to delay current > task for some microseconds. I know taskDelay() can delay current task, but > that will delay at least one tick which I think is too long for our driver. > Is there any method to delay current task for some microseconds? I a few years back Geoffrey Espin posted this a self calibrating delay routine on the vxWorks exploder. It's the next best thing to a hardware timer. The first call to the delay routine takes a long time because it then does the calibration. If you can't tolerate that you can make a dummy call when installing the device driver. You can test the accuracy on your platform with timexN. Good luck ! - --leif leiff@tsl.uu.se - --------------22C0C1F50E2107BD3FEA4243 Content-Type: text/plain; charset=us-ascii; name="delay.txt" Content-Transfer-Encoding: 7bit Content-Disposition: inline; filename="delay.txt" Submitted-by espin@idiom.com Fri Dec 5 11:35:17 1997 Submitted-by: Geoffrey Espin - - --MimeMultipartBoundary Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Marc, et al, > Submitted-by marc@onion.jhuapl.edu Fri Dec 5 07:26:18 1997 > We are using vxWorks 5.1.1 and are using a clock rate of 100 ticks/msec, > we would like to switch to a higher clock rate (1000ticks/msec) and need > to know if/how using a higher clock rate might negatively impact performance. Obviously the more interrupts per second from any source will cost you something. A 1000/sec on lots of CPUs these days is probably not a big deal. If an interrupt costs you 5 microseconds: 1000 * 5 usec = 5 millisecs of mostly wasted time every second But note that you cannot use taskDelay() from an interrupt handler! :-) Attached is a portable sub-clock tick "hard" (busy loop) delay[>] library. I can't believe how many times I've seen code like (bletch!): for (ix = 0; ix < 0x40000; ix++) ; /* spin wheels for 20 usecs (I hope) */ WRS does supply nanosleep() but is actually limited to system clock rate usually I think. Attached are my delayLib.c and delayLib.h. If you have any suggestions or improvments let me know: espin@idiom.com. Geoff - - -- Geoffrey Espin espin@idiom.com - - -- +=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+= /* delayLib.c - self-calibrating hard delay routines */ /* modification history - - -------------------- 27Mar96,espin written. */ /* DESCRIPTION This module provides "hard" [<]delay[>] routines for micro and millisecond periods. EXAMPLE .CS - - -> delayMsec (1); /@ very first call used for calibration @/ - - -> timexN delayMsec, 10 timex: 75 reps, time per rep = 9555 +/- 222 (2%) microsecs value = 59 = 0x3b = ';' - - -> .CE The routines sysClkRateGet() and tickGet() are used to calibrate the timing loop the first time either routine is called. Therefore, the first call is much longer than requested. If the system clock rate is changed, a new calibration must be explicitly made by passing a negative [<]delay[>] duration. INTERNAL A one-shot timer could provide high resolution sub-clock tick delay... but then this would be board dependent. */ #include "vxWorks.h" #include "tickLib.h" #include "sysLib.h" #define DEBUG FALSE void delayUsec (unsigned int u); void delayMsec (unsigned int m); /******************************************************************************* * * delayUsec - hard [<]delay[>] for microseconds * * RETURNS: N/A */ #if DEBUG int delayLoop = 0; #endif /* DEBUG */ void delayUsec ( unsigned int u /* # of microsecs */ ) { #if !DEBUG static int delayLoop = 0; #endif /* !DEBUG */ int ix; int iy; if (delayLoop == 0 || u == 0xffffffff) /* need calibration? */ { int maxLoop; int start = 0; int stop = 0; int mpt = (1000 * 1000) / sysClkRateGet (); /* microsecs per tick */ for (delayLoop = 1; delayLoop < 0x1000 && stop == start; delayLoop<<=1) { for (stop = start = tickGet (); start == stop; start = tickGet ()) ; /* wait for clock turn over */ delayUsec (mpt); /* single recursion */ stop = tickGet (); } maxLoop = delayLoop / 2; /* loop above overshoots */ #if DEBUG printf ("maxLoop = %d\n", maxLoop); #endif /* DEBUG */ start = 0; stop = 0; if (delayLoop < 4) delayLoop = 4; for (delayLoop /= 4; delayLoop] for milliseconds * * RETURNS: N/A */ void delayMsec ( unsigned int m /* # of millisecs */ ) { delayUsec (m * 1000); } Here's the include file: /* delayLib.h - self-calibrating hard [<]delay routines header file */ /* modification history - - -------------------- 27Mar96,espin written. */ #ifndef __INCdelayLibh #define __INCdelayLibh #if defined(__STDC__) || defined(__cplusplus) extern void delayUsec (unsigned int u); extern void delayMsec (unsigned int