From owner-vxwexplo-process Mon Mar 1 01:34:22 1999 From: David Laight Date: Mon Mar 1 01:34:26 PST 1999 Subject: Re: How do I get gcc / Tornado 1.0.1 to compile unnamed Status: O > I believe that anonymous struct and union definitions are allowed in C++ but > not in standard C. My guess is that the Microsoft compiler allowed the C++ > behavior to creep into their C compiler as a feature. Your best bet is to > bite the bullet and name the union. Agreed - good old Microsoft destroy another standard... You only need change the header file, eg: typedef struct X { union { unsigned long x_status; void * x_Pointer; } x; unsigned long other; } vxworks_t; #ifdef __cplusplus #define x #define status x_status #define Pointer x_Pointer #else #define status x.x_status #define Pointer x.x_Pointer #endif (but don't use 'x'!) David ---------------------------------------------------------------- David Laight email: dsl@tadpole.co.uk Tadpole Technology plc phone: +44 1223 278 256 Cambridge, UK fax: +44 1223 278 201 From owner-vxwexplo-process Mon Mar 1 05:49:15 1999 From: "Thomas, Justin W." Date: Mon Mar 1 05:49:18 PST 1999 Subject: Tornado on Windows 98? Status: O VxWorkers, I am currently using Tornado 1.0.1 on my NT 4.0 workstation. The target server, license server, etc. are all run from the same workstation. I would like to move the development environment to a laptop to do some field development work. I am hesitant to use NT on the laptop because I have heard that it is difficult to get all of the drivers to support the built in hardware. Does anyone have experience (positive or negative) with using Tornado on Windows 98 (or 95) as opposed to NT? I need the transition to go smoothly because I have very little time to spend working out kinks in the development environment. Thanks in advance, Justin Justin Thomas SAIC - Project Engineer ph 301 866-6705 fax 301 863-0299 justin.w.thomas@cpmx.saic.com From owner-vxwexplo-process Mon Mar 1 10:34:02 1999 From: "Maureen O'Brien" Date: Mon Mar 1 10:34:05 PST 1999 Subject: Re: Tornado on Windows 98? Status: O Hi, I'm relatively new to Tornado. But I go back and forth between a NT hosts, a 95 host and a 98 host these last several months and have experienced no problems. Setup and use on all the machines is about the same. I remember there were some small differences in setup. Our license server was set up on an NT machine. Wind River seems to encourage NT because it is more stable--meaning less likely to crash. The stability issue is an ongoing Microsoft problem. Maureen O'Brien From owner-vxwexplo-process Mon Mar 1 11:39:39 1999 From: BruceS Date: Mon Mar 1 11:39:42 PST 1999 Subject: Re: Tornado on Windows 98? Status: O VxWorks, Tornado >>>>> "Justin" == "Thomas, Justin W." Justin> I am currently using Tornado 1.0.1 on my NT 4.0 Justin> workstation. The target server, license server, etc. are Justin> all run from the same workstation. I would like to move Justin> the development environment to a laptop to do some field Justin> development work. I am hesitant to use NT on the laptop Justin> because I have heard that it is difficult to get all of Justin> the drivers to support the built in hardware. What drivers? What hardware? I am doing all of this on a laptop under NTv4.0 sp3 and the only complaints I've had can be attributed to NT (vs. Solaris/SPARC) - no problems as a result of the laptop! Justin> Does anyone have experience (positive or negative) with Justin> using Tornado on Windows 98 (or 95) as opposed to NT? I Justin> need the transition to go smoothly because I have very Justin> little time to spend working out kinks in the development Justin> environment. If you move to NT on the laptop, there should be NO kinks (other than kinks you already have!) regards :-) BruceS -- Bruce S. Skinner | skinner@drea.dnd.ca DEFENCE RESEARCH ESTABLISHMENT ATLANTIC | norstead@ns.sympatico.ca Dartmouth, Nova Scotia, CANADA, B2Y 3Z7 | (902) 426-3100 (205) "New opinions are always suspected, and usually opposed without any other reason but because they are not already common." John Locke From owner-vxwexplo-process Mon Mar 1 12:13:42 1999 From: ksamavedam@hns.com (Krishna Samavedam) Date: Mon Mar 1 12:13:45 PST 1999 Subject: vxWorks & crosswind on HPPA.. Status: O HI, We are using the ported version on vxWorks for HPPA(hp-ux10.20). If we donot use the debug environment, it works fine. But when I start suing the Crosswind, it gives me problems, esp. while single stepping through the program, crosswind hangs and vxWorks starts hogging the CPU time. This happens often, each time at a different place in the program. I tried with gdbhppa also, this problem is less frequent with this. ANy help ? Thanx Krishna Samavedam From owner-vxwexplo-process Tue Mar 2 04:00:41 1999 From: daemon@csg.lbl.gov Date: Tue Mar 2 04:00:44 PST 1999 Subject: comp.os.vxworks newsdigest Status: O Comp.Os.Vxworks Daily Digest Tue Mar 2 04:00:37 PST 1999 Subject: SENS and fei82557 ethernet Subject: MVME177 board for sale Subject: Help: Bootrom using Intel EtherExpress Pro 10/100 Subject: Re: How can I get VxWorks Real Time OS for Intel x86 Subject: Re: VxWorks memory partitions Subject: JNI for VxWorks ? Subject: Re: Tornado on Windows 98? Subject: Re: compression source code ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: SENS and fei82557 ethernet Date: Mon, 1 Mar 1999 13:12:39 -0800 From: "Glenn Dicus" Organization: Nomadix LLC Message-ID: <7bevh7$5tu$1@la-mail4.digilink.net> We are currently developing with the pc486 BSP and SENS 1.1, which runs on a K-6 processor, 430HX chipset, and two 82558 Intel 10/100 ethernet MACs. The problem that we are having is that when we run a bridging test with SmartBits (packet blaster) at 100% utilization and @ 100Mb/s ports, configured for full and half duplex, we crash the interfaces. When there is contention on one of the interfaces the muxSend() begins to fail. Characteristically, when blasting packets @ 100% utilization in one direction (fei(0) to fie(1)) traffic in the opposite direction becomes disabled. The problem occurs even more frequently when we test the 10 Mb/s interface. This is with a simple mux stub running (ie. fei(0) does a muxSend to fei(1) and vice-versa) with the default WDB agent and tnetTask() etc. We did run into a similar problem with our mux interface implementation (bridge/router). Because when there is contention on one port the pool of CFDs elements would run out on the outbound port, causing an END_BLOCK_ERROR with muxSend() failing on subsequent calls. Not releasing the packet to the ring buffer we eventually ran out of RFDs on the receiving interface, which would, in effect shut it down. What we did was to call 'netMblkClChainFree()', which the driver leaves to the protocol to do on this and other errors, when an END_BLOCK_ERRORs occured. With this fix the interface does not shutdown when blasting in one direction. Has anybody had similar problems? Has anybody tested two ethernet ports and SENS 1.1 with a SmartBits box? If so, what were your results? Thanks glenn dicus --------------------------- Newsgroups: comp.os.vxworks Subject: MVME177 board for sale Date: Sat, 27 Feb 1999 12:54:14 -0700 From: Innovative Research Technologies Organization: Innovative Research Technologies Message-ID: <36D84D66.FAF92112@innovative-research.com> Reply-To: Marty@innovative-research.com We have the following for sale: MODEL: MVME177 50 MHz MC68060 32-Bit, 8KB of Cache, MMU, and FPU 16MB ECC DRAM, On-Board SCSI, Ethernet Interface 4MB of FLASH ROM E-mail if interested or for more info. Thanks for your interest. --------------------------- Newsgroups: comp.os.vxworks Subject: Help: Bootrom using Intel EtherExpress Pro 10/100 Date: Tue, 02 Mar 1999 11:23:20 +0900 From: Young Jin Nam Organization: Postech Message-ID: <36DB4B98.BD23D9F5@postech.ac.kr> Hello forks, I have been currently trying to make a bootrom for a x86-based PC using Intel Ether Express Pro 10/100 (PCI). The target PC is connected to its host via "10Mbps Hub." I made the bootrom with "#define INCLUDE_FEI" enabled. However, the bootrom couldn't properly download the "vxWorks" from the host. The followings are the messages we got by that moment: > Attaching network interface fei0... done. > Attaching network interface lo0... done. > Loading... Also, by running "sys82557Show()" routine, we have got the following results (even though I'm NOT sure that I've correctly INITIALIZED this network board): > 82557(0): Intel EtherExpress Pro 10/100 at 0xf400 00:A0:C9:31:74:71 > CSR mem base address = fd000000, Flash mem base address = fd100000 > PCI bus no. = 0, device no. = c, function no. = 0, IRQ = 11 > Board assembly 661949-004, Physical connectors present: RJ45 > Primary interface chip i82553-C PHY #1. > General self-test: passed. > Serial sub-system self-test: passed. > Internal registers self-test: passed. > ROM checksum self-test: passed (0x49caa8d6). By the way, do I have to set some parameters to "reduce" the speed of the network interface (because we don't use the 100Mbps Hub)? Any help from you will be very appreciated. Thanks in advance. p.s. I would be also very happy if you could let me know of how to initialize the Intel EtherExpress Pro 10/100 Network board in general on a x86-PC. - -- Young Jin Nam Office: +82-562-279-5668 Fax: +82-562-279-5699 Email : yjnam@postech.ac.kr Http://www.postech.ac.kr/~yjnam PCS : 018-523-4131 System Software Laboratory Dept. of Computer Science and Engineering POSTECH, Pohang, KOREA --------------------------- Newsgroups: comp.os.vxworks Subject: Re: How can I get VxWorks Real Time OS for Intel x86 Date: 27 Feb 1999 20:39:25 GMT From: "Eric S. Riseman" Organization: The Internet Access Company, Inc. Message-ID: <01be6290$bd919540$65f677ce@LittleFoot> References: <7b93pc$5pn$1@hfc.hk.super.net> <36d83b3f.0@news.pacifier.com> The Kernel you have to buy, the rest of WindRivers tools are GNU based and are freeware. Either find the freeware on the internet, or buy the freeware from Wind (guess that means they are no longer freeware.....). Steve Doiel wrote in article <36d83b3f.0@news.pacifier.com>... > Sorry, you cannot download VxWorks you need to shell out $10K or more to > Wind River Systems, and then the send you a CDROM and some manuals. I'm not > sure about the price but it is definitely not free. > > SteveD > > XYZ wrote in message <7b93pc$5pn$1@hfc.hk.super.net>... > >Hi, > > > >Is there anybody knows where can I download the VxWorks for Intel X86 CPU > ? > > > >Thanks ! > > > >Sunny > >mailto:sunny.ng@tek.com > > > > > > > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: VxWorks memory partitions Date: 2 Mar 1999 05:51:56 GMT From: Hwa-Jin Bae Organization: PSO Systems Inc (http://www.pso.com) Message-ID: <7bfu9s$q7b$1@news.idiom.com> References: <7b4i1k$59914@overload.lbl.gov> <7b9f02$474$1@news.idiom.com> Hwa-Jin Bae wrote: > When I get a chance I will dig up some old code that essentially is > a port of this Doug Lea's malloc onto VxWorks and make it available > in my VxHacks repository. It really does wonders in terms of > avoiding fragmentation in an embedded system, and is pretty fast. I did a quick port of both BSD malloc and Doug Lea's malloc to VxWorks and it is now available in VxHack archive site which is at http://www.pso.com/vxhacks.html THis port is intended to be a complete replacement for equivalent VxWorks original malloc/free, etc. I hope you find it useful. - -- Hwa-Jin Bae PSO Systems, Inc. http://www.pso.com mailto:hjb@pso.com --------------------------- Newsgroups: comp.os.vxworks Subject: JNI for VxWorks ? Date: Tue, 02 Mar 1999 15:29:38 +0900 From: Won-Soon Kim Organization: Daewoo Electronics Co., LTD. Message-ID: <36DB8551.FD427052@phoenix.dwe.co.kr> Reply-To: soon@phoenix.dwe.co.kr Hi all !! Can I use JNI (Java Native Interface) with VxWorks ? If so, How can I do that ? I installed JDK1.1.6 & Tornado 1.0.1 for x86.. I need more ? (Let me know ).. Here is my test makefile for Tornado temp project (?) - ------(Makefile start)------ # Makefile - makefile skeleton # Copyright (C) 1996 Wind River Systems, Inc. # This file was generated via the Tornado Project Facility. CPU = I80486 TOOL = gnu ADDED_CFLAGS = -c -ID:\tornado\target\h -ID:\JDK116\include - -ID:\JDK116\include\win32 -DCPU=I80486 -Wall -pedantic include $(WIND_BASE)/target/h/make/defs.bsp include $(WIND_BASE)/target/h/make/make.$(CPU)$(TOOL) include $(WIND_BASE)/target/h/make/defs.$(WIND_HOST_TYPE) include $(WIND_BASE)/target/h/make/rules.bsp - ------(Makefile End)------ Thanks in advance.... Won-Soon Kim --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Tornado on Windows 98? Date: Tue, 02 Mar 1999 08:14:29 +0100 From: Ole Asbjorn Fadum Organization: Telenor Online Public Access Message-ID: <36DB8FD4.B25F6048@scanmar.no> References: <199903011833.KAA07284@marlin.nosc.mil> Hi, I've tried both NT and 98. Don't bother to go with 98. For the NT installation I had no serious problems, but for the 98... If you buy the WFC (Wind Foundation Classes) it would not compile during 98. At least I could not get it to. I had to power off my PC 5 times, before I give up. (Yes I have a good PC, IBM Aptiva, with 350MHz Pentium II, and 65 Mb RAM). Maureen O'Brien wrote: > Hi, > I'm relatively new to Tornado. But I go back and forth between a NT > hosts, a 95 host and a 98 host these last several months and have > experienced no problems. Setup and use on all the machines is about the > same. I remember there were some small differences in setup. Our license > server was set up on an NT machine. Wind River seems to encourage NT > because it is more stable--meaning less likely to crash. The stability > issue is an ongoing Microsoft problem. > > Maureen O'Brien --------------------------- Newsgroups: comp.arch.embedded,comp.os.vxworks,comp.realtime Subject: Re: compression source code Date: Tue, 02 Mar 1999 10:38:54 +0200 From: Zvika Perry Organization: NetVision Israel Message-ID: <36DBA39E.598491AB@lucent.com> References: <79s8hi$3r2$1@news.netway.at> <36D975BD.B69BADD1@lucent.com> Lord Apollyon wrote: > In article <36D975BD.B69BADD1@lucent.com>, Zvika Perry > wrote: > > > As we had to compress my application due to FLASH memory constraints and > > decompress it on the target, we have a gzip code modified to work over memory > > rather than a file system. > > It seems bzip does it better - about 10% in my application and I'm currently > > working on it. > > If this can be a help, I can send it (the gzip) by e-mail. > > Not sure bzip (aka Wheeler Block Sorting compressor) would be all that > practical for your embedded work... seeing as it has monstrous memory > requirements for decompression. Yes, you can contain it by using a small > bucket size on the compression side, but.... > > Have you explored Markov coders, a la PPMZ*, etc? They offer a better > squeeze than ancient LZ77/78-based "technology" without the extreme > resource requirements of W-B Sorter techniques. > > =R= Any ideas where can I find a code that implements Markov coders, a la PPMZ ? Modification of bzip is really tedious (but I desperately need it) --------------------------- End of New-News digest ********************** From owner-vxwexplo-process Tue Mar 2 07:00:45 1999 From: "Josipovici, Diane (MED)" Date: Tue Mar 2 07:00:48 PST 1999 Subject: code size inflation between PPC860 and 68360 Status: O Hi vxworkers ! I have to transfere a package (application + BSP, with vxworks 5.3) from a 68360 to a PPC860 processor. If anyone has done this, what can we expect for the code oversize with the PPC (using gcc compiler in both environment) ? Thanks for any help, > Diane JOSIPOVICI > g GE Medical Systems > ____________________________________________ > mailto:Diane.Josipovici@med.ge.com > Tel: 33 (0)1 30 70 98 29 > Fax: 33 (0)1 30 70 46 70 > > From owner-vxwexplo-process Tue Mar 2 08:40:24 1999 From: "Rasty Slutsker (STX)" Date: Tue Mar 2 08:40:28 PST 1999 Subject: BUG++ in vxWorks or I miss something? Status: O 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_01BE64CB.2F8F9912 Content-Type: text/plain Hi VxWorkers, I suspect that Priority inheritance protocol is not implemented correctly. This happens if you wrap some function OS of with mutex having SEM_INVERSION_SAFE flag ( I proved this with malloc/free/open/close). Suddenly the tasks designed to be low priority processes are boosted to some absurd priority. The following example demonstrates this phenomenon. The example consists of two tasks one on High priority and second on low, both are calling to malloc/free, the low priority task stops itself if its priority is higher than expected. Note that mutual exclusion semaphore is used by single task. I assume that malloc has some internal mechanism similar to mutual exclusion with priority inheritance, but should the low-priority task restore his original priority by leaving malloc? Thanks, Rasty /* function Main() is entry point */ #include #include #include #include #include #include int tHigh,tLow; #define PRI_HIGH (150) #define PRI_LOW (160) /* * High priority task */ void task_High(void) { void *pv; while(1) { pv=malloc(100); free(pv); taskDelay(1); } } /* * Low priority task */ void task_Low(void) { void *pv; int pri; /* * PAY Attention: mutex is used only by single task */ SEM_ID mutex=semMCreate(SEM_Q_PRIORITY|SEM_INVERSION_SAFE); while(1) { semTake(mutex,WAIT_FOREVER); pv=malloc(100); /* Check the priority of itself */ taskPriorityGet(0,&pri); if (pri!=PRI_LOW) {printf("My priority is %d should be %d, why?\n",pri,PRI_LOW);taskSuspend(tLow);} free(pv); semGive(mutex); } } int Main(void) { if ((tHigh = taskSpawn("tHigh",PRI_HIGH,VX_FP_TASK,0x1000,(FUNCPTR)task_High,0,0,0,0,0,0, 0,0,0,0))==ERROR) return tHigh; if ((tLow = taskSpawn("tLow",PRI_LOW,VX_FP_TASK,0x1000,(FUNCPTR)task_Low,0,0,0,0,0,0,0,0 ,0,0))==ERROR) return tLow; return OK; } ------_=_NextPart_001_01BE64CB.2F8F9912 Content-Type: text/html Content-Transfer-Encoding: quoted-printable BUG++ in vxWorks or I miss something?

    Hi VxWorkers,

    I suspect = that Priority inheritance protocol is not = implemented correctly. This happens if you wrap some function OS of with mutex = having SEM_INVERSION_SAFE = flag ( I proved this with malloc/free/open/close). Suddenly the = tasks designed to be low priority processes are boosted to some = absurd priority. = The following example demonstrates this phenomenon. =

    The example = consists of two tasks one on High priority and second on low, both = are calling to malloc/free, the low priority = task stops itself if its = priority is higher than expected.

    Note that mutual = exclusion semaphore is used by single task. 

    I assume that = malloc has some internal mechanism similar to mutual = exclusion with priority = inheritance, but should the = low-priority task restore = his original priority by leaving = malloc?

    Thanks,

    Rasty

    /* function = Main() is entry point */

    #include = <vxWorks.h>

    #include = <semLib.h>

    #include = <taskLib.h>

    #include = <memLib.h>

    #include = <ioLib.h>

    #include = <stdIo.h>

    int = tHigh,tLow;

    #define PRI_HIGH = (150)

    #define PRI_LOW = (160)

    /*

     * High priority = task

     */

    void = task_High(void)

    {

            void *pv;

            while(1)

            {

            =       pv=3Dmalloc(100);

            =       free(pv);

            =       taskDelay(1);

            }

    }

    /*

     * Low priority = task

     */

    void = task_Low(void)

    {

            void *pv;

            int pri;

    /*

     * PAY Attention: = mutex is used only by single task

     */

            SEM_ID = mutex=3DsemMCreate(SEM_Q_PRIORITY|SEM_INVERSION_SAFE);

            while(1)

            {

            =       semTake(mutex,WAIT_FOREVER);

            =           pv=3Dmalloc(100);

    /* Check the priority of = itself */

            =           taskPriorityGet(0,&pri);

            =           if (pri!=3DPRI_LOW)

            =             {printf("My priority is %d should be %d, = why?\n",pri,PRI_LOW);taskSuspend(tLow);}

            =           free(pv);

            =       semGive(mutex);

            }

    }


    int Main(void)

    {

            if ((tHigh =3D = taskSpawn("tHigh",PRI_HIGH,VX_FP_TASK,0x1000,(FUNCPTR)task_Hig= h,0,0,0,0,0,0,0,0,0,0))=3D=3DERROR)

            =       return = tHigh;

            if ((tLow =3D = taskSpawn("tLow",PRI_LOW,VX_FP_TASK,0x1000,(FUNCPTR)task_Low,0= ,0,0,0,0,0,0,0,0,0))=3D=3DERROR)

            =       return = tLow;

     return = OK;