m); #else extern void delayUsec (); extern void delayMsec (); #endif /* __STDC__ || __cplusplus */ #endif /* __INCdelayLibh */ - --------------22C0C1F50E2107BD3FEA4243-- --------------------------- Newsgroups: comp.os.vxworks Subject: ftpXfer question Date: Thu, 14 Sep 2000 12:46:26 GMT From: rohitkapur@my-deja.com Organization: Deja.com - Before you buy. Message-ID: <8pqhas$hmb$1@nnrp1.deja.com> Hi: I am writing an ethernet driver for the AMD79C940 LAN chip for the 386EX processor using Tornado 1.0.1. I am trying to perform simple FTP tests to verify that my ethernet driver is working correctly. I am using the 'ftpXfer' function. I wanted to know if anyone knows if there are any bugs with the 'ftpXfer' function that vxWorks 5.3 uses. If anyone can provide me assistance with this, i would appreciate it very much. Thanks, Rohit Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: How to use socket API in pSOS Date: 14 Sep 2000 14:26:27 GMT From: Jun Chen Organization: Carleton University Message-ID: <39C0E013.418D549C@chat.carleton.ca> Reply-To: jchen@chat.carleton.ca what lib should I link with the program using socket? --------------------------- Newsgroups: comp.os.vxworks Subject: Telecom 'Start-up' Equity Opportunities / RTOS Date: Thu, 14 Sep 2000 14:57:49 GMT From: careers@trisearch.com Organization: FlashNet Communications, http://www.flash.net Message-ID: <39c0e2ab.4722246@news.flash.net> This message is to advise users of VxWorks or other RTOS that we are currently looking for Software engineers with 3+ years experience in Real time embedded software design and development for Telecom or Datacom applications. Our clients are fully venture capital funded Telecom start-up companies as well as established Telecom companies. Top salaries to $120K (DOE) plus incentive bonuses with generous stock options and full benefits and relocation packages are offered for this experience. These opportunities are truly exceptional 'career enhancing' positions. TRI has played a key part in staffing some of the most successful start-up and established telecom companies in the country. If you have an interest in surveying these opportunities: Please respond in strictest confidence to: TRI 972/424-6076(fax) careers@trisearch.com www.trisearch.com --------------------------- Newsgroups: comp.os.vxworks Subject: Re: How to delay current task for some microseconds? Date: Thu, 14 Sep 2000 14:46:55 GMT From: Ted Dennison Organization: Deja.com - Before you buy. Message-ID: <8pqocm$qnm$1@nnrp1.deja.com> References: <8pkh2c$bgi$1@schbbs.mot.com> <39C0C833.30D8C2C9@tsl.uu.se> In article <39C0C833.30D8C2C9@tsl.uu.se>, leif.thuresson@tsl.uu.se wrote: > I a few years back Geoffrey Espin posted this > a self calibrating delay routine on the vxWorks exploder. > It's the next best thing to a hardware timer. ...{most code snipped} > for (iy = 0; iy < u; iy++) > { > for (ix = 0; ix < delayLoop; ix++) > ; > } 3 questions here: Don't modern processors render this type of timing completely unfeasable? With pipelinging, instruction caches and speculative execution etc. combined with possible task switches, How do you know that the later runs of this code bear any relation time-wise to the calibration run? If I want to optimize my code, what's to stop the compiler from optimizing these loops into a couple of assignments? Thirdly: What happens if the calling task get switched out for a higher priority task during the "delay"? There's no accounting for this extra time that I see. When the calling task resumes, it will keep on looping until the end. Depending on what you wanted the delay for, I guess this might not always be a big problem. - -- T.E.D. http://www.telepath.com/~dennison/Ted/TED.html Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Use of spares in WIND_TCB Date: Thu, 14 Sep 2000 11:04:18 -0400 From: "Joe Ott" Organization: Verio Message-ID: Hello, I have the need to pass an arguement to a task after it has been spawned. On inspection of the WIND_TCB I found spare1. Using spare1 would allow me to pass a one time arguement to a task with little overhead. My question is, are there some rules about using spare1 (or the other spares) that I need to be aware of? So far things work fine, but I hate to use spare1 blindly incase VxWorks uses it or will use it in the future. Joe --------------------------- Newsgroups: comp.os.vxworks Subject: [Fwd: Bootup Times ... how long????] Date: Thu, 14 Sep 2000 17:23:29 +0100 From: Derek Cassidy Organization: TCD, Ireland (post does not reflect views of Trinity College) Message-ID: <39C0FB81.594411DD@cs.tcd.ie> Derek Cassidy wrote: > Can anyone tell me how long it takes some of the RTOS's to boot up, like > > Hard Hat Linux and QNX and vxworks?? > > does anyone know of any sites on the web comparing bootup times on > different RTOS?? > > Derek --------------------------- Newsgroups: comp.os.vxworks Subject: task states Date: Fri, 15 Sep 2000 01:20:25 +0900 From: "º¯¼ºÁÖ" Organization: Korea Telecom Message-ID: <8pqtlv$eqp$1@news2.kornet.net> 1. In VxWorks, task states are READY, PEND, DELAY, SUSPEND, etc... Is not there RUNNING state? 2. At calling taskActivate(), task become RUNNING state or READY state? 3. In preemptive priority scheduling, when dose scheduler occure? 3.1 When the higher priority task become READY state, is scheduler directly occured? or any other case? 4. In round robin scheduling, Is the time called scheduler end of timeslice? 4.1 In case POSIX, althought timeslice is left when calling wait(), O/S invokes scheduler by force. In case VxWorks, when scheduler is called by force like above? or Is not there any case to make tasks running by force? 5. In signal mechanism, Dose target task become directly RUNNING state? or only READY state when signal occures? --------------------------- Newsgroups: comp.os.vxworks Subject: Re: printf of "long long" Date: Thu, 14 Sep 2000 17:32:44 GMT From: John Fusco Organization: Deja.com - Before you buy. Message-ID: <8pr23r$7oi$1@nnrp1.deja.com> References: <39B99888.6A31EEC7@mindspring.com> <8pj6lg$t2f$1@nnrp1.deja.com> <39BF86E7.9327DE2@mindspring.com> <8pofop$65f$1@nnrp1.deja.com> <39BFEC8C.1EFB7288@mindspring.com> <39C08178.C23AF43D@nortelnetworks.com> Check out the Feb 2000 issue of Embedded Systems Programming (available online at http://www.embedded.com/2000/0002/). There is an article with instructions on how to build a GNU cross compiler. I was able to follow these instructions (with some tweaks) to get a working GCC for PowerPC on my Linux machine at home. This was just for fun. I never actually tried to run any of the code with VxWorks though there's no reason it shouldn't work. BTW, "%lld" was a lucky guess. I just figured if "%ld" means long decimal, why not "%lld" for long long decimal? In article <39C08178.C23AF43D@nortelnetworks.com>, Ofer Goren wrote: > > > Al Johnston wrote: > > > Seems like a strange thing for wrs to break... but then again, %ll may > > not > > have worked in gcc 2.7.x to begin with. I am planing to build my own > > gcc/ppc sometimes in the near future... but I am putting it off at the > > moment. And hoping some other nice person will post a faq/howto... > > > > Well, maybe this nice person will be you. If you DO try to build the newer > compiler, can you share the steps required, with us? > Thanks, > > > > > thanks for the help! > > > > -al > > -- > Embedded SW Team Leader > Email : ogoren@nortelnetworks.com > Phone number: +9723-6456023 > Fax number : +9723-6479579 > ESN : 826-6023 > http://www.nortelnetworks.com > ---------------------------------------------------- > If you lend someone $20, and never see that person again, > it was probably worth the investment. > > Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: hardware to monitor 32 vxworks "consols" Date: Thu, 14 Sep 2000 12:57:17 -0500 From: Andrew Johnson Organization: APS, Argonne National Laboratory Message-ID: <39C1117D.AD435059@aps.anl.gov> References: <39BFA9AC.C24754D1@mindspring.com> <39BFC4E8.55B26F9F@aps.anl.gov> <39BFCA4A.957BDD57@mindspring.com> Hi Al, Al Johnston wrote: > > Your "blast tas" code is one reason why I wanted to set this up. > I had a failure after three days... no msg was written to the > screen (a xterm with a rsh to the target)... I think it was > a "network timeout" issue rather than a real failure, but I > am not sure. There've been some other postings here recently about network problems on PPC that might explain your fault - from what you said above it sounds like you might have experienced one of these rather than a TAS failure - my test code should recognise any faults in the TAS actions and report them. - - Andrew - -- Every great idea appears crazy to start with. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: How to use socket API in pSOS Date: Thu, 14 Sep 2000 12:20:18 -0500 From: Steven Deans Organization: Fujitsu Companies of North America Message-ID: <39C108D2.DFA402E6@fnc.fujitsu.com> References: <39C0E013.418D549C@chat.carleton.ca> Jun Chen wrote: > what lib should I link with the program using socket? Assuming pSOS builds the same for all targets. Use pNA which is part of the OS library. You have to "turn on" pNA in sys_conf.h and have installed pNA. Steven --------------------------- Newsgroups: comp.os.vxworks Subject: Re: How to add hdlc to ppp Date: Thu, 14 Sep 2000 18:22:31 GMT From: xltan@my-deja.com Organization: Deja.com - Before you buy. Message-ID: <8pr50f$bgn$1@nnrp1.deja.com> Hi, I'm interested in finding out the following with the suggested approach for doing sync hdlc via VxWorks' ppplib. Will there be a performance bottleneck on ppp over hdlc driver when it internally is based on and limited by tty type interface processing (char-by-char based) provided by VxWorks' ppplib? Is there any interoperability issue of VxWorks ppplib with other vendor's PPP implementations? The VxWorks's pppLib only implemented RFC1548 which has been obsoleted by RFC1661 and it doesn't include RFC1662 (PPP in HDLC-like framing) either. Thanks for sharing. /Stephen Subject: Re: How to add hdlc to ppp Date: 07/12/2000 Author: Nick Gagnon << previous in search · next in search >> Hi kathy, We are doing similar work here. One solution that came up it that you make your hdlc driver looks like a serial driver. Then when you will select which device you will be using for ppp it will be transparent to it. There is other solution then the one I propose you, but they are not easy path and for now is as been giving us a hell of trouble -- Regards Nick Gagnon Mitel Corporation 350 Legget Drive P.O. Box 13089 Kanata, Ontario Canada K2K 2W7 Tel: (613) 592-2122 Ext 1797 Fax: (613) 592-4784 E-mail : nick_gagnon@mitel.com "kathy" wrote in message news:8kep8u$fmf@netnews.hinet.net... > > Hi, > Now I want to add synchrous hdlc to vxworks PPP interface. > I check pppinit system call .It just supports uart. > How to do synchrous hdlc to vxworks PPP interface. > Thanks Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: tgtsvr Date: 14 Sep 2000 21:10:22 +0200 From: srinivas_000@rediffmail.com ("srinivas kumar") Organization: Mailgate.ORG Server - http://www.Mailgate.ORG Message-ID: <20000914121058.26449.qmail@mailweb8.rediffmail.com> hello everyone, whever i try to configure and launch the target server on the host win NT, the following thing is seen in the log console. Connecting to target agent... Error: rpccore backend client Timed out failed. Connecting to target agent... Error: rpccore backend client Timed out failed. please do inform me about the cause of this error and the possible solutions. thanks in advance, --srinu _________________________________________________ Get Your Free Email At, http://www.rediffmail.com For fabulous shopping deals visit: http://www.rediff.co.in/shopping/index.html - -- Posted from IDENT:qmailr@[202.54.124.153] via Mailgate.ORG Server - http://www.Mailgate.ORG --------------------------- Newsgroups: comp.os.vxworks Subject: Lint and TornadoII Date: Thu, 14 Sep 2000 21:57:13 +0100 From: "Andy Kirby" Organization: BT Internet Message-ID: <8pre3g$2tm$1@neptunium.btinternet.com> Has anyone had any experience of setting up Lint with the Tornado NT environment? I have been trying to use Gimpel FlexeLint with T2(PPC) and there seems to be an awful lot of setting up of the Lint switches to even get it pass the VXWorks headers and the GNU extensions (like __attribute__ ((packed)) ). If anyone can point me to a place where this has been done, I'd appreciate any tips. Thanks Andy --------------------------- Newsgroups: comp.os.vxworks Subject: localtime function in vxWorks Date: Thu, 14 Sep 2000 17:17:51 -0400 From: "Somesh Agrawal" Organization: 3Com Message-ID: <8prf9c$49b$1@news.ne.3com.com> Is there a way to setup local time zone to something other than "GMT", so that localtime and gmtime give different output. Thanks, S.A --------------------------- Newsgroups: comp.os.vxworks Subject: MAC address Date: Thu, 14 Sep 2000 17:51:49 -0400 From: "Martin Mayer" Organization: Lawrence Berkeley National Laboratory Message-ID: <8prikj$865$1@overload.lbl.gov> This message is in MIME format. Since your mail reader does not understand this format, some or all of this message may not be legible. - ------_=_NextPart_001_01C01E95.FC4FEA30 Content-Type: text/plain; charset="iso-8859-1" Hi, Where does vxWorks get the MAC address from when displaying it in the ifShow routine? Is it possible to modify this MAC address dynamically? Thanks, Martin Mayer - ------_=_NextPart_001_01C01E95.FC4FEA30 Content-Type: text/html; charset="iso-8859-1" MAC address

Hi,

Where does vxWorks get the MAC address from when displaying it in the ifShow routine?