    }



------_=_NextPart_001_01BE64CB.2F8F9912-- From owner-vxwexplo-process Tue Mar 2 12:03:09 1999 From: Fred Roeber Date: Tue Mar 2 12:03:13 PST 1999 Subject: Re: BUG++ in vxWorks or I miss something? Status: O Rasty Slutsker (STX)" wrote: > I suspect that Priority inheritance protocol is not implemented > correctly. This happens if you wrap some function OS of with mutex having > SEM_INVERSION_SAFE flag ( I proved this with malloc/free/open/close). > Suddenly the tasks designed to be low priority processes are boosted to some > absurd priority. The following example demonstrates this phenomenon. > The example consists of two tasks one on High priority and second on > low, both are calling to malloc/free, the low priority task stops itself if > its priority is higher than expected. > Note that mutual exclusion semaphore is used by single task. > > I assume that malloc has some internal mechanism similar to mutual > exclusion with priority inheritance, but should the low-priority task > restore his original priority by leaving malloc? The VxWorks memory allocator can use either mutex or binary semaphores. The default memory partition is protected by a mutex semaphore created with both deletion safety and priority inheritance. The behaviour you are seeing is correct from what I understand of VxWorks priority inheritance (I extended the concept to a mutliprocessor version of VxWorks so had to understand the details). VxWorks uses a somewhat "lazy" version of priority inheritance (or actually of priority restoration). While a task holds ANY priority inheritance mutex semaphore it is considered to be in a special state. If some higher priority task tries to get ANY of the priority semaphores the task holds, the priority of the task holding the semaphore is boosted to the level of the task requesting the semaphore. This is normal priority inheritance behaviour. The lazy part is that the original priority of a task holding priority inheritance semaphores isn't restored until it releases ALL such semaphores it is holding. Ideally, priority should potentially be adjusted each time a task releases any priority inheritance semaphore it is holding. I think this is what you expected. In order to make such intermediate adjustments work, the VxWorks software would have to get MUCH more complex. I remember talking to some of the HP and SUN Solaris OS designers who said that getting full priority inheritance working was one of the trickier parts of the OS. The VxWorks design is much simpler and therefor faster and smaller. I think this is a good tradeoff. BTW, the behaviour you are seeing is documented in the VxWorks Programmers Guide section on Priority Inversion. Fred -- | Fred J Roeber, BBN Systems & Technologies | | 4 John Clarke Road Middletown, RI 02842-5202 | | froeber@bbn.com 401-848-3548 | | TraceMaker product manager -> www.tracemaker.bbn.com | From owner-vxwexplo-process Tue Mar 2 14:49:16 1999 From: "Legoas, Marco A (FL51)" Date: Tue Mar 2 14:49:20 PST 1999 Subject: dosFsLib Y2K compliance Status: O VxWorks Does anyone know of a Y2K limitation on VxWorks 5.1(mvme167 BSP) dosFsLib ? When using: ->dosFsDateSet(2000,03,31) I get a file time stamp of APR-00-2000 I get the same behavior for dosFsDateTimeInstall() WRS Y2K white paper (page 8) , says that dosFsLib can handle dates through 12/31/2099?. Any ideas? Marco A. Legoas malegoas@space.honeywell.com From owner-vxwexplo-process Wed Mar 3 04:00:20 1999 From: daemon@csg.lbl.gov Date: Wed Mar 3 04:00:24 PST 1999 Subject: comp.os.vxworks newsdigest Status: O Comp.Os.Vxworks Daily Digest Wed Mar 3 04:00:16 PST 1999 Subject: Re: RPC server ? Subject: C & C++ together w/VxWorks Subject: using SENS in vxSim Subject: Re: How do I get gcc / Tornado 1.0.1 to compile unnamed unions?. Subject: Re: compression source code Subject: Re: Seeking help understanding Makefile error message when trying to Subject: Re: BUG++ in vxWorks or I miss something? Subject: Re: View Symbol Table? Subject: How to attach a non-boot driver to the vxworks network stack?. Subject: Re: VxWorks memory partitions Subject: Re: looking for IP stack Subject: Re: Dec 21143 Media Select Subject: Re: VxWorks->Linux: Last call for votes Subject: Re: VxWorks->Linux: Last call for votes Subject: Slow network performance with 5.3.1 Subject: Re: compression source code Subject: Principal Software Engineer Subject: Windview PPC Question ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: RPC server ? Date: Tue, 02 Mar 1999 15:45:16 +0100 From: Thomas Kowatsch Organization: Switching Test Solutions AG Message-ID: <36DBF97B.1A049C92@stest.ch> References: <36D5255A.3146@spacetec.no> Asbjoern Pettersen wrote: > I'm testing out rpcgen for VxWorks for PPC on a Solaris computer. > The code is generated and working on other OS'es (IRIX). > The client compile easily but the server program fails. > > I'm missing the , ,,,,, files! > The header files isn't present on /usr/wind/ directory. > > What's wrong ? > > Asbjoern Pettersen Asbjoern, I think your problem is using the rpcgen on Solaris 2.5.1 as it generates code for using network databases and rpcbind service stuff for default (also called TIRPC). VxWorks RPC is using the portmapper interface stuff (ONC RPC) which is not genrated by Solaris 2.5.1 rpcgen by default. Use the -b flag on rpcgen to generate code for portmapper interface. You should also be aware, that server side stubs generate a main() function, which is of no use under VxWorks. Also rpcTaskInit() should be called, ... If you are interested I have a port of rpcgen which generates VxWorks usable code, include enhancement of "-H name" option to specify an alternate name for main(). Tommy - -- - ------------------------------------------------------------------------ Thomas Kowatsch | - Senior Software Engineer - Switching Test Solutions AG | mailto:thomas.kowatsch@stest.ch a Wandel & Goltermann Company | web :http://www.stest.com Friesenbergstr. 75 | Phone : +41 1 454-6731 CH-8055 Zuerich | FAX : +41 1 454-6612 --------------------------- Newsgroups: comp.os.vxworks Subject: C & C++ together w/VxWorks Date: Fri, 26 Feb 1999 11:42:44 -0500 From: Bruce Clements Organization: Cabletron Systems, Inc. Message-ID: <36D6CF04.DAAC8F61@ctron.com> Reply-To: clements@ctron.com Yet another "new to WRS / gnu" question... This is probably easier said than done.. but I have a bunch of legacy "C" code and a bunch of legacy "C++" code. I need to incorporate some C++ calls into the C code (and maybe vice versa). Any ideas on how to make this as painless as possible? ? Write a bunch of mangled name interface files? Uggghh. ? Just compile the "C" code with the "C++" compiler.. How do I do this without renaming all the "C" files? there seems to be a -language option but the WRS provided makefiles manage to override whatever I try to do. ? Something else Thanks... Bruce Clements --------------------------- Newsgroups: comp.os.vxworks Subject: using SENS in vxSim Date: Tue, 02 Mar 1999 16:57:40 +0100 From: Jeroen Damen Organization: Siemens Inc. Message-ID: <36DC0A74.16C8E3FE@vnet.atea.be> Hi, When I wanted to built the BSP with SENS for vxSim, configNet.h couldn't be found. Does this file exist for vxSim and if so, where can I find it. Thanks, Jeroen --------------------------- Newsgroups: comp.os.vxworks Subject: Re: How do I get gcc / Tornado 1.0.1 to compile unnamed unions?. Date: Fri, 26 Feb 1999 17:13:51 GMT From: jyl97@my-dejanews.com Organization: Deja News - The Leader in Internet Discussion Message-ID: <7b6ko6$g13$1@nnrp1.dejanews.com> References: <7b4vbg$dfk@news.or.intel.com> Hi, You missed a ";" after closing brace of union. Try this typedef struct X { union { unsigned long status; void * Pointer; }; unsigned long other; } X_t; You will still get the warning because un-named union is not recommanded. - -- Lei In article <7b4vbg$dfk@news.or.intel.com>, "rvsaripa" wrote: > Even though this might be for a comp.lang.c newsgroup, since I am getting > this problem with Tornado, I am posting it here. > Sorry for any inconvenience. > Hi, > I have a header file that describes the following data structure as :- > > file ---> test.c > typedef struct X { > union { > unsigned long status; > void * Pointer; > } > unsigned long other; > } X_t; > > Test() > { > X_t k; > > k.status = 5; > k.Pointer = 10; > > } > > When I compile the above snippet of code in a .c file, the compiler refuses > to compile the last 2 compilable statements. > I get the following errors. > > test.c:7: warning: unnamed struct/union that defines no instances > test.c: In function `Test': > test.c:14: structure has no member named `Status' > test.c:15: structure has no member named `Pointer' > > If I just recompile the file as a .cpp file, it compiles just fine. > Microsoft's Visual C++ compiler compiles in both > .c and .cpp mode without problems. > > Question :- Is there a flag that I need to turn on for gcc to compile the > above when used in c programs?. > I cannot name the union because I have some general C code that I would like > to recompile for vxworks without changing the code. > > I am using Tornado 1.0.1 with a 486 BSP. > > Can someone suggest a fix / solution for this problem?. > > Thanks, > rk > > - -----------== Posted via Deja News, The Discussion Network ==---------- http://www.dejanews.com/ Search, Read, Discuss, or Start Your Own --------------------------- Newsgroups: comp.arch.embedded,comp.os.vxworks,comp.realtime Subject: Re: compression source code Date: Wed, 03 Mar 1999 05:35:46 +1000 From: NOsatanusSPAM@KILLSPAM.ender.SPAMDIE.com (Lord Apollyon) Organization: Acheron, Ninth Plane of HELL Message-ID: References: <79s8hi$3r2$1@news.netway.at> <36D975BD.B69BADD1@lucent.com> <36DBA39E.598491AB@lucent.com> In article <36DBA39E.598491AB@lucent.com>, Zvika Perry wrote: > Any ideas where can I find a code that implements Markov coders, a la PPMZ ? > Modification of bzip is really tedious (but I desperately need it) Give Charlie Bloom's site a look-see. http://lonester.texas.net/~cbloom/src/ppmz.html - -- Save money and retain control of your embedded cross-assembler tools! Use Alfred Arnold's excellent universal cross-assembler, ASX. Available from: --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Seeking help understanding Makefile error message when trying to make bootrom_uncmp Date: Tue, 02 Mar 1999 14:08:03 -0600 From: Brian Strickland Organization: PEI Electronics Message-ID: <36DC4522.D061F521@peimail.com> References: <7ausoq$sgi$1@nntp6.u.washington.edu> Jim, I have been told that I should comment out the MACH EXTRA line when compiling the bootrom_uncmp. I cant give youa reason why, I just know it worked for me. Brian Strickland Jim Luby wrote: > Hi, > I'm attempting to create a bootrom_uncmp file that includes an atm network interface. I've definded > what I believe to be the necessary define statements, written a network interface attach routine, and > added the object files to the Makefile MACH_EXTRAS list. If I make the vxWorks image everything runs > till completion. However when I try to make the bootrom_uncmp image I get undefined references > related to functions contained in the Wind River supplied library "libPPC604gnuvx.a. I get the > following output from "make bootrom_uncmp": > > ccppc -BC:\Tornado/host/x86-win32/lib/gcc-lib/ -M -mstrict-align -ansi -nostdinc -O2 -fvol > atile -fno-builtin -fno-for-scope -Wall -I/h -I. -IC:\Tornado\target\config\all -IC:\Tor > nado\target/h -IC:\Tornado\target/src/config -IC:\Tornado\target/src/drv -DCPU=PPC604 -DM > V2300 -DTARGET_DIR="\"mv2304.atmboot\"" ctdt.c elattach.c myAppInit.c pciIomapLib.c pci > IomapShow.c ravenAuxClk.c ravenMpic.c sl82565IntrCtl.c sysCache.c sysLib.c sysScsi.c sysSe > rial.c universe.c C:\Tornado\target\config\all/bootConfig.c C:\Tornado\target\config\all/b > ootInit.c C:\Tornado\target\config\all/dataSegPad.c C:\Tornado\target\config\all/usrConfig > .c C:\Tornado\target\config\all/version.c > depend.mv2304.atmboot > C:\Tornado\host\x86-win32\bin\cppPPC -M -E -I/h -I. -IC:\Tornado\target\config\all -IC: > \Tornado\target/h -IC:\Tornado\target/src/config -IC:\Tornado\target/src/drv -DCPU=PPC604 > -DMV2300 -DTARGET_DIR="\"mv2304.atmboot\"" romInit.s >> depend.mv2304.atmboot > C:\Tornado\host\x86-win32\bin\cppPPC -M -E -I/h -I. -IC:\Tornado\target\config\all -IC: > \Tornado\target/h -IC:\Tornado\target/src/config -IC:\Tornado\target/src/drv -DCPU=PPC604 > -DMV2300 -DTARGET_DIR="\"mv2304.atmboot\"" sysALib.s >> depend.mv2304.atmboot > ccppc -BC:\Tornado/host/x86-win32/lib/gcc-lib/ -c -mstrict-align -ansi -nostdinc -O2 -fvol > atile -fno-builtin -fno-for-scope -Wall -I/h -I. -IC:\Tornado\target\config\all -IC:\Tor > nado\target/h -IC:\Tornado\target/src/config -IC:\Tornado\target/src/drv -DCPU=PPC604 -DM > V2300 -DTARGET_DIR="\"mv2304.atmboot\"" C:\Tornado\target\config\all\bootInit.c > copy C:\Tornado\target\config\all\bootInit.c bootInit_uncmp.c > 1 file(s) copied. > ccppc -BC:\Tornado/host/x86-win32/lib/gcc-lib/ -c -mstrict-align -ansi -nostdinc -O2 -fvol > atile -fno-builtin -fno-for-scope -Wall -I/h -I. -IC:\Tornado\target\config\all -IC:\Tor > nado\target/h -IC:\Tornado\target/src/config -IC:\Tornado\target/src/drv -DCPU=PPC604 -DM > V2300 -DTARGET_DIR="\"mv2304.atmboot\"" -DUNCOMPRESS bootInit_uncmp.c > ccppc -BC:\Tornado/host/x86-win32/lib/gcc-lib/ -mstrict-align -ansi -nostdinc -O2 -fvolati > le -fno-builtin -fno-for-scope -I/h -I. -IC:\Tornado\target\config\all -IC:\Tornado\targ > et/h -IC:\Tornado\target/src/config -IC:\Tornado\target/src/drv -DCPU=PPC604 -DMV2300 -DT > ARGET_DIR="\"mv2304.atmboot\"" -P -x assembler-with-cpp -c -o romInit.o romInit.s > C:\TEMP\cca00474: Assembler messages: > C:\TEMP\cca00474:4766: Warning: operand out of range (57343 not between -32768 and 32767) > ccppc -BC:\Tornado/host/x86-win32/lib/gcc-lib/ -c -mstrict-align -ansi -nostdinc -O2 -fvol > atile -fno-builtin -fno-for-scope -Wall -I/h -I. -IC:\Tornado\target\config\all -IC:\Tor > nado\target/h -IC:\Tornado\target/src/config -IC:\Tornado\target/src/drv -DCPU=PPC604 -DM > V2300 -DTARGET_DIR="\"mv2304.atmboot\"" C:\Tornado\target\config\all\bootConfig.c > ccppc -BC:\Tornado/host/x86-win32/lib/gcc-lib/ -mstrict-align -ansi -nostdinc -O2 -fvolati > le -fno-builtin -fno-for-scope -I/h -I. -IC:\Tornado\target\config\all -IC:\Tornado\targ > et/h -IC:\Tornado\target/src/config -IC:\Tornado\target/src/drv -DCPU=PPC604 -DMV2300 -DT > ARGET_DIR="\"mv2304.atmboot\"" -P -x assembler-with-cpp -c -o sysALib.o sysALib.s > ccppc -BC:\Tornado/host/x86-win32/lib/gcc-lib/ -mstrict-align -ansi -nostdinc -O2 -fvolati > le -fno-builtin -fno-for-scope -Wall -I/h -I. -IC:\Tornado\target\config\all -IC:\Tornad > o\target/h -IC:\Tornado\target/src/config -IC:\Tornado\target/src/drv -DCPU=PPC604 -DMV23 > 00 -DTARGET_DIR="\"mv2304.atmboot\"" -c sysLib.c > ccppc -BC:\Tornado/host/x86-win32/lib/gcc-lib/ -c -mstrict-align -ansi -nostdinc -O2 -fvol > atile -fno-builtin -fno-for-scope -Wall -I/h -I. -IC:\Tornado\target\config\all -IC:\Tor > nado\target/h -IC:\Tornado\target/src/config -IC:\Tornado\target/src/drv -DCPU=PPC604 -DM > V2300 -DTARGET_DIR="\"mv2304.atmboot\"" -o version.o C:\Tornado\target\config\all/versi > on.c > ldppc -X -N -e _romInit -Ttext 00300000 \ > -o bootrom_uncmp romInit.o bootInit_uncmp.o version.o bootConfig.o \ > sysALib.o sysLib.o dec21140.obj myAppInit.o fore_atm.mv2604.o elattach.o elconfig libs > nmp.o snmpd C:\Tornado\target/lib/libPPC604gnuvx.a > C:\Tornado\target/lib/libPPC604gnuvx.a(strerror.o): In function `strerrorIf': > strerror.o(.text+0xba): undefined reference to `statSymTbl' > strerror.o(.text+0xbe): undefined reference to `statSymTbl' > make.exe: *** [bootrom_uncmp] Error 0x1 > > Done. > > My Makefile is: > > # Makefile - makefile for bsp/config/mv2304 > # > # Copyright 1984-1997 Wind River Systems, Inc. > # Copyright 1996 Motorola, Inc. > # > # modification history > # -------------------- > # 01b,15aug97,mas changed bootrom_uncmp.hex to bootrom.hex for standard RELEASE > # 01a,02jun97,dat written, from mv2604/Makefile, ver 01f. > # > # DESCRIPTION > # This file contains rules for building VxWorks for the > # Motorola MVME2300 board with a PowerPC 604 processor. > # > # INCLUDES > # makeTarget > #*/ > > CPU = PPC604 > TOOL = gnu > > TGT_DIR=$(WIND_BASE)/target > include $(TGT_DIR)/h/make/defs.bsp > include $(TGT_DIR)/h/make/make.$(CPU)$(TOOL) > include $(TGT_DIR)/h/make/defs.$(WIND_HOST_TYPE) > > ## Only redefine make definitions below this point, or your definitions will > ## be overwritten by the makefile stubs above. > > TARGET_DIR = mv2304.atmboot > VENDOR = Motorola > BOARD = MVME2300 > > EXTRA_DEFINE = -DMV2300 -DTARGET_DIR="\"$(TARGET_DIR)\"" > > RELEASE = make.dec21140 vxWorks vxWorks.st bootrom.hex > > USR_ENTRY = usrInit > > # > # The constants ROM_TEXT_ADRS, ROM_SIZE, and RAM_HIGH_ADRS are defined > # in config.h, MakeSkel, Makefile, and Makefile.* > # All definitions for these constants must be identical. > # > > ROM_TEXT_ADRS = fff00100 # ROM entry address > ROM_SIZE = 00080000 # number of bytes of ROM space > > RAM_LOW_ADRS = 00100000 # RAM text/data address > RAM_HIGH_ADRS = 00300000 # RAM text/data address > > HEX_FLAGS = -a 100 > > MACH_EXTRA = dec21140.obj myAppInit.o fore_atm.mv2604.o elattach.o elconfig libsnmp.o snmpd > > ## Only redefine make definitions above this point, or the expansion of > ## makefile target dependencies may be incorrect. > > include $(TGT_DIR)/h/make/rules.bsp > include $(TGT_DIR)/h/make/rules.$(WIND_HOST_TYPE) > > make.dec21140: > (cd ./dec21140; $(MAKE) CPU=PPC604 DIR=$(TGT_DIR)/config/$(TARGET_DIR) objs) > > Thanks for any hints as to the cause of my problem! > > Jim Luby --------------------------- Newsgroups: comp.os.vxworks Subject: Re: BUG++ in vxWorks or I miss something? Date: Tue, 02 Mar 1999 16:20:57 -0500 From: "Lawrence T. Hoff" Organization: Brookhaven National Lab Message-ID: <36DC5639.91733ED8@bnl.gov> References: <7bh6ap$dn5@overload.lbl.gov> "Rasty Slutsker (STX)" wrote: > Hi VxWorkers, > I suspect that Priority inheritance protocol is not implemented > correctly. To paraphrase Bill Clinton: that depends on what "correctly" means. :-) The VxWorks implementation is not what I would call "intuitive". Once a task's priority is boosted due to priority inheritance, it stays boosted until *all* priority-inversion-protected mutex locks are released, not just the one that caused the priority to be changed. (this is from memory, some one jump in and correct me if I'm wrong). I suspect that malloc/free use priority-inversion-protected mutex locks. I suspect that at some point, your "low priority task" gets it's priority boosted by a high priority task trying to take the malloc/free lock. However, since the "low priority task" is also holding another priority-inversion-protected mutex lock, it's priority is not lowered even when it releases the malloc/free lock. Make sense? You could prove/disprove this my releasing the priority-inversion-protected mutex lock, just before each call to malloc or free, and re-take it just after each call. If the problem never occurs, then my recollection is probably correct. My conclusion was that VxWorks priority-inversion-protected mutex locks should be used only sparingly. HTH -- Larry - -- Building 911-C, Brookhaven National Laboratory PO Box 5000, Upton, NY, 11973 http://www.rhichome.bnl.gov/~hoff, mailto:hoff@bnl.gov Phone: (516) 344-2194 FAX: (516) 344-5443 --------------------------- Newsgroups: comp.os.vxworks Subject: Re: View Symbol Table? Date: Tue, 02 Mar 1999 15:08:08 -0800 From: Lukas Piesik Organization: Porter Engineering Ltd. Message-ID: <36DC6F58.DC623A02@portereng.com> References: The shell command lkup "xxx" shows you all the symbols with "xxx" in them (e.g. xxxMyVar). - --> lkup "" should show you all the symbols. (Since every symbol contains the null string.) - - Lukas "Joseph J. Lindula" wrote: > Hello, I'm a newbie with vxWorks. Can anyone tell me if it is possible to > view the symbol table, if so how? > > Thanks, > > Joe > > jjlindula@netscape.net --------------------------- Newsgroups: comp.os.vxworks Subject: How to attach a non-boot driver to the vxworks network stack?. Date: Tue, 2 Mar 1999 15:22:27 -0800 From: "rvsaripa" Organization: Intel Corporation Message-ID: <7bhu5k$n9o@news.or.intel.com> Hi, I have a vxworks link-layer driver that I need to attach to the vxworks network stack. For the purpose of testing this driver, I load it from the shell using "ld " commands. Also, this driver is a 4.3BSD style driver. If i put an attach routine name in the netIf structure in the bootconfig.c in the tornado directories and recompile the bootload binary( this is for a 486 BSP BTW ), I get an unresolved error when the boot image loads, which I can understand. So I added an entry in the netIf structure, added my name and left everything to 0 ( including attachRtn field ). Also, I added a routine called X() in bootconfig.c that would return the address of the netif structure ( since this is not exposed in the symbol table . I cannot find it using lkup on the shell ). My plan was to call X() from my driver and patch the attachRtn field with the address of my actual attach routine before invoking usrNetIfAttach() and usrNetIfConfig() which would go off and do the right thing. The problem is that X is now NOT visible in the symbol table. My driver loads with an unresolved reference to X. Looks like everything else in bootconfig.c appears in the symbol table. How do I get X to appear in the standard symbol table. I hate using this method to put my attach routine in the netIf structure. Is there a cleaner way of achieving the same?. I do not WANT to build my driver with the vxworks image at this point. Thanks for any help, Ramakrishna Saripalli All opinions are mine & mine only. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: VxWorks memory partitions Date: 27 Feb 1999 18:53:54 GMT From: Hwa-Jin Bae Organization: Sirius Cybernetics, Sirius City branch Message-ID: <7b9f02$474$1@news.idiom.com> References: <7b4i1k$59914@overload.lbl.gov> Fred Roeber wrote: > People interested in better behaviour in a memory allocator really > should check out the allocator reference that Hwa Jin Bae provided the > other day (http://gee.cs.oswego.edu/dl/html/malloc.html). This is an > allocator that has been widely used and carefully tuned for over 10 > years. It has been widely measured by researchers. When I get a chance I will dig up some old code that essentially is a port of this Doug Lea's malloc onto VxWorks and make it available in my VxHacks repository. It really does wonders in terms of avoiding fragmentation in an embedded system, and is pretty fast. Stay tuned. - -- Hwa-Jin Bae PSO Systems, Inc. http://www.pso.com mailto:hjb@pso.com --------------------------- Newsgroups: comp.os.vxworks Subject: Re: looking for IP stack Date: 27 Feb 1999 18:55:51 GMT From: Hwa-Jin Bae Organization: Sirius Cybernetics, Sirius City branch Message-ID: <7b9f3n$474$2@news.idiom.com> References: <36D2D6AF.53FAB1D@nortelnetworks.com> Watson, Bruce (EXCHANGE:AIRPT:3181) wrote: > Does anyone have any experience with third party IP stacks for VxWorks? > I'm looking for something that is BSD 4.4 compliant (mbufs interface) > which rules out Wind River's SENS stack. Any pointers, recommendations, > or warnings appreciated. PSO Systems Inc is currently conducting a fresh port of NetBSD (4.4 BSD plus much more enhancements) TCP/IP stack onto VxWorks. Our port will be available towards middle of March for generic network devices. - -- Hwa-Jin Bae PSO Systems, Inc. http://www.pso.com mailto:hjb@pso.com --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Dec 21143 Media Select Date: 27 Feb 1999 19:02:22 GMT From: Hwa-Jin Bae Organization: Sirius Cybernetics, Sirius City branch Message-ID: <7b9ffu$474$3@news.idiom.com> References: <7auuno$59u2@overload.lbl.gov> Robert L. Perry wrote: > I've written a callback xxxMediaSelect() routine for the DEC 21143 driver on > an x86 platform. If I hard code the media type (100baseT, 10baseT, duplex > type , etc.), all is well. When I try to autosense I'm able to properly > determine the type of media, but I'm having trouble reconfiguring the chip > to the desired settings. The original WRS driver has some scary code in the > form of the xxxxInitialReset() routine, which appears to address a problem > with the silicon. A good source for MII PHY handling is found in Linux tulip.c driver. - -- Hwa-Jin Bae PSO Systems, Inc. http://www.pso.com mailto:hjb@pso.com --------------------------- Newsgroups: comp.os.vxworks Subject: Re: VxWorks->Linux: Last call for votes Date: 27 Feb 1999 19:06:06 GMT From: Hwa-Jin Bae Organization: Sirius Cybernetics, Sirius City branch Message-ID: <7b9fmu$474$4@news.idiom.com> References: <7akk8m$9i5@overload.lbl.gov> <5m1zjidpit.fsf@jtc.redbacknetworks.com> J.T. Conklin wrote: > Before WRS supports another host OS, I'd rather they spent the effort > to clean up and contribute their changes to the GNU toolchain so that > anyone can have a functional toolchain regardless of host OS they use. I agree. Especially the WTX related stuff needs to be put into mainstream GNU toolchain (gdb espeicially). - -- Hwa-Jin Bae PSO Systems, Inc. http://www.pso.com mailto:hjb@pso.com --------------------------- Newsgroups: comp.os.vxworks Subject: Re: VxWorks->Linux: Last call for votes Date: 27 Feb 1999 19:08:56 GMT From: Hwa-Jin Bae Organization: Sirius Cybernetics, Sirius City branch Message-ID: <7b9fs8$474$5@news.idiom.com> References: <7akk8m$9i5@overload.lbl.gov> <5m1zjidpit.fsf@jtc.redbacknetworks.com> Stan Shebs wrote: > I've gotten a couple minor mods. But the main stuff, no. I talked to > Todd Whitesel (hi Todd) last fall at ESC West, and he *promised* me > more changes were on the way; but so far, nothing. I think that the recent Tornado CD's have full source code for the vxgdb as distributed by WRS. Unfortunately you have to have the license for VxWorks to own one of the CD's and have to manually select an option to install the source code. However, once source is extracted I assume it's a fair game to provide back to FSF/Cygnus? Any volunteers? I do not have latest Tornado CD. - -- Hwa-Jin Bae PSO Systems, Inc. http://www.pso.com mailto:hjb@pso.com --------------------------- Newsgroups: comp.os.vxworks Subject: Slow network performance with 5.3.1 Date: Wed, 03 Mar 1999 02:56:56 GMT From: William Ross Organization: DIGEX, Inc. - Beltsville, MD - http://www.digex.net Message-ID: <36DCA562.20AAB852@empireone.net> This is a multi-part message in MIME format. - --------------58C10FF4C784BA8FF796C3AA Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Hi All, I've got a VxWorks 5.3.1 box talking to a linux machine across 100Base-T, via a crossover cable. Both the NICs are Intel EtherExpress Pro 100+ cards (using fei). Both machines are PIIs (233 and 266). With this link, I would expect to see transfer rates on the order of 80-90 MBits/sec over a TCP socket. Indeed, colleagues here have seen rates that high with 5.3.0 talking to Solaris 2.5.1. I'm using netperf to do this benchmarking. Instead, I see rate somewhere around 20 MBits/sec! I've heard many rumors concerning problems with the fei driver, and I don't intend to use SENS (a major bottleneck I hear). I know the hardware is good, as I've tested it in other machines (by this I mean NT). Furthermore, I've swapped in the older EtherExpress Pro 100, with the same results. Changing the socket size does not affect performance. Being relatively new to VxWorks, I can only assume that it's a configuration issue, or did something change between 5.3.0 and 5.3.1. Any help would be greatly appreciated. Bill Ross GE Corporate R&D - --------------58C10FF4C784BA8FF796C3AA Content-Type: text/x-vcard; charset=us-ascii; name="wrr.vcf" Content-Transfer-Encoding: 7bit Content-Description: Card for William Ross Content-Disposition: attachment; filename="wrr.vcf" begin:vcard n:Ross;William x-mozilla-html:TRUE version:2.1 email;internet:wrr@empireone.net x-mozilla-cpt:;0 fn:William Ross end:vcard - --------------58C10FF4C784BA8FF796C3AA-- --------------------------- Newsgroups: comp.arch.embedded,comp.os.vxworks,comp.realtime Subject: Re: compression source code Date: Wed, 03 Mar 1999 18:02:16 +1000 From: NOsatanusSPAM@KILLSPAM.ender.SPAMDIE.com (Lord Apollyon) Organization: Acheron, Ninth Plane of HELL Message-ID: References: <79s8hi$3r2$1@news.netway.at> <36D975BD.B69BADD1@lucent.com> <36DBA39E.598491AB@lucent.com> <7bhicu$5bc$1@remarQ.com> In article <7bhicu$5bc$1@remarQ.com>, "Pat Rogers" wrote: > That's "lonestar", rather than "lonester". You're not from Texas, are > you? :-) PrayZe GeeZUZ, pass the ammo, hell NO! :) It was an honest "cross-machine paste" type. (for the imagination impaired, that means I typed it manually reading from one machine's monitor onto a different machine's keyboard). Tongue planted firmly into cheek, =R= - -- Save money and retain control of your embedded cross-assembler tools! Use Alfred Arnold's excellent universal cross-assembler, ASX. Available from: --------------------------- Newsgroups: comp.os.vxworks Subject: Principal Software Engineer Date: Fri, 26 Feb 1999 20:27:38 GMT From: "Robertson-Surrette" Organization: Robertson-Surrette Inc. Message-ID: <01be61df$d024da70$77d1ba89@istar> If the prospect of applying leading-edge commercial voice and data communications technology to highly reliable, ruggedized and multi-level, redundancy based military products sounds challenging, read on: On behalf of one of the Ottawa region's most innovative system solutions providers, we are currently conducting a search for a Principal Engineer, Software Development. This organization has established a worldwide reputation for designing and manufacturing state-of-the-art advanced communications products. As part of an advanced development team, this individual will play a key role in defining the "next generation" of the company's leading-edge system. To be ideally suited for this role, you have a strong communications and telephony background with extensive ATM and ISDN design experience. Your network management expertise will serve you well in designing multimedia delivery capabilities using state-of-the-art technologies such as JAVA, for use in very specialized military environments. Experience with ADA Programming and/or VxWorks will be considered an assets. You possess a bachelor's degree in either Computer Science or Engineering. Having worked in the technology field for at least 7 to 8 years, you are ready for a challenging role where a small team setting fosters mentoring and learning. Here, every team member understands the difference his or her contribution makes to the product. The position requires an individual who loves to be challenged. If you, or someone you know is interested in pursuing this unique opportunity further, please contact Jim Harmon at 'jharmon@robsur.com' or by fax at (613) 749-9599. We would also welcome your call at (613) 749-9909. You can find more information about this role on our web site at 'www.robsur.com'. --------------------------- Newsgroups: comp.os.vxworks Subject: Windview PPC Question Date: Tue, 02 Mar 1999 22:02:17 -0500 From: Tim Sohacki Organization: Nortel Networks Message-ID: <36DCA639.9A026D99@slowspam.nortel.ca> I'm using Windview 2.0 on a PPC 603 target. Because of the interesting PPC interrupt architecture, WV displays a single INT 32 line for interrupts. Occasionally, I see what appears to be task code billed to INT 32. Example: I have a small program that I load that runs every 10 seconds for about 100 ms per burst; each 1 ms piece has a wvEvent() surrounding it. Every so often, I'll catch a WV trace that has a chunk of the burst associated with INT32, instead of my task. Behavior then returns to normal. I'm wondering if this is pointing to an underlying problem in my system, if it is a WV glitch, or what. Thoughts? Thanks, Tim - -- Tim Sohacki, Nortel Networks, Inc. email: sohacki@nortelnetworks.com --------------------------- End of New-News digest ********************** From owner-vxwexplo-process Wed Mar 3 08:08:02 1999 From: dit@bach.jhuapl.edu (Daryl I. Tewell) Date: Wed Mar 3 08:08:05 PST 1999 Subject: Re: BUG++ in vxWorks or I miss something? Status: O VxWorks,vxworks,vxWorks > Suddenly the tasks designed to be low priority processes are boosted to some > absurd priority. The following example demonstrates this phenomenon. Could the absurd priority be the result of taskPriorityGet() returning ERROR (and leaving pri uninitialized)? My reference manual is unclear about whether you can pass 0 to taskPriorityGet() to refer to the calling task. > BTW, the behaviour you are seeing is documented in the VxWorks > Programmers Guide section on Priority Inversion. Fred, what version of the Programmers Guide do you have? I could not find this level of detail in the (American) 5.3 manual. > You could prove/disprove this my releasing the > priority-inversion-protected > mutex lock, just before each call to malloc or free, and re-take it just > after > each call. If the problem never occurs, then my recollection is probably > correct. I don't doubt that Fred and Larry are right, but please do this experiment and let us know what happens. However, I would still use something like error = taskPriorityGet(taskIdSelf(),&pri) to be absolutely sure of your results. Not to nitpick, but you might also want to check the return value of semTake(). -Daryl Daryl.Tewell@jhuapl.edu From owner-vxwexplo-process Wed Mar 3 09:59:20 1999 From: "John R. Moore III" Date: Wed Mar 3 09:59:24 PST 1999 Subject: Re: BUG++ in vxWorks or I miss something? Status: O the vxWorks Users Group Exploder wrote: > > Submitted-by owner-vxwexplo-process Wed Mar 3 08:08:02 1999 > Submitted-by: dit@bach.jhuapl.edu (Daryl I. Tewell) > > VxWorks,vxworks,vxWorks > > > Suddenly the tasks designed to be low priority processes are boosted > to some > > absurd priority. The following example demonstrates this phenomenon. > > Could the absurd priority be the result of taskPriorityGet() returning > ERROR (and leaving pri uninitialized)? My reference manual is unclear > about whether you can pass 0 to taskPriorityGet() to refer to the > calling task. > Page 32 in the 5.3 Programmers Guide section 2.3.4.2: "...uses a convention that a task ID of 0 (zero) always implies the calling task." > > BTW, the behaviour you are seeing is documented in the VxWorks > > Programmers Guide section on Priority Inversion. > > Fred, what version of the Programmers Guide do you have? I could > not find this level of detail in the (American) 5.3 manual. Page 59 in the 5.3 Programmers Guide section 2.4.3.3 > Not to nitpick, but you might also > want to check the return value of semTake(). > Always a good thing to do, particularly in the case that the semaphore has been deleted with your tasks knowledge. -- ------------------------------------------------------------ |John R. Moore | Voice -> 703 448 1683 x229 | |Principal Engineer | FAX -> 703 893 5494 | |EMAIL: | On the WEB! | |jrm@mclean.sparta.com | http://www.mclean.sparta.com | |___________________________|______________________________| | | | | /\\ SPARTA, Inc. | | | ///\\ 7926 Jones Branch Drive | | | /////\\ Suite 900 | | | ///// \\ McLean, VA 22102 | | | / \\\\\ \\ 703-448-0120 | | | \ \\\\\ // | | | \ ////// | | | \////// SPARTA | | | \//// Pride In Performance | | | \// -------------------- | | ------------------------------------------------------------ From owner-vxwexplo-process Wed Mar 3 10:19:45 1999 From: Fred Roeber Date: Wed Mar 3 10:19:49 PST 1999 Subject: Re: BUG++ in vxWorks or I miss something? Status: O Daryl I. Tewell wrote: > > BTW, the behaviour you are seeing is documented in the VxWorks > > Programmers Guide section on Priority Inversion. > > Fred, what version of the Programmers Guide do you have? I could > not find this level of detail in the (American) 5.3 manual. In the VxWorks 5.3.1 Programmer's Guide on pg 62 in the second paragraph under Priority Inversion it documents the behaviour of a task maintaining its boosted priority until it releases all inversion safe semaphores it is holding. Fred -- | Fred J Roeber, BBN Systems & Technologies | | 4 John Clarke Road Middletown, RI 02842-5202 | | froeber@bbn.com 401-848-3548 | | TraceMaker product manager -> www.tracemaker.bbn.com | From owner-vxwexplo-process Wed Mar 3 10:33:30 1999 From: Hamid Shoaee Date: Wed Mar 3 10:33:33 PST 1999 Subject: Job Opportunities Status: O We have several immediate openings for senior software engineers at the Stanford Linear Accelerator Center (SLAC), a high energy physics research facility operated by Stanford University. This is an opportunity to join a scientific software development group supporting the operation of a particle accelerator facility as well as conducting R&D for a next generation accelerator control system. You will perform system and software development for real-time computer systems and networks, utilizing your knowledge of distributed systems and networking. Knowledge of and experience with UNIX and/or NT operating systems, real-time executives (VxWorks, etc.) C/C++ and control systems or real-time systems is also essential. The positions require a BSCS/EE/CE and several years of experience in programming and systems integration in time critical or distributed control applications. Experience with distributed systems using TCP/IP or field buses, writing instrument and hardware-related ISRs, hardware interface routines and device drivers for VxWorks-based processor boards is desired. Applicants must be able to communicate technical information and participate effectively in small teams of engineers and physicists. These positions are in Northern California, in the San Francisco Bay Area. SLAC offers competitive compensation and excellent benefits. If you wish to be considered for these positions, please send your resume to hamid@SLAC.Stanford.edu. Hamid Shoaee Stanford Linear Accelerator Center P.O.Box 4349 Stanford, CA 94309 (650) 926-2954 From owner-vxwexplo-process Thu Mar 4 04:00:41 1999 From: daemon@csg.lbl.gov Date: Thu Mar 4 04:00:45 PST 1999 Subject: comp.os.vxworks newsdigest Status: RO Comp.Os.Vxworks Daily Digest Thu Mar 4 04:00:37 PST 1999 Subject: Re: dosFsLib Y2K compliance Subject: Re: VxWorks 5.3 Interrupt Enable Subject: Re: How to attach a non-boot driver to the vxworks network stack?. Subject: Re: How can I get VxWorks Real Time OS for Intel x86 Subject: Re: Which host platform, Solaris or NT? Subject: sharing configuration files, etc. Subject: Re: VxWorks memory partitions Subject: Using IRQ3 interrupts on a PPC860DE Subject: Re: How to open an ELF file Subject: Re: How to open an ELF file Subject: STREAMS & VxWorks Subject: do_protocol_with_type() question. Subject: Re: compression source code Subject: mutiple telnet sessions w/ vxWorks ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: dosFsLib Y2K compliance Date: Wed, 03 Mar 1999 14:10:22 -0200 From: Trevor Pope Organization: Adamastor Message-ID: <36DD5EEE.CAA@dip3.ee.uct.ac.za> References: <7bhrdo$dn8@overload.lbl.gov> Legoas, Marco A (FL51) wrote: > > VxWorks > > Does anyone know of a Y2K limitation on VxWorks 5.1(mvme167 BSP) dosFsLib ? > > When using: > > ->dosFsDateSet(2000,03,31) > > I get a file time stamp of APR-00-2000 > > I get the same behavior for dosFsDateTimeInstall() > > WRS Y2K white paper (page 8) > , says that dosFsLib can > handle dates through 12/31/2099?. > > Any ideas? > > Marco A. Legoas > malegoas@space.honeywell.com If I remember correctly, I saw something similar on 5.2, but it didn't matter - the date was stored correctly internally - it was interpreted incorrectly by ls(). We are also stuck in a time warp with 5.2 - it is not worth upgrading, because we have a new development direction, so I am stuck with patching the Y2k problems with no support on the BSP either. The DS1286 clock chip used by Artesyn (Heurikon) on the Nitro260 has a problem,that Artesyn acknowledge. Dallas says it is compliant and they haven't answered my email! --------------------------- Newsgroups: comp.os.vxworks Subject: Re: VxWorks 5.3 Interrupt Enable Date: Wed, 03 Mar 1999 12:10:36 GMT From: aerdogan@hc.aselsan.com.tr Organization: Deja News - The Leader in Internet Discussion Message-ID: <7bj8ro$s1l$1@nnrp1.dejanews.com> References: <36D9E627.17BDB4A4@cs.ucdavis.edu> In article <36D9E627.17BDB4A4@cs.ucdavis.edu>, MAXi wrote: > Hi All, > > I was just wondering how I can enable interrupt (say /INT5) in VxWorks > 5.3 . > I have used "intConnect" to tell where the interrupt service routine is > so all I need to do is to actually enable the interrupt. > > TIA. > > Regards, > > MAXi > -- > ======================================================================== > > Maxwell CHEONG > University of California (Davis) Address: > Computer Engineering, Junior Rm 302, Pierce J Building > Studnet ID#: 991-023926 Tercero Hall Circle > Phone: (530) 754 0352 University of California (Davis) > > email: cheong@cs.ucdavis.edu DAVIS, CA 95616 > ======================================================================== > > The following lines enables SCC1 interrupt. Do not forget to clear in-service register when leaving the int. service routine. MsrValue = vxMsrGet(); MsrValue &= 0xFFFF7FFF; vxMsrSet(MsrValue); *CIMR(PQUICC) |= CIMR_PARAM; /* unmask SCC1 interrupt */ *SCCM1(PQUICC) = 0x1c97; /* enable all SCC events */ *CICR(PQUICC) |= CICR_IRL_LEVEL_4; *SIMASK(PQUICC) |= SIMASK_LVM4; *SIMASK(PQUICC) |= SIMASK_IRM4; MsrValue |= 0x00008000; vxMsrSet(MsrValue); Alper Erdogan - -----------== Posted via Deja News, The Discussion Network ==---------- http://www.dejanews.com/ Search, Read, Discuss, or Start Your Own --------------------------- Newsgroups: comp.os.vxworks Subject: Re: How to attach a non-boot driver to the vxworks network stack?. Date: Wed, 03 Mar 1999 13:41:58 +0000 From: Phil Ashby Organization: BT Wireplay Message-ID: <36DD3C26.3C31@wireplay.bt.co.uk> References: <7bhu5k$n9o@news.or.intel.com> rvsaripa wrote: > I have a vxworks link-layer driver that I need to attach to the vxworks > network stack. For the purpose of testing this > driver, I load it from the shell using "ld " commands. Also, this driver > is a 4.3BSD style driver. [ugly attempts to connect a dynamically loaded driver to the IP stack via netIf table :)] If you read the code for usrNetIfAttach (in bootConfig.c), you will see that it does very little except call the attach routine of the driver and pass in the arguments from the netIf table. This is simple to do from a startup script if necessary. useNetIfConfig() is similarly trivial, calling ifMaskSet, ifAddrSet and hostAdd. Just do it yourself in a script. We replaced the table-driven usrNetXxx() routines in our kernel with no ill effect (sometimes knowing that a data structure is _not_ visible outside a particular source file is useful :). HTH, - -- !--------------------------- Phil "Phlash" Ashby ---------------------! ! BT Wireplay - Lead Programmer and general technical trivia supplier ! ! Phone: 01473-644348 WWW: http://www.wireplay.com/ ! ! Snail: B81/G41, BT Labs, Martlesham Heath, Ipswich IP5 3RE, England.! --------------------------- Newsgroups: comp.os.vxworks Subject: Re: How can I get VxWorks Real Time OS for Intel x86 Date: Wed, 03 Mar 1999 10:05:18 -0500 From: Douglas Fraser Organization: Lucent Technologies Message-ID: <36DD4FAE.1BCAD88C@lucent.com> References: <7b93pc$5pn$1@hfc.hk.super.net> <36d83b3f.0@news.pacifier.com> <01be6290$bd919540$65f677ce@LittleFoot> Eric S. Riseman wrote: > > The Kernel you have to buy, the rest of WindRivers tools are GNU based and > are freeware. Either find the freeware on the internet, or buy the > freeware > from Wind (guess that means they are no longer freeware.....). To quote (loosely) Richard Stallman of FSF, Free means Freedom, not Free of cost. Freedom to modify and grow. Freedom of knowledge. When you buy GNU from Wind River, or Cygnus, wherever, you are paying for some tangible (or not) value added. You may be getting support, or tool wrappers, whatever. That is what FSF and OpenSource is all about. The freedom to experiment and learn. The freedom to modify and develop from a foundation without have to start from zero. With that freedom comes the responsibility to return your work back to the community that provided you that base. You can charge for your work, but you have to give back the code. Anything you develop independently from scratch you can use and distribute as you see fit. Hence the kernel and the host services from Wind River are proprietary. They could OpenSource those parts if they wanted to, but obviously, they don't want to. Freedom is also embodied in this web site. Some come here to ask questions, some provide answers. In the last year, it has worked well for me in both directions. Cheers! Doug Fraser Lucent Technologies --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Which host platform, Solaris or NT? Date: Wed, 03 Mar 1999 10:18:43 -0500 From: Douglas Fraser Organization: Lucent Technologies Message-ID: <36DD52D3.2DC49F83@lucent.com> References: <7be77d$fab$1@nnrp1.dejanews.com> <7bfcgb$hu6$1@nnrp1.dejanews.com> djhystad@my-dejanews.com wrote: > > In article <7be77d$fab$1@nnrp1.dejanews.com>, > eezekiel@my-dejanews.com wrote: > > Does anyone have any opinion on which host platform is preferable for the > > Tornado Host environment, Sun Solaris or Windows NT? .. snip .. > I think that tornado is awful on both platforms. Pick the Solaris platform, .. snip .. I am new to the NT platform, have used the Solaris platform for a year, and MUCH prefer the Solaris platform. More stable, more up time, better file sharing. If you have multiple developers they can set up and use the tools as they wish. NT has you all using the same config unless you do a LOT of custom BS to get around that on logging in. The launcher in Solaris behaves better than clicking through a million windows on the NT version. Aye, the list goes on. I am stuck here, so I have to move onward and live with it, but given my choice, I'ld have a BIG chunky multiprocessor Sparc host running Solaris and I'ld have my PCs running X, and that would suit me fine. I have just come to realize that Windows is not good at descerning the differences between users, whereas in UNIX everybody gets their own env, which is a blessing and a curse. I just happen to believe the blessing outweighs the curse. Doug Fraser dwfraser@lucent.com --------------------------- Newsgroups: comp.os.vxworks Subject: sharing configuration files, etc. Date: Wed, 03 Mar 1999 10:33:23 -0500 From: Douglas Fraser Organization: Lucent Technologies Message-ID: <36DD5643.C4FF5031@lucent.com> hello all! I am new to Tornado on NT/98, old to Tornado on Solaris. We have an NT serving the licenses and registry and one shared volume for boot files and some other goodies. We have Tornado distributed among our clients because building across a share was hammering our already loaded network. I have the file distribution issues resolved (for hte most part) except for the Tornado configuration things. For example, I want a .wind directory in the share based the user (aka HOME/.wind in UNIX) so the windsh.tcl doesn't have to be copied and maintained all over the place. I want a common project/build/customization directory. (I find that make is very ill behaved from the command line). I want to be able to add customizations to the target menus. I want common target server configurations. Can I do this? Is it actually documented somewhere in that ugly stack of manuals? A simple pointer (or yes/no) answer would be appreciated. Thanks. Doug Fraser dwfraser@lucent.com PS: If I could swing my current organization back to a Solaris host I would, but I can't, so I have to figure out how to live in Windows land. You've been a bad boy, Billy Gates. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: VxWorks memory partitions Date: Sun, 28 Feb 1999 11:23:13 +0300 From: Leonid Rosenboim Organization: BitBand Technologies Ltd. Message-ID: <36D8FCF1.E8EF9F7@bitband.com> References: <7akk8t$9l4@overload.lbl.gov> Keith Buchanan wrote: > [...snip...] > Geez, this is laying it on a little thick don't you think? I tend to > believe any designer who allocates and frees memory willy-nilly in a > real-time application is a bug ;o). > Used to be true, but now with Object Oriented design techniques (either C++ or good old C) this is a different world out there now. Besides, applications grew much more complex, and requires a different way of thinking. I used to share Keith's view some years ago, but I changed my mind. > > You cannot get something for nothing. The more elaborate the memory > allocation algorithms, the slower the functions that allocate and free > memory and the larger the footprint of the OS If you believe this, you should be still using bubble-sort ... ;-) In other words, I don't agree with this either - the OS is scalable, and WRS should have offered a more sophisticated memory allocation library AS AN OPTION. Much better algorithm was available from GNU emacs for over 10 years now, which is much faster indeed, and only is slightly larger in footprint. I am sure that there are many applications which would gladly pay for the extra footprint to get better allocation efficiency and less fragmentation. > [snip] > > Wind River had to compromise between complex alogrithms with a > large footprint and the current implementation. In addition, many of > the choices for algorithms are not deterministic. The functionality > provided is reasonable and my opinion of it more typical. > No, they did not have to compromise - they should offer both (or more) of them. For example, for the management of the Ready Queue, there is an option (which is turned on by default) to add an extra hash table to make Ready Q operations require constant time (its actually constant insertion time), so those applications which want to save memory and have slower kernel operation can turn this option off. So why not doing the same with memory allocation ? By the way, there is a fixed-block memory allocation library in VxWorks which is not even documented - bufLib. It was done for internal needs, but there is plenty of uses for it in applications too, but they didn't bother to include its manual pages in the product. This only confirms the feeling that many of you have that such shortcomings are simply due to lack of attention to the customers needs, or plain laziness. Adios Amigos, Leonid --------------------------- Newsgroups: comp.os.vxworks Subject: Using IRQ3 interrupts on a PPC860DE Date: Wed, 3 Mar 1999 12:24:22 -0500 From: "Rich Duszczak" Organization: Monmouth Internet Message-ID: <7bjps5$atn$1@news.monmouth.com> Hello and thanks for reading my message. I am running vxWorks on our proprietary development board using an MPC860DE processor with a bus speed of 25MHz. I am trying to install a handler to service an external interrupt generated on /IRQ3. Here is the relevant code: VOIDFUNCPTR irq3handler(int *unused) { /* irq3Count++ ; logMsg("IRQ3 Asserted (%d)\n", irq3Count, 0, 0, 0, 0, 0) ;*/ } void irq3init() { *SIUMCR(vxImmrGet()) &= ~(SIUMCR_DPC) ; START_TIMING /* Put IRQ3 into non-asserted state */ intConnect(IV_IRQ3, (VOIDFUNCPTR)irq3handler, 0) ; /* intDisable(IV_IRQ3) ;*/ intEnable(IV_IRQ3) ; } void irq3(int state) { switch(state) { case 0: STOP_TIMING /* Assert IRQ3 */ break; case 1: START_TIMING /* De-assert IRQ3 */ break; case 2: STOP_TIMING START_TIMING break; } } irq3handler: I have tried irq3handler as both an empty function and with a simple logMsg call. From what I can see it never reaches this code. irq3init: Setting the DPC bit in the SIUMCR register configures the appropriate multiplexed pins on the processor as /IRQx. START_TIMING is a macro that forces /IRQ3 high (non-asserted). I then connect to and enable/disable the interrupt. When disabled, I can use the irq3 function to toggle the /IRQ3 line. When the interrupt is enabled and I try to assert /IRQ3 the board locks and the ICE reports a check stop exception at address 0x22fcc. I used the command 'lkAddr 0x22fcc' to find out that 0x22fcc is the beginning of excIntHandle(). My question is what could cause the system to fail at that point when interrupts are enabled and a handler is installed for /IRQ3? I have verified that the /IRQ3 line is being toggled (it is wired to a general purpose output pin so I can assert it when I want to) and that an interrupt is being generated by the processor. I've tried playing with the SIU interrupt registers but nothing has changed. Thank you in advance for any help you can provide, Rich Duszczak --------------------------- Newsgroups: comp.os.vxworks Subject: Re: How to open an ELF file Date: Wed, 3 Mar 1999 21:18:16 +0200 From: "RonenO" Organization: Unspecified Organization Message-ID: <7bk22l$avm$1@news.ibm.net.il> References: <7bk1ga$nss$1@news.ibm.net.il> hi hagit IBM News server wrote in message <7bk1ga$nss$1@news.ibm.net.il>... >We would like to download (using TFTP) an ELF file (VxWorks kernel with an >application linked together) into the target memory and then open it in >order to run it. >Did anyone implement the algorythem of converting an ELF file into an actual >code, or maybe used a VxWorks utility that do so? > >Hagit. > > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: How to open an ELF file Date: Wed, 03 Mar 1999 14:35:32 -0500 From: Douglas Fraser Organization: Lucent Technologies Message-ID: <36DD8F04.6B46BC3A@lucent.com> References: <7bk1ga$nss$1@news.ibm.net.il> IBM News server wrote: > > We would like to download (using TFTP) an ELF file (VxWorks kernel with an > application linked together) into the target memory and then open it in > order to run it. > Did anyone implement the algorythem of converting an ELF file into an actual > code, or maybe used a VxWorks utility that do so? > > Hagit. If you are trying to load your kernel also, you need to call bootElfInit(), open the Elf file image and call bootLoadModule(), then jump to the resulting entry point. For examples on how this is done, take a look at the code in bootConfig.c in $WIND_BASE/target/config/all. Good luck! Doug Fraser dwfraser@lucent.com --------------------------- Newsgroups: comp.os.vxworks Subject: STREAMS & VxWorks Date: 3 Mar 1999 21:19:41 GMT From: robinson@see-my-sig.com (Jim Robinson) Organization: Somewhere out in the boondocks ... Message-ID: <7bk91d$am7$1@vancouver.ipsg.mot.com> Reply-To: robinson@see-my-sig!.com As STREAMS was originally written for a non-realtime OS, I am wondering whether anyone has encountered problems using it in a realtime environment. For example, does the STREAMS scheduler negatively interact with VxWorks' scheduler? Are there STREAMS algorithms or data structures that are less than efficient? If anyone has any positive or negative experiences with which to respond, it would be appreciated. - -- Jim Robinson robinson@wdg. X mot X .com # remove spaces and Xs in address --------------------------- Newsgroups: comp.os.vxworks Subject: do_protocol_with_type() question. Date: Wed, 3 Mar 1999 13:58:17 -0800 From: "rvsaripa" Organization: Intel Corporation Message-ID: <7bkdk6$sl1@news.or.intel.com> In the type field for do_protocol_with_type() function , when calling this function, do I pass the type after converting it to host order ( by calling NTOHS) or do I pass it as it is ?. The reason I ask is because I have 2 ENET drivers with me for which I have sources. One driver appears to do the conversion before calling this function and another does not appear to do that. On a different note, how do I compile ping utility to be part of the vxworks image?. I tried "Configure BSP", chose INCLUDE_PING from the excluded options list ( it was part of the list ) and then chose build to build the boot ROM ( bootrom_uncmp ) and the vxworks image ( vxworks ). But when I try lkup "ping", it says no ping found. How do I enable ping?. Thanks, rk --------------------------- Newsgroups: comp.os.vxworks Subject: Re: compression source code Date: Tue, 02 Mar 1999 07:45:46 GMT From: mmcquade@aa.net Organization: Alternate Access Incorporated Message-ID: <36db967c.1540705@news.aa.net> References: <79s8hi$3r2$1@news.netway.at> <36D975BD.B69BADD1@lucent.com> On Tue, 02 Mar 1999 00:01:41 +1000, NOsatanusSPAM@KILLSPAM.ender.SPAMDIE.com (Lord Apollyon) wrote: >Not sure bzip (aka Wheeler Block Sorting compressor) would be all that >practical for your embedded work... seeing as it has monstrous memory >requirements for decompression. Yes, you can contain it by using a small >bucket size on the compression side, but.... > Someone in this group / on this list, has ported BZIP, I had an email dialogue with him about a year ago. I remember him saying it was alot of work to port BZIP to VxWorks, and as I remember, it wasn't a big win. Maybe this person lurks in the background, and will come forth and post the results. This could be added to the FAQ. Mike --------------------------- Newsgroups: comp.os.vxworks Subject: mutiple telnet sessions w/ vxWorks Date: Thu, 4 Mar 1999 11:05:38 +0200 From: Message-ID: <36de6f42.0@news.barak.net.il> Hi, We are trying to create a managed box using vxWorks. Part of the package should be a telnet server that should support multiple clients. my understanding is that the telnetd that comes with vxWorks can handle only one client connected. Is this correct? If so Is there a third party telnetd software that will support multiple clients? Thanks, Rami --------------------------- End of New-News digest ********************** From owner-vxwexplo-process Thu Mar 4 11:32:39 1999 From: "Alan T. Pickering" Date: Thu Mar 4 11:32:43 PST 1999 Subject: Signal Processing Library Status: RO Fellow Vxworkers. Does anyone know of a vender who sells a COTS version of a Signal Processing Algorithm Library which has been optimized for a PowerPC single board computer ? My application will be running under VxWorks. So far, I have come up with precious little in terms of vender choices. Yes, I know that using an SBC platform for signal processing functions is not the smartest choice of hardware. But, the architecture has been dicatated to me. Allan Pickering Raytheon (Portsmouth, RI) From owner-vxwexplo-process Thu Mar 4 17:57:23 1999 From: Fred Roeber Date: Thu Mar 4 17:57:26 PST 1999 Subject: Re: Signal Processing Library Status: RO Al Pickering asked: > Does anyone know of a vender who sells a COTS version of a > Signal Processing Algorithm Library which has been optimized for > a PowerPC single board computer ? My application will be running > under VxWorks. So far, I have come up with precious little in terms > of vender choices. > > Yes, I know that using an SBC platform for signal processing functions > is not the smartest choice of hardware. But, the architecture has been > dicatated to me. Al, A lot of the "array processor" vendors are actually now pushing PPC processors as the processor of choice. Mercury, Sky, and CSPI all have PPC versions of their signal processing platforms. I don't know if any of them would sell there signal processing libraries in an unbundled form but you might ask them. Just a suggestion. Fred | Fred J Roeber, BBN Systems & Technologies | | 4 John Clarke Road Middletown, RI 02842-5202 | | froeber@bbn.com 401-848-3548 | | TraceMaker product manager -> www.tracemaker.bbn.com | From owner-vxwexplo-process Fri Mar 5 04:00:31 1999 From: daemon@csg.lbl.gov Date: Fri Mar 5 04:00:35 PST 1999 Subject: comp.os.vxworks newsdigest Status: RO Comp.Os.Vxworks Daily Digest Fri Mar 5 04:00:26 PST 1999 Subject: Re: vxworks network IP address just started failing. Subject: Re: Experiences with Ethernet-Chip SMC91C94 Subject: Re: Problems with printf on IDT 64474 chip Subject: Having problems using ether_output() . Can someone help?. Subject: Question on the ethertype field in the 802.3 header Subject: My ex thought she was cool 57368 Subject: Re: BUG++ in vxWorks or I miss something? Subject: Command list Subject: a question about vx-work Subject: Re: Wind River' Tornado 1.0.1 for Windows NT is bugged Subject: Re: mutiple telnet sessions w/ vxWorks Subject: VxWorks Test (Ignore) Subject: Cache for dosFs Subject: mkdir with DOS not seen by VxWorks ll Subject: ada95 and vxworks Subject: VME Programming Book Subject: Re: IP address change.. ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: vxworks network IP address just started failing. Date: Thu, 4 Mar 1999 13:40:56 -0000 From: "John Cresswell" Organization: Customer of Planet Online Message-ID: <7bm2jj$t5q$1@newsreader2.core.theplanet.net> References: <7akr9o$9i7@overload.lbl.gov> Sounds as if the MAC address you are using msb broadcast bit set Scott Peters wrote in message <7akr9o$9i7@overload.lbl.gov>... >Hello vx-ers: > >We just started to getting the following message on all our machines: >0xfc95b8 (tNetTask): arp: ether address is broadcast for IP address a010547! >We have not change the software and we even disconnect from our company Lan. >We just started seeing this as of 2/18/97. > >Has anybody seen this? > >Please email me directly, I am not current on the exploder. > >Thank you very much >Scott.Peters@agai.com > > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Experiences with Ethernet-Chip SMC91C94 Date: Thu, 4 Mar 1999 13:44:07 -0000 From: "John Cresswell" Organization: Customer of Planet Online Message-ID: <7bm2pe$t8m$1@newsreader2.core.theplanet.net> References: <7a9f6j$tdj$1@news03.btx.dtag.de> Heribert Volkhausen wrote in message <7a9f6j$tdj$1@news03.btx.dtag.de>... >Hi, > >does somebody have made experiences in using an ethernet adapter based on >SMC's 91C94. Yes ! > >Please give me a comment. Seems to function ok > >Thanks in advance >Heribert Volkhausen > > > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Problems with printf on IDT 64474 chip Date: Thu, 04 Mar 1999 15:49:44 GMT From: don.blake@lmco.com (Donald R. Blake) Organization: Lockheed Martin Federal Systems, Owego, NY Message-ID: <36deaaf2.12583486@news.owg.fs.lmco.com> References: <7beshd$do1@overload.lbl.gov> I've tried twice to send this via the VxWorks exploder with no success so I'll try posting to the newsgroup. I ran into a similar problem last year when doing some benchmarks of a custom network driver with SENS. I eventually narrowed the problem down to a printf with floating point. The problem didn't occur if I changed the printf to sprintf or I use fixed point instead of floating point. The problem only occurred after I sent at least one UDP/IP packet via my custom network driver. My test case did nothing but a bunch of floating point printf's. It would always hang on the 28th printf - 27 printf's worked okay. We passed this on to WindRiver but never heard anything back. We're not using SENS at this time, so I haven't look into the problem further. We've received a couple of SENS updates and have not tried to recreate the problem since. Don Lockheed Martin Federal Systems Owego, NY On Fri, 26 Feb 1999 21:08:35 -0800, "Dmitri Varsanofiev" wrote: >Dear VxWorkers, > >We are running VxWorks on a new IDT chip, 64474. The vanilla R4000 BSP >(4650/4700 variety, CPU=R4000) seems to work well, with one notable >exception: any attempt to printf a floating point number causes a task to >hang up. > >Did anyone experience anything similar? Any hints will be appreciated. --------------------------- Newsgroups: comp.os.vxworks Subject: Having problems using ether_output() . Can someone help?. Date: Thu, 4 Mar 1999 11:25:06 -0800 From: "rvsaripa" Organization: Intel Corporation Message-ID: <7bmp1a$18i@news.or.intel.com> Hi, I am new to vxworks so excuse me for the verbosity here because I would like to give as much info as possible so that you can find out what I am doing wrong here. I am using Tornado 1.0.1 with a 486 BSP. I am writing a link-layer driver that sends/receives IP frames encapsulated in 802.3 format ( with MAC addresses ). I can attach properly with ether_attach because when I do ping with an IP address on my subnet, my output function does get called. I think I have a parameter mismatch with ether_output ( either numbers/ types of parameters ). In my xxxoutput() routine, I have the following :- struct YYY { struct arpcom IDR; // other stuff } IfHandle; int xxxOutput( struct ifnet *pIf, struct mbuf *pmbufchain, SOCKADDR *pDst ) { int retValue; logMsg("xxxOutput() : entered calling ether_output() with if=%x mbuf=%x dst=%x arpcom=%x\n", (int)pIf, (int)pmbufchain, (int)pDst, (int)&(IfHandle.IDR), 0, 0 ); retValue = ether_output( pIf, pmbufchain, pDst, (FUNCPTR)IfStartOutput, &(IfHandle.IDR) ); if ( retValue ) { logMsg("xxxOutput() : ether_output() returned value=%d\n",retValue,0,0,0,0,0); } return retValue; } IfStartOutput( int unit ) { // do something here } In my init() routine, I set the following flags , IFF_UP and IFF_RUNNING When I start a ping , my xxxOutput() gets called and call goes to ether_output() . Then I get a GPF in ether_output() at address 0x150369. When I disassemble ether_output , I have the following assembly code. The first instruction here is the entry point for ether_output. 001502d8 55 PUSH EBP 001502d9 89 e5 MOV EBP, ESP 001502db 83 ec 24 SUB ESP, 36 001502de 57 PUSH EDI 001502df 56 PUSH ESI 001502e0 53 PUSH EBX 001502e1 8b 75 08 MOV ESI, [EBP+8] 001502e4 8b 5d 14 MOV EBX, [EBP+20] 001502e7 c7 45 f0 00 00 00 00 MOV [EBP-16], 0x0 001502ee 8b 55 0c MOV EDX, [EBP+12] 001502f1 89 55 ec MOV [EBP-20], EDX 001502f4 c7 45 e8 00 00 00 00 MOV [EBP-24], 0x0 001502fb 8b 4a 18 MOV ECX, [EDX+24] 001502fe 89 4d e4 MOV [EBP-28], ECX 00150301 89 75 e0 MOV [EBP-32], ESI 00150304 66 OSIZE 00150305 8b 46 1a MOV EAX, [ESI+26] 00150308 66 OSIZE 00150309 83 e0 41 AND EAX, 65 0015030c 66 OSIZE 0015030d 83 f8 41 CMP EAX, 65 00150310 74 16 JE _ether_output + 0x50 00150312 c7 45 f0 3e 00 00 00 MOV [EBP-16], 0x3e 00150319 e9 fa 02 00 00 JMP _ether_output + 0x340 0015031e 8d 76 00 LEA ESI, [ESI+0] 00150321 8d b4 26 00 00 00 00 LEA ESI, [ESI+0] 00150328 e8 9b e2 ff ff CALL _tickGet 0015032d 89 46 58 MOV [ESI+88], EAX 00150330 89 df MOV EDI, EBX 00150332 85 ff TEST EDI, EDI 00150334 0f 84 ae 00 00 00 JE _ether_output + 0x110 0015033a f6 47 34 01 TEST [EDI+52], 0x1 0015033e 75 1a JNE _ether_output + 0x82 00150340 6a 01 PUSH 1 00150342 8b 55 10 MOV EDX, [EBP+16] 00150345 52 PUSH EDX 00150346 e8 ed ae 00 00 CALL _rtalloc1 0015034b 83 c4 08 ADD ESP, 8 0015034e 89 c7 MOV EDI, EAX 00150350 89 fb MOV EBX, EDI 00150352 85 ff TEST EDI, EDI 00150354 74 3b JE _ether_output + 0xb9 00150356 66 OSIZE 00150357 ff 4f 36 DEC [EDI+54] 0015035a f6 47 34 02 TEST [EDI+52], 0x2 0015035e 74 48 JE _ether_output + 0xd0 00150360 83 7f 4c 00 CMP [EDI+76], 0 00150364 74 14 JE _ether_output + 0xa2 00150366 8b 7f 4c MOV EDI, [EDI+76] 00150369 f6 47 34 01 TEST [EDI+52], 0x1 -----> Where the GPF occurs. 0015036d 75 39 JNE _ether_output + 0xd0 0015036f 57 PUSH EDI 00150370 e8 b3 b0 00 00 CALL _rtfree 00150375 83 c4 04 ADD ESP, 4 00150378 89 df MOV EDI, EBX 0015037a 6a 01 PUSH 1 At instruction 0x1502e4, ebx is loaded with the address of my actual output routine ( which is address of IfStartOutput() ). Before that, esi points to my driver's struct ifnet At instruction 0x150305, eax is loaded with the value of the if_flags field in the struct ifnet structure which is 0x41 which means IFF_UP and IFF_RUNNING. The test for equality at 0x150310 is true and we jump to 0x150328. As we trace the flow of execution from instruction 0x150330 ( which loads the address of IfStartOutput into edi from ebx ), it is very clear that this function thinks edi points to a data address whereas it actually points to a code address ) as it starts testing offsets into this address for a specific value. When control comes to 0x150366, edi is loaded with contents of edi+76 and then in the very next instruction it GPFs. Can anyone tell me if my usage of ether_output() is correct?. I picked up the code from an ethernet driver. Also, from the Tornado manuals , looks like I do not have to use ether_output() because all it does is ARP work. Is there some code that I can do this myself without having to call ether_output(). I know I have to call arpresolve() because ether_output() seems to call this because some C code to replace the functionality of ether_output() would be useful. Thanks a lot for any and all help, Ramakrishna Saripalli. --------------------------- Newsgroups: comp.os.vxworks Subject: Question on the ethertype field in the 802.3 header Date: Thu, 4 Mar 1999 13:03:49 -0800 From: "rvsaripa" Organization: Intel Corporation Message-ID: <7bmuqe$64p@news.or.intel.com> Hi, I have an Ethernet-type driver . When I get an IP packet to send , I put the destination MAC address (6 bytes) + the source MAC address ( 6 bytes ) followed by 2 bytes of the ethertype field ( 806 for ARP and 800 for IP ). I think I should do HTONS conversion on this type field before putting it in the buffer but the DEC Ethernet driver does not seem to do it. Look at the if_dcFast.c file in the tornado installation for Tornado 1.0.1 for a 486 BSP. How does this work?. Does the card take care of this?. I am pretty sure that HTONS has to be done on the ethertype field but can someone answer the above question?. BTW, does HTONS has to be done for ARP type too ?. Thanks a bunch, Rk All opinions are mine & mine only. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: BUG++ in vxWorks or I miss something? Date: Thu, 04 Mar 1999 14:05:17 -0800 From: mlang@fst.jpl.nasa.gov (M Lang) Organization: JPL Message-ID: References: <7bjn69$do2@overload.lbl.gov> In article <7bjn69$do2@overload.lbl.gov>, dit@bach.jhuapl.edu (Daryl I. Tewell) wrote: > VxWorks,vxworks,vxWorks > > > Suddenly the tasks designed to be low priority processes are boosted to some > > absurd priority. The following example demonstrates this phenomenon. > > Could the absurd priority be the result of taskPriorityGet() returning > ERROR (and leaving pri uninitialized)? My reference manual is unclear > about whether you can pass 0 to taskPriorityGet() to refer to the > calling task. 0 is ok as passing NULL as the task id means the calling task. I just tried it on a Heurikon PPC 740 board running vxWorks 5.3.1 and I saw the same behavior (the low priority task having a prio 160 got promoted to 150). Looks like this "absurd priority" was caused from inside the malloc call. = Minh = --------------------------- Newsgroups: comp.os.vxworks Subject: Command list Date: Thu, 4 Mar 1999 18:33:17 -0600 From: "Richard Kimball" Message-ID: Is there a source for a full list of commands available in VxWorks? Thanks, Rich --------------------------- Newsgroups: comp.os.vxworks Subject: a question about vx-work Date: Thu, 04 Mar 1999 20:03:25 -0500 From: Lei Sun Organization: University of Central Florida Message-ID: <36DF2D5D.47A1@pegasus.cc.ucf.edu> Hi: I am a new user in vx-work and tornado development tools. I have configured Tornado properly and written a small program.It simply create 10 tasks and print out the task ID individually. However I can't get the expected result. It only print out the main function message,say"main program begain..." and then program exit normally. anybody can tell me the reason ? By the way, I have redirect the I/O and the printed message should appears in Virtual console, I am using VME167 as BSP and host machine is sparc station. attatched is my code.(very simple) #include "vxWorks.h" #include "taskLib.h" #include "stdio.h" #define TASKNB 10 void a_task(void); /*************************/ void multiTask(void) { int i; printf("Task begain....\n"); // for (i=0;i Organization: Nortel Networks Message-ID: <36DF24F5.1595EC16@slowspam.nortel.ca> References: <7bn25a$jeu$1@nslave1.tin.it> Geez, I dunna. Besides the fact that the installation process is a little clumsy, we didn't have much trouble getting it set up (on HP-UX, on Win95, on WinNT workstation, on Solaris). What troubles are you experiencing? Some tips: If you're using a pre-Windview 2.0 (ZC-10) CD-ROM (at least for PPC target), always install SENS after everything else. Be careful to install with the license server set up correctly (that is, floating if it is floating, etc.). You probably need to install the Portmapper. Make sure the registry and license servers are running where you've told the installations they are running; you can check under Programs->Tornado->Registry&License NOTE: You'll still need to reboot occasionally, but typically not directly due to Tornado. Wolfy wrote: > > I installed Tornado 1.0.1 ( 80x86) for Windows on 3 PC: > 2 NT 4.0 Workstation and a Win98 laptop; > > It is about 10 days I am trying to make them work but I am only experimentig > that one installation on WinNT4.0 is working at 60%, another one on WinNT > 4.0 too ( with SP3 ) is absolutly not working. ( installed from the same > CD on an twin machine ( the only difference is that one machine has 2 NTFS > partitions while the other one has 1 FAT16 and 1 NTFS ). > The Win98 version is at the moment the only one that let me compile vxWorks > kernel, > Yesterday I installed the licence manager 3 times on an NT4.0 Server and, at > last, after 2 reboots has started working. > At the moment the licence manager is THE ONLY THING THAT WORKS PROPERLY .... > I have the feeling ( and something more really ) that this development > system is a true "banana". > > I only hope that vxWorks operating system may work much better than its > development system. > > Does anyone experimented a similar case ? > > Wolfy - -- Tim Sohacki, Nortel Networks, Inc. email: sohacki@nortelnetworks.com --------------------------- Newsgroups: comp.os.vxworks Subject: Re: mutiple telnet sessions w/ vxWorks Date: Thu, 04 Mar 1999 17:35:09 -0800 From: Jerry Hudgins Organization: Western Digital Design, Marin County, California Message-ID: <36DF34CD.977D253C@e-farm.com> References: <36de6f42.0@news.barak.net.il> r wrote: > > If so Is there a third party telnetd software that will support multiple > clients? You might be interested in the reentrant rshell client/server offered by Real-Time Innovations as part of their RTILib package. I've used it, and it works fine. See: http://www.rti.com/ - -jch --------------------------- Newsgroups: comp.os.vxworks Subject: VxWorks Test (Ignore) Date: Wed, 03 Mar 1999 10:06:24 -0800 From: Edward LaPier Organization: Lockheed Martin Federal Systems, Owego, NY Message-ID: <36DD7A20.7B21A9E8@lmco.com> VxWorks Test (Ignore) --------------------------- Newsgroups: comp.os.vxworks Subject: Cache for dosFs Date: Thu, 4 Mar 1999 23:23:11 -0500 From: "David Garner" Message-ID: <7bnm7v$ph2$1@winter.news.rcn.net> Does anyone know of a caching utility that can be integrated into dosFsLib? Are there any vendors that market such products? If I read a one meg file four times in a row I would like the last three reads to obtain the file data completely from dram. Thanks for any help. Dave Garner vxworks tornado --------------------------- Newsgroups: comp.os.vxworks Subject: mkdir with DOS not seen by VxWorks ll Date: Fri, 05 Mar 1999 04:58:49 GMT From: ndh@InternetSmiths.com (ndh) Keywords: mkdir, dosFs, DOS, Message-ID: folks, When I do a DOS mkdir using MS DOS 6.22, Win95 or WinNT 4.0, then take this SCSI disk to a VxWorks system, ll "/sd0" does not see the directories. I even tried all UPPER as well as lower case. A mkdir "/sd0/subdir" creates a subdirectory that both systems can see. Any ideas as to what might be causing this problem? thanks in advance, nigel nigel dyson-hudson, InternetSmiths, ndh@InternetSmiths.com "mad dogs and Englishmen" and I am both! --------------------------- Newsgroups: comp.os.vxworks Subject: ada95 and vxworks Date: Fri, 5 Mar 1999 02:19:03 -0600 From: "Michael Garrett" Message-ID: I would like to hear from someone who has ada95 / vxworks experience. Gnat or Aonix. We may use it in our next product and would like some feedback. Michael Garrett --------------------------- Newsgroups: comp.os.vxworks Subject: VME Programming Book Date: Fri, 05 Mar 1999 08:49:28 +0000 From: Stephen A Walsh Organization: Marconi Electronic Systems Message-ID: <36DF9A98.27D4@gecm.com> Can anyone recommend a good book on programming a VME interface. It would help if it was written from a Software rather that Hardware point of view. TIA Steve - -- ******************************************* Stephen A Walsh Senior Software Engineer Radar and Countermeasures Systems Marconi Electronic Systems ******************************************* Opinions are usually mine and mine alone ******************************************* - -- ******************************************* Stephen A Walsh Senior Software Engineer Radar and Countermeasures Systems Marconi Electronic Systems ******************************************* Opinions are usually mine and mine alone ******************************************* --------------------------- Newsgroups: comp.os.vxworks Subject: Re: IP address change.. Date: Fri, 05 Mar 1999 12:26:09 +0100 From: Olivier STOLZENBERG Organization: PROSYS S.A. Message-ID: <36DFBF51.43E1E1AF@cur-archamps.fr> References: <990219014357.ZM22009@dcn49.hns.com> Krishna Samavedam a écrit: > HI VxWorkers, > I have a basic Question, Can the IP address be changed when vxWorks up and running, > with out rebooting ? if so how ? > > thanx in advance, > Krishna Samavedam Use the fellowing instruction ifAddrSet((char * )ifEtherName, (char *)c_AddresseIP); for more information look online manuals or books....... O. STOLZENBERG Prosys S.A. --------------------------- End of New-News digest ********************** From owner-vxwexplo-process Fri Mar 5 08:12:49 1999 From: Roger Cruz Date: Fri Mar 5 08:12:52 PST 1999 Subject: Re: Question on the ethertype field in the 802.3 header Status: O I ported this code to an x86 platform and I found the same problems. You need to put the HTONS there. I believe there were other similar problems (the EEPROM read for the MAC address I think). Search the dejanews.com for my postings. I recall another major problem was a register write of CSR6 which the specs indicates a RESET must be performed after and the driver didn't do it. I hope this helps. Roger From: "rvsaripa" I have an Ethernet-type driver . When I get an IP packet to send , I put the destination MAC address (6 bytes) + the source MAC address ( 6 bytes ) followed by 2 bytes of the ethertype field ( 806 for ARP and 800 for IP ). I think I should do HTONS conversion on this type field before putting it in the buffer but the DEC Ethernet driver does not seem to do it. Look at the if_dcFast.c file in the tornado installation for Tornado 1.0.1 for a 486 BSP. How does this work?. Does the card take care of this?. I am pretty sure that HTONS has to be done on the ethertype field but can someone answer the above question?. BTW, does HTONS has to be done for ARP type too ?. vxworks, VxWorks PS: Remove "no_junk" from email address -- Roger Cruz cruz_nojunk@xyplex.com Xyplex Networks w: 978-952-4783 295 Foster Street f: 978-952-4887 Littleton, MA 01460 From owner-vxwexplo-process Fri Mar 5 10:41:12 1999 From: Hakan Karakas Date: Fri Mar 5 10:41:16 PST 1999 Subject: Re: comp.os.vxworks newsdigest Status: O try this.... > ll "/sd0/" Hakan Karakas h.karakas@ieee.org --------------------------- Newsgroups: comp.os.vxworks Subject: mkdir with DOS not seen by VxWorks ll Date: Fri, 05 Mar 1999 04:58:49 GMT From: ndh@InternetSmiths.com (ndh) Keywords: mkdir, dosFs, DOS, Message-ID: folks, When I do a DOS mkdir using MS DOS 6.22, Win95 or WinNT 4.0, then take this SCSI disk to a VxWorks system, ll "/sd0" does not see the directories. I even tried all UPPER as well as lower case. A mkdir "/sd0/subdir" creates a subdirectory that both systems can see. Any ideas as to what might be causing this problem? thanks in advance, nigel From owner-vxwexplo-process Fri Mar 5 10:44:35 1999 From: Hakan Karakas Date: Fri Mar 5 10:44:39 PST 1999 Subject: VME Bus... Status: O This book may help... VME BUS A Practical Companion Steve Heath ISBN 0 7506 1750 0 Hakan Karakas h.karakas@ieee.org --------------------------- Newsgroups: comp.os.vxworks Subject: VME Programming Book Date: Fri, 05 Mar 1999 08:49:28 +0000 From: Stephen A Walsh Organization: Marconi Electronic Systems Message-ID: <36DF9A98.27D4@gecm.com> Can anyone recommend a good book on programming a VME interface. It would help if it was written from a Software rather that Hardware point of view. TIA Steve From owner-vxwexplo-process Sat Mar 6 04:00:31 1999 From: daemon@csg.lbl.gov Date: Sat Mar 6 04:00:34 PST 1999 Subject: comp.os.vxworks newsdigest Status: O Comp.Os.Vxworks Daily Digest Sat Mar 6 04:00:27 PST 1999 Subject: Re: Using IRQ3 interrupts on a PPC860DE Subject: Re: Wind River' Tornado 1.0.1 for Windows NT is bugged Subject: Driver for Local PCI Bus Subject: How to open an ELF file Subject: Codewright Subject: Re: mutiple telnet sessions w/ vxWorks Subject: Re: Having problems using ether_output() . Can someone help?. Subject: Re: Question on the ethertype field in the 802.3 header Subject: Re: WindNet SNMP - Are multiple agent tasks possible Subject: Re: Which host platform, Solaris or NT? Subject: Re: Question on the ethertype field in the 802.3 header Subject: Re: Tornado on Windows 98? Subject: flash file system ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: Using IRQ3 interrupts on a PPC860DE Date: Wed, 3 Mar 1999 09:56:46 -0800 From: "John Finley" Organization: Kivala Systems, Inc. Message-ID: <7bjsuc$1is@dfw-ixnews5.ix.netcom.com> References: <7bjps5$atn$1@news.monmouth.com> Howdy. It's been a while since I worked with this level stuff on the 860, but I can offer at least one comment. If the IRQ is edge sensitive, you have to ack it in the ISR by setting (I think) its bit in SIPEND. If it isn't edge sensitive, I think you have to disable it (in general). Without either of those, the first int will send it off in the weeds. If that doesn't help, perhaps I can put together a little program that works on our board. I have a timer example I've sent to some folks, in case that would be useful, but perhaps you are beyond that. Good luck, John - -------------------------------------------------- John Finley Kivala Systems, Inc. Project Manager (619) 689-0032 john@kivala.com http://www.kivala.com - -------------------------------------------------- VxTool - Build Tornado Apps with MS Visual Basic - -------------------------------------------------- Rich Duszczak wrote in message <7bjps5$atn$1@news.monmouth.com>... >Hello and thanks for reading my message. > >I am running vxWorks on our proprietary development board using an MPC860DE >processor with a bus speed of 25MHz. I am trying to install a handler to >service an external interrupt generated on /IRQ3. Here is the relevant >code: > > >VOIDFUNCPTR irq3handler(int *unused) >{ >/* irq3Count++ ; > logMsg("IRQ3 Asserted (%d)\n", irq3Count, 0, 0, 0, 0, 0) ;*/ >} > >void irq3init() >{ > *SIUMCR(vxImmrGet()) &= ~(SIUMCR_DPC) ; > START_TIMING /* Put IRQ3 into non-asserted state */ > > intConnect(IV_IRQ3, (VOIDFUNCPTR)irq3handler, 0) ; > >/* intDisable(IV_IRQ3) ;*/ > intEnable(IV_IRQ3) ; >} > >void irq3(int state) >{ > switch(state) > { > case 0: > STOP_TIMING /* Assert IRQ3 */ > break; > > case 1: > START_TIMING /* De-assert IRQ3 */ > break; > > case 2: > STOP_TIMING > START_TIMING > break; > } >} > >irq3handler: >I have tried irq3handler as both an empty function and with a simple logMsg >call. From what I can see it never reaches this code. > >irq3init: >Setting the DPC bit in the SIUMCR register configures the appropriate >multiplexed pins on the processor as /IRQx. START_TIMING is a macro that >forces /IRQ3 high (non-asserted). I then connect to and enable/disable the >interrupt. When disabled, I can use the irq3 function to toggle the /IRQ3 >line. When the interrupt is enabled and I try to assert /IRQ3 the board >locks and the ICE reports a check stop exception at address 0x22fcc. I used >the command 'lkAddr 0x22fcc' to find out that 0x22fcc is the beginning of >excIntHandle(). > >My question is what could cause the system to fail at that point when >interrupts are enabled and a handler is installed for /IRQ3? > >I have verified that the /IRQ3 line is being toggled (it is wired to a >general purpose output pin so I can assert it when I want to) and that an >interrupt is being generated by the processor. I've tried playing with the >SIU interrupt registers but nothing has changed. > >Thank you in advance for any help you can provide, > >Rich Duszczak > > > --------------------------- Newsgroups: comp.arch.embedded,comp.os.vxworks Subject: Re: Wind River' Tornado 1.0.1 for Windows NT is bugged Date: Fri, 5 Mar 1999 14:31:04 +0000 From: Mike Davies Message-ID: <7sC$fQAoq+32Ew9$@noco.demon.co.uk> References: <7bn25a$jeu$1@nslave1.tin.it> In article <7bn25a$jeu$1@nslave1.tin.it>, Wolfy writes >I installed Tornado 1.0.1 ( 80x86) for Windows on 3 PC: >2 NT 4.0 Workstation and a Win98 laptop; > >It is about 10 days I am trying to make them work but I am only experimentig >that one installation on WinNT4.0 is working at 60%, another one on WinNT >4.0 too ( with SP3 ) is absolutly not working. ( installed from the same >CD on an twin machine ( the only difference is that one machine has 2 NTFS >partitions while the other one has 1 FAT16 and 1 NTFS ). >The Win98 version is at the moment the only one that let me compile vxWorks >kernel, >Yesterday I installed the licence manager 3 times on an NT4.0 Server and, at >last, after 2 reboots has started working. >At the moment the licence manager is THE ONLY THING THAT WORKS PROPERLY .... >I have the feeling ( and something more really ) that this development >system is a true "banana". It is fair to say that WRS seem to care more about getting the licensing working than they care about the product. Make sure that you have the patch installed : when you start the Tornado environment pull down the help dialog box and look at the version details. There should be an entry there for something like "cumulative patch x86". Tornado is not really usable without it so be warned. >I only hope that vxWorks operating system may work much better than its >development system. It does. It would have to in order to be useful :-) - -- Mike Davies --------------------------- Newsgroups: comp.os.vxworks Subject: Driver for Local PCI Bus Date: Tue, 2 Mar 1999 12:32:42 -0500 From: "Bruce Clements" Organization: Cabletron Systems Inc Message-ID: <7bh74h$o65$1@mer-news.ctron.com> Can anyone enumerate what the steps are to writing a driver for a local PCI bus to create an internal interface and hook it into the VxWorks IP stack? We have data packets coming in from an external source and they need to be directed over the PCI bus to our CPU. This CPU will also have an ethernet chip attached to it with a externally visible IP. Processes will then want to bind on specific UDP ports to get the data. I was wondering what I need to override/implement and what the hooks are to get the VxWorks to know my driver is present without screwing up the external interface. Thanks... Bruce Clements --------------------------- Newsgroups: comp.os.vxworks Subject: How to open an ELF file Date: Wed, 3 Mar 1999 21:08:39 +0200 From: "IBM News server" Organization: Unspecified Organization Message-ID: <7bk1ga$nss$1@news.ibm.net.il> We would like to download (using TFTP) an ELF file (VxWorks kernel with an application linked together) into the target memory and then open it in order to run it. Did anyone implement the algorythem of converting an ELF file into an actual code, or maybe used a VxWorks utility that do so? Hagit. --------------------------- Newsgroups: comp.os.vxworks Subject: Codewright Date: Fri, 05 Mar 1999 16:40:46 -0500 From: Douglas Fraser Organization: Lucent Technologies Message-ID: <36E04F5E.CFF9F278@lucent.com> Any one on this list using Codewright in conjunction with Tornado? I know this has got to be as bad as asking "which is better, emacs or vi" in a UNIX forum, but I am looking for opinions and actual usage. Someone here is very enthusiastic from prior non-Tornado experience. I am looking for a little help before I order a dozen copies and get myself in up to my eyeballs. TIA Doug Fraser dwfraser@lucent.com --------------------------- Newsgroups: comp.os.vxworks Subject: Re: mutiple telnet sessions w/ vxWorks Date: Fri, 05 Mar 1999 16:50:35 -0500 From: Garo Senyildiz Organization: Nortel Message-ID: <36E051AB.530BBDB@nortel.ca> References: <36de6f42.0@news.barak.net.il> Ask windRiver. They got a deamon for multiple clients. They are not officialy supporting it though... Garo r wrote: > > Hi, > We are trying to create a managed box using vxWorks. Part of the package > should be a telnet server that should support multiple clients. > my understanding is that the telnetd that comes with vxWorks can handle only > one client connected. Is this correct? > If so Is there a third party telnetd software that will support multiple > clients? > > Thanks, > Rami --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Having problems using ether_output() . Can someone help?. Date: Fri, 05 Mar 1999 17:29:58 -0800 From: "T. Venkata Ramana" Organization: Accelerated Networks, Inc. Message-ID: <36E08516.82130366@hotmail.com> References: <7bmp1a$18i@news.or.intel.com> The correct protype for ether_output is : int ether_output(ifp, m0, dst, rt0) register struct ifnet *ifp; struct mbuf *m0; struct sockaddr *dst; struct rtentry *rt0; Ignore this message if you have already received one of my reply. I had some problems sending reply. > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Question on the ethertype field in the 802.3 header Date: Fri, 05 Mar 1999 17:07:10 -0800 From: "T. Venkata Ramana" Organization: Accelerated Networks, Inc. Message-ID: <36E07FBE.2A285E61@hotmail.com> References: <7bmuqe$64p@news.or.intel.com> As far as I know, the endian type in VxWorks is Big Endian which happens to be the network byte order. So there is no need to do HTONS and everything seems to work. rvsaripa wrote: > Hi, > I have an Ethernet-type driver . When I get an IP packet to send , I put > the destination MAC address (6 bytes) + the source MAC address ( 6 bytes ) > followed by 2 bytes of the ethertype field ( 806 for ARP and 800 for IP ). I > think I should do HTONS conversion on this type field before putting it in > the buffer but the DEC Ethernet driver does not seem to do it. Look at the > if_dcFast.c file in the tornado installation for Tornado 1.0.1 for a 486 > BSP. How does this work?. Does the card take care of this?. > > I am pretty sure that HTONS has to be done on the ethertype field but can > someone answer the above question?. BTW, does HTONS has to be done for ARP > type too ?. > > Thanks a bunch, > Rk > > All opinions are mine & mine only. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: WindNet SNMP - Are multiple agent tasks possible Date: Fri, 05 Mar 1999 18:34:19 -0800 From: "T. Venkata Ramana" Organization: Accelerated Networks, Inc. Message-ID: <36E0942A.BF294AF2@hotmail.com> References: <01be6516$d6b7fae0$2a9cd126@salmon> Snmp agent uses UDP protocol and listens for requests on port number 161. A server can bind to only one port. You cannot run multiple agents on the same target system. I cant think of a reason as to why you need multiple snmp agents? Venkata Tom Salmon wrote: > Hello everyone, > > I am using WindNet SNMPv1/v2c to create an agent for a > 68K Tornado 5.3 system. By default the WindNet product seems to create > a single tSnmp task. Does anyone know how to create more than > one of these SNMP agent tasks? --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Which host platform, Solaris or NT? Date: 1 Mar 1999 18:35:54 GMT From: "Kelly Hornsby" Organization: AT&T WorldNet Services Message-ID: <7bemma$gul@bgtnsc01.worldnet.att.net> References: <7be77d$fab$1@nnrp1.dejanews.com> I've used both, and I like both. Since GNU is now available on WinXX, along with Emacs, I like to use the WinXX station. On the other hand, it is MUCH easier to do shell processing and makefiles on a Solaris box. I vote for the NT box for the UI, but for Solaris for the ease of file processing and executable creation. eezekiel@my-dejanews.com wrote in message <7be77d$fab$1@nnrp1.dejanews.com>... >Does anyone have any opinion on which host platform is preferable for the >Tornado Host environment, Sun Solaris or Windows NT? > >I think that the WRS user interface support is better on NT. Any other >considerations? > >Thanks, >Ephraim Ezekiel >Quarry Technologies >eezekiel@quarrytech.com > >-----------== Posted via Deja News, The Discussion Network ==---------- >http://www.dejanews.com/ Search, Read, Discuss, or Start Your Own --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Question on the ethertype field in the 802.3 header Date: 6 Mar 1999 05:34:31 GMT From: Hwa-Jin Bae Organization: PSO Systems Inc (http://www.pso.com) Message-ID: <7bqep7$lqs$1@news.idiom.com> References: <7bmuqe$64p@news.or.intel.com> <36E07FBE.2A285E61@hotmail.com> T. Venkata Ramana wrote: > As far as I know, the endian type in VxWorks is Big Endian which > happens to be the network byte order. So there is no need to > do HTONS and everything seems to work. The endian depends on machine (cpu) architecture, not the OS like VxWorks. - -- Hwa-Jin Bae PSO Systems, Inc. http://www.pso.com mailto:hjb@pso.com --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Tornado on Windows 98? Date: Thu, 4 Mar 1999 13:39:41 -0000 From: "John Cresswell" Organization: Customer of Planet Online Message-ID: <7bm2hc$t4n$1@newsreader2.core.theplanet.net> References: <9B8FE94E7CB1D211BA85006097C9E09E9A60@US-Pax.mail.saic.com> I would not recommend Tornado 1.0.1 on Windows 98 but Torando 2.0 might support it. We use NT running on both desktops & laptops with out problems. Thomas, Justin W. wrote in message <9B8FE94E7CB1D211BA85006097C9E09E9A60@US-Pax.mail.saic.com>... >VxWorkers, > >I am currently using Tornado 1.0.1 on my NT 4.0 workstation. The target >server, license server, etc. are all run from the same workstation. I would >like to move the development environment to a laptop to do some field >development work. I am hesitant to use NT on the laptop because I have >heard that it is difficult to get all of the drivers to support the built in >hardware. > >Does anyone have experience (positive or negative) with using Tornado on >Windows 98 (or 95) as opposed to NT? I need the transition to go smoothly >because I have very little time to spend working out kinks in the >development environment. > > >Thanks in advance, > >Justin > >Justin Thomas >SAIC - Project Engineer >ph 301 866-6705 >fax 301 863-0299 >justin.w.thomas@cpmx.saic.com > > --------------------------- Newsgroups: comp.os.vxworks Subject: flash file system Date: Sat, 06 Mar 1999 01:03:41 -0800 From: phoebeli@tsoft.com Message-ID: <36E0EF6D.2162EB4E@tsoft.com> Reply-To: phoebeli@tsoft.com This is a multi-part message in MIME format. - --------------BB832A91AFB407A7DCDB1109 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Anyone has a shareware version of flash file system for flash memory chips? Dean dwang@andanets.com - --------------BB832A91AFB407A7DCDB1109 Content-Type: text/x-vcard; charset=us-ascii; name="vcard.vcf" Content-Transfer-Encoding: 7bit Content-Description: Card for Phoebe Li Content-Disposition: attachment; filename="vcard.vcf" begin: vcard fn: Phoebe Li n: Li;Phoebe email;internet: phoebeli@tsoft.com x-mozilla-cpt: ;0 x-mozilla-html: FALSE version: 2.1 end: vcard - --------------BB832A91AFB407A7DCDB1109-- --------------------------- End of New-News digest ********************** From owner-vxwexplo-process Sun Mar 7 04:00:15 1999 From: daemon@csg.lbl.gov Date: Sun Mar 7 04:00:19 PST 1999 Subject: comp.os.vxworks newsdigest Status: O Comp.Os.Vxworks Daily Digest Sun Mar 7 04:00:12 PST 1999 Subject: Defrag utility for dos file system? Subject: Re: Wind River' Tornado 1.0.1 for Windows NT is bugged Subject: Re: Tornado on Windows 98? ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Defrag utility for dos file system? Date: Sat, 6 Mar 1999 11:20:14 -0800 From: Randy Riggins Message-ID: <54C40CA7FBB3D111B0D200A0C9AB28C102AAAF2A@earth.telestream.net> Sender: Randy Riggins Does anyone know of or is willing to sell a vxWorks based defrag utility for the dos file system? Thanks, randy riggins --------------------------- Newsgroups: comp.arch.embedded,comp.os.vxworks Subject: Re: Wind River' Tornado 1.0.1 for Windows NT is bugged Date: Sat, 6 Mar 1999 20:07:16 -0000 From: "Darren J Longhorn" Organization: Customer of Planet Online Message-ID: <7bs1vp$52u$1@newsreader1.core.theplanet.net> References: <7bn25a$jeu$1@nslave1.tin.it> <7sC$fQAoq+32Ew9$@noco.demon.co.uk> Hmm, I don't see that on mine. Where is it exactly? Mike Davies wrote in message <7sC$fQAoq+32Ew9$@noco.demon.co.uk>... >Make sure that you have the patch installed : when you >start the Tornado environment pull down the help dialog >box and look at the version details. There should be an >entry there for something like "cumulative patch x86". >Tornado is not really usable without it so be warned. > > > >-- >Mike Davies --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Tornado on Windows 98? Date: Sun, 07 Mar 1999 03:01:05 GMT From: John or Jenn Organization: Who, us organized??? Message-ID: References: <36de6f42.0@news.barak.net.il> <36DF34CD.977D253C@e-farm.com> <36E04F5E.CFF9F278@lucent.com> <36E051AB.530BBDB@nortel.ca> In article <7bm2hc$t4n$1@newsreader2.core.theplanet.net> John Cresswell, john.cresswell@pace.co.uk writes: >I would not recommend Tornado 1.0.1 on Windows 98 but Torando 2.0 might >support it. We use NT running on both desktops & laptops with out problems. > Ummm...why not? Have you tried it and run into problems? Just superstitious? (or, most likely, just woldn¹t recommend Win 98 for anything? ;-) John --------------------------- End of New-News digest ********************** From owner-vxwexplo-process Sun Mar 7 17:00:50 1999 From: "Bill W.G. Shi" Date: Sun Mar 7 17:00:54 PST 1999 Subject: vxworks snmp Status: O vxWorks Hi, can anyone tell me if this "WindNet SNMPv1/v2c" part of "SENS"? Bill W.G. Shi at BBT Inc. E-Mail Address: wgshi@bbti.com.cn From owner-vxwexplo-process Mon Mar 8 04:00:52 1999 From: daemon@csg.lbl.gov Date: Mon Mar 8 04:00:56 PST 1999 Subject: comp.os.vxworks newsdigest Status: O Comp.Os.Vxworks Daily Digest Mon Mar 8 04:00:48 PST 1999 Subject: Re: Wind River' Tornado 1.0.1 for Windows NT is bugged Subject: Re: Slow network performance with 5.3.1 Subject: Re: ei82596End users? Subject: Announcement: malloc/free replacement for VxWorks Subject: Re: a question about vx-work Subject: Re: compression source code Subject: Re: ada95 and vxworks Subject: Re: VxWorks memory partitions Subject: Re: Anyone has some examples for ramDrv? Subject: Re: Slow network performance with 5.3.1 Subject: Netware NCP server source Subject: Re: select() with Wind Msg Queues Subject: Help. Shell Subject: Re: about device driver Subject: Re: flash file system ------------------------------------------------------- Newsgroups: comp.arch.embedded,comp.os.vxworks Subject: Re: Wind River' Tornado 1.0.1 for Windows NT is bugged Date: Sun, 7 Mar 1999 11:47:40 +0000 From: Mike Davies Message-ID: References: <7bn25a$jeu$1@nslave1.tin.it> <7sC$fQAoq+32Ew9$@noco.demon.co.uk> <7bs1vp$52u$1@newsreader1.core.theplanet.net> In article <7bs1vp$52u$1@newsreader1.core.theplanet.net>, Darren J Longhorn writes >Hmm, I don't see that on mine. Where is it exactly? > >Mike Davies wrote in message <7sC$fQAoq+32Ew9$@noco.demon.co.uk>... >>Make sure that you have the patch installed : when you >>start the Tornado environment pull down the help dialog >>box and look at the version details. There should be an >>entry there for something like "cumulative patch x86". >>Tornado is not really usable without it so be warned. >> >> >> >>-- >>Mike Davies OK, on Tornado 1.01 targeted for the 80x86 there is a standard looking help menu with an "about" item that displays a dialog box. This dialog box has a button that displays the Tornado SW version and configuration details in a list box. (I'm at home and I can't remember the button's name, sorry). If you apply the patch after installing Tornado then the first item in the list box will be words to the effect of "cumulative patch x86" or something similar. If it doesn't say this then you don't have the patch. And Tornado, compiler and all is bugged beyond use. Why didn't your WRS representative explain this to you ? I'd ask them, con brio. Regards, - -- Mike Davies --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Slow network performance with 5.3.1 Date: 3 Mar 1999 17:03:22 GMT From: jcl@apl.washington.edu (Jim Luby) Organization: APL-UW Message-ID: <7bjq0q$iqc$1@nntp6.u.washington.edu> References: <36DCA562.20AAB852@empireone.net> In article <36DCA562.20AAB852@empireone.net>, wrr@empireone.net says... > >This is a multi-part message in MIME format. >--------------58C10FF4C784BA8FF796C3AA >Content-Type: text/plain; charset=us-ascii >Content-Transfer-Encoding: 7bit > > >Hi All, > > I've got a VxWorks 5.3.1 box talking to a linux machine across > 100Base-T, via a crossover cable. Both the NICs are Intel > EtherExpress Pro 100+ cards (using fei). Both machines are > PIIs (233 and 266). With this link, I would expect to see > transfer rates on the order of 80-90 MBits/sec over a TCP socket. > Indeed, colleagues here have seen rates that high with 5.3.0 talking > to Solaris 2.5.1. I'm using netperf to do this benchmarking. > > Instead, I see rate somewhere around 20 MBits/sec! I've heard many > rumors concerning problems with the fei driver, and I don't intend to > use SENS (a major bottleneck I hear). I know the hardware is good, > as I've tested it in other machines (by this I mean NT). Furthermore, > I've swapped in the older EtherExpress Pro 100, with the same results. > Changing the socket size does not affect performance. Being relatively > new to VxWorks, I can only assume that it's a configuration issue, or > did something change between 5.3.0 and 5.3.1. Any help would be >greatly > appreciated. > > Bill Ross > GE Corporate R&D Bill, You might try increasing the the size of the socket transfer buffers (modify SO_SNDBUF option in call to function setsockopt()). I increased the size of this parameter and got substantial performance improvements for tcp transfer between my single board PPC system running vxworks 5.3.1 and a Sun system. I tested a range of power of two buffer sizes from 256 to 256K. For my system the performance peaked when the parameter was set to 32768 ON BOTH SYSTEMS. Good luck! Jim --------------------------- Newsgroups: comp.os.vxworks Subject: Re: ei82596End users? Date: Wed, 03 Mar 1999 15:36:05 -0500 From: Jeff Daly Organization: Sanders A Lockheed Martin Company Message-ID: <36DD9D35.F8262975@lmco.com> References: <36DC1FA5.37F3BEAD@lmco.com> Reply-To: jeffrey.j.daly@lmco.com This is a multi-part message in MIME format. - --------------DDC2B7356A629411F5124172 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit doh! answering my own message: the documentation says the loadstring should be defined as unit:ivec:sysbus:membase:ntfds:nrfds when it should be: (my original post was incorrect) ivec:sysbus:membase:ntfds:nrfds apparantly the unit is prepended somewhere along the way. also, the ntfds:nrfds field(s) should be decimal because the ParseInit() function in the driver does an atoi() on them, not strtoul() as the documentation says. so now i have the driver being installed correctly and all, but i get an address error (load) exception (i'm using MIPS btw) in the ipintr() function. anyone use this interface specifically with MIPS? i've downloaded and installed both the SENS 1.1 patch and the Windview 1.0.1 patch (not currently using Windview, tho) any help is appreciated Jeff Daly wrote: > anyone have success using the ei82596End interface? i'm having a little > trouble. drop me a line. > btw, i've figured out that the SENS documentation for the ei82596EndLoad > function is incorrect. > it appears that the sysBus and memBase fields of the loadstring are > reversed. memBase comes > before. > > ------------------------------------------------------------------------ > > Jeff Daly > SW Guy > Sanders, A Lockheed Martin Co. > > Jeff Daly > SW Guy > Sanders, A Lockheed Martin Co. > Netscape Conference Address > Netscape Conference DLS Server > B-sharps > Additional Information: > Last Name Daly > First Name Jeff > Version 2.1 - --------------DDC2B7356A629411F5124172 Content-Type: text/x-vcard; charset=us-ascii; name="vcard.vcf" Content-Transfer-Encoding: 7bit Content-Description: Card for Jeff Daly Content-Disposition: attachment; filename="vcard.vcf" begin: vcard fn: Jeff Daly n: Daly;Jeff org: Sanders, A Lockheed Martin Co. email;internet: jeffrey.j.daly@lmco.com title: SW Guy note: B-sharps x-mozilla-cpt: ;0 x-mozilla-html: FALSE version: 2.1 end: vcard - --------------DDC2B7356A629411F5124172-- --------------------------- Newsgroups: comp.os.vxworks Subject: Announcement: malloc/free replacement for VxWorks Date: Wed, 03 Mar 1999 22:14:41 GMT From: Hwa Jin Bae @ PSO Systems Inc Organization: Deja News - The Leader in Internet Discussion Message-ID: <7bkc87$t2e$1@nnrp1.dejanews.com> Hi! I am not sure if this got through. My news server has been flakey lately. I have done a port of two memory allocator libraries that can be used to replace VxWorks original memory allocator routines (malloc, free, etc.). There are two algorithms: BSD style one and Doug Lea's algorithm. You can choose to use either one via an ifdef. I use Doug Lea's allocator which has good fragmentation avoidance behavior and is also efficient and fast. The source code is at http://www.pso.com/vxhacks.html Hwa Jin Bae PSO Systems Inc http://www.pso.com mailto:hjb@pso.com - -----------== Posted via Deja News, The Discussion Network ==---------- http://www.dejanews.com/ Search, Read, Discuss, or Start Your Own --------------------------- Newsgroups: comp.os.vxworks Subject: Re: a question about vx-work Date: Sun, 07 Mar 1999 19:47:01 GMT From: kam@plex.nl (Gerald van Kampen) Organization: Plex -- a public access Internet provider Message-ID: <920835918.408180@WHIRLwind.plex.nl> References: <36DF2D5D.47A1@pegasus.cc.ucf.edu> In article <36DF2D5D.47A1@pegasus.cc.ucf.edu>, Lei Sun wrote: >Hi: > I am a new user in vx-work and tornado development tools. >I have configured Tornado properly and written a small program.It simply >create 10 tasks and print out the task ID individually. However I can't >get the expected result. It only print out the main function >message,say"main program begain..." and then program exit normally. >anybody can tell me the reason ? > By the way, I have redirect the I/O and the printed message should >appears in Virtual console, I am using VME167 as BSP and host machine is >sparc station. >attatched is my code.(very simple) > >#include "vxWorks.h" >#include "taskLib.h" >#include "stdio.h" > >#define TASKNB 10 > >void a_task(void); > > >/*************************/ > >void multiTask(void) >{ > int i; > printf("Task begain....\n"); >// for (i=0;i// { > if >((taskSpawn("a_task",0x100,VX_NO_STACK_FILL,200,(FUNCPTR)a_task,0,0,0,0,0,0,0,0 >,0,0))==ERROR) > printf("Error.Can't creat task %d\n",i); >// } > taskDelay(5); > printf("MAIN Task endn....\n"); >} >/*************************/ >void a_task(void) >{ > int i; > printf("Task Id=%d\n",taskIdSelf()); >// for (i=1;i<50000;i++) >// { >// printf("looping...\n"); >// } >// exit(0); > >} When using printf() in a task I think you really need more than 200 bytes of stack space. Try checkStack() or your information browser to see if you have a stack overflow situation. If so this results in unpredictable results. Gerald. --------------------------- Newsgroups: comp.arch.embedded,comp.os.vxworks,comp.realtime Subject: Re: compression source code Date: Sun, 28 Feb 1999 18:58:38 +0200 From: Zvika Perry Organization: NetVision Israel Message-ID: <36D975BD.B69BADD1@lucent.com> References: <79s8hi$3r2$1@news.netway.at> mash wrote: > Does anyone has a source code for ASCII Data compression or decompression ?? > > Thanx for any help > mash As we had to compress my application due to FLASH memory constraints and decompress it on the target, we have a gzip code modified to work over memory rather than a file system. It seems bzip does it better - about 10% in my application and I'm currently working on it. If this can be a help, I can send it (the gzip) by e-mail. zvperry@lucent.com --------------------------- Newsgroups: comp.os.vxworks Subject: Re: ada95 and vxworks Date: Sun, 07 Mar 1999 20:59:11 GMT From: deischen@my-dejanews.com Organization: Deja News - The Leader in Internet Discussion Message-ID: <7bupas$h4b$1@nnrp1.dejanews.com> References: In article , "Michael Garrett" wrote: > I would like to hear from someone who has ada95 / vxworks experience. Gnat > or Aonix. > We may use it in our next product and would like some feedback. > > Michael Garrett We use GNAT for VxWorks m68k. It uses the same GNU toolset and environment as shipped with VxWorks (gcc, g++, gdb, etc). We've had good luck with the GNAT compiler, so much so that we dropped the Rational Apex Ada system for GNAT. Our developers are much happier now :-) GNAT is freely available, and with a little work you can build your own GNAT/VxWorks cross-compiler. See http://www.unm.edu/~krisinsk/gnu/ for more info. I recommend puchasing GNATWorks, the professional edition, in order to get support and the most recent version of the GNAT/VxWorks cross-compiler. Dan Eischen - -----------== Posted via Deja News, The Discussion Network ==---------- http://www.dejanews.com/ Search, Read, Discuss, or Start Your Own --------------------------- Newsgroups: comp.os.vxworks Subject: Re: VxWorks memory partitions Date: 28 Feb 1999 17:15:31 GMT From: Hwa-Jin Bae Organization: PSO Systems Inc (http://www.pso.com) Message-ID: <7bbtjj$rmt$1@news.idiom.com> References: <199902190926.JAA08729@dsl-2.tadpole.co.uk> <36CD5189.28EB49ED@bitband.com> <36CDB722.1E7B6549@pso.com> <36D8F319.DCA2B5CB@bitband.com> Leonid Rosenboim wrote: > Just a brief follow-up: [snip... something about memPartAlloc hack to overcome memory problem] Yes, it is a well known hack to avoid the memory fragmentation problem. Many people use this hack. However, it is hardly a clean solution. You can still fragment the partition that you have created, and the threshold of bad fragmentation behavior will persist. There are better solutions. - -- Hwa-Jin Bae PSO Systems, Inc. http://www.pso.com mailto:hjb@pso.com --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Anyone has some examples for ramDrv? Date: Sun, 07 Mar 1999 23:44:29 GMT From: johnchann@home.net (johnchann) Organization: @Home Network Message-ID: References: <7bneag$r0e$1@news2.kornet.net> In article <7bneag$r0e$1@news2.kornet.net>, "À±°­½Ä" wrote: >Hi, Kang-Sik Yoon....in Korea. > >If there is anyone who has some examples for ramDrv, please let me have >them... > >Thanks in advance.... > > This is a very very simple thing to use. The fist function creates a RAM drive and puts a DOS file system on it. The second logs some printout to it as an example. JRC #include #include #include #include #include #include BLK_DEV *pBlkDev; /* Pointer to the block device structure */ DOS_VOL_DESC *pDosVolDesc; /* Pointer to the DOS volume descriptor */ int logFd; /* fd to log to */ /* Make the RAM drive we will use later */ int makeWwwDev ( void ) { /* Make a ram drive */ if ( (int)( pBlkDev = ramDevCreate ( 0, /* malloc the memory required */ 1024, /* bytes per block */ 2048, /* Blocks per track */ 2048, /* Blocks per device */ 0 /* Blocks to skip at start */ ) ) == ERROR ) { printf("RAM device create failed\n\n"); return (ERROR); } printf("Made ramDrive block device, pBlkDev=$%08X\n",(unsigned)pBlkDev); /* Make the dos file system on the new disk */ if ( (int)( pDosVolDesc = dosFsMkfs ( "/www", /* volume name to use */ pBlkDev /* pointer to block device struct */ ) ) == ERROR ) { printf("DOS device create failed\n\n"); return (ERROR); } printf("Made DOS file system, pDosVolDesc=$%08X\n\n",(unsigned)pDosVolDesc); return (OK); } int logTest ( void) { FILE *inFile; /* File pointer for recovering data */ char inStr[132]; /* String to put it in */ int lineNum = 0; /* Line during print */ logFd = open("/www/gugus.dat", 0x0201, 0644); /* Open file read/write/create */ printf("Log file descriptor is %d\n", logFd); /* Tell user what we'll use */ printf("Starting log of stdout\n"); ioTaskStdSet(0,1,logFd); /* Log stdout for this task */ ifShow("ei"); /* Do something that prints */ ioTaskStdSet(0,1,1); /* Stop logging for this task */ close(logFd); /* Close the log file */ printf("Log terminated and file closed\n\n\n"); printf("Re-opening log file and dumping...\n\n"); inFile = fopen("/www/gugus.dat","r"); /* Open the file to read */ while ( (int)fgets(inStr,131,inFile) != 0 ) /* While the read works */ { printf("inFile:%02d > %s",lineNum++,inStr); /* Print the line */ } fclose(inFile); /* Close the file pointer */ printf("Done!\n"); return (0); } --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Slow network performance with 5.3.1 Date: Thu, 04 Mar 1999 02:24:08 GMT From: netperf@my-dejanews.com Organization: Deja News - The Leader in Internet Discussion Message-ID: <7bkqs0$9ta$1@nnrp1.dejanews.com> References: <36DCA562.20AAB852@empireone.net> [Mime junk deleted...might want to disable the vcards in nettnews posts...] > Instead, I see rate somewhere around 20 MBits/sec! I've heard many > rumors concerning problems with the fei driver, and I don't intend to > use SENS (a major bottleneck I hear). I know the hardware is good, > as I've tested it in other machines (by this I mean NT). Furthermore, > I've swapped in the older EtherExpress Pro 100, with the same results. > Changing the socket size does not affect performance. Being relatively If you've changed the socket size on *both* ends, so you've set both -s and - -S of netperf, and the performance remains unchanged, you might want to look at the CPU utilization of your systems, and also the link and TCP stats. Look for TCP retransmissions and also look for link-level errors. You might also try to cut and paste the netperf output and command line(s) to your posts so we can see the actual socket sizes being used and such. rick jones http://www.netperf.org/ raj in the cup subdomain of the hp subdomain of com i use dejanews to find posts with certain keywords instead of subscribing to zillions of groups. if the post contains netperf, I will see it :) - -----------== Posted via Deja News, The Discussion Network ==---------- http://www.dejanews.com/ Search, Read, Discuss, or Start Your Own --------------------------- Newsgroups: comp.os.vxworks Subject: Netware NCP server source Date: Mon, 08 Mar 1999 04:01:50 GMT From: Jerry Marcinko Organization: @Home Network Message-ID: <36E34AE2.50687EFB@home.com> Hello all, I am looking for source code to a Netware NCP server which runs under VxWorks or I can port to VxWorks. I know about the mars_nwe server for Linux, but I am trying to find more than one so I can evaluate them side by side. Thanks - --- Jerry Marcinko Email: jmarcinko1@home.com --------------------------- Newsgroups: comp.os.vxworks Subject: Re: select() with Wind Msg Queues Date: Thu, 25 Feb 1999 14:33:56 GMT From: fritzensslin@my-dejanews.com Organization: Deja News - The Leader in Internet Discussion Message-ID: <7b3n0j$s6r$1@nnrp1.dejanews.com> References: <36d509a6.26060352@10.250.5.25> In article <36d509a6.26060352@10.250.5.25>, ak@igm-group.com (Andy Kriechhammer) wrote: > Is there a chance to implement a select() mechanism to wait on (two) > Wind message queues simultaneously without polling the state of the > queues permanently ? > I found some functions in selectLib, but it seems to me, that they > only work with filedescriptors (including pipes). That's right. No filedescriptor - no select. If your design requires one task listening on two queues, and you can afford a little overhead, you might consider adding another task, listening on one of the queues and forwarding the data to the other queue. In this case, you may need to add some kind of multiplexing indicator to the other queue's objects. > > Any help is appreciated > > Andy Kriechhammer, igm Austria > HTH, Friedrich Ensslin - ------------------------------------------------------- fritzensslin@my-dejanews.c - -----------== Posted via Deja News, The Discussion Network ==---------- http://www.dejanews.com/ Search, Read, Discuss, or Start Your Own --------------------------- Newsgroups: comp.os.vxworks Subject: Help. Shell Date: Mon, 08 Mar 1999 14:10:12 +0900 From: Aaron Jang Organization: Integra Telecom Co., Ltd. Seoul KOREA Message-ID: <36E35BB4.E85DB973@integra.co.kr> Help. Shell In my target board, the target shell always answer "undefined symbol:xxx". How can I fix it? I was added "#define INCLUDE_SHELL" to the configdb.h file. And the target showed to me the prompt "->", but commands are can't work. - -- Aaron Jang (aaron@integra.co.kr,"ÀåÀÎÁø") Integra Telecom Co., Ltd. http://www.integra.co.kr Hwayoung Bdg., 112-2, Garak-Dong, Tel : +82-2-449-6851 Songpa-Gu, Seoul, 138-160, Korea Fax : +82-2-431-3741 --------------------------- Newsgroups: comp.os.vxworks Subject: Re: about device driver Date: Mon, 08 Mar 1999 10:57:29 +0000 From: Phil Ashby Organization: BT Wireplay Message-ID: <36E3AD19.5F6D@wireplay.bt.co.uk> References: <36E098C2.4D26@pegasus.cc.ucf.edu> Lei Sun wrote: > the Vx-Work I/O system. But how do I install the driver into the vx-work > system? I am using MVE 167 computer and power up via the network, How > can I let the driver installed as soon as the system is power up. Assuming you do not want the driver available in the bootrom, you can: [1] Call the driver initialisation routines from usrRoot() [modify usrConfig.c], then rebuild the kernel. [2] Call the driver initialisation routines from a startup script. - -- !--------------------------- Phil "Phlash" Ashby ---------------------! ! BT Wireplay - Lead Programmer and general technical trivia supplier ! ! Phone: 01473-644348 WWW: http://www.wireplay.com/ ! ! Snail: B81/G41, BT Labs, Martlesham Heath, Ipswich IP5 3RE, England.! --------------------------- Newsgroups: comp.os.vxworks Subject: Re: flash file system Date: Mon, 08 Mar 1999 11:01:23 +0000 From: Phil Ashby Organization: BT Wireplay Message-ID: <36E3AE03.52EC@wireplay.bt.co.uk> References: <36E0EF6D.2162EB4E@tsoft.com> phoebeli@tsoft.com wrote: > Anyone has a shareware version of flash file system > for flash memory chips? You are unlikely to find a fully fledged FFS as shareware. My colleauge (Ian Love) wrote a flash programming library for MVME230x boards which can be found on DejaNews. This is a reasonable framework to start from.. Perhaps we should organise a GPL'ed FFS project? - -- !--------------------------- Phil "Phlash" Ashby ---------------------! ! BT Wireplay - Lead Programmer and general technical trivia supplier ! ! Phone: 01473-644348 WWW: http://www.wireplay.com/ ! ! Snail: B81/G41, BT Labs, Martlesham Heath, Ipswich IP5 3RE, England.! --------------------------- End of New-News digest ********************** From owner-vxwexplo-process Tue Mar 9 04:00:40 1999 From: daemon@csg.lbl.gov Date: Tue Mar 9 04:00:44 PST 1999 Subject: comp.os.vxworks newsdigest Status: O Comp.Os.Vxworks Daily Digest Tue Mar 9 04:00:36 PST 1999 Subject: Re: Help. Shell Subject: Using RSH ... Subject: Q: SENS buffer scheme Subject: Problem building kernel - Ultra604 Subject: Tcl 7.6/8.0.x for VxWorks 5.3 Subject: Re: Using RSH ... Subject: Re: Wind River' Tornado 1.0.1 for Windows NT is bugged Subject: Re: WindNet SNMP (was: Are multiple agent tasks possible) Subject: Re: Codewright Subject: Re: Wind River' Tornado 1.0.1 for Windows NT is bugged Subject: Anyone has some examples for ramDrv? Subject: Looking for x86 PCI bus BSP Subject: Wind River' Tornado 1.0.1 for Windows NT is bugged Subject: Re: ada95 and vxworks Subject: DHCP table configuration for BOOTP client Subject: Re: Performance impact of C++ Subject: Re: Performance impact of C++ Subject: WIND_TCB structure description Subject: Re: Tornado on Windows 98? Subject: New site! Subject: Help! - Interprocess comms problem using message queues ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: Help. Shell Date: Mon, 08 Mar 1999 21:57:15 +0900 From: Aaron Jang Organization: Integra Telecom Co., Ltd. Seoul KOREA Message-ID: <36E3C92B.4CDC16C6@integra.co.kr> References: <36E35BB4.E85DB973@integra.co.kr> <36E3AEC3.560@wireplay.bt.co.uk> Thank you for your interesting about my question Phil Ashby. But I got a new problem. The compiler display this... == vxWorks.tmp(.text+0x1ef6): undefined reference to `standTblSize' vxWorks.tmp(.text+0x1f0c): undefined reference to `standTblSize' vxWorks.tmp(.text+0x1f18): undefined reference to `standTbl' vxWorks.tmp(.text+0x1f30): undefined reference to `standTblSize' make: *** [vxWorks] Error 1 == Please help me. - -- Aaron Jang (aaron@integra.co.kr,"ÀåÀÎÁø") Integra Telecom Co., Ltd. http://www.integra.co.kr Hwayoung Bdg., 112-2, Garak-Dong, Tel : +82-2-449-6851 Songpa-Gu, Seoul, 138-160, Korea Fax : +82-2-431-3741 --------------------------- Newsgroups: comp.os.vxworks Subject: Using RSH ... Date: Mon, 08 Mar 1999 22:32:30 +0900 From: Aaron Jang Organization: Integra Telecom Co., Ltd. Seoul KOREA Message-ID: <36E3D16E.229BAFC4@integra.co.kr> Hello there. In my target(MC68360 custom board) the RSH functions are always display this message at the end of processing. And some functions are failed. == bright:stty: : Invalid argument == ..bright is my host(RSH server) name. Failed functions are related open() function. But rm(), ls(), cd(), etc... are passed. Please help me to overcome this problem. Thanks. - -- Aaron Jang (aaron@integra.co.kr,"ÀåÀÎÁø") Integra Telecom Co., Ltd. http://www.integra.co.kr Hwayoung Bdg., 112-2, Garak-Dong, Tel : +82-2-449-6851 Songpa-Gu, Seoul, 138-160, Korea Fax : +82-2-431-3741 --------------------------- Newsgroups: comp.os.vxworks Subject: Q: SENS buffer scheme Date: Mon, 08 Mar 1999 13:57:19 +0000 From: Jim Chapman Organization: Xstreamis plc Message-ID: <36E3D73F.9808CB3E@xstreamis.com> This is a multi-part message in MIME format. - --------------EC180FAE657558DD7235EBBD Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit SENS device drivers are passed M_BLKs to transmit. Can anyone tell me if the M_BLK always references only one (possibly multi-cluster) packet, or can the MUX send multiple packets to the driver in a single call to the END's transmit routine by chaining M_BLK_HDR's together using the mNextPkt field? typedef struct mHdr { struct mBlk * mNext; /* next buffer in chain */ struct mBlk * mNextPkt; /* next chain in queue/record */ char * mData; /* location of data */ int mLen; /* amount of data in this mBlk */ UCHAR mType; /* type of data in this mBlk */ UCHAR mFlags; /* flags; see below */ USHORT reserved; } M_BLK_HDR; I guess what I want to know is if END drivers need to do anything with mNextPkt. Thanks - -Jim - --------------EC180FAE657558DD7235EBBD Content-Type: text/x-vcard; charset=us-ascii; name="jim.chapman.vcf" Content-Transfer-Encoding: 7bit Content-Description: Card for Jim Chapman Content-Disposition: attachment; filename="jim.chapman.vcf" begin:vcard n:Chapman;Jim tel;fax:+44-1865-784004 tel;work:+44-1865-784569 x-mozilla-html:TRUE url:http://www.xstreamis.com org:Xstreamis plc;R&D version:2.1 email;internet:jim.chapman@xstreamis.com adr;quoted-printable:;;Magdalen Centre=0D=0AOxford Science Park=0D=0A;OXFORD;OXON;OX4 4GA;England x-mozilla-cpt:;-1 fn:Jim Chapman end:vcard - --------------EC180FAE657558DD7235EBBD-- --------------------------- Newsgroups: comp.os.vxworks Subject: Problem building kernel - Ultra604 Date: Thu, 04 Mar 1999 18:07:00 GMT From: matthieu.sevestre@lemel.Fr Organization: World Online France News server Message-ID: <36dec828.1890158@news.worldonline.fr> Hello, I recently had Tornado/VxWorks for a Motorola Ultra604 target (PPC 606). I wanted to connect the host to the target through the serial port #2 of my Ultra5 station with Solaris 2.6 so, I modified the config.h and tried to rebuild the 'vxWorks' file. The problem is that I have buiding error (the Tornado patch is installed). Here is the error: Neurix:make vxWorks ccppc -B/home/sevestre/wind/host/sun4-solaris2/lib/gcc-lib/ -c - -mstrict-align -ansi -nostdinc -O2 -fvolatile -fno-builtin - -fno-for-scope -Wall -I/h -I. - -I/home/sevestre/wind/target/config/all -I/home/sevestre/wind/target/h - -I/home/sevestre/wind/target/src/config - -I/home/sevestre/wind/target/src/drv -DCPU=PPC604 -DULTRA - -DTARGET_DIR="\"ultra604\"" /home/sevestre/wind/target/config/all/usrConfig.c -o usrConfig.o In file included from /home/sevestre/wind/target/src/config/usrExtra.c:208, from /home/sevestre/wind/target/config/all/usrConfig.c:107: /home/sevestre/wind/target/src/config/../../src/config/usrNetwork.c:262: syntax error before `mbufConfig' /home/sevestre/wind/target/src/config/../../src/config/usrNetwork.c:264: `NUM_INIT_MBUFS' undeclared here (not in a function) /home/sevestre/wind/target/src/config/../../src/config/usrNetwork.c:264: initializer element for `mbufConfig' is not constant /home/sevestre/wind/target/src/config/../../src/config/usrNetwork.c:264: `NUM_MBUFS_TO_EXPAND' undeclared here (not in a function) /home/sevestre/wind/target/src/config/../../src/config/usrNetwork.c:264: warning: excess elements in scalar initializer after `mbufConfig' /home/sevestre/wind/target/src/config/../../src/config/usrNetwork.c:264: `MAX_MBUFS' undeclared here (not in a function) /home/sevestre/wind/target/src/config/../../src/config/usrNetwork.c:264: warning: excess elements in scalar initializer after `mbufConfig' /home/sevestre/wind/target/src/config/../../src/config/usrNetwork.c:264: warning: excess elements in scalar initializer after `mbufConfig' /home/sevestre/wind/target/src/config/../../src/config/usrNetwork.c:265: warning: excess elements in scalar initializer after `mbufConfig' /home/sevestre/wind/target/src/config/../../src/config/usrNetwork.c:265: warning: data definition has no type or storage class /home/sevestre/wind/target/src/config/../../src/config/usrNetwork.c:267: syntax error before `clusterConfig' /home/sevestre/wind/target/src/config/../../src/config/usrNetwork.c:269: `NUM_INIT_CLUSTERS' undeclared here (not in a function) /home/sevestre/wind/target/src/config/../../src/config/usrNetwork.c:269: initializer element for `clusterConfig' is not constant /home/sevestre/wind/target/src/config/../../src/config/usrNetwork.c:269: `NUM_CLUSTERS_TO_EXPAND' undeclared here (not in a function) /home/sevestre/wind/target/src/config/../../src/config/usrNetwork.c:269: warning: excess elements in scalar initializer after `clusterConfig' /home/sevestre/wind/target/src/config/../../src/config/usrNetwork.c:269: `MAX_CLUSTERS' undeclared here (not in a function) /home/sevestre/wind/target/src/config/../../src/config/usrNetwork.c:269: warning: excess elements in scalar initializer after `clusterConfig' /home/sevestre/wind/target/src/config/../../src/config/usrNetwork.c:269: warning: excess elements in scalar initializer after `clusterConfig' /home/sevestre/wind/target/src/config/../../src/config/usrNetwork.c:270: warning: excess elements in scalar initializer after `clusterConfig' /home/sevestre/wind/target/src/config/../../src/config/usrNetwork.c:270: warning: data definition has no type or storage class /home/sevestre/wind/target/src/config/../../src/config/usrNetwork.c: In function `usrNetInit': /home/sevestre/wind/target/src/config/../../src/config/usrNetwork.c:522: warning: passing arg 2 of `usrPPPInit' makes integer from pointer without a cast /home/sevestre/wind/target/src/config/../../src/config/usrNetwork.c:522: too few arguments to function `usrPPPInit' /home/sevestre/wind/target/src/config/../../src/config/usrNetwork.c:530: warning: passing arg 2 of `usrSlipInit' makes integer from pointer without a cast /home/sevestre/wind/target/src/config/../../src/config/usrNetwork.c:530: too few arguments to function `usrSlipInit' /home/sevestre/wind/target/src/config/../../src/config/usrNetwork.c:543: too few arguments to function `usrBpInit' /home/sevestre/wind/target/src/config/../../src/config/usrNetwork.c:546: warning: passing arg 2 of `usrNetIfAttach' makes integer from pointer without a cast /home/sevestre/wind/target/src/config/../../src/config/usrNetwork.c:546: too few arguments to function `usrNetIfAttach' /home/sevestre/wind/target/src/config/../../src/config/usrNetwork.c:548: warning: passing arg 2 of `usrNetIfConfig' makes integer from pointer without a cast /home/sevestre/wind/target/src/config/../../src/config/usrNetwork.c:548: warning: passing arg 4 of `usrNetIfConfig' makes pointer from integer without a cast /home/sevestre/wind/target/src/config/../../src/config/usrNetwork.c:548: too few arguments to function `usrNetIfConfig' /home/sevestre/wind/target/src/config/../../src/config/usrNetwork.c:566: warning: passing arg 2 of `usrNetIfAttach' makes integer from pointer without a cast /home/sevestre/wind/target/src/config/../../src/config/usrNetwork.c:566: too few arguments to function `usrNetIfAttach' /home/sevestre/wind/target/src/config/../../src/config/usrNetwork.c:568: warning: passing arg 2 of `usrNetIfConfig' makes integer from pointer without a cast /home/sevestre/wind/target/src/config/../../src/config/usrNetwork.c:568: warning: passing arg 4 of `usrNetIfConfig' makes pointer from integer without a cast /home/sevestre/wind/target/src/config/../../src/config/usrNetwork.c:568: too few arguments to function `usrNetIfConfig' /home/sevestre/wind/target/src/config/../../src/config/usrNetwork.c:578: warning: passing arg 2 of `usrNetIfAttach' makes integer from pointer without a cast /home/sevestre/wind/target/src/config/../../src/config/usrNetwork.c:578: too few arguments to function `usrNetIfAttach' /home/sevestre/wind/target/src/config/../../src/config/usrNetwork.c:580: warning: passing arg 2 of `usrNetIfConfig' makes integer from pointer without a cast /home/sevestre/wind/target/src/config/../../src/config/usrNetwork.c:580: warning: passing arg 4 of `usrNetIfConfig' makes pointer from integer without a cast /home/sevestre/wind/target/src/config/../../src/config/usrNetwork.c:580: too few arguments to function `usrNetIfConfig' /home/sevestre/wind/target/src/config/../../src/config/usrNetwork.c:744: warning: passing arg 2 of `usrNetIfAttach' makes integer from pointer without a cast /home/sevestre/wind/target/src/config/../../src/config/usrNetwork.c:744: too few arguments to function `usrNetIfAttach' /home/sevestre/wind/target/src/config/../../src/config/usrNetwork.c:745: warning: passing arg 2 of `usrNetIfConfig' makes integer from pointer without a cast /home/sevestre/wind/target/src/config/../../src/config/usrNetwork.c:745: too few arguments to function `usrNetIfConfig' /home/sevestre/wind/target/src/config/../../src/config/usrNetwork.c: At top level: /home/sevestre/wind/target/src/config/../../src/config/usrNetwork.c:1057: conflicting types for `usrNetIfAttach' /home/sevestre/wind/target/h/usrConfig.h:282: previous declaration of `usrNetIfAttach' /home/sevestre/wind/target/src/config/../../src/config/usrNetwork.c:1154: conflicting types for `usrNetIfConfig' /home/sevestre/wind/target/h/usrConfig.h:292: previous declaration of `usrNetIfConfig' /home/sevestre/wind/target/src/config/../../src/config/usrNetwork.c:1214: conflicting types for `usrBpInit' /home/sevestre/wind/target/h/usrConfig.h:280: previous declaration of `usrBpInit' /home/sevestre/wind/target/src/config/../../src/config/usrNetwork.c:1321: conflicting types for `usrSlipInit' /home/sevestre/wind/target/h/usrConfig.h:284: previous declaration of `usrSlipInit' /home/sevestre/wind/target/src/config/../../src/config/usrNetwork.c:1397: conflicting types for `usrPPPInit' /home/sevestre/wind/target/h/usrConfig.h:286: previous declaration of `usrPPPInit' make: *** [usrConfig.o] Error 1 Neurix: Is somebody able to help me solving the problem ? Thank you . Matthieu SEVESTRE - France --------------------------- Newsgroups: comp.os.vxworks,comp.lang.tcl Subject: Tcl 7.6/8.0.x for VxWorks 5.3 Date: Mon, 08 Mar 1999 16:58:09 +0100 From: Hans van der Meij Organization: European Space Agency Message-ID: <36E3F391.F927813D@estec.esa.nl> Hi there, Can anyone give me a pointer to VxWorks 5.3 ports of Tcl 7.6 or higher? Thanks in advance, Hans --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Using RSH ... Date: Mon, 08 Mar 1999 09:07:46 -0800 From: Scott Kamara Organization: AP Labs Message-ID: <36E403E2.72F7317C@pop.sttl.uswest.net> References: <36E3D16E.229BAFC4@integra.co.kr> Reply-To: spk@uswest.net On bootup, when the kernel is being loaded, your .cshrc is being executed by the rsh daemon (rshd) on your host and the stdout,stderr is being directed into vxWorks target memory and thus causing corruption. This could be the cause if your original symbol table problems as well. To fix this, modify the the very top of your .cshrc file in your host's home directory to contain: # # Skip remaining setup if not # an interactive shell. # (note: this is 'csh' syntax) # if ($?term == 0) exit Aaron Jang wrote: > > Hello there. > > In my target(MC68360 custom board) the RSH functions are > always display this message at the end of processing. > And some functions are failed. > > == > bright:stty: : Invalid argument > == > ..bright is my host(RSH server) name. > > Failed functions are related open() function. > But rm(), ls(), cd(), etc... are passed. > > Please help me to overcome this problem. > Thanks. > > -- > > Aaron Jang (aaron@integra.co.kr,"ÀåÀÎÁø") > > Integra Telecom Co., Ltd. http://www.integra.co.kr > Hwayoung Bdg., 112-2, Garak-Dong, Tel : +82-2-449-6851 > Songpa-Gu, Seoul, 138-160, Korea Fax : +82-2-431-3741 - -- - ----------------------- Scott Kamara AP Data Systems Seattle Engr. Office (206) 525-8592 scott@aplabs.com --------------------------- Newsgroups: comp.arch.embedded,comp.os.vxworks Subject: Re: Wind River' Tornado 1.0.1 for Windows NT is bugged Date: Mon, 8 Mar 1999 17:08:39 -0000 From: "Darren J Longhorn" Organization: Customer of Planet Online Message-ID: <7c108v$f5$1@newsreader2.core.theplanet.net> References: <7bn25a$jeu$1@nslave1.tin.it> <7sC$fQAoq+32Ew9$@noco.demon.co.uk><7bs1vp$52u$1@newsreader1.core.theplanet.net> We definately have nothing like this! Does it only apply when your target is x86? We are using NT as host but our target is sh3. Our experience of WindRiver support has been laughable. We usually fix problems that we report before they do! The only thing they fixed was when the clocks went back and the license server stopped 'serving licenses'. Mike Davies wrote in message ... >In article <7bs1vp$52u$1@newsreader1.core.theplanet.net>, Darren J >Longhorn writes >>Hmm, I don't see that on mine. Where is it exactly? >> >>Mike Davies wrote in message <7sC$fQAoq+32Ew9$@noco.demon.co.uk>... >>>Make sure that you have the patch installed : when you >>>start the Tornado environment pull down the help dialog >>>box and look at the version details. There should be an >>>entry there for something like "cumulative patch x86". >>>Tornado is not really usable without it so be warned. >>> >>> >>> >>>-- >>>Mike Davies > >OK, on Tornado 1.01 targeted for the 80x86 there is a standard >looking help menu with an "about" item that displays a dialog >box. This dialog box has a button that displays the Tornado >SW version and configuration details in a list box. (I'm at >home and I can't remember the button's name, sorry). > >If you apply the patch after installing Tornado then the first >item in the list box will be words to the effect of "cumulative >patch x86" or something similar. > >If it doesn't say this then you don't have the patch. And >Tornado, compiler and all is bugged beyond use. > >Why didn't your WRS representative explain this to you ? I'd >ask them, con brio. > >Regards, > > >-- >Mike Davies --------------------------- Newsgroups: comp.os.vxworks Subject: Re: WindNet SNMP (was: Are multiple agent tasks possible) Date: Mon, 08 Mar 1999 10:02:09 -0700 From: Mike Stimpson Organization: Utah Scientific, Inc. Message-ID: <36E40291.2781@utsci.com> References: <01be6516$d6b7fae0$2a9cd126@salmon> <36E0942A.BF294AF2@hotmail.com> T. Venkata Ramana wrote: > > Snmp agent uses UDP protocol and listens for requests on port > number 161. A server can bind to only one port. You > cannot run multiple agents on the same target system. I cant > think of a reason as to why you need multiple snmp agents? > > Venkata > > Tom Salmon wrote: > > > Hello everyone, > > > > I am using WindNet SNMPv1/v2c to create an agent for a > > 68K Tornado 5.3 system. By default the WindNet product seems to create > > a single tSnmp task. Does anyone know how to create more than > > one of these SNMP agent tasks? Two questions: - - Is WindNet SNMPv1/v2c solid? Is there another SNMP implementation that is more solid? Are there implementations that should be avoided? - - How much space (code and memory) does SNMP consume? Please E-mail replies as well, as my news feed is quite unreliable. Thanks! - -- Mike Stimpson (mstimp@utsci.com) Opinions expressed are not necessarily those of the management... The more I work as an engineer, the more I disbelieve the theory of evolution. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Codewright Date: Mon, 08 Mar 1999 19:19:02 GMT From: HJohnson@ZoneAutomation.Com (Hugh Johnson) Organization: Zone Automation Message-ID: References: <36E04F5E.CFF9F278@lucent.com> In article <36E04F5E.CFF9F278@lucent.com>, Douglas Fraser wrote: >Any one on this list using Codewright in conjunction with >Tornado? I know this has got to be as bad as asking >"which is better, emacs or vi" in a UNIX forum, but I >am looking for opinions and actual usage. Someone here >is very enthusiastic from prior non-Tornado experience. >I am looking for a little help before I order a dozen >copies and get myself in up to my eyeballs. > >TIA > >Doug Fraser >dwfraser@lucent.com Yes, we are doign so here at Zone Automation. If you need any help with setup of CodeWright with the WR tools, let me know. - -- HJohnson@ZoneAutomtion.Com --------------------------- Newsgroups: comp.arch.embedded,comp.os.vxworks Subject: Re: Wind River' Tornado 1.0.1 for Windows NT is bugged Date: Mon, 8 Mar 1999 20:45:43 +0000 From: Mike Davies Message-ID: References: <7bn25a$jeu$1@nslave1.tin.it> <7sC$fQAoq+32Ew9$@noco.demon.co.uk> <7bs1vp$52u$1@newsreader1.core.theplanet.net> <7c108v$f5$1@newsreader2.core.theplanet.net> In article <7c108v$f5$1@newsreader2.core.theplanet.net>, Darren J Longhorn writes >We definately have nothing like this! >Does it only apply when your target is x86? Sorry, I don't know - I've only used the x86 target. >We are using NT as host but our target is sh3. > >Our experience of WindRiver support has been laughable. >We usually fix problems that we report before they do! >The only thing they fixed was when the clocks went back and the license >server stopped 'serving licenses'. That's them OK :-) Get the licences sorted and who gives a fuck about the product ? I made a caustic remark about the quality of WRS's support to an experienced colleague when I first started using VxWorks , his jaundiced reply was : "You mean they didn't get back to you to tell you they didn't know how to fix it ?" This an exact description of WRS support. What are they playing at ? When they next answer the phone say the word "sourceware" then carry on without making any other reference to it |-) Good Luck, - -- Mike Davies --------------------------- Newsgroups: comp.os.vxworks Subject: Anyone has some examples for ramDrv? Date: Fri, 5 Mar 1999 11:04:01 +0900 From: "À±°­½Ä" Organization: Korea Telecom Message-ID: <7bneag$r0e$1@news2.kornet.net> Hi, Kang-Sik Yoon....in Korea. If there is anyone who has some examples for ramDrv, please let me have them... Thanks in advance.... --------------------------- Newsgroups: comp.os.vxworks Subject: Looking for x86 PCI bus BSP Date: Mon, 08 Mar 1999 16:32:28 -0500 From: colbyp Organization: Raytheon Systems Company Message-ID: <36E441EC.9912AEC7@res.raytheon.com> I'm looking for a vendor of a good generic PC BSP that can handle a motherboard with a PCI bus. Is one of the Wind River generic PC/AT BSP modules OK? We want to target our initial development toward a PC with PCI based peripherals and eventually move to a CompactPCI based x86 SBC. Thanks pcolby@res.raytheon.com --------------------------- Newsgroups: comp.arch.embedded,comp.os.vxworks Subject: Wind River' Tornado 1.0.1 for Windows NT is bugged Date: Thu, 4 Mar 1999 23:39:30 +0100 From: "Wolfy" Organization: TIN Message-ID: <7bn25a$jeu$1@nslave1.tin.it> I installed Tornado 1.0.1 ( 80x86) for Windows on 3 PC: 2 NT 4.0 Workstation and a Win98 laptop; It is about 10 days I am trying to make them work but I am only experimentig that one installation on WinNT4.0 is working at 60%, another one on WinNT 4.0 too ( with SP3 ) is absolutly not working. ( installed from the same CD on an twin machine ( the only difference is that one machine has 2 NTFS partitions while the other one has 1 FAT16 and 1 NTFS ). The Win98 version is at the moment the only one that let me compile vxWorks kernel, Yesterday I installed the licence manager 3 times on an NT4.0 Server and, at last, after 2 reboots has started working. At the moment the licence manager is THE ONLY THING THAT WORKS PROPERLY .... I have the feeling ( and something more really ) that this development system is a true "banana". I only hope that vxWorks operating system may work much better than its development system. Does anyone experimented a similar case ? Wolfy --------------------------- Newsgroups: comp.os.vxworks Subject: Re: ada95 and vxworks Date: Tue, 09 Mar 1999 00:13:24 GMT From: deischen@my-dejanews.com Organization: Deja News - The Leader in Internet Discussion Message-ID: <7c1p2t$2m7$1@nnrp1.dejanews.com> References: <7bupas$h4b$1@nnrp1.dejanews.com> <#JxqksRa#GA.342@nih2naaa.prod2.compuserve.com> In article <#JxqksRa#GA.342@nih2naaa.prod2.compuserve.com>, "Michael Garrett" wrote: > > >We use GNAT for VxWorks m68k. It uses the same GNU toolset and environment > >as shipped with VxWorks (gcc, g++, gdb, etc). > > Is the GDB Debugger that ships with VxWorks fully Kernel Aware at the Ada > layer?? Or is there an Gnat version that is? All Ada tasks in GNAT (and I would expect in other Ada->VxWorks compilers) are implemented as VxWorks tasks. The modified GDB from ACT is ada-aware, and should work the same way as if you were developing tasks written in C/C++ for VxWorks. > I have to do some investigating of Wind River Tools. Is there task profiling > available for the Ada Versions? Again, whatever tools you use for profiling VxWorks tasks written in C/C++ should work the same with Ada tasks. I guess you're talking about WindView? We haven't needed to use that yet, but it should work regardless of whether your tasks are written in C, C++, or Ada. Dan Eischen - -----------== Posted via Deja News, The Discussion Network ==---------- http://www.dejanews.com/ Search, Read, Discuss, or Start Your Own --------------------------- Newsgroups: comp.protocols.tcp-ip,comp.os.vxworks Subject: DHCP table configuration for BOOTP client Date: Thu, 25 Feb 1999 17:52:02 +0100 From: Carsten Boeckmann Organization: Ericsson Eurolab Deutschland GmbH Message-ID: <36D57FB1.13410D7D@ericsson.com> Reply-To: eedcabo@eede.ericsson.se Hi, we 've setup a DHCP server, which is able to reply to requests of BOOTP clients (in this case a Motorola board running VxWorks). Our problem is, how to configure the DHCP table, so that the BOOTP client's hostname is sent with the reply. Our actual DHCP table entry looks like this: name m :Include=net:\ :UTCoffst=<..>:\ :TFTPsrvN=:\ :BootFile=:\ :BootSrvA=:\ :BootSrvN=:\ :Hostname: (Specific fields are replaced with an <..> entry) With this we currently get all information (IP addresses of client and host, name of boot file, name of boot server) via BOOTP requests. Router and domain can be extracted from the bootp's vendor field also. The client's hostname is stored in a DNS database, so why we don't get the BOOTP client's hostname here ? Thanks in advance, Carsten --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Performance impact of C++ Date: Mon, 08 Mar 1999 21:12:34 -0500 From: Mark Clayton Organization: Shore.Net/Eco Software, Inc; (info@shore.net) Message-ID: <36E48392.DCE2B62F@shore.net> References: Bill Pringlemeir wrote: > Many CPUs support the Green Hills compiler. Green Hills provide PJ Plaugers I would recommend you NOT use green hills unless you're working on a PowerPC. I use gh on an SH-1 and it's terrible. Tech support is no help as well. Mark --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Performance impact of C++ Date: 8 Mar 1999 18:44:22 -0800 From: gillies@cs.ubc.ca (Donald Gillies) Organization: Computer Science, University of British Columbia, Canada Message-ID: <7c21u6$2fc$1@cascade.cs.ubc.ca> References: <36E48392.DCE2B62F@shore.net> Mark Clayton writes: >Bill Pringlemeir wrote: >> Many CPUs support the Green Hills compiler. Green Hills provide PJ Plaugers >I would recommend you NOT use green hills unless you're working on a PowerPC. >I use gh on an SH-1 and it's terrible. Tech support is no help as well. >Mark What makes you think that things are any better with Green Hills on the PowerPC ??? - - Don --------------------------- Newsgroups: comp.os.vxworks Subject: WIND_TCB structure description Date: Tue, 09 Mar 1999 09:37:04 +0200 From: Alex Shteinfeld Organization: ECI Telecom Message-ID: <36E4CF9F.E5D497A5@ecitele.com> Hello VxWorkers! I am looking for the complete description of WIND_TCB structure (in addition to the comments in the taskLib.h file). Actually I want to write something like Unix "ps" command and I'm trying to understand how to retreive such information from the TCB. Thanks in advance Alex Shteinfeld e-mail: shtein@ecitele.com --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Tornado on Windows 98? Date: Tue, 09 Mar 1999 08:08:42 GMT From: johnchann@home.net (johnchann) Organization: @Home Network Message-ID: References: <36de6f42.0@news.barak.net.il> <36DF34CD.977D253C@e-farm.com> <36E04F5E.CFF9F278@lucent.com> <36E051AB.530BBDB@nortel.ca> On the one hand I just wouldn't recomend 95 or 98 for anything except games, on the other hand NT certainly works much better. Solaris works much better. What in particular? It's less reliable and just doesn't work as well when it isn't crashing. JRC In article , John or Jenn wrote: >In article <7bm2hc$t4n$1@newsreader2.core.theplanet.net> John Cresswell, >john.cresswell@pace.co.uk writes: >>I would not recommend Tornado 1.0.1 on Windows 98 but Torando 2.0 might >>support it. We use NT running on both desktops & laptops with out problems. >> >Ummm...why not? > >Have you tried it and run into problems? >Just superstitious? > >(or, most likely, just woldn¹t recommend Win 98 for anything? ;-) > >John --------------------------- Newsgroups: comp.os.vxworks Subject: New site! Date: 9 Mar 1999 13:12:25 +0300 From: nobody@zdes.net Message-ID: <36e4f409.0@news.telekom.ru> New site! Welcome! http://195.58.59.2/go.htm Look and tell me what you think --------------------------- Newsgroups: comp.realtime,comp.os.vxworks Subject: Help! - Interprocess comms problem using message queues Date: Tue, 09 Mar 1999 11:08:12 +0000 From: Simon Roberts Message-ID: <36E5011C.D426C2A6@roke.co.uk> *Any* help/suggestions which might help us to solve the following problem would be much appreciated! Thanks! Here is a summary of the problem: Target: 5 x Motorola MVME 2306 PowerPC cards (300MHz PPC604, 32 MB RAM). OS: VxWorks 5.3.1 System: 20 or so tasks on each processor, interprocessor communication using shared memory message queues (VxMP). Estimated 140+ messages / second passing through shared memory. We have a random problem which manifests itself on long runs (many hours). After the problem occurs we find that one of the slave processors has exceeded the spin-lock re-try limit to access a shared memory resource (i.e. a message Q). The VxWorks manual states that after such an event, VxMP is not guaranteed to work properly until the system is reset. We suspect that this comes about owing to the master processor failing to service a shared memory request for some reason. Thanks for your time! Please reply to: matthew.jones@roke.co.uk or simon.roberts@roke.co.uk --------------------------- End of New-News digest ********************** From owner-vxwexplo-process Tue Mar 9 07:10:41 1999 From: nmehrotra@lucent.com Date: Tue Mar 9 07:10:45 PST 1999 Subject: vxWorks 5.2 DOS FS structures corruptions Status: O Hi folks, We run vxWorks 5.2 on a MIPS platforms and we have a SCSI disk for code/data storage. We're encountering a problem here where some application code of ours seems to corrupting the dos FS structures in memory and this eventually causes the hard disk to become corrupted when these structures are written to disk. The problem initially shows up in a strange way. Issue an "ls" from the shell in a directory where there are files and it comes up empty. Do "ls" again and it shows the directory contents. Do "ll" and it gives an "IO Error - can't stat file S_dosFsLib_FILE_NOT_FOUND" error message. Do it again and it works. Change directory and do "ls" and it shows the contents of the previous directory. Has anyone heard of/or seen this problem before? I'd like to figure out what is the location and pattern of corruption to try and isolate the offending code. Also, the MIPS BSP doesn't support memory protection; any ideas/suggestions on how we could go about trying to try and detect the corrupting code/task? thanks, nitin nmehrotra@lucent.com From owner-vxwexplo-process Tue Mar 9 07:58:47 1999 From: Rainer Stelzer Date: Tue Mar 9 07:58:51 PST 1999 Subject: Fwd: Board-Detection in VMEbus-Systems Status: O > Hi, > > I am trying to implement an automatic board detection in a VMEbus-System, > > as i did very often before in 68k assembler. This is my algorithm: > > 1.) Install a Buserror-Exception handler, which only increments an > error counter "ErrCnt" > 2.) Set ErrCnt=0 > 3.) Make a single access to the hardware you are testing for. > 4.) Check ErrCnt: > If it is zero then the board is present. > If it is not zero then there is no board. > 5.) Restore the old Buserror-Exception handler. > > In old assembler times there was no problem, once the special stack > format was considered. > But I don't know how to do it with VxWorks in a simple, clean and > reliable manner, because there seems to be no possibility to register > my own Buserror-Exception routine for the short time of board detection > only: > * intVecConnect(...) is not suitable Buserror-Exceptions > * execAddHook(...) executes normal exception handling first and > afterwards my handler. But the normal exception handling stops my task > which is testing for the board. > > So how can I implement my algorithm or something doing the same with > vxWorks? > Any hint or sample code is appreciated. > > Environment: > VMEbus-System > a single PEP VSBC32 CPU (68360 processor) > serveral passive Slave Boards > VxWorks/Tornado NT4.0 > Language C > > Thanks for your help. > > -- > > =================================================================== > Fraunhofer Institut fuer Mikroelektronische Schaltungen und Systeme > > FHG IMS Telefon: +49 (0) 203 / 3783-0 > Dipl.-Ing. G. Koenigsmann Durchwahl -229 > Abt. SYS1 Fax : +49 (0) 203 / 3783-266 > Finkenstrasse 61 Telex : 855 793 uni du d > D-47057 Duisburg Email : gerhard@ims.fhg.de > =================================================================== From owner-vxwexplo-process Tue Mar 9 09:14:20 1999 From: Martin Schrape Date: Tue Mar 9 09:14:25 PST 1999 Subject: Re: Job Opportunities Status: O Does anyone knows Hamids correct email address. hamid@SLAC.Stanford.edu does not work. Martin Hamid Shoaee wrote: > > We have several immediate openings for senior software engineers at the > Stanford Linear Accelerator Center (SLAC), a high energy physics research > facility operated by Stanford University. This is an opportunity to join a > scientific software development group supporting the operation of a particle > accelerator facility as well as conducting R&D for a next generation > accelerator control system. You will perform system and software development > for real-time computer systems and networks, utilizing your knowledge of > distributed systems and networking. Knowledge of and experience with UNIX > and/or NT operating systems, real-time executives (VxWorks, etc.) C/C++ and > control systems or real-time systems is also essential. > The positions require a BSCS/EE/CE and several years of experience in > programming and systems integration in time critical or distributed control > applications. Experience with distributed systems using TCP/IP or field > buses, writing instrument and hardware-related ISRs, hardware interface > routines and device drivers for VxWorks-based processor boards is desired. > Applicants must be able to communicate technical information and participate > effectively in small teams of engineers and physicists. > These positions are in Northern California, in the San Francisco Bay Area. > SLAC offers competitive compensation and excellent benefits. If you wish to > be considered for these positions, please send your resume to > hamid@SLAC.Stanford.edu. > > Hamid Shoaee > Stanford Linear Accelerator Center > P.O.Box 4349 > Stanford, CA 94309 > (650) 926-2954 -- --------------------------------------------------------------- Dipl.-Phys. Martin Schrape schrape@atomika.com Senior Software Engineer Tel. +49 89 315 891 34 Atomika Instruments GmbH Fax +49 89 315 59 21 Bruckmannring 40 http://www.atomika.com 85764 Oberschleissheim/Munich, Germany --------------------------------------------------------------- From owner-vxwexplo-process Tue Mar 9 09:50:20 1999 From: "Michael Baumann" Date: Tue Mar 9 09:50:24 PST 1999 Subject: RE: Board-Detection in VMEbus-Systems Status: O VxWorks Well, we do it via vxMemProbe, to a read address on the board. -- Michael Baumann Optivus Technology Inc.|Loma Linda University Medical Center San Bernardino, California. (909)799-8308 |Internet: baumann@llumc.edu > Submitted-by: Rainer Stelzer > > > > > > Hi, > > > > I am trying to implement an automatic board detection in a > VMEbus-System, > > > > as i did very often before in 68k assembler. This is my algorithm: > > > > 1.) Install a Buserror-Exception handler, which only increments an > > error counter "ErrCnt" > > 2.) Set ErrCnt=0 > > 3.) Make a single access to the hardware you are testing for. > > 4.) Check ErrCnt: > > If it is zero then the board is present. > > If it is not zero then there is no board. > > 5.) Restore the old Buserror-Exception handler. > > > > In old assembler times there was no problem, once the special stack > > format was considered. > > > But I don't know how to do it with VxWorks in a simple, clean and > > reliable manner, because there seems to be no possibility to register > > my own Buserror-Exception routine for the short time of board detection > > only: > > * intVecConnect(...) is not suitable Buserror-Exceptions > > * execAddHook(...) executes normal exception handling first and > > afterwards my handler. But the normal exception handling stops my task > > which is testing for the board. > > > > So how can I implement my algorithm or something doing the same with > > vxWorks? > > Any hint or sample code is appreciated. > > > > Environment: > > VMEbus-System > > a single PEP VSBC32 CPU (68360 processor) > > serveral passive Slave Boards > > VxWorks/Tornado NT4.0 > > Language C > > > > Thanks for your help. > > > From owner-vxwexplo-process Tue Mar 9 10:12:29 1999 From: Bill Brown Date: Tue Mar 9 10:12:33 PST 1999 Subject: Re: Fwd: Board-Detection in VMEbus-Systems Status: O No problem! See vxMemProbe()in the Reference Manual. vxWorks does all of the "dirty work" for you. Hope this helps. the vxWorks Users Group Exploder wrote: > > Submitted-by owner-vxwexplo-process Tue Mar 9 07:58:47 1999 > Submitted-by: Rainer Stelzer > > > Hi, > > > > I am trying to implement an automatic board detection in a VMEbus-System, > > > > as i did very often before in 68k assembler. This is my algorithm: > > > > 1.) Install a Buserror-Exception handler, which only increments an > > error counter "ErrCnt" > > 2.) Set ErrCnt=0 > > 3.) Make a single access to the hardware you are testing for. > > 4.) Check ErrCnt: > > If it is zero then the board is present. > > If it is not zero then there is no board. > > 5.) Restore the old Buserror-Exception handler. > > > > In old assembler times there was no problem, once the special stack > > format was considered. > > > But I don't know how to do it with VxWorks in a simple, clean and > > reliable manner, because there seems to be no possibility to register > > my own Buserror-Exception routine for the short time of board detection > > only: > > * intVecConnect(...) is not suitable Buserror-Exceptions > > * execAddHook(...) executes normal exception handling first and > > afterwards my handler. But the normal exception handling stops my task > > which is testing for the board. > > > > So how can I implement my algorithm or something doing the same with > > vxWorks? > > Any hint or sample code is appreciated. > > > > Environment: > > VMEbus-System > > a single PEP VSBC32 CPU (68360 processor) > > serveral passive Slave Boards > > VxWorks/Tornado NT4.0 > > Language C > > > > Thanks for your help. -- Disclaimer: Any opinions are my own and have | -bill nothing to do with the official policy or the | wlbrown@lbl.gov management of L.B.N.L, who probably couldn't | Berkeley, CA care less about employees who play with trains. | aka wlbrown@netcom.com From owner-vxwexplo-process Tue Mar 9 12:16:30 1999 From: "Kerry Williams" Date: Tue Mar 9 12:16:34 PST 1999 Subject: MMU initialization speed issues Status: O Hello fellow vxWorkers, We have an app which maps large, contiguous blocks of PCI space prior to running. Currently, we are modifying sysPhysMemDesc[] to map in those blocks. Unfortunately, initializing MMU pages is a long slow process. We are using a Pentium class processor and a modified pc486 bsp. Does anyone know of any short cuts to mapping large contiguous blocks of cPci space? I know in 68K VME land that one could poke the TTRs (transparent translation registers) to map the space all at once. I'd like to do something similar for x86/cPci if possible. Any help/ideas a greatly appreciated! TIA, Kerry * vxWorks * Tornado * Wind River System * spam filter buster * -------------------------------------------- Kerry Williams Software Design Engineer Nortel Networks Broadband Wireless Access Division 14 Fultz Boulevard Winnipeg, Manitoba R3Y 1V3 kwilliams@broadbandnetworks.com ESN: 771-2407 PH : (204)631-2407 From owner-vxwexplo-process Wed Mar 10 04:00:25 1999 From: daemon@csg.lbl.gov Date: Wed Mar 10 04:00:30 PST 1999 Subject: comp.os.vxworks newsdigest Status: O Comp.Os.Vxworks Daily Digest Wed Mar 10 04:00:21 PST 1999 Subject: Re: Performance impact of C++ Subject: BSP failure when using HP/SDS solution Subject: BSP failure when using HP/SDS solution Subject: PCMCIA pcata Cards Subject: Re: Wind River' Tornado 1.0.1 for Windows NT is bugged Subject: Help urgently!! Subject: Re: a question about vx-work Subject: Looking for embedded DHCP server for PSOS (using a 68302) Subject: Help! - Subject: about device driver Subject: question about load vxworks Subject: Consultant available Subject: Re: IP address change.. Subject: Re: Spawn 2 tasks? Subject: Re: question about load vxworks Subject: y2k Question revisited Subject: Virtual Console Subject: Hard reset on MPC860 ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: Performance impact of C++ Date: Tue, 09 Mar 1999 12:29:13 GMT From: mcarr92365@my-dejanews.com Organization: Deja News - The Leader in Internet Discussion Message-ID: <7c346j$6h1$1@nnrp1.dejanews.com> References: Beware of EC++, a "standard" being pushed by Green Hills that cripples the language. One of the points of standard C++ is that you only pay (in performance/resources) for features you use. Why write crippled code today to take advantage of one company's compiler, when your competitor is writing standard C++ code, faster than you (because he doesn't have to write work-arounds and rearchitect his design to make up for things that are missing in EC++) and will be able to take advantage of better compilers tomorrow? - --Mike p.s. Yes, I agree that the gcc compiler that ships with Tornado 1.0.1 is long-in-the-tooth (to say the least), but with add-ons (like STLport and Cygnus-derived basic_string) and a little knowledge of template instantiation, you can get a little extra mileage out of it. In article , "Pankaj Datta" wrote: > We are evaluating using C++ for our project. I have come across a few > articles in trade magazines and otherwise which describe how certain C++ > features may adversely impact performance i.e. execution time and/or code > size. This mail is meant to invite comments on the impact on performance > using the GNU C++ compiler provided as part of Tornado 1.0.1. > > * Have the included libraries been optimized for speed or size? > > * Are there any WindRiver guidelines for using C++? > > * Can the GNU compiler be put in Embedded C++ (EC++) mode, informing the > user when the EC++ feature set has been transcended? > > * Are there better compilers (and libraries) than the one included in > Tornado for the x86 platform? > > I hope to learn from the experience of those who've "been there, done that" > and now recommend/dissuade others to do so. > > Thanks, > > Pankaj > > - -----------== Posted via Deja News, The Discussion Network ==---------- http://www.dejanews.com/ Search, Read, Discuss, or Start Your Own --------------------------- Newsgroups: comp.os.vxworks Subject: BSP failure when using HP/SDS solution Date: Tue, 09 Mar 1999 13:23:11 +0200 From: Rami Gideoni Organization: Chromatis Message-ID: <36E5049F.F32252D6@chromatis.com> Hi, I'm using the following setup : 1. Motorola FADS860 evaluation board, both the mother and daughter boards are PILOT. 2. MPC860SAR version C.1 3. WindRiver Tornado 1.0.1 VxWorks (Licence number 17755), ADS860 BSP with the FADS860 patch BSP. 4. HP BDM 5. SDS SingleStep Version 7.4 connected to the HP BDM by TCP/IP I burned down the BSP - it's works well, thanks you - --BUT-- When I try to run the code from the BDM, the BSP returns "data access" error Any ideas ? Thanks in advance Rami Gideoni - -- \\\||||||/// (o o) -----------oOO--(_)--OOo----------- Rami Gideoni Chromatis Networks Inc. Email : rami_gideoni@chromatis.com Phone : 972-3-9231030 Ext. 210 Fax : 972-3-9231050 ------------------------------------ --------------------------- Newsgroups: comp.os.vxworks Subject: BSP failure when using HP/SDS solution Date: Tue, 09 Mar 1999 13:22:54 +0200 From: Rami Gideoni Organization: Chromatis Message-ID: <36E5048E.2A8906B4@chromatis.com> Hi, I'm using the following setup : 1. Motorola FADS860 evaluation board, both the mother and daughter boards are PILOT. 2. MPC860SAR version C.1 3. WindRiver Tornado 1.0.1 VxWorks (Licence number 17755), ADS860 BSP with the FADS860 patch BSP. 4. HP BDM 5. SDS SingleStep Version 7.4 connected to the HP BDM by TCP/IP I burned down the BSP - it's works well, thanks you - --BUT-- When I try to run the code from the BDM, the BSP returns "data access" error Any ideas ? Thanks in advance Rami Gideoni - -- \\\||||||/// (o o) -----------oOO--(_)--OOo----------- Rami Gideoni Chromatis Networks Inc. Email : rami_gideoni@chromatis.com Phone : 972-3-9231030 Ext. 210 Fax : 972-3-9231050 ------------------------------------ --------------------------- Newsgroups: comp.unix.solaris,comp.os.vxworks,comp.sys.sun.hardware Subject: PCMCIA pcata Cards Date: Tue, 09 Mar 1999 07:58:24 -0500 From: jeffrey.m.creem@lmco.com (Jeff Creem) Organization: Jeff Creem Message-ID: I am having a problem with a PCMCIA ATA card that I use on both SPARC Solaris 2.6 (with the SBUS PCMCIA adapter and/or built in PCMCIA of an RDI Ultrabook) and vxworks (real-time OS running on a single board computer). The card contains a DOS filesystem. At times, after writing to the card under vxWorks, I can no longer mount the card under Solaris. (I get a mount: /dev/dsk/c1d1s0:c is not a DOS filesystem.) (If i run truss on the mount all I see is that mount() errno is EINVAL) This does not appear to be the traditional vxWorks issue of vxWorks overwriting the partition table information at the header of the card since as near as I can tell the partition information on the card is not changed and it looks like the driver is properly offsetting from the start of the card. If I use mtools on the SUN I can still read and write all of the files. If I run chkdsk on a PC it says there are no problems with the card. Once this happens the only thing that fixes it is to reformat it on a PC (since there does not appear to be solaris support for formatting PC ata cards). I've got the latest vxWorks DOSFs patches installed and it does not appear to help. I am leaning towards believing that this is a Solaris issue but certainly vxWorks is doing something to the card that makes Solaris unhappy. My next path is to try to get the card on a PC with Norton so I can take a look at the disk structure a little more but this will take a few days so I thought I'd ask if anyone has seen this sort of thing. (Still hard to believe how poor pcata cards are supported under Solaris, apart from this the Sun Also spits out unhandled interrupts each time a new card is inserted and certain combinations of cards makes the Sun Panic....Anyone know if this got any better in Solaris 7) Jeff Creem --------------------------- Newsgroups: comp.arch.embedded,comp.os.vxworks Subject: Re: Wind River' Tornado 1.0.1 for Windows NT is bugged Date: Fri, 05 Mar 1999 15:53:27 GMT From: bruce@CUTMEOUTscimisys.com (Bruce Wedding) Organization: NeoSoft, Inc. Message-ID: <36dffd07.897959@news.neosoft.com> References: <7bn25a$jeu$1@nslave1.tin.it> On Thu, 4 Mar 1999 23:39:30 +0100, "Wolfy" wrote: >I installed Tornado 1.0.1 ( 80x86) for Windows on 3 PC: >2 NT 4.0 Workstation and a Win98 laptop; > >It is about 10 days I am trying to make them work but I am only experimentig >that one installation on WinNT4.0 is working at 60%, another one on WinNT >4.0 too ( with SP3 ) is absolutly not working. ( installed from the same >CD on an twin machine ( the only difference is that one machine has 2 NTFS >partitions while the other one has 1 FAT16 and 1 NTFS ). >The Win98 version is at the moment the only one that let me compile vxWorks >kernel, >Yesterday I installed the licence manager 3 times on an NT4.0 Server and, at >last, after 2 reboots has started working. >At the moment the licence manager is THE ONLY THING THAT WORKS PROPERLY .... >I have the feeling ( and something more really ) that this development >system is a true "banana". > >I only hope that vxWorks operating system may work much better than its >development system. > >Does anyone experimented a similar case ? I spent the best part of yesterday getting Tornado 1.0.1 running on my new laptop. It is all up and running but I'm really getting fed up with their paranoid copy protection BS. I have wasted countless hours because of license keys. I have to get a new one everytime I change locations, which is often. Then, I had to get special permission from our sales rep to transfer my licenses from my desktop to my new laptop. The entire problem is the license and copy protection stuff, and if Wind River doesn't get it out of there, CE is going to eat their lunch. Bruce Wedding Software Manager bruce@scimisys.com Scientific MicroSystems, Inc. http://www.scimisys.com Tomball, Texas --------------------------- Newsgroups: comp.os.vxworks Subject: Help urgently!! Date: Tue, 09 Mar 1999 11:50:27 -0500 From: Lei Sun Organization: University of Central Florida Message-ID: <36E55153.4B03@pegasus.cc.ucf.edu> This is a multi-part message in MIME format. - --------------46142D362668 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Hi: I am developing device driver for VMI/VME4514A analogue data acquisition board using Vx-Work and Tornado tools. My BSP is MVME 167,host is sun ultrasparc station. I have done evry thing in order. As first step, I didn't write actual code in xxRead(),xxWrite().. routine. however ,in the xxInit()routine I wrote: LOCAL STATUS VMI4514AInit(WORD *addr) { printf("The base address of VMI4514A is %d\n",addr); *(addr+1)=0x4000; printf("Fail Led is off and auto-scanning mode is on\n"); } According to the hardware documentation, the board memory mapped into VMEbus at xxxx0000. The offset of 02H is Control Status Register(CSR), if the bit 14 is set to 1,the "Fail" Led will be off and auto-scanning mode will be on. So I compiled the program and loaded into the target,and in debug mode ,I pass the bass address of the board(0x0000) to the install function and run it. Everything seems in order,but the "fail" Led is still on. I can't descript the whole procedure. If anyone have experience on device driver, or on VMI/VME4514A,contact me please(ls00722@pegasus.cc.ucf.edu). I need your help. I have been confused with this problem for a week. I attached the whole code,if interested. you may read it and give me any sugguestion. thank you! - --------------46142D362668 Content-Type: text/plain; charset=us-ascii; name="vmi4514a.c" Content-Transfer-Encoding: 7bit Content-Disposition: inline; filename="vmi4514a.c" #include "vxWorks.h" #include "iosLib.h" #include "vmi4514a.h" LOCAL int VMI4514ADrvNum; LOCAL Channel_Desc vmi4514a_channel[2] ; /*0-> output,1->input */ LOCAL int VMI4514AOpen(Channel_Desc *channel,char *remainder,int mode) { printf("Entered in open\n"); return((int)channel); } LOCAL VMI4514AClose(Channel_Desc *channel) { printf("Entered in close\n"); return(OK); } LOCAL VMI4514ARead(Channel_Desc *channel, char *buffer,int nBytes) { printf("Entered in read\n"); return(OK); } LOCAL VMI4514AWrite(Channel_Desc *channel, char *buffer, int nBytes) { printf("Entered in write\n"); return(OK); } LOCAL int VMI4514AIoctl(Channel_Desc * channel, int requestCode) { printf("Entered in ioctl\n"); return(OK); } LOCAL STATUS VMI4514AInit(WORD *addr) { printf("The base address of VMI4514A is %d\n",addr); *(addr+1)=0x4000; printf("Fail Led is off and auto-scanning mode is on\n"); } LOCAL STATUS VMI4514ADrv(WORD *addr) { /* check if driver is already installed */ if (VMI4514ADrvNum > 0) { printf("Already installed ...\n"); return(ERROR); } /* install the driver */ printf ("Installing the VMI4514A driver\n"); VMI4514ADrvNum=iosDrvInstall(0,0,VMI4514AOpen,VMI4514AClose,VMI4514ARead,VMI4514AWrite,VMI4514AIoctl); /* Init the borad */ VMI4514AInit(addr); /* return OK or ERROR */ return (VMI4514ADrvNum == ERROR ? ERROR : OK); } /******************/ LOCAL STATUS VMI4514ADevCreate(char * name) { static i=0; int status; Channel_Desc *temp; /*check if driver installed */ if (VMI4514ADrvNum<=0) { errnoSet(S_ioLib_NO_DRIVER); return (ERROR); } if (i>=2) i=0; temp= &vmi4514a_channel[i++]; if ((status=iosDevAdd(&(temp->devHdr),name,VMI4514ADrvNum)) == ERROR) { printf("ERROR. device %s not added to the device list\n\n",name); return (ERROR); } return (status); } /**********************************************************/ void VMI4514AInstall (int address) { WORD *base =(WORD *) address; char devName[16]; int status; if (VMI4514ADrv(base) == ERROR) { printf("\n Error installing VMI4514A Driver\n"); exit(1); } // printf("Progam has created driver number\n"); vmi4514a_channel[0].base=base+10; vmi4514a_channel[1].csr =(base+1); vmi4514a_channel[0].in_use=FALSE; sprintf(vmi4514a_channel[0].name, "%s","outp"); sprintf(devName,"%s","/vmi4514a/outp"); status=VMI4514ADevCreate(devName); vmi4514a_channel[1].base=base+20; vmi4514a_channel[1].csr=(base+1); vmi4514a_channel[1].in_use=FALSE; sprintf(vmi4514a_channel[1].name, "%s","inp"); sprintf(devName,"%s","/vmi4514a/inp"); status=VMI4514ADevCreate(devName); // printf("\n Installing the VMI4514A board channels...\n\n\"); printf("channel /vmi4514a/outp is output channel.\n"); printf("channel /vmi4514a/inp is input channel.\n\n"); } /****************************/ STATUS VMI4514ARemove(void) { int status; printf("Deleting the devicces...\n"); iosDevDelete(&vmi4514a_channel[0].devHdr); iosDevDelete(&vmi4514a_channel[1].devHdr); printf("Removing the VMI4514A driver...\n"); status = iosDrvRemove(VMI4514ADrvNum,TRUE); VMI4514ADrvNum = 0; return(status); } - --------------46142D362668-- --------------------------- Newsgroups: comp.os.vxworks Subject: Re: a question about vx-work Date: Fri, 05 Mar 1999 13:01:58 -0500 From: Lei Sun Organization: University of Central Florida Message-ID: <36E01C16.57AF@pegasus.cc.ucf.edu> References: <36DF2D5D.47A1@pegasus.cc.ucf.edu> Charles H. Chapman wrote: > > On Thu, 04 Mar 1999 20:03:25 -0500, Lei Sun wrote: > >Hi: > > I am a new user in vx-work and tornado development tools. > >I have configured Tornado properly and written a small program.It simply > >create 10 tasks and print out the task ID individually. However I can't > >get the expected result. It only print out the main function > >message,say"main program begain..." and then program exit normally. > >anybody can tell me the reason ? > > By the way, I have redirect the I/O and the printed message should > >appears in Virtual console, I am using VME167 as BSP and host machine is > >sparc station. > >attatched is my code.(very simple) > > >((taskSpawn("a_task",0x100,VX_NO_STACK_FILL,200,(FUNCPTR)a_task,0,0,0,0,0, > 0,0,0,0,0))==ERROR) > > I don't know if this is the cause of your problem (I haven't tried an > experiment to confirm it) but task priorities range from 0 to 255. You > specified 0x100 = 256 as the task priority. Maybe taskSpawn doesn't > return ERROR in that case but also doesn't spawn the task since 256 is > lower than the lowest allowed priority. You might try increasing the > stack size a bit too. > > Chuck I forgot to tell you, Using windbrowser,I find that tasks have been created successfully. Everything is in order,except that task refuse to print out it's message. Someone tell me that I should redirect I/O in task. but as far as I have issue I/O redir command in Windsh. and I have got message from main function, I don't think that will be helpful. another thing is that since the whole program terminate normally, why do the 10 tasks still stay in memory? I tried it in Unix, as far as the main program is terminited ,the task will be killed automatically. do I have to include task kill comand in program? Lei sun --------------------------- Newsgroups: comp.dcom.net-management,comp.protocols.snmp,comp.protocolc.tcp-ip,comp.realtime,comp.sources,comp.os.vxworks Subject: Looking for embedded DHCP server for PSOS (using a 68302) Date: Tue, 09 Mar 1999 22:21:10 +0000 From: Tim Meagher Organization: Netoids, Inc. Message-ID: <36E59ED6.C925D0B4@netoids.com> Please advise. Tim Meagher Sr. Engineer NexTone Communications Rockville, MD 20850 240-453-6305 tmeagher@netoids.com --------------------------- Newsgroups: comp.os.vxworks Subject: Help! - Date: Tue, 09 Mar 1999 17:22:46 -0500 From: Peter Message-ID: <36E59F35.A43CE7A8@metrolink.net> I am hoping that someone on this newsgroup can help me. I am very new to vxWorks (less than 6 months) and have come across a run-time error that I am puzzled why it occurs. We are in the process of developing an application that basically receives serial data from several sources and then combines data from all of these sources for distribution to other systems. The error that we have come across occurs with a use of memcpy. In once instance of using the memcpy we receive a data access error. When this is substituted with a for loop to do the copy, the error does not occur. Can anyone provide an explanation of this or give me an idea of what to look into? Thanks in advance, Peter Software Engineer CSR PAFB - ----------------------------------------------- E-mail: funonh2o@metrolink.net (home) - ----------------------------------------------- --------------------------- Newsgroups: comp.os.vxworks Subject: about device driver Date: Fri, 05 Mar 1999 21:53:54 -0500 From: Lei Sun Organization: University of Central Florida Message-ID: <36E098C2.4D26@pegasus.cc.ucf.edu> I come up with another question, I have writen a small device driver for VMI4514A data acquisition board,Just for testing and understanding the Vx-Work I/O system. But how do I install the driver into the vx-work system? I am using MVE 167 computer and power up via the network, How can I let the driver installed as soon as the system is power up. Lei sun --------------------------- Newsgroups: comp.os.vxworks Subject: question about load vxworks Date: Wed, 10 Mar 1999 09:10:37 +0800 From: "SRD Zha Junhai" Organization: SBELL Message-ID: <36e5c74f.0@dnews.sbell.com.cn> Hi, everybody I just met a question when I use the ads860 board and loading the vxWorks from host. Here are the parameters : Attaching network interface cpm0... done. Subnet Mask: 0xfffffff8 Attaching network interface lo0... done. Loading... Error loading file: errno = 0xd0003 The procedure is very short. Could you tell me what caused this problem and the means of errno( 0xd0003 ), by the way, please tell me how can I find the error's mean. And I want to know if anyone tried to select from both socket Fd and pipe Fd. I tried it, but select seems can only find the pipe's activity. Thank you for your future response Best Regards Zha Junhai --------------------------- Newsgroups: comp.os.vxworks Subject: Consultant available Date: Tue, 09 Mar 1999 17:47:16 -0800 From: sainfeld@postoffice.pacbell.net Organization: SBC Internet Services Message-ID: <36E5CF24.AEB5F443@postoffice.pacbell.net> Reply-To: sainfeld@pacbell.net I am available for consulting contracts. I am interested in vxWorks/Tornado on any platform. C, C++, Java. Please contact me by email if you would like to have a copy of my resume. Thank you ! Jean-Pierre Sainfeld FistLink Consulting Services, Inc. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: IP address change.. Date: Sun, 28 Feb 1999 21:44:43 +0100 From: Robert Nachbaur Organization: Swisscom IP+ (post doesn't reflect views of Swisscom) Message-ID: <36D9AABB.CBA8B825@vienna.at> References: <7aja2p$9i4@overload.lbl.gov> <36ce39c4.0@news.pacifier.com> How do you change the ip adress under win95/winnt without re-booting? I think this is only possible if you are using dhcp, isn't it? Robert Steve Doiel wrote: > Keith Buchanan wrote in message <7aja2p$9i4@overload.lbl.gov>... > >> I have a basic Question, Can the IP address be changed when > >> vxWorks up and running, > >> with out rebooting ? if so how ? > > > >Unlike Windoze, it sure can. Use the ifAddrSet and ifMaskSet API > functions. > > > Gee. I often change the IP address of my windows system without re-booting. > But I didn't manage to figure out the right combinations of ifAddrSet, > ifMaskSet functions to use after VxWorks was up and running to change the IP > address under VxWorks. > > Hmmm... > SteveD --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Spawn 2 tasks? Date: Tue, 09 Mar 1999 19:47:55 -0800 From: Scott Johnson Organization: National Association for the Advancement of Computer Geeks Message-ID: <36E5EB6B.61D5@nospam.aracnet.com> References: Reply-To: sj_nospam@nospam.aracnet.com Joseph J. Lindula wrote: > > Hello, I'm new at programming in VxWorks so this will be a simple > question. I want a program to spawn two more tasks when it executes. Can I > just use a for loop and have the taskSpawn() only execute if its the > original task? Hmm. It sounds like you are confusing taskSpawn with the UNIX call fork(). fork() essentially creates a "clone", more or less, of the original process; both processes return from the fork() call. (The parent returns the PID of the child, the child returns 0). taskSpawn() in vxWorks works differently; you specify a callback function to be invoked in the new task; the OS creates the new task and calls that function. When that function returns, the task dies. Only the parent task returns from taskSpawn. taskSpawn is similar, in its operation, to the Win32 API CreateThread(). The callback function passed to taskSpawn() can be any function, it doesn't have to be your "main" function...so you don't need to have taskSpawn recursively invoke the function that invoked taskSpawn. (In fact, that probably isn't a good idea....) :) Scott - -- ======================================================================== Scott Johnson Patriot, Beer-Drinker, Geek GO BLAZERS Don't take yourself too seriously. Nobody else does......... ======================================================================== --------------------------- Newsgroups: comp.os.vxworks Subject: Re: question about load vxworks Date: Tue, 09 Mar 1999 19:19:25 -0800 From: "T. Venkata Ramana" Organization: Accelerated Networks, Inc. Message-ID: <36E5E4BD.B808DA17@hotmail.com> References: <36e5c74f.0@dnews.sbell.com.cn> You can look for errnos in vxworks header files target\h\errno.h. The error displayed indicates that the system is not able to locate the file specified in the bootline on the server from which it is trying to download. Select works on both socket fd & pipe fd. Venkata SRD Zha Junhai wrote: > Hi, everybody > > I just met a question when I use the ads860 board and loading the > vxWorks from host. > Here are the parameters : > Attaching network interface cpm0... done. > Subnet Mask: 0xfffffff8 > Attaching network interface lo0... done. > Loading... > Error loading file: errno = 0xd0003 > > The procedure is very short. Could you tell me what caused this > problem and the means of errno( 0xd0003 ), by the way, please tell > me how can I find the error's mean. > > And I want to know if anyone tried to select from both socket Fd > and pipe Fd. I tried it, but select seems can only find the pipe's > activity. > > Thank you for your future response > > Best Regards > > Zha Junhai --------------------------- Newsgroups: comp.os.vxworks Subject: y2k Question revisited Date: Wed, 10 Mar 1999 06:20:53 GMT From: TheZenith Organization: Deja News - The Leader in Internet Discussion Message-ID: <7c5300$ufk$1@nnrp1.dejanews.com> Hello, I need your expertise. What percentage of the following MPU's (micro processors) do you believe will: 1) Fail on the y2k rollover (cease to operate normally) 2) Produce non-critical errors on the y2k rollover (and continue to operate) 3) Operate normally Also, on average, how long would it take to find a faulty system, and replace it with a new y2k compliant system, and test it (in hours)? There are roughly 2.3 billion MPUs sold worldwide from 1980-1999. The US has roughly 70% of these, or 1.61 billion. Facts from: http://www.semichips.org/stats/mpu.htm http://www.semichips.org/stats/shares.htm http://www.semichips.org/stats/shares2.htm - -----------== Posted via Deja News, The Discussion Network ==---------- http://www.dejanews.com/ Search, Read, Discuss, or Start Your Own --------------------------- Newsgroups: comp.os.vxworks Subject: Virtual Console Date: Wed, 10 Mar 1999 16:54:53 +0900 From: "À±°­½Ä" Organization: Korea Telecom Message-ID: <7c58bd$6or$1@news3.kornet.nm.kr> How are you!! My name is Kang-Sik Yoon, working for Daewoo Elec. Co. LTD., in Korea. I want to know how to make "Virtual Console" be displayed in Target-Server Window. Of course, I have tried to change parameters associated with "Virtual Console" column in Target Server Tool bar and seen "-V" option when executing target-server. But, finally failed to install "Virtual Console" in Target-Server Window..... If there are any people knowing this solution, please let me know.. Thanks in advance for kind people giving solutions to me for the future... --------------------------- Newsgroups: comp.os.vxworks Subject: Hard reset on MPC860 Date: 9 Mar 1999 23:28:13 GMT From: sborho@ststech.com (Steve Borho) Organization: STARNET, L.L.C. Message-ID: Reply-To: sborho@ststech.com If someone's figured out how to force an MPC860 to do a hard reset through software, please let me in on the secret. Thanks. - -- Steve Borho --------------------------- End of New-News digest ********************** From owner-vxwexplo-process Wed Mar 10 07:03:56 1999 From: William Sheehan Date: Wed Mar 10 07:04:00 PST 1999 Subject: workQPanic in vxWorks/apexworks Ada Status: O Dear vxWorkers, I have 2 problems, that I don't know if they are related. What I'm asking for is a process to investigate the problems. Any suggestions to help my investigation will be appreciated. The platform is a 200Mhz PPC SBC (64MB RAM) with a dual Mil-Std-1553B PCI daughter card and a Discretes card communicating to the SBC via Parallel Port interrupts. Externally, we have 10Mbs ethernet, BC on 1553B, RT on 1553B, discrete signals into Discretes card (the only active signal at this time is a 1hz clock interrupt). The Application is Rational's ApexWorks Ada83 (3.0.0.b) with a variety of tasking rates. It is currently loaded into the SBC and kicked off with "-> sp __start". Problem #1: I'm running an Ada application on a 200Mhz PowerPC SBC when in random times, I see: workQPanic: Kernel work queue overflow. Press any key to stop auto-boot... 7 6.... The areas to investigate are possibly(low to high probability): 1. Dual Mil-Std-1553B PCI communications 2. sysClkRateSet 1000 (1000Hz clock for 1ms tasking resolution) 3. Parallel Port ISR7 communicating with a Discretes card (uses vads v_i_interrupts). 4. Possible "bad" address w.r.t. 1553 communications. 