Is it possible to modify this MAC address dynamically?

Thanks,
Martin Mayer

- ------_=_NextPart_001_01C01E95.FC4FEA30-- --------------------------- Newsgroups: comp.os.vxworks Subject: Dynamically changing IP addresses Date: Thu, 14 Sep 2000 17:50:20 -0400 From: "Martin Mayer" Organization: Lawrence Berkeley National Laboratory Message-ID: <8prikk$866$1@overload.lbl.gov> This message is in MIME format. Since your mail reader does not understand this format, some or all of this message may not be legible. - ------_=_NextPart_001_01C01E95.C7270F00 Content-Type: text/plain; charset="iso-8859-1" Hi, I would like to write a command that will change the IP address that vxWorks uses. Any hints? Thanks, Martin Mayer - ------_=_NextPart_001_01C01E95.C7270F00 Content-Type: text/html; charset="iso-8859-1" Dynamically changing IP addresses

Hi,

I would like to write a command that will change the IP address that vxWorks uses.

Any hints?

Thanks,
Martin Mayer



- ------_=_NextPart_001_01C01E95.C7270F00-- --------------------------- Newsgroups: comp.os.vxworks Subject: Anyone running RIP V2 out there? Date: Thu, 14 Sep 2000 16:09:40 -0700 From: Carlos Rimola Organization: Lawrence Berkeley National Laboratory Message-ID: <4.3.1.2.20000914143424.00b47e08@10.1.0.201> Is anyone running VxWorks RIP in version 2 mode? I just found a bug in the latest Wind River code that I can't imagine anyone being able to successfully run RIP in V2 mode. The bug is in the ripSplitPacket function and will result in routes that expire after 120 seconds and a possible page fault due to a 0 ifp address being stored in the route entry. E-mail me if you want more details and a fix.. In a related note - if you were to ship a new router product with RIP support, what would you default its operation (supply) to? RIP V1 or V2? Which one is more popular? (I noticed that the RIP that ships with Linux defaults to RIP 1 mode). Carlos Rimola Sr. Software Engineer Praxon, Inc. --------------------------- Newsgroups: comp.os.vxworks Subject: Posting Date: Thu, 14 Sep 2000 16:54:27 -0700 From: "Jay Salerno-nonTRW" Organization: Lawrence Berkeley National Laboratory Message-ID: <8prplj$bcb$1@overload.lbl.gov> I have a SBS POWERDENALI board. Has anyone been able to get vxworks to boot from the 2nd 8Mbyte flash part? Or, does anyone know of a TFFS driver for the intel 28F160 flash memory part? Thanks in advance Jay Salerno email: jay.salerno-nonTrw@trrw.com phone: 310-764-6310 --------------------------- Newsgroups: comp.os.vxworks Subject: How can I config local IP address in pSOS? Date: 15 Sep 2000 00:45:05 GMT From: Jun Chen Organization: Carleton University Message-ID: <39C17112.72847306@chat.carleton.ca> Reply-To: jchen@chat.carleton.ca --------------------------- Newsgroups: comp.os.vxworks Subject: how to allocate a block of physically contiguous memory Date: Fri, 15 Sep 2000 02:54:24 +0100 From: "White Weng" Organization: Motorola Message-ID: <8prvgg$h8h$1@schbbs.mot.com> Hi, anyone help me? Can cacheDmaMalloc() allocate a block of physically contiguous memory, and this block includes 3 pages, in addition these 3 pages must be physically contiguous? If there are no 3 physically contiguous pages in current memory but there are some more pages which are not physically configuous, it will return NULL. I wonder if the function can accomplish this operation. If it can that is to say, it can not only handle cache-saft operation but also physically contiguous memory allocation. If it can not, who can? Thanks very much! --------------------------- Newsgroups: comp.os.vxworks Subject: synchronizing local and host sym tables Date: Fri, 15 Sep 2000 02:20:03 GMT From: "Dragan Ostojic" Organization: Sympatico Message-ID: Hi, I can't get synchronizing target and host symbol table to work. I boot off the flash on the target and then I start target server from the host. When I start target server it doesn't successfully synchronize the tables. If somebody has done this before I would be very grateful for help. Regards, Dragan. --------------------------- Newsgroups: comp.os.vxworks Subject: re: hardware to monitor 32 vxworks "consols" Date: Thu, 14 Sep 2000 22:23:21 -0400 (EDT) From: Fred Roeber