5. vNetTask - possible large ethernet traffic, in a "lab" environment. I'll add more to the list, if offered. Normally, the processor's time taken by ISR activity is 1%. The parallel port ISR does not use semaphores. The processor is performing several levels of Ada tasking. Problem #2: I try to make a single load image containing vxWorks and an ADA application (main.omf). Whether I use all the code above in problem #1 -or- one Put_Line("Hello...") I get this on boot-up after Loading the image from the server: -> Exception at interrupt Level: machine check Exception next instruction address: 0x003ca25c Machine Status Register: 0x0049030 Condition Register: 0x4400c40 Regs at 0x3d3520 Press any key to stop auto-boot... < machine reboots> P.S. I could use any help in debugging the memory dumped at 0x3d3520. P.S. I've waited too long for Rational to help out. They struck out looking... Thank You in advance for any help that you may volunteer. ========================================= Bill Sheehan LMFS Yonkers 914-968-2500 x2316 william.j.sheehan@lmco.com "Duck!! It hit the fan!" ========================================= From owner-vxwexplo-process Wed Mar 10 07:25:28 1999 From: Fred Roeber Date: Wed Mar 10 07:25:31 PST 1999 Subject: Re: Help! - Status: O Peter asked: > We are in the process of developing an application that basically > receives serial data from several sources and then combines data from > all of these sources for distribution to other systems. The error that > we have come across occurs with a use of memcpy. In once instance of > using the memcpy we receive a data access error. When this is > substituted with a for loop to do the copy, the error does not occur. There isn't really enough data here to say what the problem is. Would need to know what the target platform is, and what type of memory areas are being copied to and from. I suspect that the problem is due to the fact that the memcpy function is moving data in larger chunks than just bytes. Most versions of the VxWorks have specialized support for the bcopy routine (which is called by memcpy) that tries to speed up the task of copying data by using larger accesses to move data bytes. On some machines, this will even use specialized assembler instructions like MOVE16 on the 68040 to move data in big blocks. The problem is, if you are copying to or from anything other than local memory, these optimizations could cause problems. I have seen cards where you can't use MOVE16 to write to VME memory areas without risking getting bus errors. Of course there are lots of other potential problem issues as well. If the byte by byte copy works, I would say stick with it. Fred -- | Fred J Roeber, BBN Systems & Technologies | | 4 John Clarke Road Middletown, RI 02842-5202 | | froeber@bbn.com 401-848-3548 | | TraceMaker product manager -> www.tracemaker.bbn.com | From owner-vxwexplo-process Wed Mar 10 07:48:22 1999 From: Bill Mills Date: Wed Mar 10 07:48:26 PST 1999 Subject: RE: question about load vxworks Status: O This message is in MIME format. Since your mail reader does not understand this format, some or all of this message may not be legible. ------_=_NextPart_000_01BE6B0D.CA4F61D0 Content-Type: text/plain; charset="ISO-8859-1" Zha, These answers are off the top of my head, but.... The error probably means that the loader does not like the file format. The loader for our 860 based system won't load a file if we have stripped the symbols out of it (using objcopy). (Also make sure you are loading the right file, it should be an elf file.) I don't know where to find the meaning of the error code, I did look one time. If anyone else knows, please let us know. As for socket & pipe in select, yes we have done it and it worked (also on an 860 platform). I don't think there were any "tricks". >Hi, everybody > > I just met a question when I use the ads860 board and loading the >vxWorks from host. > Here are the parameters : > Attaching network interface cpm0... done. > Subnet Mask: 0xfffffff8 > Attaching network interface lo0... done. > Loading... > Error loading file: errno = 0xd0003 > > The procedure is very short. Could you tell me what caused this > problem and the means of errno( 0xd0003 ), by the way, please tell > me how can I find the error's mean. > > And I want to know if anyone tried to select from both socket Fd > and pipe Fd. I tried it, but select seems can only find the pipe's >activity. ====================================== Wm. A. Mills Principal Engineer, VoP Systems Telogy Networks, Inc 20250 Century Blvd Germantown, MD 20874 301-515-6548 Direct 301-515-8580 Main 301-515-7954 Fax bmills@telogy.com http://www.telogy.com ------_=_NextPart_000_01BE6B0D.CA4F61D0 Content-Type: application/ms-tnef Content-Transfer-Encoding: base64 eJ8+IjsPAQaQCAAEAAAAAAABAAEAAQeQBgAIAAAA5AQAAAAAAADoAAEIgAcAGAAAAElQTS5NaWNy b3NvZnQgTWFpbC5Ob3RlADEIAQWAAwAOAAAAzwcDAAoACgAyADoAAwBcAQEggAMADgAAAM8HAwAK AAoAMgA6AAMAXAEBCYABACEAAAAyRDlBNDM2NEVERDREMjExOTQzRTAwMTA0QjlFMDMxMQD5BgEE gAEAIAAAAFJFOiBxdWVzdGlvbiBhYm91dCBsb2FkIHZ4d29ya3MAqAsBDYAEAAIAAAACAAIAAQOQ BgCACQAALQAAAAsAAgABAAAAQAA5AJDxAsoNa74BHgBwAAEAAAAgAAAAUkU6IHF1ZXN0aW9uIGFi b3V0IGxvYWQgdnh3b3JrcwACAXEAAQAAABYAAAABvmsNs1RkQ5ox1O0R0pQ+ABBLngMRAAAeADFA AQAAAAYAAABCSUxMTQAAAAMAGkAAAAAAHgAwQAEAAAAGAAAAQklMTE0AAAADABlAAAAAAAIBCRAB AAAA0AQAAMwEAAB5BwAATFpGdcoAzzwDAAoAcmNwZzEyNeIyA0N0ZXgFQQEDAff/CoACpAPkBxMC gA/zAFAEVj8IVQeyESUOUQMBAgBjaOEKwHNldDIGAAbDESX2MwRGE7cwEiwRMwjvCfe2OxgfDjA1 ESIMYGMAULMLCQFkMzYWUAunYwEwdCBaE+AsCqIKhAqAVNJoB5BlIABxdwSQBCDLCsAecG8BICB0 HkAfgIRvcB9BIG15IB5AAGFkLCBidXQuuyEBHX0gBJADYAXAcANgemIBoGwgUAeABiIfkGGfBUAf khewIJASgWRvB5HEbm8FQGxpax+xH6E+ZgMQJZEFsADAIPAgIJ8iAiQlJgEfQAhwIDgcMYMi0BQQ ZCBzeXMOsOhtIHcCICclASQxHoDfJaQGkCjAHnAT4HYecCiA/QURcCgxH5IoYAbSBCAIYEsFQCAR aQVAKHUAkG4SZx9AYmoFoHB5KakmYChBK8BvIDBhJUH2cwhwHnB5CGAfAyQiLMLzH5IFEGdoBUAl siCwLGHUc2gIYGwoQGIeciIwpmwfcCWyLikmcEkkge0o8msk4Afgdx5AHyEf0P0loW4rFCMyLMMf dCJEBaA3AQAgsDJhaShAF7Bva90fQG4fsQdxJmFJH3AAcN8uoDbBMaAeYTLicyCwC1B/IIAeYSXQ BUAsoDLTITtBtwQgJ0It0GMlQAVAJiKQnwUgKdEDoBQQJdBjdCCw/nkHkSomMoEp0QVAAHAoQNss YSjQciVAKEAoB0At0f8CIDFiJ8ILUQAwJhEtcTJX/x+QC4A2kB+RHyEewR5yIFAmIiqxO0BzIiE7 PkhKaSCwZSpwcnkG4GTeeUNVQ1UmcDJCaiygBUCzB4A90SBxClAogGk/Mb8zMQOgMmAsoCVUIJBz J8MfJDALID3jLylDVXZ4V+8+cTrBA2EgYG8ogCE1RWPeSEHjM2IfoQqxYUYhHtL6OkUGQQJAANBB ISzgNsDudD5iO/FNMWYA0B5wDfAcbTAhAT1jS0dTdWIHTrEF0CgQazogMHh7ASBSEzhNn06vJAJQ DyDuTC80IRdFYEUiUy8mJbLnUdAiQSTgID1R4QmQWSDfFTBEryICIqFPoGQuYgQA/iBEEjDCACAm YAhRMREuoj0OsGwDIAeAMyEjsWNh/0eBKxIEAEUGIqIl0CiwPfIvNBYr0R9wWHMoWOYgKccgsSBQ H5J3YXk4p1zy/0UGXUEw4AfgXcBHQjPHIkPeJwQgIzJLRkUJQTPhMmD/YhACMDOCMuMp8TeVKrEr Ar8t4Dw0SqQG4B+QOxZGCzH/RSQ98juzasAmYDJgaLQsYP8gs2k2FBAooAQgZAICICMBvzPHO7Jl QENVANA28HYsYHp5ITs9cU9yX3KRHXRX6m0mYEEmYE0DECvAHXTyUAUQbmMFIAdAVyAs0IsLgAng ciCwVm9QFFHLKIJeZVQxoG9nIFAHwPtO0ziRSXVQHXQB0A4wFlGNCfB0CHAgUEJsdmrVTkcEkAOB H9B3biCwTUREIAHQODc0HXQzADAxLTUxNS02wDU0OCBEaRggPHDxe9w4NTgWUFGQC4F760w3OXzA arBheB10YlptdIJAXPF3oS4FoG0DHXQwEHRwOi8vdy2CcC6BDh16fYSAAwDeP69vAAALAAeACCAG AAAAAADAAAAAAAAARgAAAAADhQAAAAAAAAMAAIAIIAYAAAAAAMAAAAAAAABGAAAAABCFAAAAAAAA AwADgAggBgAAAAAAwAAAAAAAAEYAAAAAUoUAAPATAAAeACyACCAGAAAAAADAAAAAAAAARgAAAABU hQAAAQAAAAQAAAA4LjUAAwAtgAggBgAAAAAAwAAAAAAAAEYAAAAAAYUAAAAAAAALADaACCAGAAAA AADAAAAAAAAARgAAAAAOhQAAAAAAAAMAN4AIIAYAAAAAAMAAAAAAAABGAAAAABGFAAAAAAAAAwA5 gAggBgAAAAAAwAAAAAAAAEYAAAAAGIUAAAAAAAAeAEmACCAGAAAAAADAAAAAAAAARgAAAAA2hQAA AQAAAAEAAAAAAAAAHgBKgAggBgAAAAAAwAAAAAAAAEYAAAAAN4UAAAEAAAABAAAAAAAAAB4AS4AI IAYAAAAAAMAAAAAAAABGAAAAADiFAAABAAAAAQAAAAAAAAALAIqACyAGAAAAAADAAAAAAAAARgAA AAAAiAAAAAAAAAsAjIALIAYAAAAAAMAAAAAAAABGAAAAAAWIAAAAAAAACwBfgQggBgAAAAAAwAAA AAAAAEYAAAAABoUAAAAAAAADAPE/CQQAAAMA/T/kBAAAAwAmAAAAAAADADYAAAAAAAMAgBD///// AgFHAAEAAAA4AAAAYz1VUzthPSA7cD1UZWxvZ3kgTmV0d29ya3M7bD1UTklOVDAzLTk5MDMxMDE1 NTA1OFotMzI3MgAeADhAAQAAAAYAAABCSUxMTQAAAB4AOUABAAAABgAAAEJJTExNAAAAQAAHMHD6 AsoNa74BQAAIMNBhT8oNa74BHgA9AAEAAAABAAAAAAAAAB4AHQ4BAAAAIAAAAFJFOiBxdWVzdGlv biBhYm91dCBsb2FkIHZ4d29ya3MAHgA1EAEAAAA8AAAAPEYwMkMwOTJFNzFCREQyMTE4OUJEMDA2 MDk3QzlDMTFGMDZFNEMzQHRuaW50MDMudGVsb2d5LmNvbT4ACwApAAAAAAALACMAAAAAAAMABhBc 9C5JAwAHEFsEAAADABAQAAAAAAMAERAAAAAAHgAIEAEAAABlAAAAWkhBLFRIRVNFQU5TV0VSU0FS RU9GRlRIRVRPUE9GTVlIRUFELEJVVFRIRUVSUk9SUFJPQkFCTFlNRUFOU1RIQVRUSEVMT0FERVJE T0VTTk9UTElLRVRIRUZJTEVGT1JNQVRUSAAAAAACAX8AAQAAADwAAAA8RjAyQzA5MkU3MUJERDIx MTg5QkQwMDYwOTdDOUMxMUYwNkU0QzNAdG5pbnQwMy50ZWxvZ3kuY29tPgDkSg== ------_=_NextPart_000_01BE6B0D.CA4F61D0-- From owner-vxwexplo-process Wed Mar 10 07:58:36 1999 From: "Blake, Don" Date: Wed Mar 10 07:58:40 PST 1999 Subject: RE: Board-Detection in VMEbus-Systems Status: O > I am trying to implement an automatic board detection in a VMEbus-System, > > as i did very often before in 68k assembler. This is my algorithm: > > 1.) Install a Buserror-Exception handler, which only increments an > error counter "ErrCnt" > 2.) Set ErrCnt=0 > 3.) Make a single access to the hardware you are testing for. > 4.) Check ErrCnt: > If it is zero then the board is present. > If it is not zero then there is no board. > 5.) Restore the old Buserror-Exception handler. > > In old assembler times there was no problem, once the special stack > format was considered. > But I don't know how to do it with VxWorks in a simple, clean and > reliable manner, because there seems to be no possibility to register > my own Buserror-Exception routine for the short time of board detection > only: I've done something similar in VxWorks but it is board specific. Basically what I've done (and this is on a PowerPC board) is: 1. disable machine check 2. access the hardware 3. check for machine check 4. if machine check, then reset machine check - board not present 5. else, board is present Of course the disabling, checking and resetting of machine check is going to be specific to your particular processor and board. I've never used the 68K so I can't help you there. Don Lockheed Martin Federal Systems Owego, NY From owner-vxwexplo-process Wed Mar 10 08:42:04 1999 From: Fred Roeber Date: Wed Mar 10 08:42:07 PST 1999 Subject: Re: workQPanic in vxWorks/apexworks Ada Status: O William Sheehan asked: > Problem #1: > I'm running an Ada application on a 200Mhz PowerPC SBC when in random > times, I see: > > workQPanic: Kernel work queue overflow. > > Press any key to stop auto-boot... > 7 > 6.... > > > The areas to investigate are possibly(low to high probability): > 1. Dual Mil-Std-1553B PCI communications > 2. sysClkRateSet 1000 (1000Hz clock for 1ms tasking resolution) > 3. Parallel Port ISR7 communicating with a Discretes card (uses vads > v_i_interrupts). > 4. Possible "bad" address w.r.t. 1553 communications. > 5. vNetTask - possible large ethernet traffic, in a "lab" environment. Since the Work Queue panic happens only periodically, I would assume it is a "design" problem versus a hardware control problem. Work Queue panics result when the 64 entry work queue used by VxWorks to queue up "deferred work" overflows. Stuff gets added to this queue when interrupt handlers call any VxWorks functions that can't be executed immediately from within the interrupt handler. This includes most "communication type" functions like semaphore posting or message queue sends. It sounds like at times you are just throwing too many interrupts at the PPC. You need to check what communication functions are being called within the interrupt handlers (including any watchdog timer routines). To the extent you can, defer work from interrupt handlers to task level handlers that you signal via a semaphore. I would suggest that the first thing you do is get WindView and use it in "post mortem" mode to investigate the processing activity that lead up to the crash. For all the CPU performance, PPC's aren't good I/O processors; they aren't very efficient at handling interrupts. Something like a 68K or i960 can often do better at I/O handling even though it runs at a lower clock rate. Since you are probably stuck with the HW you are using, you need to reduce what gets deferred from the interrupt handler. One thing that is often done is signalling a semaphore each time you get an interrupt. If multiple calls to signal the same semaphore happen during the same sequence of interrupt processing, multiple requests to "do the same action" will get queued up, wasting work queue space. In such cases, you can use a global flag to control things so you only signal the semaphore once. Similarly, if you are using message queues to post requests from the interrupt handler to other tasks, you can have multiple requests deferred. In this case you can "write your own" message queue handling that uses a ring buffer and semaphore (with a global flag) to queue up the requests without deferring the work. This is definitely messy but can be done. > I'll add more to the list, if offered. > Normally, the processor's time taken by ISR activity is 1%. The > parallel port ISR does not > use semaphores. The processor is performing several levels of Ada > tasking. Use WindView to see what is really being done. Are any of your Ada tasks really attached as interrupt handlers that you don't know about? > Problem #2: > I try to make a single load image containing vxWorks and an ADA > application (main.omf). > Whether I use all the code above in problem #1 -or- one > Put_Line("Hello...") I get this on > boot-up after Loading the image from the server: > > -> > > Exception at interrupt Level: > machine check > Exception next instruction address: 0x003ca25c > Machine Status Register: 0x0049030 > Condition Register: 0x4400c40 > Regs at 0x3d3520 > > Press any key to stop auto-boot... > < machine reboots> Again, since you got as far as a VxWorks prompt, I would suggest you use WindView in post mortem mode to capture the OS activity that lead up to the crash. It's been a number of years since I have traced through the initialization sequence for fully resolved VADS Ada images but there are definitely some tricky things that go on. HTH, Fred -- | Fred J Roeber, BBN Systems & Technologies | | 4 John Clarke Road Middletown, RI 02842-5202 | | froeber@bbn.com 401-848-3548 | | TraceMaker product manager -> www.tracemaker.bbn.com | From owner-vxwexplo-process Wed Mar 10 10:40:23 1999 From: "Legoas, Marco A (FL51)" Date: Wed Mar 10 10:40:27 PST 1999 Subject: GNU compiler optimization control Status: O VxWorks, Tornado Does anyone know of a way of controlling optimization (on & off) in different areas of code of a single source file? . ...Under GNU, I could not find compiler directives that could be used within the source file for this purpose., other than at the "gcc" or Makefile level. ... Will appreciate any pointers. Marco A. Legoas malegoas@space.honeywell.com From owner-vxwexplo-process Thu Mar 11 04:00:43 1999 From: daemon@csg.lbl.gov Date: Thu Mar 11 04:00:47 PST 1999 Subject: comp.os.vxworks newsdigest Status: O Comp.Os.Vxworks Daily Digest Thu Mar 11 04:00:40 PST 1999 Subject: Looking for help: SIZE OF SERIAL BUFFERS Subject: "Relocation value does not fit in 24 bits" errors Subject: Re: Tcl 7.6/8.0.x for VxWorks 5.3 Subject: Re: Looking for help: SIZE OF SERIAL BUFFERS Subject: Re: Spawn 2 tasks? Subject: Shared memory holding bus Subject: Re: Wind River' Tornado 1.0.1 for Windows NT is bugged Subject: Re: Virtual Console Subject: Re: 3COM EtherLink III on x86 Subject: Adding Flash Type? Subject: SCC and SMC Drivers for MPC-860T Processor on Motorola FADS Board Subject: Problem with Moto MPC860SAR FADS board ATM 25 PHY... Subject: Debugger Support For Ethernet & Kernel drivers Subject: View Symbol Table? Subject: Re: Hard reset on MPC860 Subject: Re: ada95 and vxworks Subject: Re: Hard reset on MPC860 Subject: How to get system time and date with ms resolution Subject: Re: Is more than one Default IP Gateway supported in vxWorks 1.0 and 1.0.1? Subject: Re: How to get system time and date with ms resolution Subject: Re: Looking for help: SIZE OF SERIAL BUFFERS Subject: Re: Performance impact of C++ ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Looking for help: SIZE OF SERIAL BUFFERS Date: Wed, 10 Mar 1999 14:01:42 +0100 From: "M.Novaro" Organization: Unknown Message-ID: <7c5q9k$i6v$1@menelao.polito.it> Reply-To: "M.Novaro" Hi everybody. I am new in VxWorks, and I shoud write a source code to write on standard serial line of a FORCE40 board. In particular, my source code should write on the serial line of the board with retry if the line is busy (e.g. the bytes required to write are more than the free size of the buffer). So, I should write something like this: * get free size of buffer * if (freeSizeOfBuffer > number of bytes to write) write else retry. Is there any way to know the free size of transmission buffer in the serial line? Is there any ioctl() function to use? Many thanks in advance for any help. ///, //// \ /, / /. \ /, _/ /. \_ /_/ /. \__/_ < /<<< \_\_ /,)^>>_._ \ (/ \\ /\\\ // ```` - --------------((`------------------------------- Marco Novaro marconovaro@technologist.com "Try not. Do. Or do not. There is no try" - Yoda. --------------------------- Newsgroups: comp.os.vxworks Subject: "Relocation value does not fit in 24 bits" errors Date: Wed, 10 Mar 1999 09:20:04 -0500 From: "Charles H. Curley" Organization: Dynacs Engineering Co., Inc. Message-ID: <36E67F94.241C43ED@orion.ksc.nasa.gov> We've recently converted a project from C to C++ under Tornado 1.0.1, runtime 5.3.1, gcc version cygnus-2.7.2-960126, PowerPC 604 (MVME2604) and have been having problems using the target shell. Specifically it gives the subject error when loading from the target shell. Loading via a windshell works fine however. On WindSurf they give explanation as to why this happens and claim that applying patch 22767 (which adds -mlongcall to gcc) should fix the problem. Our system has this patch applied but the problems still exists. Yes, we are compiling using the -mlongcall directive. Before using the - -mlongcall we would get boat loads of relocation errors, now we only get a few (< 12). Is there a way to tell which symbols are giving the reloaction errors? We have a common library that gives the error called vmeLib. When we do "ld < vmeLib.out" we get 9 "Relocation ..." errors and a "nmppc - --undefined-only vmeLIb.out" gives: __builtin_delete __builtin_new __builtin_vec_new memcpy memset read semDelete semGive semMCreate semTake strlen strncpy write - -- Charles Curley, Senior Software Engineer, Dynacs Engineering Co., Inc. Engineering Development Contract, Kennedy Space Center, Florida --------------------------- Newsgroups: comp.os.vxworks,comp.lang.tcl Subject: Re: Tcl 7.6/8.0.x for VxWorks 5.3 Date: Wed, 10 Mar 1999 16:44:01 +0100 From: Volker Hetzer Organization: Siemens Inc. Message-ID: <36E69341.8A2A541C@sni.de> References: <36E3F391.F927813D@estec.esa.nl> Hans van der Meij wrote: > > Hi there, > > Can anyone give me a pointer to VxWorks 5.3 ports of Tcl 7.6 or higher? Me too please! Thanks a lot! Volker --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Looking for help: SIZE OF SERIAL BUFFERS Date: Wed, 10 Mar 1999 10:27:27 -0800 From: Bob Irwin Organization: Systems Interface Software Message-ID: <36E6B98F.66D06ADE@systemsinterface.com> References: <7c5q9k$i6v$1@menelao.polito.it> M.Novaro wrote: > > Is there any way to know the free size of transmission buffer in the serial > line? Is there any ioctl() function to use? > > Many thanks in advance for any help. Refer to your 'tyLib' man page or reference for IOCTLs such as FIONWRITE. - -- Bob Irwin bobi@systemsinterface.com Ph:281-484-6661 Fax:281-484-8988 Systems Interface Software, 10851 Scarsdale Suite 210, Houston TX 77089 --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Spawn 2 tasks? Date: Wed, 10 Mar 1999 10:15:55 -0700 From: Mike Stimpson Organization: Utah Scientific, Inc. Message-ID: <36E6A8CB.41C6@utsci.com> References: <36E5EB6B.61D5@nospam.aracnet.com> Scott Johnson wrote: [snip] > taskSpawn() in vxWorks works differently; you specify a callback > function to be invoked in the new task; the OS creates the new task > and calls that function. When that function returns, the task dies. Not quite. Experience shows that the task quits running when the function passed to taskSpawn() return, but the task still shows up in the task list. You have to do an explicit taskDelete() to get rid of it. If you intend to terminate a task by "falling off the end" of the function passed to taskSpawn(), you should probably end that function with something like: taskDelete (taskIdSelf()); - -- Mike Stimpson (mstimp@utsci.com) Opinions expressed are not necessarily those of the management... The more I work as an engineer, the more I disbelieve the theory of evolution. --------------------------- Newsgroups: comp.os.vxworks Subject: Shared memory holding bus Date: Wed, 10 Mar 1999 16:04:52 GMT From: Bill Oszust Organization: Northrop Grumman Corporation, Baltimore, MD Message-ID: <36E69823.1BA67BDC@ny.essd.northgrum.com> Sender: news%mother@mdnews.md.essd.northgrum.com It appears that the shared memory objects in vxWorks are hogging the vme bus. We are using a MVME2307 and a Antares 1397 I/O board. The 1397 board is timing out when requesting the bus. This 1397 board is not running vxWorks only the 2307s are. The 1397 onboard firmware is not changeable. When the bus is viewed with a bus analyzer most of the vme transactions are vxWorks shared memory objects related. These transactions hold onto the bus regardless of other boards requesting access to the bus. We have tryed various settings in the MAST_CTL and MISC_CTL register on Universe II for controlling vme access. None have proved to be successful. It was also noted that vxWorks was transferring across the bus using 16 bit transfers. Our shared memory area is an A32 vme address space so there should be no reason why the shared memory objects are doing 16 bit transfers. Doing it this way takes up more bus time. Also some of the vxWorks bus transaction I have looked at on the bus analyzer take up to 200uS to complete if not more. The 1397 Antares I/O board is made by Visicom Labs. It has a VME bus timeout of 102uS. This is not a configurable timeout and it must be able to get at the bus before this timeout elapses. Any help on this problem would be appreciated. Thank you in advance, Bill Oszust Northrop Grumman Norden Systems 65 Marcus Drive Melville, NY 11747 516-719-4775 516-719-4601 (FAX) \ --------------------------- Newsgroups: comp.arch.embedded,comp.os.vxworks Subject: Re: Wind River' Tornado 1.0.1 for Windows NT is bugged Date: Wed, 10 Mar 1999 19:34:35 +0000 From: Mike Davies Message-ID: References: <7bn25a$jeu$1@nslave1.tin.it> <7sC$fQAoq+32Ew9$@noco.demon.co.uk> <7bs1vp$52u$1@newsreader1.core.theplanet.net> <7c108v$f5$1@newsreader2.core.theplanet.net> In article , johnchann writes >Like they say, most of what you find on the internet is junk, this included. > >Well congratulations on this significant expose! Interestingly it is a >mandatory patch so that must mean it's carefully concealed. But no, there it >is in plain view on their web site! Just after a maintenance release and ahead >of the daylight saving patch. > Well, actually John, my remark was about why WRS support hadn't mentioned it. They didn't mention it to me for a prolonged period when I was having problems either. >The cumulative patch is to be applied to old installations and does not apply >to all architectures. ARM and ColdFire post date the cumulative patch for >example. Later installations apply the patch automatically. Then it either doesn't apply to the posters's target, or else he's been provided with a non-"Later installation". My reason for telling him about it was that nobody at WRS told me about the patch (which has had some positive impact on the Tornado BSOD problem under NT) - my assumption was that the same misfortune had befallen him. If support from WRS was better then I guess he'd have had other places to ask for help, hmm ? > >To quote the web site (which you both have access to since you are bitching >about support):- > LOL :-) Well, I thought I'd check the web site to see if I could find this patch. When I did then all the reasons I don't make use of WRS online support came back to me : 1 I went to "http://www.wrs.co.uk" There's something fishey about this site so I thought I'd try .com instead :-) 2 "http://www.wrs.com" - This may not be WRS's fault but from the UK this afternoon it was painfully slow :-/ 3 Oh yes, I remember now, you need a password to get onto the support pages. What did I use last time ? Oh ! It doesn't work any more :-/ 3 Let's try to obtain a new user ID for Windsurf. Of course it wants my product licence number ! (well I suppose that's common for a support web page) 4 Back to Outlook for the licence number (mailed to me at the end of Jan this year and valid from Jan 26 1999 without an expiry date) 5 Type in the licence number. "Your Licence has expired". :-/ :-/ :-/ 6 "Fill in your details for ..." followed by a form with around 40 fields. 7 As so often before using WRS support my preferred option was : "It'll be quicker just not to bother" 8 Got on with what I'm paid for muttering under my breath about WRS support. 9 The colleague I quoted before ("You mean they didn't get back to you to tell you they didn't know how to fix it ?") offered this remark : "... typical ..." Which it is. You've accused me of bitching and I object to that. If WRS looked at customer support as less of a nuisance and more as being their bread and butter (and jam, given the price) then people would complain about it less. The products WRS sell are not cheap, and the support has been of a lower quality than I have been used to. Your response also highlights the nature of the problem at WRS : They care about the licensing, not about the customers' problems. I make the assumption that you work for WRS though you don't say so. If you ignore customer complaints ("bitching") then you won't solve the problems. I was trying to help a fellow user who was proclaiming difficulties in a public forum. No-one from WRS offered to step in instead. Typical. - -- Mike Davies --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Virtual Console Date: Wed, 10 Mar 1999 11:13:28 -0800 From: "T. Venkata Ramana" Organization: Accelerated Networks, Inc. Message-ID: <36E6C458.73664558@hotmail.com> References: <7c58bd$6or$1@news3.kornet.nm.kr> Yo may redirect the output to virtual console by calling these : vf0=open("/vio/0",2,0) ioGlobalStdSet(0,vf0) ioGlobalStdSet(1,vf0) ioGlobalStdSet(2,vf0) This works. Venkata "À±°­½Ä" wrote: > How are you!! > > My name is Kang-Sik Yoon, working for Daewoo Elec. Co. LTD., in Korea. > > I want to know how to make "Virtual Console" be displayed in Target-Server > Window. > Of course, I have tried to change parameters associated with "Virtual > Console" column in Target Server Tool bar and seen "-V" option when > executing target-server. > But, finally failed to install "Virtual Console" in Target-Server > Window..... > > If there are any people knowing this solution, please let me know.. > > Thanks in advance for kind people giving solutions to me for the future... --------------------------- Newsgroups: comp.os.vxworks Subject: Re: 3COM EtherLink III on x86 Date: Wed, 10 Mar 1999 21:47:40 +0100 From: "Thomas Hieber" Organization: tesion Communikationsnetze Suedwest GmBH & Co.KG, Stuttgart, Germany Message-ID: <7c6m9g$ibj$1@stu1id2.tesion.de> References: <7c464i$2btq$1@news.gate.net> Bob Beers schrieb in Nachricht <7c464i$2btq$1@news.gate.net>... >I suspect that this is a well known and often solved problem, but I don't >have the answer yet, so I'm asking the experts: > >Less than 5% of the time, on power-up, my x86 target hangs during >the usrNetInit for subject card (elt0). I have never had any trouble after >initialization completes, just every now and then I get a > >Attaching network interface elt0...failed. > >A power cycle or soft reboot usually results in a successful load, >but I'd like to avoid even these occasional hangs. I haven't collected >much data since during development it's pretty easy to clear the problem, >but once it's in the field, I'd like the product to boot every time. > >What's the story? I saw the same behaviour on a SMC Elite 16C Ultra. Switching to a SMC Elite 16 solved the problem. Then I switched to the 3Com Etherlink III on a 486. Worked without any problems. Then I switched to a 586 Target and it would never recognize the the card, though that card on that board was functioning perfectly both with Linux and DOS. Now I have switched to a NE2000 card and it works again. I think you will have to accept, that most drivers for network-cards on vxWorks are buggy, one way or another. If one card doesn't work, try to get another card. At least this works for me. What is really annoying is the fact, that WRS is not very interested in fixing these bugs. (As it is not interested in fixing all those bugs that are still in Tornado btw.) Thomas - -- The above E-Mail adress is invalid. Use thieber at bluewin dot de instead. --------------------------- Newsgroups: comp.os.vxworks Subject: Adding Flash Type? Date: Wed, 10 Mar 1999 15:17:12 -0600 From: "Bruce Nehlsen" Organization: UUNET Message-ID: <7c6nct$2oi$1@ffx2nh5.news.uu.net> Greetings! Has anyone gone through the exercise of adding a flash type to the devices supported by vxWorks? I'm going through adding the AM29LV160, and any pointers would be appreciated. Thanks in Advance, Bruce Nehlsen. --------------------------- Newsgroups: comp.os.vxworks Subject: SCC and SMC Drivers for MPC-860T Processor on Motorola FADS Board Date: 10 Mar 1999 22:30:45 GMT From: "Rich Glossop" Message-ID: <01be6b45$b43a7d00$6114e481@richg2.xetron.com> Does anyone know of existing driver source for the SCC and SMC serial channels on the MPC-860T processor with the FADS board ? Our BSP did not supply SCC drivers and only sets up and uses one of the two SMC channels on the uP. We need to use all SCCs and both SMCs. Please E-mail me at richg@xetron.com --------------------------- Newsgroups: comp.os.vxworks Subject: Problem with Moto MPC860SAR FADS board ATM 25 PHY... Date: Sun, 28 Feb 1999 16:11:08 -0600 From: "J. Estes" Organization: uma Message-ID: <36D9BEFC.A2BC4531@uma.com> Has anyone ever used an MPC860FADS board w/ the ATM25 PHY successfully? I have tried without any success to get ours to run. The OC3 interface works with no problem, and I am beginning to wonder if the ATM25 ever worked. Our HP broadband test set says it is not working-no cells in or out on ATM25 UTP, works fine on OC3. Any suggestions? Perhaps the interface chips require some programming that is not mentioned in the system manual? - -- Regardz(tm), Joel Estes jestes@boschtelecominc.com "The early bird gets the worm, but the second mouse gets the cheese." - --Author unknown --------------------------- Newsgroups: comp.os.vxworks Subject: Debugger Support For Ethernet & Kernel drivers Date: Wed, 10 Mar 1999 16:25:22 -0800 From: Ganesh Sundaram Message-ID: <36E70D71.7F8DDC55@NOSPAM.metabyte.com> Hi, I am new to the VxWorks operating system, so excuse me if this is a repeat/naive question. I wanted to know if Tornado supports "source level debugging for ethernet drivers" ? I am looking for debugging support similar to what SoftIce provides for Windows NT driver debugging. I will be developing an ethernet driver under vxWorks and would be very much interested to know if this is possible. On what kind of hardware platform is the source level debugging supported? What is the setup required to acheive this? Does the basic Tornado environment support source level debugging of ISR's? Any information or direction to obtain the setup required for debugging is greatly appreciated. Thanks in advance. - -Ganesh --------------------------- Newsgroups: comp.os.vxworks Subject: View Symbol Table? Date: Thu, 25 Feb 1999 13:20:13 -0800 From: "Joseph J. Lindula" Organization: California State University, Chico Message-ID: Hello, I'm a newbie with vxWorks. Can anyone tell me if it is possible to view the symbol table, if so how? Thanks, Joe jjlindula@netscape.net --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Hard reset on MPC860 Date: Wed, 10 Mar 1999 21:44:00 -0600 From: Pierre Olivier Organization: Netcom Canada Message-ID: <36E73C00.5913@netcom.ca> References: <36E6AD16.7624@wireplay.bt.co.uk> Reply-To: polivier@netcom.ca Phil Ashby wrote: > > Steve Borho wrote: > > > > If someone's figured out how to force an MPC860 to do a hard reset through > > software, please let me in on the secret. > > This may not help, however: we induce a hard reset on our MVME230x by > instructing the Universe chip to toggle the SYSRESET line..... > -- > !--------------------------- Phil "Phlash" Ashby ---------------------! > ! BT Wireplay - Lead Programmer and general technical trivia supplier ! > ! Phone: 01473-644348 WWW: http://www.wireplay.com/ ! > ! Snail: B81/G41, BT Labs, Martlesham Heath, Ipswich IP5 3RE, England.! Generically speaking, and without involving special hardware, there are two sure ways to reset the CPU: - - Supposing the watchdog is activated (and it should be activated, for failfast operation), you simply stop feeding the watchdog. This will of course have a delay of up to the period of the watchdog. - - For immediate reset, you simply have to send the MPC860 in checkstop state, by accessing an unmapped (outside any CS) memory address in write or read. Of course, if you have a checkstop handler (exc 0x00200), you must let through the access to this particular address. - --------------------------------------------- Pierre Olivier Montreal (QC) Canada polivier@netcom.ca (. .) - ------------------------------oOO-(_)-OOo---- --------------------------- Newsgroups: comp.os.vxworks Subject: Re: ada95 and vxworks Date: Sun, 7 Mar 1999 22:23:55 -0600 From: "Michael Garrett" Message-ID: <#JxqksRa#GA.342@nih2naaa.prod2.compuserve.com> References: <7bupas$h4b$1@nnrp1.dejanews.com> >We use GNAT for VxWorks m68k. It uses the same GNU toolset and environment >as shipped with VxWorks (gcc, g++, gdb, etc). Is the GDB Debugger that ships with VxWorks fully Kernel Aware at the Ada layer?? Or is there an Gnat version that is? I have to do some investigating of Wind River Tools. Is there task profiling available for the Ada Versions? Michael. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Hard reset on MPC860 Date: Wed, 10 Mar 1999 21:20:52 -0800 From: Scott Johnson Organization: National Association for the Advancement of Computer Geeks Message-ID: <36E752B4.2898@nospam.aracnet.com> References: <36E6AD16.7624@wireplay.bt.co.uk> <36E73C00.5913@netcom.ca> Reply-To: sj_nospam@nospam.aracnet.com Pierre Olivier wrote: > > Phil Ashby wrote: > > > > Steve Borho wrote: > > > > > > If someone's figured out how to force an MPC860 to do a hard reset through > > > software, please let me in on the secret. > > > > This may not help, however: we induce a hard reset on our MVME230x by > > instructing the Universe chip to toggle the SYSRESET line..... > > -- > > !--------------------------- Phil "Phlash" Ashby ---------------------! > > ! BT Wireplay - Lead Programmer and general technical trivia supplier ! > > ! Phone: 01473-644348 WWW: http://www.wireplay.com/ ! > > ! Snail: B81/G41, BT Labs, Martlesham Heath, Ipswich IP5 3RE, England.! > > Generically speaking, and without involving special hardware, there > are two sure ways to reset the CPU: Do these do a hard reset--ie one that (among other things) puts the memory controller back into "boot" mode (all addresses go through CS0, and BR0 and OR0 are set according to what is sampled on the hard machine configuration word???) I was under the impression that these did a somewhat softer reset--and not completely resetting the CPU. On our project, we provide an external register that will pull reset, thus giving us a true "hard" reset. > - Supposing the watchdog is activated (and it should be activated, for > failfast operation), you simply stop feeding the watchdog. This will > of course have a delay of up to the period of the watchdog. > > - For immediate reset, you simply have to send the MPC860 in checkstop > state, by accessing an unmapped (outside any CS) memory address in > write or read. > Of course, if you have a checkstop handler (exc 0x00200), you must > let through the access to this particular address. Careful, you're confusing the "checkstop" condition (in which the processor halts) with the "machine check exception". There is no such thing as a "checkstop exception", if the necessary conditions for checkstop occur, the processor halts, and can only be unhalted by a reset. (In other words, there is no way to recover.) Machine check exceptions (bus errors on other architectures) are, of course, trappable and recoverable (at least at the system level. Applications that machine check may or may not be recoverable.) Confounding the issue is the fact that if the correct bit (ME) in the MSR is not set, a machine check condition causes a checkstop. But the two are two different things. Scott ======================================================================== Scott Johnson Patriot, Beer-Drinker, Geek GO BLAZERS Don't take yourself too seriously. Nobody else does......... ======================================================================== --------------------------- Newsgroups: comp.os.vxworks Subject: How to get system time and date with ms resolution Date: Thu, 11 Mar 1999 07:14:43 +0100 From: "Torsten Trzeciak" Organization: mannesmann.de Message-ID: <7c7n0s$l2b@as1fw.mannesmann.de> Hello, I like to get the system time with a ms resolution. Is there a kind of getTime function? Torsten --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Is more than one Default IP Gateway supported in vxWorks 1.0 and 1.0.1? Date: Mon, 08 Mar 1999 12:24:56 -0500 From: Tim Sohacki Organization: Nortel (Northern Telecom) Message-ID: <36E407E8.CE4B0EF4@nortelnetworks.com> References: <36e3f0db.655031181@nntp.concentric.net> fyi, I think you mean vxWorks 5.3 or 5.3.1 (Tornado is the host development environment, vxWorks the target-based OS.) Anyway, will adding an additional route to the routing table suffice? You can use routeAdd() to do that. But, how does vxWorks behave is you have two entries in the table for "0.0.0.0" ... I suppose that is the real question. To add a default gateway to the routing table: - -> routeAdd("0", "47.0.0.0") Les Cheong wrote: > > We have a vxWorks application on an embedded server that currently > supports one Default IP Gateway address. A customer wants to support > a secondary one (for redundancy purposes). > > Is this supported under vxWorks 1.0 or 1.0.1? > > Thanks, all. > > Les Cheong > lcheong@netbrowser.com - -- Tim Sohacki, Nortel Networks email: sohacki@nortelnetworks.com phone: (919) 991-2228 [ESN 35-12228] fax: (919) 991-4126 [ESN 35-14126] --------------------------- Newsgroups: comp.os.vxworks Subject: Re: How to get system time and date with ms resolution Date: Thu, 11 Mar 1999 10:43:39 +0100 From: "M.Novaro" Organization: Unknown Message-ID: <7c8321$t6g$1@menelao.polito.it> References: <7c7n0s$l2b@as1fw.mannesmann.de> Reply-To: "M.Novaro" Torsten Trzeciak ha scritto nel messaggio <7c7n0s$l2b@as1fw.mannesmann.de>... >Hello, >I like to get the system time with a ms resolution. Is there a kind of >getTime function? > >Torsten Try the clock_gettime() procedure: it should return the time till ns (!) Regards, Marco --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Looking for help: SIZE OF SERIAL BUFFERS Date: Thu, 11 Mar 1999 10:40:49 +0100 From: "M.Novaro" Organization: Unknown Message-ID: <7c82sm$bm5$1@menelao.polito.it> References: <7c5q9k$i6v$1@menelao.polito.it> <36E6B98F.66D06ADE@systemsinterface.com> Reply-To: "M.Novaro" Bob Irwin wrote > >Refer to your 'tyLib' man page or reference for IOCTLs such as FIONWRITE. > Thanks, but I already know the use of FIONWRITE... The problem was in the way to determine the number of _free_ bytes in the buffer (FIONWRITE returns the number of bytes yet to be written, isn't it?). Thanks again, Marco Novaro --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Performance impact of C++ Date: Mon, 08 Mar 1999 23:08:23 GMT From: Bill Pringlemeir Message-ID: References: Sender: bpringle@DeadDuck Many CPUs support the Green Hills compiler. Green Hills provide PJ Plaugers embedded C++ library. I have heard that it is very good library but have no practical experience. The GCC version that comes with VxWorks is antiquated. (I think somebody has said that before on this NG...) Don't look for the latest features from the GCC version shipped with VxWorks. You may also try a DejaNews search. I believe that many of your questions have been asked before. Bill Pankaj> * Have the included libraries been optimized for speed or Pankaj> size? Pankaj> * Are there any WindRiver guidelines for using C++? Pankaj> * Can the GNU compiler be put in Embedded C++ (EC++) mode, Pankaj> informing the user when the EC++ feature set has been Pankaj> transcended? Pankaj> * Are there better compilers (and libraries) than the one Pankaj> included in Tornado for the x86 platform? Pankaj> I hope to learn from the experience of those who've "been Pankaj> there, done that" and now recommend/dissuade others to do Pankaj> so. - -- Brave New Waves, "http://www.radio.cbc.ca/programs/bnw/" --------------------------- End of New-News digest ********************** From owner-vxwexplo-process Thu Mar 11 05:30:07 1999 From: Arand Thomas Date: Thu Mar 11 05:30:11 PST 1999 Subject: Tool for detectiong memory access violations? Status: O Hi vxWorks world, we are looking for a tool that is able to detect memory access violation (e.g. writing to a memory location that hasn't been allocated before, reading from a memory location that has been freed before, reading a location that wasn't initialised before, ...). We know about CodeTest, but that covers only a part of our requirements. The tool should approximately do what 'purify' (from Rational) does on UNIX systems. Does anyone has ideas? Thanks and Ciao, Thomas ---------------------------------------------------------------------- Thomas Arand Siemens AG Tel. +49 - 89 - 722 42296 Fax. +49 - 89 - 722 26572 thomas.arand@icn.siemens.de From owner-vxwexplo-process Thu Mar 11 10:16:58 1999 From: "Bruce A. Wilson" Date: Thu Mar 11 10:17:02 PST 1999 Subject: RE: question about load vxworks Status: O Use the printErrno function from the shell to get a description as shown below: -> printErrno 0xd0003 0xd0003 = S_iosLib_INVALID_FILE_DESCRIPTOR >Submitted-by owner-vxwexplo-process Wed Mar 10 07:48:22 1999 >Submitted-by: Bill Mills < > >... I don't know where to find the meaning of >the error code, I did look one time. If anyone else knows, please let us >know. > >> >> Loading... >> Error loading file: errno = 0xd0003 >> >> The procedure is very short. Could you tell me what caused this >> problem and the means of errno( 0xd0003 ), by the way, please tell >> me how can I find the error's mean. >> Bruce A. Wilson U.C. Lawrence Livermore National Lab P.O. Box 808 (L-054) Livermore, CA 94550 voice: (925) 422-4145 fax: (925) 422-9905 From owner-vxwexplo-process Fri Mar 12 04:00:47 1999 From: daemon@csg.lbl.gov Date: Fri Mar 12 04:00:51 PST 1999 Subject: comp.os.vxworks newsdigest Status: O Comp.Os.Vxworks Daily Digest Fri Mar 12 04:00:43 PST 1999 Subject: Re: Question on the ethertype field in the 802.3 header Subject: Re: question about load vxworks Subject: Re: How to get system time and date with ms resolution Subject: VxWorks 5.3 Interrupt Enable Subject: Re: Tcl 7.6/8.0.x for VxWorks 5.3 Subject: Re: Help. Shell Subject: Interrupt Handlers in ROM-Resident images Subject: Re: Overload new operator -- solved and thanks Subject: Interrupt Handlers in ROM-Resident images Subject: SENS and sockLib Subject: Re: Wind River' Tornado 1.0.1 for Windows NT is bugged Subject: Re: How to get system time and date with ms resolution Subject: Re: workQPanic in vxWorks/apexworks Ada Subject: Copy from host to target FS over serial WDB? Subject: Re: Help. Shell Subject: vxWorks SCSI support for Adaptec 1520 Subject: Interrupt Handlers in ROM-Resident images Subject: Attaching a CLI to telnet Subject: Re: question about load vxworks Subject: File I/O ? Subject: Re: VxWorks 5.3 Interrupt Enable Subject: Tornado and JTAG Subject: Re: Performance impact of C++ Subject: Re: Using IRQ3 interrupts on a PPC860DE Subject: Re: File I/O ? Subject: Re: workQPanic in vxWorks/apexworks Ada Subject: For memory protection... Subject: Re: How to get system time and date with ms resolution Subject: DDD with Tornado? Subject: Re: File I/O ? Subject: Re: For memory protection... Subject: 3COM EtherLink III on x86 ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: Question on the ethertype field in the 802.3 header Date: Mon, 08 Mar 1999 09:58:22 -0700 From: Mike Stimpson Organization: Utah Scientific, Inc. Message-ID: <36E401AE.167E@utsci.com> References: <7bmuqe$64p@news.or.intel.com> <36E07FBE.2A285E61@hotmail.com> T. Venkata Ramana wrote: > > As far as I know, the endian type in VxWorks is Big Endian which > happens to be the network byte order. So there is no need to > do HTONS and everything seems to work. 1). I'm not sure that vxWorks is Big Endian on the Intel (x86) chips. 2). You should probably do an HTONS anyway, even if it's a no-op, if you care about writing portable code. Neither of these points necessarily answers the original question, which was: > rvsaripa wrote: > > > Hi, > > I have an Ethernet-type driver . When I get an IP packet to send , I put > > the destination MAC address (6 bytes) + the source MAC address ( 6 bytes ) > > followed by 2 bytes of the ethertype field ( 806 for ARP and 800 for IP ). I > > think I should do HTONS conversion on this type field before putting it in > > the buffer but the DEC Ethernet driver does not seem to do it. Look at the > > if_dcFast.c file in the tornado installation for Tornado 1.0.1 for a 486 > > BSP. How does this work?. Does the card take care of this?. > > > > I am pretty sure that HTONS has to be done on the ethertype field but can > > someone answer the above question?. BTW, does HTONS has to be done for ARP > > type too ?. > > > > Thanks a bunch, > > Rk > > > > All opinions are mine & mine only. - -- Mike Stimpson (mstimp@utsci.com) Opinions expressed are not necessarily those of the management... The more I work as an engineer, the more I disbelieve the theory of evolution. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: question about load vxworks Date: Thu, 11 Mar 1999 07:56:17 -0800 From: Edward LaPier Organization: Lockheed Martin Federal Systems, Owego, NY Message-ID: <36E7E7A1.F2297B10@lmco.com> References: <36e5c74f.0@dnews.sbell.com.cn> <36E680AA.6D1@ti.com> Just to add to the possible reasons you may be getting this error: 1) Check that the Subnet Mask is correct. I was working with a MVME1600 in the lab and the subnet mask needed to be changed manually in the boot params. Ed --------------------------- Newsgroups: comp.os.vxworks Subject: Re: How to get system time and date with ms resolution Date: 11 Mar 1999 13:18:58 GMT From: chc@nasa2.ksc.nasa.gov (Charles H. Chapman) Organization: NASA, Kennedy Space Center Message-ID: References: <7c7n0s$l2b@as1fw.mannesmann.de> <7c8321$t6g$1@menelao.polito.it> On Thu, 11 Mar 1999 10:43:39 +0100, M.Novaro wrote: > >Torsten Trzeciak ha scritto nel messaggio ><7c7n0s$l2b@as1fw.mannesmann.de>... >>Hello, >>I like to get the system time with a ms resolution. Is there a kind of >>getTime function? >> > >Try the clock_gettime() procedure: it should return the time till ns (!) Well, it may return a value that represents the time in ns but it still only has a resolution of system clock ticks, typically 1/60 second. Chuck --------------------------- Newsgroups: comp.os.vxworks Subject: VxWorks 5.3 Interrupt Enable Date: Sun, 28 Feb 1999 16:58:15 -0800 From: MAXi Organization: UC Davis Fault Tolerant Systems (Research Assistant) Message-ID: <36D9E627.17BDB4A4@cs.ucdavis.edu> Hi All, I was just wondering how I can enable interrupt (say /INT5) in VxWorks 5.3 . I have used "intConnect" to tell where the interrupt service routine is so all I need to do is to actually enable the interrupt. TIA. Regards, MAXi - -- ======================================================================== Maxwell CHEONG University of California (Davis) Address: Computer Engineering, Junior Rm 302, Pierce J Building Studnet ID#: 991-023926 Tercero Hall Circle Phone: (530) 754 0352 University of California (Davis) email: cheong@cs.ucdavis.edu DAVIS, CA 95616 ======================================================================== --------------------------- Newsgroups: comp.os.vxworks,comp.lang.tcl Subject: Re: Tcl 7.6/8.0.x for VxWorks 5.3 Date: Thu, 11 Mar 1999 12:32:38 +0000 From: Martin Schrape Organization: Atomika Instruments GmbH Message-ID: <36E7B7E6.BFE23AD5@atomika.com> References: <36E3F391.F927813D@estec.esa.nl> <36E69341.8A2A541C@sni.de> Volker Hetzer wrote: > > Hans van der Meij wrote: > > > > Hi there, > > > > Can anyone give me a pointer to VxWorks 5.3 ports of Tcl 7.6 or higher? > Me too please! I'm not the first one who worked on a TCL8.0 port but I found the one from Chris Frank incomplete. See: This was the starting point for me. Currently I have a tcl8.0p1 port. The test suite still crashes at some point. I use TCL as a shell via telnet. Before an public release I would like to jump to TCL8.1 and integrate the thread interface. Don't hesitate to contact me if you require more information. Martin - --------------------------------------------------------------- Dipl.-Phys. Martin Schrape schrape@atomika.com Senior Software Engineer Tel. +49 89 315 891 34 Atomika Instruments GmbH Fax +49 89 315 59 21 Bruckmannring 40 http://www.atomika.com 85764 Oberschleissheim/Munich, Germany - --------------------------------------------------------------- "We're back to the times when men were men and wrote their own device drivers." (Linus Torwald) --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Help. Shell Date: Mon, 08 Mar 1999 12:19:15 -0500 From: Tim Sohacki Organization: Nortel (Northern Telecom) Message-ID: <36E40693.660AAD70@nortelnetworks.com> References: <36E35BB4.E85DB973@integra.co.kr> <36E3AEC3.560@wireplay.bt.co.uk> <36E3C92B.4CDC16C6@integra.co.kr> You need to build vxWorks.st instead of vxWorks. Aaron Jang wrote: > = > Thank you for your interesting about my question Phil Ashby. > But I got a new problem. > The compiler display this... > = > =3D=3D > vxWorks.tmp(.text+0x1ef6): undefined reference to `standTblSize' > vxWorks.tmp(.text+0x1f0c): undefined reference to `standTblSize' > vxWorks.tmp(.text+0x1f18): undefined reference to `standTbl' > vxWorks.tmp(.text+0x1f30): undefined reference to `standTblSize' > make: *** [vxWorks] Error 1 > =3D=3D > = > Please help me. > = > -- > = > Aaron Jang (aaron@integra.co.kr,"=C0=E5=C0=CE=C1=F8") > = > Integra Telecom Co., Ltd. http://www.integra.co.kr > Hwayoung Bdg., 112-2, Garak-Dong, Tel : +82-2-449-6851 > Songpa-Gu, Seoul, 138-160, Korea Fax : +82-2-431-3741 - -- = Tim Sohacki, Nortel Networks email: sohacki@nortelnetworks.com --------------------------- Newsgroups: comp.os.vxworks Subject: Interrupt Handlers in ROM-Resident images Date: Mon, 01 Mar 1999 01:12:23 GMT From: mmigdol@bigfoot.com Organization: Deja News - The Leader in Internet Discussion Message-ID: <7bcphi$c1k$1@nnrp1.dejanews.com> Hi all, I'm working on a ROM-resident version of the 486 BSP for an embedded project. When I call intConnect to hook up an interrupt handler, I'm seeing the system to writes to the ROM area associated with the symbol excCallTable, which is accessed during the interrupt "stub" before my handler is called. It appears that this symbol, which is defined in excLiba.o was mistakenly located in the text segment as a const instead of the data segment. I'm guessing that either I'm going to need a "fixed" version of the library file from WRS or come up with a way to move that object from the text segment to the data segment in the library file myself. Of course, going RAM-resident is an option, but I'd rather not if I can avoid it. Has anyone come across this before? Thanks, Michael - -----------== Posted via Deja News, The Discussion Network ==---------- http://www.dejanews.com/ Search, Read, Discuss, or Start Your Own --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Overload new operator -- solved and thanks Date: Mon, 01 Mar 1999 09:18:25 +0800 From: "T.C. Tung" Organization: DCI HiNet Message-ID: <36D9EAE0.207ECE1F@zeon.com.tw> References: <36D25670.48D09BC5@zeon.com.tw> <36D28718.F8A8E778@zeon.com.tw> <36D3FA09.5A66FA3D@email.com> <36D53CBA.C4609AD8@zeon.com.tw> <36D59070.41C6@utsci.com> Mike Stimpson wrote: > Because you are using delete rather than delete[] and the compiler > assumes, based on this, that you are handing it a pointer to a *single* > object instead of to an *array* of objects. > ............................... > should be: > delete[] p; // now destructor will be called 10 times. You arr right, when add the delete[], all problems are gone. > Quoting from Stroustrup, "The C++ Programming Language", Second Edition, > pp. 175-6: > > "The destructor must be called for each element of an array when that > array is destroyed. This is done implicitly for arrays that are not > allocated using new. However, this cannot be done implicitly for arrays > on the free store because the compiler cannot distinguish the pointer to > a single object from a pointer to the first element of an array of > objects." > > And the solution is to use delete[] for arrays and delete for single > objects. If only I have read the book in advance...... > Note that, according to Stroustrup (p.577), "The global operator new() > and operator delete() are used for arrays of class objects." That is, > if I understand this statement correctly, you can replace new and delete > for basic types (such as ints), and for objects, and for arrays of basic > types, but not for arrays of objects. > evolution. No, we can also overload the new and delete for arrays of objects. Thanks again, T.C. --------------------------- Newsgroups: comp.os.vxworks Subject: Interrupt Handlers in ROM-Resident images Date: Mon, 01 Mar 1999 01:12:07 GMT From: mmigdol@bigfoot.com Organization: Deja News - The Leader in Internet Discussion Message-ID: <7bcph3$c1f$1@nnrp1.dejanews.com> Hi all, I'm working on a ROM-resident version of the 486 BSP for an embedded project. When I call intConnect to hook up an interrupt handler, I'm seeing the system to writes to the ROM area associated with the symbol excCallTable, which is accessed during the interrupt "stub" before my handler is called. It appears that this symbol, which is defined in excLiba.o was mistakenly located in the text segment as a const instead of the data segment. I'm guessing that either I'm going to need a "fixed" version of the library file from WRS or come up with a way to move that object from the text segment to the data segment in the library file myself. Of course, going RAM-resident is an option, but I'd rather not if I can avoid it. Has anyone come across this before? Thanks, Michael - -----------== Posted via Deja News, The Discussion Network ==---------- http://www.dejanews.com/ Search, Read, Discuss, or Start Your Own --------------------------- Newsgroups: comp.os.vxworks Subject: SENS and sockLib Date: Sun, 28 Feb 1999 18:24:05 -0700 From: Dan George Message-ID: <36D9EC35.720432A4@nsmicron.net> Reply-To: nsdgeorge@nsmicron.net SENS allows me to add my own protocol but how to I get to that module via the sockets library? I was guessing that if I called socket(AF_MYPROTOCOL, ...) I could get to my protocol but I don't see any documentation on this. I need to have some way of registering my protocol module with the sockets library. Otherwise, how does it know what to do with AF_MYPROTOCOL? Maybe I have to write my own sockets interface for my protocol? Dan --------------------------- Newsgroups: comp.arch.embedded,comp.os.vxworks Subject: Re: Wind River' Tornado 1.0.1 for Windows NT is bugged Date: Thu, 11 Mar 1999 15:46:35 GMT From: John or Jenn Organization: Who, us organized??? Message-ID: References: <7c108v$f5$1@newsreader2.core.theplanet.net> In article Mike Davies, mike_davies@noco.demon.co.uk writes: >4 Back to Outlook for the licence number (mailed > to me at the end of Jan this year and valid from > Jan 26 1999 without an expiry date) > >5 Type in the licence number. > "Your Licence has expired". :-/ :-/ :-/ Is your maintenance up to date? --------------------------- Newsgroups: comp.os.vxworks Subject: Re: How to get system time and date with ms resolution Date: Thu, 11 Mar 1999 07:54:01 -0800 From: Scott Kamara Organization: AP Labs Message-ID: <36E7E719.A1247CB1@pop.sttl.uswest.net> References: <7c7n0s$l2b@as1fw.mannesmann.de> Reply-To: spk@uswest.net > Well, it may return a value that represents the time in ns but it still > only has a resolution of system clock ticks, typically 1/60 second. That is right. The clock_gettime() support is based on the resolution of the posix master clock which is equivalent to the system clock (sysClk). This clk resolution, by typical default (bsp's do vary this default), is 1/60 second (16 ms). So, you can achieve 16 ms resolution. One can increase this resolution by doing sysClkRateSet(). BUT!, when you do this, you must accompany this by calling clock_setres() with an arg that matches the new resolution just set using sysClkRateSet(). If you do not do this, clock_gettime() will be using the wrong resolution value to determine time. Fundamentally, depending on the bsp, to get time-of-day, one must incorporate time-of-day support into the vxworks/bsp via a rtc and using the posix/ansi time functions. We sell this kind of 'bsp extension' utility, along with others, for the Motorola 68k/PPC boards for a nominal fee (including source). If you would like to know more or look at our documentation for this, let me know via email (scott@aplabs.com). Torsten Trzeciak wrote: > > Hello, > I like to get the system time with a ms resolution. Is there a kind of > getTime function? > > Torsten - -- - ----------------------- Scott Kamara AP Data Systems Seattle Engr. Office (206) 525-8592 scott@aplabs.com --------------------------- Newsgroups: comp.os.vxworks Subject: Re: workQPanic in vxWorks/apexworks Ada Date: Thu, 11 Mar 1999 08:00:06 -0800 From: Scott Kamara Organization: AP Labs Message-ID: <36E7E886.1165FCF8@pop.sttl.uswest.net> References: <7c629o$acq1@overload.lbl.gov> Reply-To: spk@uswest.net Check out the 'pretty good' faq on workQPanic at the WRS web site. (customer support:wind surf:key word search:workQPanic). I have seen workQPanic in my systems when a device (VME device in my case) goes nuts and issues many,many interrupts per second. The isr typically was not handshaking with the device correctly. William Sheehan wrote: > > Dear vxWorkers, > > I have 2 problems, that I don't know if they are related. What I'm > asking for is a process > to investigate the problems. Any suggestions to help my investigation > will be appreciated. > > The platform is a 200Mhz PPC SBC (64MB RAM) with a dual Mil-Std-1553B > PCI daughter card and a Discretes card communicating to the SBC via > Parallel Port interrupts. Externally, we have 10Mbs ethernet, BC on > 1553B, RT on 1553B, discrete signals into Discretes card (the only > active signal at this time is a 1hz clock interrupt). > The Application is Rational's ApexWorks Ada83 (3.0.0.b) with a variety > of tasking rates. It is currently loaded into the SBC and kicked off > with "-> sp __start". > > Problem #1: > I'm running an Ada application on a 200Mhz PowerPC SBC when in random > times, I see: > > workQPanic: Kernel work queue overflow. > > Press any key to stop auto-boot... > 7 > 6.... > > > The areas to investigate are possibly(low to high probability): > 1. Dual Mil-Std-1553B PCI communications > 2. sysClkRateSet 1000 (1000Hz clock for 1ms tasking resolution) > 3. Parallel Port ISR7 communicating with a Discretes card (uses vads > v_i_interrupts). > 4. Possible "bad" address w.r.t. 1553 communications. > 5. vNetTask - possible large ethernet traffic, in a "lab" environment. > > I'll add more to the list, if offered. > Normally, the processor's time taken by ISR activity is 1%. The > parallel port ISR does not > use semaphores. The processor is performing several levels of Ada > tasking. > > Problem #2: > I try to make a single load image containing vxWorks and an ADA > application (main.omf). > Whether I use all the code above in problem #1 -or- one > Put_Line("Hello...") I get this on > boot-up after Loading the image from the server: > > -> > > Exception at interrupt Level: > machine check > Exception next instruction address: 0x003ca25c > Machine Status Register: 0x0049030 > Condition Register: 0x4400c40 > Regs at 0x3d3520 > > Press any key to stop auto-boot... > < machine reboots> > > P.S. I could use any help in debugging the memory dumped at 0x3d3520. > P.S. I've waited too long for Rational to help out. They struck out > looking... > > Thank You in advance for any help that you may volunteer. > > ========================================= > Bill Sheehan > LMFS Yonkers > 914-968-2500 x2316 > william.j.sheehan@lmco.com > "Duck!! It hit the fan!" > ========================================= - -- - ----------------------- Scott Kamara AP Data Systems Seattle Engr. Office (206) 525-8592 scott@aplabs.com --------------------------- Newsgroups: comp.os.vxworks Subject: Copy from host to target FS over serial WDB? Date: Thu, 11 Mar 99 16:21:37 GMT From: nobody@nowhere Organization: NASA Goddard Space Flight Center -- Greenbelt, Maryland USA Message-ID: <7c8qf6$d0l@post.gsfc.nasa.gov> Is it possible to copy files from a host operating system disk to a vxWorks target disk over a serial WDB link? Usually when I am using the networked version of tornado, the host name shows up in the 'devs' list and I can copy files from the host to the target using the tornado shell. Now I have stripped down the kernel without networking and I would like to copy a file from the host to the target using the host tornado shell over a serial WDB link. Is this possible? Do I need networking code to do this anyway? Thanks, Alan Cudmore ( please remove the x's from the my e-mail address ) Alan.P.Cudmore.1xxxxx@gsfc.nasa.gov --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Help. Shell Date: Mon, 08 Mar 1999 15:41:26 -0500 From: Douglas Fraser Organization: Lucent Technologies Message-ID: <36E435F6.F91DAC7E@lucent.com> References: <36E35BB4.E85DB973@integra.co.kr> <36E3AEC3.560@wireplay.bt.co.uk> Phil Ashby wrote: > > Aaron Jang wrote: > > I was added "#define INCLUDE_SHELL" > > to the configdb.h file. > > And the target showed to me the prompt "->", > > but commands are can't work. > > You need the symbol table included for the shell to work. > #define INCLUDE_SYM_TBL > #define INCLUDE_STAT_SYM_TBL > #define INCLUDE_STANDALONE_SYM_TBL He may also want to continue downloads for now, so instead, INCLUDE_NET_SYM_TBL and INCLUDE_SYM_TBL_SYNC. If you set teh synch button on the target server, you can load from target or wind shell and keep things sane. Cheers. Doug Fraser dwfraser@lucent.com > > HTH, > -- > !--------------------------- Phil "Phlash" Ashby ---------------------! > ! BT Wireplay - Lead Programmer and general technical trivia supplier ! > ! Phone: 01473-644348 WWW: http://www.wireplay.com/ ! > ! Snail: B81/G41, BT Labs, Martlesham Heath, Ipswich IP5 3RE, England.! --------------------------- Newsgroups: comp.os.vxworks Subject: vxWorks SCSI support for Adaptec 1520 Date: Mon, 08 Mar 1999 21:06:52 GMT From: moyer_dick@my-dejanews.com Organization: Deja News - The Leader in Internet Discussion Message-ID: <7c1e55$oq9$1@nnrp1.dejanews.com> Wind River tells me it has SCSI 2 support for Adaptec AIC 7800. I need support for an Adaptec 1520 chip used by Ziatech in their SCSI controller. Does anyone know whether the 7800 chip is backward compatible to the 1520? Dick Moyer 650-604-0817 dmoyer@mail.arc.nasa.gov - -----------== Posted via Deja News, The Discussion Network ==---------- http://www.dejanews.com/ Search, Read, Discuss, or Start Your Own --------------------------- Newsgroups: comp.os.vxworks Subject: Interrupt Handlers in ROM-Resident images Date: Mon, 01 Mar 1999 01:12:18 GMT From: mmigdol@bigfoot.com Organization: Deja News - The Leader in Internet Discussion Message-ID: <7bcphe$c1i$1@nnrp1.dejanews.com> Hi all, I'm working on a ROM-resident version of the 486 BSP for an embedded project. When I call intConnect to hook up an interrupt handler, I'm seeing the system to writes to the ROM area associated with the symbol excCallTable, which is accessed during the interrupt "stub" before my handler is called. It appears that this symbol, which is defined in excLiba.o was mistakenly located in the text segment as a const instead of the data segment. I'm guessing that either I'm going to need a "fixed" version of the library file from WRS or come up with a way to move that object from the text segment to the data segment in the library file myself. Of course, going RAM-resident is an option, but I'd rather not if I can avoid it. Has anyone come across this before? Thanks, Michael - -----------== Posted via Deja News, The Discussion Network ==---------- http://www.dejanews.com/ Search, Read, Discuss, or Start Your Own --------------------------- Newsgroups: comp.os.vxworks Subject: Attaching a CLI to telnet Date: 11 Mar 1999 13:49:02 PST From: Chip Roberson Organization: Cerent Corporation (http://www.cerent.com) Message-ID: <36E83A4E.2B00DB4A@Cerent.com> Hello, I was looking in WindSurf for an answer to my problem and didn't find a satisfactory answer. So, I've come to you. I have a client application that wants to connect to our box using telnet and talk to a server speaking an ASCII-based language. What I'd like to do is attach a telnet daemon to a specific port and our server on the back of the telnet daemon. Basically, I just want to do the same thing Wind River did with the windShell and let telnet negotiate the terminal characteristics. >From what I have read in the docs and on the WindSurf web site, this doesn't seem possible with the WRS telnet daemon. Suggestions? Thanks, Chip - -- Charles S. Roberson This message may contain two attachments: My Business vCard and My S/MIME Digital ID --------------------------- Newsgroups: comp.os.vxworks Subject: Re: question about load vxworks Date: Thu, 11 Mar 1999 22:54:46 GMT From: radcliff@jps.net Organization: Deja News - The Leader in Internet Discussion Message-ID: <7c9hjh$tf5$1@nnrp1.dejanews.com> References: <36e5c74f.0@dnews.sbell.com.cn> For the error number, you need to know how the error codes are set up. The error codes are a 32 bit number. The high 2-bytes is the module that the error occured in. These are defined in vwModNum.h. 0xD is iosLib. The lower 2 bytes is the specific error that that module encountered, defined in the header file for that module (in this case iosLib.h). 3 is invalid file descriptor-- I think the code that loads the kernel might not be checking the return value from open for ERROR. Which brings us to why you got the error. I would need to see more parameters than the output of the load, but if you happen to be loading, as it seems, from the network you are probably trying to use FTP. You should start by making sure that you have the FTP server set up correctly with the same user name and password that you are giving the target, make sure that you have access to the directory and that the path that you give the target is correct. You can set the FTP server that comes with Tornado to do more verbose logging to help you out with that. The short of the explanation is that it can't open or find the file. In article <36e5c74f.0@dnews.sbell.com.cn>, "SRD Zha Junhai" wrote: > Hi, everybody > > I just met a question when I use the ads860 board and loading the > vxWorks from host. > Here are the parameters : > Attaching network interface cpm0... done. > Subnet Mask: 0xfffffff8 > Attaching network interface lo0... done. > Loading... > Error loading file: errno = 0xd0003 > > The procedure is very short. Could you tell me what caused this > problem and the means of errno( 0xd0003 ), by the way, please tell > me how can I find the error's mean. > > And I want to know if anyone tried to select from both socket Fd > and pipe Fd. I tried it, but select seems can only find the pipe's > activity. > > Thank you for your future response > > Best Regards > > Zha Junhai > > - -----------== Posted via Deja News, The Discussion Network ==---------- http://www.dejanews.com/ Search, Read, Discuss, or Start Your Own --------------------------- Newsgroups: comp.os.vxworks Subject: File I/O ? Date: Thu, 11 Mar 1999 15:27:03 -0800 From: "Joseph J. Lindula" Organization: California State University, Chico Message-ID: Hello, I'm new to programming in Vxworks and this is a simple question about file i/o. Here's what I've done, FILE * fd; fd = open("joes_file.txt", O_CREAT|O_WRONLY, 0777); fprintf (fd, "Hello from task %d \n",taskIdSelf() ); When I run the program, the target machine prints, joes_file: The system cannot find the file specified. Am I useing the correct syntax? Are files created on the target or are they created on the remote computer? Our target is basically a motherboard and a floppy drive. If anyone can point me in proper direction or can recommend some good web site I'd greatly appreciate it. Thanks, Joe jjlindula@netscape.net --------------------------- Newsgroups: comp.os.vxworks Subject: Re: VxWorks 5.3 Interrupt Enable Date: Thu, 11 Mar 1999 21:36:18 -0700 From: Michael Oehrli Organization: EarthLink Network, Inc. Message-ID: <36E899C2.D941C5A3@email.com> References: <36D9E627.17BDB4A4@cs.ucdavis.edu> <7bj8ro$s1l$1@nnrp1.dejanews.com> aerdogan@hc.aselsan.com.tr wrote: > In article <36D9E627.17BDB4A4@cs.ucdavis.edu>, > MAXi wrote: > > Hi All, > > > > I was just wondering how I can enable interrupt (say /INT5) in VxWorks > > 5.3 . > > I have used "intConnect" to tell where the interrupt service routine is > > so all I need to do is to actually enable the interrupt. > > > > TIA. > > > > Regards, > > > > MAXi > > -- > > ======================================================================== > > > > Maxwell CHEONG > > University of California (Davis) Address: > > Computer Engineering, Junior Rm 302, Pierce J Building > > Studnet ID#: 991-023926 Tercero Hall Circle > > Phone: (530) 754 0352 University of California (Davis) > > > > email: cheong@cs.ucdavis.edu DAVIS, CA 95616 > > ======================================================================== > > > > > > The following lines enables SCC1 interrupt. Do not forget to clear in-service > register when leaving the int. service routine. > > MsrValue = vxMsrGet(); > MsrValue &= 0xFFFF7FFF; > vxMsrSet(MsrValue); > > *CIMR(PQUICC) |= CIMR_PARAM; /* unmask SCC1 interrupt */ > *SCCM1(PQUICC) = 0x1c97; /* enable all SCC events */ > > *CICR(PQUICC) |= CICR_IRL_LEVEL_4; > > *SIMASK(PQUICC) |= SIMASK_LVM4; > > *SIMASK(PQUICC) |= SIMASK_IRM4; > > MsrValue |= 0x00008000; > vxMsrSet(MsrValue); > > Alper Erdogan > > -----------== Posted via Deja News, The Discussion Network ==---------- > http://www.dejanews.com/ Search, Read, Discuss, or Start Your Own regarding enabling interrupts after you've called IntConnect(), you would invoke sysIntEnable(). - -- Michael Oehrli Senior Software Engineer Raytheon Optical Systems, Inc. http://wwp.mirabilis.com/10888574 --------------------------- Newsgroups: comp.os.vxworks Subject: Tornado and JTAG Date: Tue, 09 Mar 1999 09:43:31 -0600 From: "John R. Cobarruvias" Organization: NASA/JSC Message-ID: <36E541A3.F64B740@orbitworld.net> Has anyone had experience with using the Tornado tools with a JTAG connect? I understand this requires a third party tool. Is this correct? How much would this cost? johncoby@orbitworld.net --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Performance impact of C++ Date: Tue, 09 Mar 1999 15:24:13 GMT From: Bill Pringlemeir Message-ID: References: <36E48392.DCE2B62F@shore.net> Sender: bpringle@DeadDuck The Plauger EC++ library has several versions. You can take libraries with or without exceptions handling and with or without templates. The original poster seemed to be concerned with efficiency. I suppose I should have recommended the Dinkumware?? library instead of the Green Hills compiler. I have actually used the Diab Data and the VxWorks/Tornado GCC for the PowerPC. Actually, Diab's tech support was really good from what I experienced. They were probably one of the most responsive companies I have dealt with. I have no idea about the SH-1. I have only limited use of VxWorks on an x86. Sorry If I mislead anyone... Bill >>>>>> Bill Pringlemeir wrote: >> Many CPUs support the Green Hills compiler. Green Hills >> provide PJ Plaugers Mark> I would recommend you NOT use green hills unless you're Mark> working on a PowerPC. I use gh on an SH-1 and it's Mark> terrible. Tech support is no help as well. - -- Brave New Waves, "http://www.radio.cbc.ca/programs/bnw/" --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Using IRQ3 interrupts on a PPC860DE Date: Thu, 11 Mar 1999 21:56:12 -0700 From: Michael Oehrli Organization: EarthLink Network, Inc. Message-ID: <36E89E6C.7F80317C@email.com> References: <7bjps5$atn$1@news.monmouth.com> Rich Duszczak wrote: > Hello and thanks for reading my message. > > I am running vxWorks on our proprietary development board using an MPC860DE > processor with a bus speed of 25MHz. I am trying to install a handler to > service an external interrupt generated on /IRQ3. Here is the relevant > code: > > VOIDFUNCPTR irq3handler(int *unused) > { > /* irq3Count++ ; > logMsg("IRQ3 Asserted (%d)\n", irq3Count, 0, 0, 0, 0, 0) ;*/ > } > > void irq3init() > { > *SIUMCR(vxImmrGet()) &= ~(SIUMCR_DPC) ; > START_TIMING /* Put IRQ3 into non-asserted state */ > > intConnect(IV_IRQ3, (VOIDFUNCPTR)irq3handler, 0) ; > > /* intDisable(IV_IRQ3) ;*/ > intEnable(IV_IRQ3) ; > } > > void irq3(int state) > { > switch(state) > { > case 0: > STOP_TIMING /* Assert IRQ3 */ > break; > > case 1: > START_TIMING /* De-assert IRQ3 */ > break; > > case 2: > STOP_TIMING > START_TIMING > break; > } > } > > irq3handler: > I have tried irq3handler as both an empty function and with a simple logMsg > call. From what I can see it never reaches this code. > > irq3init: > Setting the DPC bit in the SIUMCR register configures the appropriate > multiplexed pins on the processor as /IRQx. START_TIMING is a macro that > forces /IRQ3 high (non-asserted). I then connect to and enable/disable the > interrupt. When disabled, I can use the irq3 function to toggle the /IRQ3 > line. When the interrupt is enabled and I try to assert /IRQ3 the board > locks and the ICE reports a check stop exception at address 0x22fcc. I used > the command 'lkAddr 0x22fcc' to find out that 0x22fcc is the beginning of > excIntHandle(). > > My question is what could cause the system to fail at that point when > interrupts are enabled and a handler is installed for /IRQ3? > > I have verified that the /IRQ3 line is being toggled (it is wired to a > general purpose output pin so I can assert it when I want to) and that an > interrupt is being generated by the processor. I've tried playing with the > SIU interrupt registers but nothing has changed. > > Thank you in advance for any help you can provide, > > Rich Duszczak i am not familiar with your board, but i do have some general comments regarding your use of intConnect and intEnable. the first argument of intConnect is (VOIDFUNCPTR *), while the argument to intEnable is an int (level). you seem to be treating these two arguments as identical, but they usually aren't. here's an example of how i've set up an ISR: #include #include #define INTR_LEVEL 0x07 #define INTR_VECTOR_BASE 0xa8 . . . intConnect((VOIDFUNCPTR *) INUM_TO_IVEC(INTR_LEVEL + INTR_VECTOR_BASE), (VOIDFUNCPTR) semGive, (int) semId); sysIntEnable(INTR_LEVEL); . . . my ISR is actually just a semGive. it gives a semaphore which another process is waiting on. INUM_TO_IVEC() is a macro/function define in either iv.h or intLib.h. you should note that the interrupt vector is INTR_LEVEL+INTR_VECTOR_BASE. - -- Michael Oehrli Senior Software Engineer Raytheon Optical Systems, Inc. http://wwp.mirabilis.com/10888574 --------------------------- Newsgroups: comp.os.vxworks Subject: Re: File I/O ? Date: Fri, 12 Mar 1999 06:07:16 GMT From: bobarmstrong@home.net (Robert Armstrong) Organization: @Home Network Message-ID: References: Either you didn't cd to a useable directory or you used cd on the host in windShell rather than on the target. Either use @cd, @pwd and @ls to make them work in the target context, give a full path name for your file or put the cd in the application too. Bob In article , "Joseph J. Lindula" wrote: >Hello, I'm new to programming in Vxworks and this is a simple question >about file i/o. Here's what I've done, > >FILE * fd; >fd = open("joes_file.txt", O_CREAT|O_WRONLY, 0777); >fprintf (fd, "Hello from task %d \n",taskIdSelf() ); > >When I run the program, the target machine prints, >joes_file: The system cannot find the file specified. > >Am I useing the correct syntax? > >Are files created on the target or are they created on the remote >computer? Our target is basically a motherboard and a floppy drive. >If anyone can point me in proper direction or can recommend some good web >site I'd greatly appreciate it. > >Thanks, > >Joe > > >jjlindula@netscape.net > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: workQPanic in vxWorks/apexworks Ada Date: Fri, 12 Mar 1999 06:23:18 GMT From: bobarmstrong@home.net (Robert Armstrong) Organization: @Home Network Message-ID: References: <7c629o$acq1@overload.lbl.gov> What Scott says is correct, a task is unlikely to do this. Basically the kernel is not being allowed enough time to clear kernel mode work out of the queue. If you have a logic analyzer try hooking it to the IRQ lines. You could try WindView, if it's just gradually filling up you may see why, other than that it's imagination and hardware tools time. Bob In article <7c629o$acq1@overload.lbl.gov>, William Sheehan wrote: >Dear vxWorkers, > >I have 2 problems, that I don't know if they are related. What I'm >asking for is a process >to investigate the problems. Any suggestions to help my investigation >will be appreciated. > >The platform is a 200Mhz PPC SBC (64MB RAM) with a dual Mil-Std-1553B >PCI daughter card and a Discretes card communicating to the SBC via >Parallel Port interrupts. Externally, we have 10Mbs ethernet, BC on >1553B, RT on 1553B, discrete signals into Discretes card (the only >active signal at this time is a 1hz clock interrupt). >The Application is Rational's ApexWorks Ada83 (3.0.0.b) with a variety >of tasking rates. It is currently loaded into the SBC and kicked off >with "-> sp __start". > > >Problem #1: >I'm running an Ada application on a 200Mhz PowerPC SBC when in random >times, I see: > > workQPanic: Kernel work queue overflow. > > Press any key to stop auto-boot... > 7 > 6.... > > >The areas to investigate are possibly(low to high probability): >1. Dual Mil-Std-1553B PCI communications >2. sysClkRateSet 1000 (1000Hz clock for 1ms tasking resolution) >3. Parallel Port ISR7 communicating with a Discretes card (uses vads >v_i_interrupts). >4. Possible "bad" address w.r.t. 1553 communications. >5. vNetTask - possible large ethernet traffic, in a "lab" environment. > >I'll add more to the list, if offered. >Normally, the processor's time taken by ISR activity is 1%. The >parallel port ISR does not >use semaphores. The processor is performing several levels of Ada >tasking. > >Problem #2: >I try to make a single load image containing vxWorks and an ADA >application (main.omf). >Whether I use all the code above in problem #1 -or- one >Put_Line("Hello...") I get this on >boot-up after Loading the image from the server: > > -> > > Exception at interrupt Level: > machine check > Exception next instruction address: 0x003ca25c > Machine Status Register: 0x0049030 > Condition Register: 0x4400c40 > Regs at 0x3d3520 > > Press any key to stop auto-boot... > < machine reboots> > >P.S. I could use any help in debugging the memory dumped at 0x3d3520. >P.S. I've waited too long for Rational to help out. They struck out >looking... > >Thank You in advance for any help that you may volunteer. > >========================================= >Bill Sheehan >LMFS Yonkers >914-968-2500 x2316 >william.j.sheehan@lmco.com >"Duck!! It hit the fan!" >========================================= > > > > --------------------------- Newsgroups: comp.os.vxworks Subject: For memory protection... Date: Fri, 12 Mar 1999 15:35:59 +0900 From: "À±°­½Ä" Organization: Korea Telecom Message-ID: <7cacej$r23$1@news3.kornet.nm.kr> Hi! I'm Kang-Sik Yoon in Korea. I recently have tested for memory protection. That is, when using the memory allocation routines of VxWorks, VxWorks cannot always gaurantee that the memory allocated through the routines is allocated continously, and almost all of allocated memories are most likely fragmented in the entire memory. What we are going to do now is to write data stream(e.g, bitstream files) into a continous free block, not fragmented block. So, we would like know how we can allocate a specific continuous memory block into target memory so that even kernel cannot control the allocated block. In other words, it is the very memory protection problem or to secure user-defined and continuous block. Please let me know if anyone has solutions on the problem. Thanks in advance..... --------------------------- Newsgroups: comp.os.vxworks Subject: Re: How to get system time and date with ms resolution Date: Fri, 12 Mar 1999 06:10:31 GMT From: bobarmstrong@home.net (Robert Armstrong) Organization: @Home Network Message-ID: References: <7c7n0s$l2b@as1fw.mannesmann.de> If you want an accurate clock try using NTP to synchronize a clock run from the aux clock and set the aux clock rate high. The system clock should not be run fast or you will thrash the kernel and waste performance. Bob In article <7c7n0s$l2b@as1fw.mannesmann.de>, "Torsten Trzeciak" wrote: >Hello, >I like to get the system time with a ms resolution. Is there a kind of >getTime function? > >Torsten > > --------------------------- Newsgroups: comp.os.vxworks Subject: DDD with Tornado? Date: Fri, 12 Mar 1999 09:35:34 +0100 From: Marcus Karlsson Organization: Ericsson Erisoft AB Message-ID: <36E8D1D6.40B8EFB@lu.erisoft.se> This is a multi-part message in MIME format. - --------------3D62CEC1DCA711565459626C Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Could someone tell me how to use ddd instead of crosswind in tornado? Thanks, Marcus - --------------3D62CEC1DCA711565459626C Content-Type: text/x-vcard; charset=us-ascii; name="Marcus.Karlsson.vcf" Content-Transfer-Encoding: 7bit Content-Description: Card for Marcus Karlsson Content-Disposition: attachment; filename="Marcus.Karlsson.vcf" begin:vcard n:Karlsson;Marcus tel;home:+46-920-13579 tel;work:+46-920-202809 x-mozilla-html:TRUE url:http://www.epl.ericsson.se/~eplmaka org:Ericsson Erisoft AB;Wideband Cellular Systems version:2.1 email;internet:Marcus.Karlsson@ericsson.com title:W-CDMA System Enginer adr;quoted-printable:;;Landsgatan 3=0D=0A972 41;Luleå;;;Sweden x-mozilla-cpt:;27648 fn:Marcus Karlsson end:vcard - --------------3D62CEC1DCA711565459626C-- --------------------------- Newsgroups: comp.os.vxworks Subject: Re: File I/O ? Date: Fri, 12 Mar 1999 10:28:32 +0000 From: Phil Ashby Organization: BT Wireplay Message-ID: <36E8EC50.5BBB@wireplay.bt.co.uk> References: Joseph J. Lindula wrote: > FILE * fd; > fd = open("joes_file.txt", O_CREAT|O_WRONLY, 0777); > fprintf (fd, "Hello from task %d \n",taskIdSelf() ); Did you really use open(), not fopen()? If you want to use raw file descriptor I/O then fdprintf() should help things along a bit :) - -- !--------------------------- Phil "Phlash" Ashby ---------------------! ! BT Wireplay - Lead Programmer and general technical trivia supplier ! ! Phone: 01473-644348 WWW: http://www.wireplay.com/ ! ! Snail: B81/G41, BT Labs, Martlesham Heath, Ipswich IP5 3RE, England.! --------------------------- Newsgroups: comp.os.vxworks Subject: Re: For memory protection... Date: Fri, 12 Mar 1999 10:33:37 +0000 From: Phil Ashby Organization: BT Wireplay Message-ID: <36E8ED81.7C0C@wireplay.bt.co.uk> References: <7cacej$r23$1@news3.kornet.nm.kr> Kang-Sik Yoon wrote: > What we are going to do now is to write data stream(e.g, bitstream files) > into a continous free block, not fragmented block. So, we would like know > how we can allocate a specific continuous memory block into target memory so > that even kernel cannot control the allocated block. > In other words, it is the very memory protection problem or to secure > user-defined and continuous block. You can either: [1] malloc() a block of the required size early in the program startup (where it is more likely to work). [2] reserve a block of memory at the top of system RAM when building the kernel by putting '#define USER_RESERVED_MEM ' in your config.h file. Option 2 will guarantee the location of the memory and will also ensure that the kernel does not touch it. You may need to adjust the physical memory descriptor table (sysPhysMemDesc[]) in sysLib.c such that this block of RAM is still accessible via the MMU. HTH, - -- !--------------------------- Phil "Phlash" Ashby ---------------------! ! BT Wireplay - Lead Programmer and general technical trivia supplier ! ! Phone: 01473-644348 WWW: http://www.wireplay.com/ ! ! Snail: B81/G41, BT Labs, Martlesham Heath, Ipswich IP5 3RE, England.! --------------------------- Newsgroups: comp.os.vxworks Subject: 3COM EtherLink III on x86 Date: Tue, 9 Mar 1999 17:07:39 -0500 From: "Bob Beers" Organization: CyberGate, Inc. Message-ID: <7c464i$2btq$1@news.gate.net> I suspect that this is a well known and often solved problem, but I don't have the answer yet, so I'm asking the experts: Less than 5% of the time, on power-up, my x86 target hangs during the usrNetInit for subject card (elt0). I have never had any trouble after initialization completes, just every now and then I get a Attaching network interface elt0...failed. A power cycle or soft reboot usually results in a successful load, but I'd like to avoid even these occasional hangs. I haven't collected much data since during development it's pretty easy to clear the problem, but once it's in the field, I'd like the product to boot every time. What's the story? Bob IPAXS Corp. --------------------------- End of New-News digest ********************** From owner-vxwexplo-process Sat Mar 13 04:00:52 1999 From: daemon@csg.lbl.gov Date: Sat Mar 13 04:00:57 PST 1999 Subject: comp.os.vxworks newsdigest Status: O Comp.Os.Vxworks Daily Digest Sat Mar 13 04:00:47 PST 1999 Subject: Re: Attaching a CLI to telnet Subject: Anybody using MPC8240 or MPC8260? Subject: Re: DDD with Tornado? Subject: Where to find docs on x86 GNU assembler Subject: Re: How to get system time and date with ms resolution Subject: Re: "Relocation value does not fit in 24 bits" errors Subject: Re: Where to find docs on x86 GNU assembler Subject: Spawn 2 tasks? Subject: Re: Help urgently!! Subject: Re: File I/O ? Subject: Re: File I/O ? Subject: Re: Virtual Console Subject: Re: File I/O ? Subject: Re: Anybody using MPC8240 or MPC8260? Subject: Re: Help urgently!! Subject: MAC query Subject: Re: question about load vxworks Subject: Re: Hard reset on MPC860 Subject: Re: MAC query Subject: vxWorks and typeahead (serial and telnet) Subject: Re: "Relocation value does not fit in 24 bits" errors Subject: advantages of SENS Subject: Re: WindNet SNMP (was: Are multiple agent tasks possible) Subject: Re: Debugger Support For Ethernet & Kernel drivers Subject: Disk suggestions and question Subject: Re: PCMCIA pcata Cards ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: Attaching a CLI to telnet Date: 12 Mar 1999 13:11:53 GMT From: chc@nasa2.ksc.nasa.gov (Charles H. Chapman) Organization: NASA, Kennedy Space Center Message-ID: References: <36E83A4E.2B00DB4A@Cerent.com> On 11 Mar 1999 13:49:02 PST, Chip Roberson wrote: >I was looking in WindSurf for an answer to my problem and didn't find >a satisfactory answer. So, I've come to you. > >I have a client application that wants to connect to our box using >telnet and talk to a server speaking an ASCII-based language. What >I'd like to do is attach a telnet daemon to a specific port and our >server on the back of the telnet daemon. Basically, I just want >to do the same thing Wind River did with the windShell and let >telnet negotiate the terminal characteristics. > >>From what I have read in the docs and on the WindSurf web site, >this doesn't seem possible with the WRS telnet daemon. > >Suggestions? I'd pattern your program after the TCP/IP networking example they give in the VxWorks Programmer's Guide (pp. 224-229 in the 5.3 version of that manual). Chuck --------------------------- Newsgroups: comp.os.vxworks Subject: Anybody using MPC8240 or MPC8260? Date: Fri, 12 Mar 1999 13:33:34 GMT From: lourens@mecalc.co.za (Lourens Geldenhuys) Organization: Mecalc (Pty) Ltd Message-ID: <36e91698.28046093@modem.mecalc.co.za> Reply-To: lourens@mecalc.co.za Hi 1. Is anybody running VxWorks on the Motorola MPC8240 or MPC8260? 2. Does a BSP exist for any boards with these devices? Regards Lourens Geldenhuys ================================================================ 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: DDD with Tornado? Date: Fri, 12 Mar 1999 14:46:20 +0100 From: Wolfgang Meisl Organization: VIC - VLT Instrument Consortium Message-ID: <36E91AAC.D49192B@bigbang.usm.uni-muenchen.de> References: <36E8D1D6.40B8EFB@lu.erisoft.se> Marcus Karlsson wrote: > > Could someone tell me how to use ddd instead of crosswind in tornado? Simple, at least if you use the GNU toolkit. You just have to find out the name of the underlying debugger. In my case it is called gdbm68k (It is located in the same directory as crosswind). Start ddd as follows: ddd --debugger gdbm68k Works pretty fine. The only drawbacks are due to the old revision (4.12) of my debugger. If you want e.g. symbol completion, you must compile gdb yourself. I use 4.16. Of course you loose the wtx interface this way and you also need a kernel that supports the rdb debugging interface. But IMHO rdb is more useful, anyhow. Ever tried with crosswind to catch a task that gets spawned dynamically? ciao, Wolfgang - -------------------------------------------------------------------------- Wolfgang Meisl Universitaets-Sternwarte Muenchen wmeisl@bigbang.usm.uni-muenchen.de Tel ++49-+89-2180-5969 http://bigbang.usm.uni-muenchen.de:8002/~wmeisl/ - -------------------------------------------------------------------------- --------------------------- Newsgroups: comp.os.vxworks Subject: Where to find docs on x86 GNU assembler Date: Fri, 12 Mar 1999 13:53:40 GMT From: "Rick Stone" Organization: PSINet Message-ID: My company is new to VxWorks. We're porting a large body of code to VxWorks from our previous system. Most of this code is C, but a small amount is 80x86 assembler. There are detail differences in the assembler syntax that are driving us batty. (For example, I'd hate to tell you how long it took to figure out that hex constants had to start with a '$' character!) Pleeeeeeze..., Can anyone tell me where to find a manual describing this assembler? --------------------------- Newsgroups: comp.os.vxworks Subject: Re: How to get system time and date with ms resolution Date: Fri, 12 Mar 1999 06:43:24 -0800 From: Scott Kamara Organization: AP Labs Message-ID: <36E9280C.720D2F93@pop.sttl.uswest.net> References: <7c7n0s$l2b@as1fw.mannesmann.de> Reply-To: spk@uswest.net > If you want an accurate clock try using NTP to synchronize a clock run from > the aux clock and set the aux clock rate high. The system clock should not be > run fast or you will thrash the kernel and waste performance. Definitely. Upping the sys clk rate to gain 1 ms resolution is not advisable since much internal kernel activity would be triggered and wasted. But, with the aux clk, there is a price to pay too. To get 1 ms resolution on the aux clk means 1000 'tick' interrupts per second are generated and must be serviced. But, its better than using the sys clk for this. > Bob > > In article <7c7n0s$l2b@as1fw.mannesmann.de>, "Torsten Trzeciak" > wrote: > >Hello, > >I like to get the system time with a ms resolution. Is there a kind of > >getTime function? > > > >Torsten > > > > - -- - ----------------------- Scott Kamara AP Data Systems Seattle Engr. Office (206) 525-8592 scott@aplabs.com --------------------------- Newsgroups: comp.os.vxworks Subject: Re: "Relocation value does not fit in 24 bits" errors Date: Wed, 10 Mar 1999 10:05:49 -0500 From: Charles Curley Organization: Dynacs Engineering Co., Inc. Message-ID: <36E68A4D.C5BBC94B@orion.ksc.nasa.gov> References: <36E67F94.241C43ED@orion.ksc.nasa.gov> After messing with this some more I've discovered a command called objectdump. We've never had these kind of problem before so don't have a whole lot of experience digging thru object files. Do a "objectdumpppc - -reloc vmeLib.out | grep REL24" gives the following: 000001c0 R_PPC_REL24 __builtin_delete 000001ec R_PPC_REL24 __builtin_delete 00000230 R_PPC_REL24 __builtin_delete 000002e8 R_PPC_REL24 __builtin_vec_new 00003654 R_PPC_REL24 __builtin_new 0000372c R_PPC_REL24 __builtin_new 00003c48 R_PPC_REL24 __builtin_new 00003c7c R_PPC_REL24 __builtin_new 0000419c R_PPC_REL24 strlen which corresponds to the number of "Relocation ..." error message we are seeing. In effort to remove these I added a -fno-builtin to the compile line and that got rid of the strlen, but the new and delete calls still remain relative. How can I force them to be longcalls? - -- Charles Curley, Senior Software Engineer, Dynacs Engineering Co., Inc. Engineering Development Contract, Kennedy Space Center, Florida --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Where to find docs on x86 GNU assembler Date: Fri, 12 Mar 1999 16:51:35 +0000 From: Phil Ashby Organization: BT Wireplay Message-ID: <36E94617.3C31@wireplay.bt.co.uk> References: Rick Stone wrote: > Pleeeeeeze..., Can anyone tell me where to find a manual describing this > assembler? I found a tutorial here (in about 2 minutes with Altavista :): http://www.gameprog.com/codex/tut/att_asm.html No doubt there will be others. I seem to remember that somebody produced a conversion tool (probably on Linux). - -- !