Organization: Lawrence Berkeley National Laboratory Message-ID: <8ps472$f3e$1@overload.lbl.gov> On Thu, 14 Sep 2000, Al Johnston wrote: > We would like to put together some hardware that would provide > access to up to 32 cpu cards vxworks "console" ports. we are > running vxworks on mv2700/ppc cards. The proposed "monitoring > hardware" would connect to the "com1" port on up to 32 of these > cards. It should be able to display a virtual window for > each processor. We use Lantronix terminal servers for this. They are neat little boxes (3" x 7" x 5") that have an ethernet connection and some number of serial port connections (e.g 8, 12 or 16). You hook the console serial lines on target boards to different serial ports on the Lantronix box. To see what is happening on the serial port, you simply telnet to the Lantronix box (over the ethernet connection) passing it a port number. You then see all serial traffic just as if you were connected directly. Different people can telnet to each of the different ports. Works great for us. Far better than rlogin method since you get to see all the bootup/crash type messages. The only down side is the Lantronix boxes aren't cheap (I think they are like $1K). Sorry, I don't buy the things for our labs so don't have the contact info. Should be able to find it on the web though. Fred | Fred J Roeber, BBN Systems & Technologies | | 4 John Clarke Road Middletown, RI 02842-5202 | | froeber@bbn.com 401-848-3548 | | Division Scientist, High Performance Computing | --------------------------- Newsgroups: comp.os.vxworks Subject: TFFS problem Date: Fri, 15 Sep 2000 05:18:34 GMT From: yanghch@my-deja.com Organization: Deja.com - Before you buy. Message-ID: <8psbes$p3h$1@nnrp1.deja.com> I want to mount True FFS on my flash device which is MT28F400B1wg.And my BSP is MCP750,but its TFFS doesn't support the flash device.The code it provides is for 16bit flash device,but the MT28F400B1Wg is 32bit,so it doesn't work.If you can give me some advice,I will be very glad. Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: how to allocate a block of physically contiguous memory Date: Fri, 15 Sep 2000 05:47:53 GMT From: john_94501@my-deja.com Organization: Deja.com - Before you buy. Message-ID: <8psd65$qvj$1@nnrp1.deja.com> References: <8prvgg$h8h$1@schbbs.mot.com> Hi, Simple answer to this... all memory allocations are physically contiguous since the memory is 1:1 mapped (i.e. physical address = virtual address). If you ask for 12K (3 pages), then you'll get 12K contiguous in both virtual and physical space. HTH, John... In article <8prvgg$h8h$1@schbbs.mot.com>, "White Weng" wrote: > Hi, anyone help me? > > Can cacheDmaMalloc() allocate a block of physically contiguous memory, > and this block includes 3 pages, in addition these 3 pages must be > physically contiguous? If there are no 3 physically contiguous pages in > current memory but there are some more pages which are not physically > configuous, it will return NULL. I wonder if the function can accomplish > this operation. If it can that is to say, it can not only handle cache-saft > operation but also physically contiguous memory allocation. If it can not, > who can? > > Thanks very much! > > Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Use of spares in WIND_TCB Date: Fri, 15 Sep 2000 05:52:00 GMT From: john_94501@my-deja.com Organization: Deja.com - Before you buy. Message-ID: <8psddr$r96$1@nnrp1.deja.com> References: Hi, In general, you should be OK using any of the 4 spares. They are not used by the core OS. There are some optional products and third party products that make use of one or more of the spares however. So, use a macro somewhere that lets you change it easily in case you buy an optional product later that uses the one you've chosen! HTH, John... PS There are also two slots labelled 'reserved' and I don't know of anything that uses those... In article , "Joe Ott" wrote: > Hello, > > I have the need to pass an arguement to a task after it has been spawned. On > inspection of the WIND_TCB I found spare1. Using spare1 would allow me to > pass a one time arguement to a task with little overhead. > > My question is, are there some rules about using spare1 (or the other > spares) that I need to be aware of? So far things work fine, but I hate to > use spare1 blindly incase VxWorks uses it or will use it in the future. > > Joe > > Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: task states Date: Fri, 15 Sep 2000 05:59:05 GMT From: john_94501@my-deja.com Organization: Deja.com - Before you buy. Message-ID: <8psdr2$riq$1@nnrp1.deja.com> References: <8pqtlv$eqp$1@news2.kornet.net> Hi, In article <8pqtlv$eqp$1@news2.kornet.net>, "º¯¼ºÁÖ" wrote: > 1. In VxWorks, task states are READY, PEND, DELAY, SUSPEND, etc... > Is not there RUNNING state? There is no 'running' state; the highest priority 'ready' task runs. > 2. At calling taskActivate(), task become RUNNING state or READY state? Ready. > 3. In preemptive priority scheduling, when dose scheduler occure? > 3.1 When the higher priority task become READY state, is scheduler directly > occured? or any other case? Not sure I understand your question, but basically any event in the OS that could cause the highest priority ready task to change (e.g. semGive(), interrupt, taskResume() and so on) will call the scheduler. At that point, the highest priority ready task will run. > 4. In round robin scheduling, Is the time called scheduler end of > timeslice? Yes, because an event that could change the highest priority ready task has occurred (timeslice expiration). > 4.1 In case POSIX, althought timeslice is left when calling wait(), O/S > invokes scheduler by force. > In case VxWorks, when scheduler is called by force like above? > or Is not there any case to make tasks running by force? If the current running task blocks (i.e. changes from ready to pend, delay or even suspend) then the scheduler will of course be called since something needs to work out which task to run next... > 5. In signal mechanism, Dose target task become directly RUNNING state? or > only READY state when signal occures? The signal handler will run with the same priority as the task which received it. There is no running state. HTH, John... > Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: How can I config local IP address in pSOS? Date: Fri, 15 Sep 2000 08:20:26 +0200 From: "Martin Raabe" Organization: T-Online Message-ID: <8psfbt$kh7$15$1@news.t-online.com> References: <39C17112.72847306@chat.carleton.ca> Hello Jun, wrong newsgroup. Try comp.os.psos - -- Martin Raabe --------------------------- Newsgroups: comp.os.vxworks Subject: Re: How to use socket API in pSOS Date: Fri, 15 Sep 2000 08:23:55 +0200 From: "Martin Raabe" Organization: T-Online Message-ID: <8psfbu$kh7$15$2@news.t-online.com> References: <39C0E013.418D549C@chat.carleton.ca> Hello Jun, please use the newsgroup comp.os.psos The answer is, that the Network is named pNA+. There is a config file in each project called sys_conf.h where you can switch pNA+ on by #define SC_PNA YES If you have pNA+ installed (standard pRISM+ delivery contains it) then everything to be done is to recompile/link by psosmake clean;psosmake The system library contains pNA+ libraries by default. - -- Martin Raabe Jun Chen schrieb in im Newsbeitrag: 39C0E013.418D549C@chat.carleton.ca... > what lib should I link with the program using socket? --------------------------- Newsgroups: comp.os.vxworks Subject: Re: how to allocate a block of physically contiguous memory Date: Fri, 15 Sep 2000 07:49:01 +0100 From: "White Weng" Organization: Motorola Message-ID: <8psgp0$19$1@schbbs.mot.com> References: <8prvgg$h8h$1@schbbs.mot.com> <8psd65$qvj$1@nnrp1.deja.com> Thanks very much first, But if the physical memory are equivalent to virtual memory, why those macro are difined, such as CACHE_DMA_VIRT_TO_PHYS(address) and CACHE_DMA_PHYS_TO_VIRT(address). Regard! White Weng john_94501@my-deja.com wrote in message <8psd65$qvj$1@nnrp1.deja.com>... >Hi, > >Simple answer to this... all memory allocations are physically >contiguous since the memory is 1:1 mapped (i.e. physical address = >virtual address). If you ask for 12K (3 pages), then you'll get 12K >contiguous in both virtual and physical space. > >HTH, > >John... > >In article <8prvgg$h8h$1@schbbs.mot.com>, > "White Weng" wrote: >> Hi, anyone help me? >> >> Can cacheDmaMalloc() allocate a block of physically contiguous >memory, >> and this block includes 3 pages, in addition these 3 pages must be >> physically contiguous? If there are no 3 physically contiguous pages >in >> current memory but there are some more pages which are not physically >> configuous, it will return NULL. I wonder if the function can >accomplish >> this operation. If it can that is to say, it can not only handle >cache-saft >> operation but also physically contiguous memory allocation. If it can >not, >> who can? >> >> Thanks very much! >> >> > > >Sent via Deja.com http://www.deja.com/ >Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: VGA controller on MBX860 Date: Fri, 15 Sep 2000 09:53:04 GMT From: francescotorre1965@my-deja.com Organization: Deja.com - Before you buy. Message-ID: <8psrht$9j2$1@nnrp1.deja.com> Hello, we are using a MBX860 (Motorola) with vxworks os; it is equipped with a QSPAN cpu to pci bridge; we need to configure the card in order to plug a PC104+ VGA controller (with CT 65550). We have found the pciDevConfig function but we don't understand if it is enough to correctly configure the chip and how we have to set the parameters (mem space, io space, irq). We need also to integrate the WindRiver (Zinc) UGL module purchased with the os ..., what we have to do? Since we're very unskilled with PCI and PCI peripheral configuration we'd like to know where find documentation on these arguments (downloadable manuals, web sites, software examples and so on). Thanks! Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: how to allocate a block of physically contiguous memory Date: Fri, 15 Sep 2000 11:08:15 +0100 From: Phil Ashby Organization: Internet Designers Limited - a BT company Message-ID: <39C1F50F.BD8EA004@ashbysoft.cix.co.uk> References: <8prvgg$h8h$1@schbbs.mot.com> <8psd65$qvj$1@nnrp1.deja.com> <8psgp0$19$1@schbbs.mot.com> WRS recommend that physical and virtual memory are the same, but it isn't a requirement to run VxWorks. Also shared memory buses have memory mapping arrangements (eg: VME address spaces) which need to be dealt with. Phil. White Weng wrote: > > Thanks very much first, > > But if the physical memory are equivalent to virtual memory, why those macro > are difined, such as CACHE_DMA_VIRT_TO_PHYS(address) and > CACHE_DMA_PHYS_TO_VIRT(address). > > Regard! > > White Weng > > john_94501@my-deja.com wrote in message <8psd65$qvj$1@nnrp1.deja.com>... > >Hi, > > > >Simple answer to this... all memory allocations are physically > >contiguous since the memory is 1:1 mapped (i.e. physical address = > >virtual address). If you ask for 12K (3 pages), then you'll get 12K > >contiguous in both virtual and physical space. > > > >HTH, > > > >John... > > > >In article <8prvgg$h8h$1@schbbs.mot.com>, > > "White Weng" wrote: > >> Hi, anyone help me? > >> > >> Can cacheDmaMalloc() allocate a block of physically contiguous > >memory, > >> and this block includes 3 pages, in addition these 3 pages must be > >> physically contiguous? If there are no 3 physically contiguous pages > >in > >> current memory but there are some more pages which are not physically > >> configuous, it will return NULL. I wonder if the function can > >accomplish > >> this operation. If it can that is to say, it can not only handle > >cache-saft > >> operation but also physically contiguous memory allocation. If it can > >not, > >> who can? > >> > >> Thanks very much! > >> > >> > > > > > >Sent via Deja.com http://www.deja.com/ > >Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Lint and TornadoII Date: Fri, 15 Sep 2000 12:25:30 +0200 From: "Matthias Guelck" Organization: Heidelberg Digital Message-ID: <8pstlq$vfb$1@news.hdpp.de> References: <8pre3g$2tm$1@neptunium.btinternet.com> Hello Andy, we are using PC-lint 7.5 to check our VxWorks projects for two years now, using a lint parameter file with the following contents: - ----------------------------------------------------------------------------- - -A /* ANSI-C checking */ +fdi /* begin include search at including file */ +fcp /* use c++ not only for cpp and cxx ext. */ +fcu /* character is unsigned */ +fva /* variable argument size */ +fwc /* wchar_t is built in type */ +ppw(option) /* add preprocessor word