--------------------------- Phil "Phlash" Ashby ---------------------! ! BT Wireplay - Lead Programmer and general technical trivia supplier ! ! Phone: 01473-644348 WWW: http://www.wireplay.com/ ! ! Snail: B81/G41, BT Labs, Martlesham Heath, Ipswich IP5 3RE, England.! --------------------------- Newsgroups: comp.os.vxworks Subject: Spawn 2 tasks? Date: Tue, 9 Mar 1999 19:05:59 -0800 From: "Joseph J. Lindula" Organization: California State University, Chico Message-ID: Hello, I'm new at programming in VxWorks so this will be a simple question. I want a program to spawn two more tasks when it executes. Can I just use a for loop and have the taskSpawn() only execute if its the original task? Thanks, Joe jjlindula@netscape.net --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Help urgently!! Date: Tue, 9 Mar 1999 19:36:25 -0800 From: "Steve Doiel" Message-ID: <36e5e747.0@news.pacifier.com> References: <36E55153.4B03@pegasus.cc.ucf.edu> Just guessing... but look into vmBaseStateSet in order to mark the region of memory associated with the hardware as non-cached. You may very well be updating a value in cache that just isn't getting written to the physical device. SteveD Lei Sun wrote in message <36E55153.4B03@pegasus.cc.ucf.edu>... >Hi: > I am developing device driver for VMI/VME4514A analogue data >acquisition board using Vx-Work and Tornado tools. My BSP is MVME >167,host is sun ultrasparc station. > I have done evry thing in order. As first step, I didn't write actual >code in xxRead(),xxWrite().. routine. however ,in the xxInit()routine I >wrote: > LOCAL STATUS VMI4514AInit(WORD *addr) >{ > printf("The base address of VMI4514A is %d\n",addr); > *(addr+1)=0x4000; > printf("Fail Led is off and auto-scanning mode is on\n"); >} > According to the hardware documentation, the board memory mapped into >VMEbus at xxxx0000. The offset of 02H is Control Status Register(CSR), >if the bit 14 is set to 1,the "Fail" Led will be off and auto-scanning >mode will be on. > So I compiled the program and loaded into the target,and in debug mode >,I pass the bass address of the board(0x0000) to the install function >and run it. Everything seems in order,but the "fail" Led is still on. >I can't descript the whole procedure. If anyone have experience on >device driver, or on VMI/VME4514A,contact me >please(ls00722@pegasus.cc.ucf.edu). I need your help. I have been >confused with this problem for a week. > > I attached the whole code,if interested. you may read it and give me >any sugguestion. > >thank you! > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: File I/O ? Date: Fri, 12 Mar 1999 13:16:54 +0000 From: Martin Schrape Organization: Atomika Instruments GmbH Message-ID: <36E913C6.9C3EB95C@atomika.com> References: Hi Joe, this is really confusing, but if you read the reference manunal carefully you learn that only on NFS devices open() works with O_CREAT. To create a file on other devices like e.g. a local DOS file system use creat() FILE * fd; fd = creat("joes_file.txt", O_WRONLY); fprintf (fd, "Hello from task %d \n",taskIdSelf() ); should work. So open() in VxWorks is not a POSIX function. - -Martin Joseph J. Lindula wrote: > > Hello, I'm new to programming in Vxworks and this is a simple question > about file i/o. Here's what I've done, > > FILE * fd; > fd = open("joes_file.txt", O_CREAT|O_WRONLY, 0777); > fprintf (fd, "Hello from task %d \n",taskIdSelf() ); > > When I run the program, the target machine prints, > joes_file: The system cannot find the file specified. > > Am I useing the correct syntax? > > Are files created on the target or are they created on the remote > computer? Our target is basically a motherboard and a floppy drive. > If anyone can point me in proper direction or can recommend some good web > site I'd greatly appreciate it. > > Thanks, > > Joe > > jjlindula@netscape.net - -- - --------------------------------------------------------------- Dipl.-Phys. Martin Schrape schrape@atomika.com Senior Software Engineer Tel. +49 89 315 891 34 Atomika Instruments GmbH Fax +49 89 315 59 21 Bruckmannring 40 http://www.atomika.com 85764 Oberschleissheim/Munich, Germany - --------------------------------------------------------------- --------------------------- Newsgroups: comp.os.vxworks Subject: Re: File I/O ? Date: Fri, 12 Mar 1999 18:35:48 GMT From: Ayyadevara Seetharma Sarma Message-ID: <36E95E4B.A171274@x.holontech.com> References: Reply-To: sarma@holontech.com hi FILE * fd; fd = fopen("joes_file.txt", rw); fprintf (fd, "Hello from task %d \n",taskIdSelf() ); I think above will work sarma Joseph J. Lindula wrote: > Hello, I'm new to programming in Vxworks and this is a simple question > about file i/o. Here's what I've done, > > FILE * fd; > fd = open("joes_file.txt", O_CREAT|O_WRONLY, 0777); > fprintf (fd, "Hello from task %d \n",taskIdSelf() ); > > When I run the program, the target machine prints, > joes_file: The system cannot find the file specified. > > Am I useing the correct syntax? > > Are files created on the target or are they created on the remote > computer? Our target is basically a motherboard and a floppy drive. > If anyone can point me in proper direction or can recommend some good web > site I'd greatly appreciate it. > > Thanks, > > Joe > > jjlindula@netscape.net --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Virtual Console Date: Wed, 10 Mar 1999 11:58:24 -0500 From: Lei Sun Organization: University of Central Florida Message-ID: <36E6A4B0.417@pegasus.cc.ucf.edu> References: <7c58bd$6or$1@news3.kornet.nm.kr> I also have this problem a couple of days ago. but I figure out the name of the host I am using(Sun sparc Station) and put the name into the "Console Display" textbox in the target create form. It works. you also need to redirect the I/O to your virtul console. hope this helpful. - ---------- lei sun University of Central Florida U.S.A À±°­½Ä wrote: > > How are you!! > > My name is Kang-Sik Yoon, working for Daewoo Elec. Co. LTD., in Korea. > > I want to know how to make "Virtual Console" be displayed in Target-Server > Window. > Of course, I have tried to change parameters associated with "Virtual > Console" column in Target Server Tool bar and seen "-V" option when > executing target-server. > But, finally failed to install "Virtual Console" in Target-Server > Window..... > > If there are any people knowing this solution, please let me know.. > > Thanks in advance for kind people giving solutions to me for the future... --------------------------- Newsgroups: comp.os.vxworks Subject: Re: File I/O ? Date: Fri, 12 Mar 1999 17:22:45 -0500 From: Peter Message-ID: <36E993B5.26B45832@metrolink.net> References: "Joseph J. Lindula" wrote: > > FILE * fd; > fd = open("joes_file.txt", O_CREAT|O_WRONLY, 0777); > fprintf (fd, "Hello from task %d \n",taskIdSelf() ); > The filename might need to meet the 8.3 filenaming convention. Try renaming the file to something like "joesfile.txt" > > When I run the program, the target machine prints, > joes_file: The system cannot find the file specified. > jjlindula@netscape.net - -- Peter - ----------------------------------------------- E-mail: funonh2o@metrolink.net (home) http://www.geocities.com/Colosseum/Arena/2475/ - ----------------------------------------------- --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Anybody using MPC8240 or MPC8260? Date: Fri, 12 Mar 1999 17:23:57 -0500 From: Garo Senyildiz Organization: Nortel Message-ID: <36E993FD.32E24D17@nortel.ca> References: <36e91698.28046093@modem.mecalc.co.za> Below is what we got from our sales rep wrt MPC8260. Hope it helps. Garo Gentlemen: We have decided to produce a limited Early Adopter Release for customers who must have the BSP now for Motorola's PowerQUICC 2 board; according to the factory, it must be the WRS sales teams that will support the EAR, until full availability is reached in June. The EAR is targeted for mid-March, and is defined as follows. Supported MPC8260 & VADS Board Features for EAR: 1. Basic EC603e CPU core with D-CACHE, I-CACHE, and MMU 2. SIU and CPM support: memory controller and interrupt controller support 3. 16-64 MB SDRAM DIMM. 16, 32, and 64 MB configurations will be tested. This is the memory off the 60x bus. 4. 2 RS-232 Ports - Ports SCC1 and SCC2 are connected to onboard connectors. Both serial ports will be supported. 5. 10 Mbs FCC Ethernet - FCC2 is wired to an RJ-45 connector on board. Only 10Mbs will be supported for EAR. Note, the 100Mbs is not the same as the 860T and therefore there is no leverage there. 6. Sys, aux, timestamp clocks will be supported 7. The DMA capability will only exist for supported devices, not to the virtual DMA feature. Not supported: 1. 4MB SDRAM on PCI Local Bus 2. 8-32MB Flash SIMM support as the primary boot device. Motorola VBUG will provide the downloading utility. No Physical bootROMs will be shipped. Will test 8, 16, and 32 MB systems. 3. 100Mbit ethernet option. 4. Basic I2C support for automatic DRAM detection and initialization. If any of you want early access, please let me know; we must get a special form signed for EAR downloads. Any other comments or requests can be directed to Dave or I. Lourens Geldenhuys wrote: > > Hi > > 1. Is anybody running VxWorks on the Motorola MPC8240 or MPC8260? > > 2. Does a BSP exist for any boards with these devices? > > Regards > Lourens Geldenhuys > > ================================================================ > 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: Help urgently!! Date: Wed, 10 Mar 1999 07:18:21 GMT From: johnchann@home.net (johnchann) Organization: @Home Network Message-ID: <11pF2.12435$FZ5.3039@news.rdc1.sfba.home.com> References: <36E55153.4B03@pegasus.cc.ucf.edu> It is rather unlikely that your card is actually located at address 0x00000000 on the VME bus. I notice that you define the address from the documentation as 0xXXXX0000. I'd expect that XXXX to be configured somehow. Leaving that aside there is generally a mapping between the local and the bus addresses. You should be using:- sysLocalToBusAdrs(int adrsSpace,char *localAdrs, char **pBusAdrs) To work out the local address of your card, the VME interface is NOT necessarily a 1:1 mapping. HTH JRC In article <36E55153.4B03@pegasus.cc.ucf.edu>, Lei Sun wrote: >This is a multi-part message in MIME format. > >--------------46142D362668 >Content-Type: text/plain; charset=us-ascii >Content-Transfer-Encoding: 7bit > >Hi: > I am developing device driver for VMI/VME4514A analogue data >acquisition board using Vx-Work and Tornado tools. My BSP is MVME >167,host is sun ultrasparc station. > I have done evry thing in order. As first step, I didn't write actual >code in xxRead(),xxWrite().. routine. however ,in the xxInit()routine I >wrote: > LOCAL STATUS VMI4514AInit(WORD *addr) >{ > printf("The base address of VMI4514A is %d\n",addr); > *(addr+1)=0x4000; > printf("Fail Led is off and auto-scanning mode is on\n"); >} > According to the hardware documentation, the board memory mapped into >VMEbus at xxxx0000. The offset of 02H is Control Status Register(CSR), >if the bit 14 is set to 1,the "Fail" Led will be off and auto-scanning >mode will be on. > So I compiled the program and loaded into the target,and in debug mode >,I pass the bass address of the board(0x0000) to the install function >and run it. Everything seems in order,but the "fail" Led is still on. >I can't descript the whole procedure. If anyone have experience on >device driver, or on VMI/VME4514A,contact me >please(ls00722@pegasus.cc.ucf.edu). I need your help. I have been >confused with this problem for a week. > > I attached the whole code,if interested. you may read it and give me >any sugguestion. > >thank you! > >--------------46142D362668 >Content-Type: text/plain; charset=us-ascii; name="vmi4514a.c" >Content-Transfer-Encoding: 7bit >Content-Disposition: inline; filename="vmi4514a.c" > >#include "vxWorks.h" >#include "iosLib.h" >#include "vmi4514a.h" > >LOCAL int VMI4514ADrvNum; > >LOCAL Channel_Desc vmi4514a_channel[2] ; /*0-> output,1->input */ > >LOCAL int VMI4514AOpen(Channel_Desc *channel,char *remainder,int mode) >{ > printf("Entered in open\n"); > return((int)channel); >} > >LOCAL VMI4514AClose(Channel_Desc *channel) >{ > printf("Entered in close\n"); > return(OK); >} > >LOCAL VMI4514ARead(Channel_Desc *channel, char *buffer,int nBytes) >{ > printf("Entered in read\n"); > return(OK); >} > >LOCAL VMI4514AWrite(Channel_Desc *channel, char *buffer, int nBytes) >{ > printf("Entered in write\n"); > return(OK); >} > >LOCAL int VMI4514AIoctl(Channel_Desc * channel, int requestCode) >{ > printf("Entered in ioctl\n"); > return(OK); >} > >LOCAL STATUS VMI4514AInit(WORD *addr) >{ > printf("The base address of VMI4514A is %d\n",addr); > *(addr+1)=0x4000; > printf("Fail Led is off and auto-scanning mode is on\n"); >} > >LOCAL STATUS VMI4514ADrv(WORD *addr) >{ >/* check if driver is already installed */ >if (VMI4514ADrvNum > 0) > { printf("Already installed ...\n"); > return(ERROR); > } >/* install the driver */ > >printf ("Installing the VMI4514A driver\n"); >VMI4514ADrvNum=iosDrvInstall(0,0,VMI4514AOpen,VMI4514AClose,VMI4514ARead,VMI451 >4AWrite,VMI4514AIoctl); > >/* Init the borad */ > >VMI4514AInit(addr); >/* return OK or ERROR */ > >return (VMI4514ADrvNum == ERROR ? ERROR : OK); >} > >/******************/ > >LOCAL STATUS VMI4514ADevCreate(char * name) >{ > static i=0; > int status; > Channel_Desc *temp; > >/*check if driver installed */ > > if (VMI4514ADrvNum<=0) > { > errnoSet(S_ioLib_NO_DRIVER); > return (ERROR); > } > > if (i>=2) > i=0; > temp= &vmi4514a_channel[i++]; > if ((status=iosDevAdd(&(temp->devHdr),name,VMI4514ADrvNum)) == ERROR) > { > printf("ERROR. device %s not added to the device list\n\n",name); > return (ERROR); > } > return (status); > >} > > >/**********************************************************/ > > >void VMI4514AInstall (int address) >{ > WORD *base =(WORD *) address; > char devName[16]; > int status; > > if (VMI4514ADrv(base) == ERROR) > { > printf("\n Error installing VMI4514A Driver\n"); > exit(1); > } >// printf("Progam has created driver number\n"); > vmi4514a_channel[0].base=base+10; > vmi4514a_channel[1].csr =(base+1); > vmi4514a_channel[0].in_use=FALSE; > sprintf(vmi4514a_channel[0].name, "%s","outp"); > sprintf(devName,"%s","/vmi4514a/outp"); > status=VMI4514ADevCreate(devName); > vmi4514a_channel[1].base=base+20; > vmi4514a_channel[1].csr=(base+1); > vmi4514a_channel[1].in_use=FALSE; > sprintf(vmi4514a_channel[1].name, "%s","inp"); > sprintf(devName,"%s","/vmi4514a/inp"); > status=VMI4514ADevCreate(devName); > > // printf("\n Installing the VMI4514A board channels...\n\n\"); > printf("channel /vmi4514a/outp is output channel.\n"); > printf("channel /vmi4514a/inp is input channel.\n\n"); >} > >/****************************/ > >STATUS VMI4514ARemove(void) >{ > int status; > printf("Deleting the devicces...\n"); > iosDevDelete(&vmi4514a_channel[0].devHdr); > iosDevDelete(&vmi4514a_channel[1].devHdr); > printf("Removing the VMI4514A driver...\n"); > status = iosDrvRemove(VMI4514ADrvNum,TRUE); > VMI4514ADrvNum = 0; > return(status); >} > >--------------46142D362668-- > --------------------------- Newsgroups: comp.os.vxworks Subject: MAC query Date: Sat, 13 Mar 1999 01:53:39 GMT From: reperjesi@my-dejanews.com Organization: Deja News - The Leader in Internet Discussion Message-ID: <7ccgf0$gl7$1@nnrp1.dejanews.com> Hi vxworkers, I'd like to get the ethernet controller's Ethernet Address (6 byte unique ID). Any hints? Thanks, Richard - -----------== Posted via Deja News, The Discussion Network ==---------- http://www.dejanews.com/ Search, Read, Discuss, or Start Your Own --------------------------- Newsgroups: comp.os.vxworks Subject: Re: question about load vxworks Date: Wed, 10 Mar 1999 08:24:42 -0600 From: Frank Minich Organization: NCP Message-ID: <36E680AA.6D1@ti.com> References: <36e5c74f.0@dnews.sbell.com.cn> Reply-To: fminich@ti.com SRD Zha Junhai wrote: > > Hi, everybody > > I just met a question when I use the ads860 board and loading the > vxWorks from host. > Here are the parameters : > Attaching network interface cpm0... done. > Subnet Mask: 0xfffffff8 > Attaching network interface lo0... done. > Loading... > Error loading file: errno = 0xd0003 You don't show enough of the boot sequence to indicate whether you are attempting to use rsh or ftp to obtain the boot file. I will assume that you have specified the "ftp password (pw)", and so are attempting to use ftp. I have been more successful using ftp than rsh. The errno isn't particularly useful in determining the underlying cause of the problem. The error indication means that the bootrom image couldn't load the vxWorks file you specified as the boot file. Possible reasons include: 1. the interrupt level and/or I/O address of the ethernet interface doesn't match the settings which are built into your bootrom image. 2. your host has a different IP address than you specified as the "host inet(h)" IP address 3. you specified a gateway inet (g), but the gateway isn't responding 4. your host isn't running ftpd (assuming that you are attempting to use ftp to get the boot file) 5. your host doesn't allow logins by the user (u) and ftp password (pw) you configured 6. the "filename" you specified doesn't exist, or is not accessible by the user you specified 7. there's no network connection between your machine and the host Your best bet is to make sure you can access the boot file you are trying to load via ftp from another unix or windows computer on the network, using the exact same host IP address, username, password, and boot file path/filename. If this works, then the problem has to be 1 or 7 above. Regards, Frank Minich --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Hard reset on MPC860 Date: Wed, 10 Mar 1999 17:34:14 +0000 From: Phil Ashby Organization: BT Wireplay Message-ID: <36E6AD16.7624@wireplay.bt.co.uk> References: Steve Borho wrote: > > If someone's figured out how to force an MPC860 to do a hard reset through > software, please let me in on the secret. This may not help, however: we induce a hard reset on our MVME230x by instructing the Universe chip to toggle the SYSRESET line..... - -- !--------------------------- Phil "Phlash" Ashby ---------------------! ! BT Wireplay - Lead Programmer and general technical trivia supplier ! ! Phone: 01473-644348 WWW: http://www.wireplay.com/ ! ! Snail: B81/G41, BT Labs, Martlesham Heath, Ipswich IP5 3RE, England.! --------------------------- Newsgroups: comp.os.vxworks Subject: Re: MAC query Date: Fri, 12 Mar 1999 23:00:08 -0600 From: petekz@cyberramp.net (Pete Kockritz) Organization: Cyberramp (using Airnews.net! at Internet America) Message-ID: <5914149BDA7E7622.2718831C80A97E46.7849779FAA73C516@library-proxy.airnews.net> References: <7ccgf0$gl7$1@nnrp1.dejanews.com> In article <7ccgf0$gl7$1@nnrp1.dejanews.com>, reperjesi@my-dejanews.com wrote: > Hi vxworkers, > > > I'd like to get the ethernet controller's Ethernet Address > (6 byte unique ID). Any hints? Assuming your BSP supports it STATUS sysEnetAddrGet ( UINT8 unit, UINT8 * addr ) will return it. Regards, Pete - -- +-----------------------------------------------------------------+ | Pete Kockritz mailto:petekz@cyberramp.net | | This space for rent. | +-----------------------------------------------------------------+ --------------------------- Newsgroups: comp.os.vxworks,comp.realtime Subject: vxWorks and typeahead (serial and telnet) Date: Thu, 11 Mar 1999 04:31:25 GMT From: Rob Faulds Organization: Deja News - The Leader in Internet Discussion Message-ID: <7c7gut$4bd$1@nnrp1.dejanews.com> Hello, I'm having quite a bit of trouble with type-ahead for my CLI sessions under vxWorks. I actually have the problems on both serial (console) and telnet sessions but likely for differing reasons. For the serial case: Has anyone successfully implemented rock-solid typeahead on a vxWorks serial port? In my case if there is a lot of output on the serial line then the input side is spotty (actually the input seems to be ignored and then the later characters are received). For example, if I type: thequickbrownfoxjumpedoverthelazydog while a large amount of scrolling output displays -- all that shows up at the input prompt is "lazydog" or so depending on how quickly I enter the input. Some of my vxWorks based server displays are quite long. I need to be able to allow users to *realiably* type the next command(s) while they wait. For the telnet case: The delivery of control-characters to a vxWorks telnet session is abysmal and confusing. I don't have my exact tick count experiment data here but when I last tried this a ^C entered from the telnet client was taking ~1 second to arrive at my FIOPROTOHOOK function. This is fairly easy to demonstrate. If you have OPT_TANDEM (XON/XOFF) enabled on your telnet session you can try this... + pick a functions that displays a lot of output (or use repeatRun() with ifShow()). + get the input ready on the shell command line but don't hit Enter-key + arrange your fingers to hold down the Ctrl-key and hover over the M and S keys. + enter Ctrl-M Ctrl-S as fast as you can. + watch incredulously as oodles of output scrolls by before the Ctrl-S causes XOFF and the holding of output In my serial session case (same CLI session but on tty vs. pty) the Ctrl-C effect in my FIOPROTOHOOK is *immediate* and a Ctrl-S (XOFF) is also nearly immediate. In these, paths though the FIOPROTOHOOK is called right from the sio driver ISR. In the telnet/pty case these is no ISR and so somehow the response *really* lags. Also my sio driver (I can't remember what UART is on my PPC604 platform) seems very slow. At 9600 baud it seems to display at about 4800 baud. Compared to competing products my vxWorks based server CLI sessions look ridiculous speed-wise...Ugh! How can I speed this up? Are there known, gross deficiencies with the sio driver or ttyLib functions? Do I have to rewite the sio driver? Any ideas on the other stuff? I have windView and can instrument the receipt of the Ctlr-C or Ctrl-S through all the layers if you think it would be enlightening. Should I sniff the delivery of the Ctrl-C, Ctrl-S on the Ethernet to ensure they arrive in a timely manner? Thanks, - -Rob - -----------== Posted via Deja News, The Discussion Network ==---------- http://www.dejanews.com/ Search, Read, Discuss, or Start Your Own --------------------------- Newsgroups: comp.os.vxworks Subject: Re: "Relocation value does not fit in 24 bits" errors Date: 10 Mar 1999 15:33:02 GMT From: ag@devsoft.devsoft.com (Armin Gruner) Organization: Customer of SpaceNet GmbH Message-ID: References: <36E67F94.241C43ED@orion.ksc.nasa.gov> <36E68A4D.C5BBC94B@orion.ksc.nasa.gov> In article <36E68A4D.C5BBC94B@orion.ksc.nasa.gov>, Charles Curley wrote: >After messing with this some more I've discovered a command called >objectdump. We've never had these kind of problem before so don't have a >whole lot of experience digging thru object files. Do a "objectdumpppc >-reloc vmeLib.out | grep REL24" gives the following: > >000001c0 R_PPC_REL24 __builtin_delete >000001ec R_PPC_REL24 __builtin_delete >00000230 R_PPC_REL24 __builtin_delete >000002e8 R_PPC_REL24 __builtin_vec_new >00003654 R_PPC_REL24 __builtin_new >0000372c R_PPC_REL24 __builtin_new >00003c48 R_PPC_REL24 __builtin_new >00003c7c R_PPC_REL24 __builtin_new >0000419c R_PPC_REL24 strlen Hi, we had similar problems here. We assume that the compiler fix does not cover libgcc.a -- and g++ rely on implementation of "new" and "delete" in libgcc.a ... If you want to avoid that these things from libgcc get linked into your application, replace them by your "own" implementation: - --- cut here --- #define MESSAGE "Virtual memory exceeded in `new'\n" /* void * operator new (size_t sz) */ void * __builtin_new(int sz) { void *p; /* malloc (0) is unpredictable; avoid it. */ if (sz == 0) sz = 1; p = (void *) malloc (sz); if (p == 0) { logMsg(MESSAGE, 0, 0, 0, 0, 0, 0); _exit(-1); } return p; } void * __builtin_vec_new (int sz) { return __builtin_new (sz); } /* void operator delete (void *ptr) */ void __builtin_delete (void *ptr) { if (ptr) free (ptr); } void __builtin_vec_delete (void *ptr) { __builtin_delete (ptr); } #undef MESSAGE #define MESSAGE "pure virtual method called\n" void __pure_virtual () { logMsg(MESSAGE, 0, 0, 0, 0, 0, 0); _exit (-1); } - --- cut here --- Regards Armin >which corresponds to the number of "Relocation ..." error message we are >seeing. In effort to remove these I added a -fno-builtin to the compile >line and that got rid of the strlen, but the new and delete calls still >remain relative. How can I force them to be longcalls? > >-- >Charles Curley, Senior Software Engineer, Dynacs Engineering Co., Inc. >Engineering Development Contract, Kennedy Space Center, Florida > > - -- Armin Gruner Forschung und Entwicklung - -- Armin Gruner, /dev Software GmbH Fon +49-89-286 598 43 Gabelsberger Str. 51 mailto:ag@devsoft.com D-80333 Muenchen http://www.devsoft.com --------------------------- Newsgroups: comp.os.vxworks Subject: advantages of SENS Date: Mon, 1 Mar 1999 08:16:52 +0100 From: "Markus Mitterer" Organization: Customer of EUnet Austria Message-ID: <7bdevc$rh3$1@fleetstreet.Austria.EU.net> Hi all, when I look at this newsgroup, I see a lot of people using SENS. I don't have this optional product but I must be able to transmit much data via ethernet. So my questions are: Is SENS much faster (and how much) than the default WRS network stack? Are there any other advantages of SENS? thanks in advance Markus Mitterer, STK --------------------------- Newsgroups: comp.os.vxworks Subject: Re: WindNet SNMP (was: Are multiple agent tasks possible) Date: Wed, 10 Mar 1999 17:38:32 GMT From: "Tom Salmon" Organization: RT Logic Message-ID: <01be6b1c$d0663c90$2a9cd126@salmon> References: <01be6516$d6b7fae0$2a9cd126@salmon> <36E0942A.BF294AF2@hotmail.com> <36E40291.2781@utsci.com> The SNMPv1/v2c agent seems to work ok, but the documentation associated with it is HORRIBLE (TERRIBLE, AWFUL, am I getting the point across?) Mike Stimpson wrote in article <36E40291.2781@utsci.com>... > Two questions: > > - Is WindNet SNMPv1/v2c solid? Is there another SNMP implementation > that is more solid? Are there implementations that should be avoided? > > - How much space (code and memory) does SNMP consume? > > Please E-mail replies as well, as my news feed is quite unreliable. > > Thanks! > > -- > Mike Stimpson (mstimp@utsci.com) > > Opinions expressed are not necessarily those of the management... > > The more I work as an engineer, the more I disbelieve the theory of > evolution. > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Debugger Support For Ethernet & Kernel drivers Date: Sat, 13 Mar 1999 09:16:14 GMT From: bobarmstrong@home.net (Robert Armstrong) Organization: @Home Network Message-ID: References: <36E70D71.7F8DDC55@NOSPAM.metabyte.com> You can do this using the serial backend. If you need to debug the interrupts you can do that in system mode, again via the serial backend. Your chances of debugging ethernet whilst using it as the debug link are pretty slim :) Bob In article <36E70D71.7F8DDC55@NOSPAM.metabyte.com>, Ganesh Sundaram wrote: >Hi, > >I am new to the VxWorks operating system, so excuse me if this is a >repeat/naive question. I wanted to know if Tornado supports "source >level debugging for ethernet drivers" ? I am looking for debugging >support similar to what SoftIce provides for Windows NT driver >debugging. I will be developing an ethernet driver under vxWorks and >would be very much interested to know if this is possible. > >On what kind of hardware platform is the source level debugging >supported? What is the setup required to acheive this? Does the basic >Tornado environment support source level debugging of ISR's? > >Any information or direction to obtain the setup required for debugging >is greatly appreciated. > >Thanks in advance. >-Ganesh > --------------------------- Newsgroups: comp.os.vxworks Subject: Disk suggestions and question Date: Wed, 10 Mar 1999 19:44:07 GMT From: radcliff@jps.net Organization: Deja News - The Leader in Internet Discussion Message-ID: <7c6i24$8h4$1@nnrp1.dejanews.com> A couple of people have posted questions on this group regarding corruption of the disk. I have seen most of the problems before (including some problems that are different). I saw quite a few of the problems while messing around with changing the parameters of the disk. When I attempted to use a disk that I had initialized with long file name support and then tried to use it as a disk with short file names only set, or vise versa- with a disk that was initialized as short file name only that was accessed as a long file name disk. What I discovered was that there was no really good way to change the parameters once the disk was initialized! Once the call to dosFsMkfs was called, I never changed the parameters of the disk. And further, I gave up trying to move the disk (PCMCIA ATA) to DOS computer. The best thing is to become aware of what the parameters that the disk was created with and what you are accessing it as! My question deals with file creation. When a file is created, no blocks are associated with it. As you write data to the disk, the file system assigns free blocks as needed. As with all disks, a single block access is much slower per byte than a multi-block access, but when a file is created for the first time it is always accessed block by block as it expands. There is a function (an ioctl call) that attempts to allocate a file at a certain size which seems fairly fast. The only problem is it requires the blocks to be all contiguous. This means that if your disk is fragmented, you can't use this function to allocate the file ahead of time (if you know how big you want), even if there is plenty of room in non-contiguous space on the device. Does anyone know of a way to efficiently allocate a file of a specific file ahead of time that is not necessarily contiguous? I think it should behave the same way as the ioctl call but doesn't necessarily need to have a huge chunk of the disk contiguously free. Thanks! cory - -----------== Posted via Deja News, The Discussion Network ==---------- http://www.dejanews.com/ Search, Read, Discuss, or Start Your Own --------------------------- Newsgroups: comp.unix.solaris,comp.os.vxworks,comp.sys.sun.hardware Subject: Re: PCMCIA pcata Cards Date: Wed, 10 Mar 1999 19:07:49 GMT From: coryrad@my-dejanews.com Organization: Deja News - The Leader in Internet Discussion Message-ID: <7c6ftt$6jq$1@nnrp1.dejanews.com> References: Are you sure there are no other VxWorks specific parameters that you are setting in the file system. I don't know if the lower case only parameter is incompatible with the other incarnations of dos file systems, but I do know that vxworks long file names are definitely cause for problems with other systems. Basically if you use any long file names with VxWorks dosFs (at least the original version) the disk will no longer be dos compatible. Try running dosFsConfigShow on the once you have written to it under vxworks and check the parameters for dos compatiblity. In article , jeffrey.m.creem@lmco.com (Jeff Creem) wrote: > > > I am having a problem with a PCMCIA ATA card that I use on both > SPARC Solaris 2.6 (with the SBUS PCMCIA adapter and/or built in PCMCIA of an > RDI Ultrabook) and vxworks (real-time OS running on a single board computer). > > The card contains a DOS filesystem. At times, after writing to the > card under vxWorks, I can no longer mount the card under Solaris. > (I get a mount: /dev/dsk/c1d1s0:c is not a DOS filesystem.) > (If i run truss on the mount all I see is that mount() errno is EINVAL) > > This does not appear to be the traditional vxWorks issue of vxWorks > overwriting the partition table information at the header of the card since > as near as I can tell the partition information on the card is not changed and > it looks like the driver is properly offsetting from the start of the card. > > If I use mtools on the SUN I can still read and write all of the files. > If I run chkdsk on a PC it says there are no problems with the card. > > Once this happens the only thing that fixes it is to reformat it on a > PC (since there does not appear to be solaris support for formatting PC ata > cards). > > I've got the latest vxWorks DOSFs patches installed and it does not > appear to help. I am leaning towards believing that this is a Solaris > issue but certainly vxWorks is doing something to the card that makes > Solaris unhappy. My next path is to try to get the card on a PC > with Norton so I can take a look at the disk structure a little more but > this will take a few days so I thought I'd ask if anyone has seen this sort > of thing. > > (Still hard to believe how poor pcata cards are supported under Solaris, > apart from this the Sun Also spits out unhandled interrupts each time > a new card is inserted and certain combinations of cards makes the Sun > Panic....Anyone know if this got any better in Solaris 7) > > Jeff Creem > - -----------== Posted via Deja News, The Discussion Network ==---------- http://www.dejanews.com/ Search, Read, Discuss, or Start Your Own --------------------------- End of New-News digest ********************** From owner-vxwexplo-process Sat Mar 13 07:21:10 1999 From: LJR Date: Sat Mar 13 07:21:13 PST 1999 Subject: 8240/8260 Support Status: O Lorens, If you need a BSP for the 8240/8260, the other guys, Integrated Systems (pSOS), have this done and shipping. LJR At 04:00 AM 3/13/99 -0800, you wrote: >Submitted-by owner-vxwexplo-process Sat Mar 13 04:00:52 1999 >Submitted-by: daemon@csg.lbl.gov > >Comp.Os.Vxworks Daily Digest Sat Mar 13 04:00:47 PST 1999 >Newsgroups: comp.os.vxworks >Subject: Anybody using MPC8240 or MPC8260? >Date: Fri, 12 Mar 1999 13:33:34 GMT >From: lourens@mecalc.co.za (Lourens Geldenhuys) >Organization: Mecalc (Pty) Ltd >Message-ID: <36e91698.28046093@modem.mecalc.co.za> >Reply-To: lourens@mecalc.co.za > >Hi > >1. Is anybody running VxWorks on the Motorola MPC8240 or MPC8260? > >2. Does a BSP exist for any boards with these devices? > >Regards > Lourens Geldenhuys From owner-vxwexplo-process Sun Mar 14 04:00:21 1999 From: daemon@csg.lbl.gov Date: Sun Mar 14 04:00:25 PST 1999 Subject: comp.os.vxworks newsdigest Status: O Comp.Os.Vxworks Daily Digest Sun Mar 14 04:00:17 PST 1999 Subject: How can I get VxWorks Real Time OS for Intel x86 Subject: Re: select() with Wind Msg Queues Subject: Re: Hard reset on MPC860 Subject: Re: compression source code Subject: TrueFFS Subject: SNTP & vxWorks Subject: Re: Which host platform, Solaris or NT? Subject: purposefully crashing VxWorks Subject: Re: purposefully crashing VxWorks Subject: ei82596End users? Subject: Re: compression source code Subject: Re: VxWorks memory partitions Subject: Re: dosFsLib Y2K compliance ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: How can I get VxWorks Real Time OS for Intel x86 Date: Sat, 27 Feb 1999 23:47:22 +0800 From: "XYZ" Organization: Hong Kong Supernet Limited Message-ID: <7b93pc$5pn$1@hfc.hk.super.net> Hi, Is there anybody knows where can I download the VxWorks for Intel X86 CPU ? Thanks ! Sunny mailto:sunny.ng@tek.com --------------------------- Newsgroups: comp.os.vxworks Subject: Re: select() with Wind Msg Queues Date: Mon, 01 Mar 1999 09:50:40 +0000 From: Martin Schrape Organization: Atomika Instruments GmbH Message-ID: <36DA62F0.524329D0@atomika.com> References: <36d509a6.26060352@10.250.5.25> <36d659c6.25371662@10.250.5.25> Andy Kriechhammer wrote: > > > After doing some short tests with pipes, the next question came up > very quick: How can I get rid of a pipe ? > > I did this: > > pipeDevCreate(PIPENAME, MAXMSGS, MAXSIZE); > > open(PIPENAME, O_RDWR, 0); > > ... some write() calls > ... som read() calls from another task > > close(pd); > > unlink(PIPENAME); > Use iosDrvShow to see if you driver implements a delete function. If there is noone remove (=unlink) do nothing, but even if there is one there is no garantie. Then you have to use iosDevFind() to look up the DEV_HDR and then call iosDevDelete(). - - martin - --------------------------------------------------------------- Dipl.-Phys. Martin Schrape schrape@atomika.com Senior Software Engineer Tel. +49 89 315 891 34 Atomika Instruments GmbH Fax +49 89 315 59 21 Bruckmannring 40 http://www.atomika.com 85764 Oberschleissheim/Munich, Germany - --------------------------------------------------------------- --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Hard reset on MPC860 Date: Wed, 10 Mar 1999 15:05:58 -0600 From: "Bruce Nehlsen" Organization: UUNET Message-ID: <7c6mns$23k$1@ffx2nh5.news.uu.net> References: We can supposedly do that by having the Software Watchdog enabled (SWE in SYPCR register), and failing to kick it. Note that by default, vxWorks disables this (romInit.s) Happy Coding, Bruce Nehlsen Steve Borho wrote in message ... >If someone's figured out how to force an MPC860 to do a hard reset through >software, please let me in on the secret. > >Thanks. > >-- >Steve Borho --------------------------- Newsgroups: comp.arch.embedded,comp.os.vxworks,comp.realtime Subject: Re: compression source code Date: Tue, 02 Mar 1999 00:01:41 +1000 From: NOsatanusSPAM@KILLSPAM.ender.SPAMDIE.com (Lord Apollyon) Organization: Acheron, Ninth Plane of HELL Message-ID: References: <79s8hi$3r2$1@news.netway.at> <36D975BD.B69BADD1@lucent.com> In article <36D975BD.B69BADD1@lucent.com>, Zvika Perry wrote: > As we had to compress my application due to FLASH memory constraints and > decompress it on the target, we have a gzip code modified to work over memory > rather than a file system. > It seems bzip does it better - about 10% in my application and I'm currently > working on it. > If this can be a help, I can send it (the gzip) by e-mail. Not sure bzip (aka Wheeler Block Sorting compressor) would be all that practical for your embedded work... seeing as it has monstrous memory requirements for decompression. Yes, you can contain it by using a small bucket size on the compression side, but.... Have you explored Markov coders, a la PPMZ*, etc? They offer a better squeeze than ancient LZ77/78-based "technology" without the extreme resource requirements of W-B Sorter techniques. =R= --------------------------- Newsgroups: comp.os.vxworks Subject: TrueFFS Date: Mon, 01 Mar 1999 14:05:57 GMT From: smooney@my-dejanews.com Organization: Deja News - The Leader in Internet Discussion Message-ID: <7be6rt$f45$1@nnrp1.dejanews.com> We are evaluating WindRivers flash file system TrueFFS and are running into a problem getting it running. We are using a PPC 860T processor and 4M of AMD29LV160B flash on a 32-bit bus, so thats two 16-bit flash chips interleaved to get the 32-bit access. WindRivers doesn't supply any 32bit MTD's and the first level of customer support says they have never run in this configuration, (although the sales support feel they have.) My attempts to modify the existing MTD's have been unsuccessful, has anyone ever used TrueFFS with 32-bit access? Any pointers would be appreciated. - -Stephen Mooney - -----------== Posted via Deja News, The Discussion Network ==---------- http://www.dejanews.com/ Search, Read, Discuss, or Start Your Own --------------------------- Newsgroups: comp.os.vxworks Subject: SNTP & vxWorks Date: Mon, 01 Mar 1999 15:24:20 +0100 From: Jeroen Damen Organization: Siemens Inc. Message-ID: <36DAA314.11487497@vnet.atea.be> Hi, Has anyone tested the sntpcLib already? If so, can you send me an example of the usage of this library? Thanks, Jeroen --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Which host platform, Solaris or NT? Date: Tue, 02 Mar 1999 00:48:14 GMT From: djhystad@my-dejanews.com Organization: Deja News - The Leader in Internet Discussion Message-ID: <7bfcgb$hu6$1@nnrp1.dejanews.com> References: <7be77d$fab$1@nnrp1.dejanews.com> In article <7be77d$fab$1@nnrp1.dejanews.com>, eezekiel@my-dejanews.com wrote: > Does anyone have any opinion on which host platform is preferable for the > Tornado Host environment, Sun Solaris or Windows NT? > > I think that the WRS user interface support is better on NT. Any other > considerations? > I think that tornado is awful on both platforms. Pick the Solaris platform, load Tornado and promptly forget that it exists, and use the command line to do everything you desire. Maybe I am a Luddite. But I am a very productive Luddite. > Thanks, > Ephraim Ezekiel > Quarry Technologies > eezekiel@quarrytech.com > > -----------== Posted via Deja News, The Discussion Network ==---------- > http://www.dejanews.com/ Search, Read, Discuss, or Start Your Own > - -----------== Posted via Deja News, The Discussion Network ==---------- http://www.dejanews.com/ Search, Read, Discuss, or Start Your Own --------------------------- Newsgroups: comp.os.vxworks Subject: purposefully crashing VxWorks Date: Sat, 13 Mar 1999 19:06:51 -0500 From: "Michael J. Abowd" Organization: CMU Software Engineering Message-ID: <7ceuin$k3t$1@goldenapple.srv.cs.cmu.edu> Hi, I am working on a project that needs to consistently create a type of OS failures in VxWorks running on a Motorola 68040. Create a process that cause the OS to hang altogether. At this point, the machine would need to rebooted. If this processes could be created by calling a POSIX call, that would be very handy, but it's not required. Any suggestions on how to create this failure? I have noticed that when I execute a while(true) loop, the OS hangs, but don't think it should and was wondering if there were other ways to do this. Thanks Much, Michael --------------------------- Newsgroups: comp.os.vxworks Subject: Re: purposefully crashing VxWorks Date: Sat, 13 Mar 1999 17:21:28 -0800 From: Scott Johnson Organization: National Association for the Advancement of Computer Geeks Message-ID: <36EB0F18.6A70@nospam.aracnet.com> References: <7ceuin$k3t$1@goldenapple.srv.cs.cmu.edu> Reply-To: sj_nospam@nospam.aracnet.com Michael J. Abowd wrote: > > Hi, > > I am working on a project that needs to consistently create a type of > OS failures in VxWorks running on a Motorola 68040. Create a process > that cause the OS to hang altogether. At this point, the machine > would need to rebooted. If this processes could be created by calling > a POSIX call, that would be very handy, but it's not required. > > Any suggestions on how to create this failure? I have noticed that > when I execute a while(true) loop, the OS hangs, but don't think it > should and was wondering if there were other ways to do this. "Crashing" vxWorks is ridiculously easy, though how to do it depends on what sort of crash you want: 1) Do you want the processor hung in a loop, unable to do any useful work, but still fetching and executing instructions. For that, intLock(); for (;;) {} works nicely. 2) Do you want to see the processor halted--ie forced into a state where it no longer fetches instructions, and cannot be recovered by any means other than a reset? On a 68k, an easy way to do that is to generate a "double fault"--cause an (fatal) exception to occur (such as a bus error), and have the exception handler generate another exception. The 68k deals with this scenario by halting... for example: int *evilAddress = ??? /* address that does not exist on your board * and will cause a bus error if * dereferenced.) */ /* An evil bus error handler that halts the processor */ void evilExcHandler (int vector) { (void) vector; *evilAddress = 0; /* Cause a second bus error in the handler, * generating a double fault and halting the * processor. */ } /* The function to call to halt the processor */ void die (void) { /* Install our "evil" bus error handler */ excSetVec (__EXC_BUS_FAULT, evilExcHandler); /* Now cause a bus fault; this will invoke the above evil * handler which will cause a double fault. */ *evilAddress = 0; } Or something like that, where __EXC_BUS_FAULT is the name of the 68K bus error handler; see your BSP documentation/source for details. Just curious--what are you doing this for? The fact that vxWorks, a striclty supervisor-mode OS with no memory protection whatsoever, is easy to crash has already been proven... Scott > Thanks Much, > Michael - -- ======================================================================== Scott Johnson Patriot, Beer-Drinker, Geek GO BLAZERS Don't take yourself too seriously. Nobody else does......... ======================================================================== --------------------------- Newsgroups: comp.os.vxworks Subject: ei82596End users? Date: Tue, 02 Mar 1999 12:28:05 -0500 From: Jeff Daly Organization: Sanders A Lockheed Martin Company Message-ID: <36DC1FA5.37F3BEAD@lmco.com> Reply-To: jeffrey.j.daly@lmco.com This is a multi-part message in MIME format. - --------------8725BF077EA21B71DBD80CFE Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit anyone have success using the ei82596End interface? i'm having a little trouble. drop me a line. btw, i've figured out that the SENS documentation for the ei82596EndLoad function is incorrect. it appears that the sysBus and memBase fields of the loadstring are reversed. memBase comes before. - --------------8725BF077EA21B71DBD80CFE Content-Type: text/x-vcard; charset=us-ascii; name="vcard.vcf" Content-Transfer-Encoding: 7bit Content-Description: Card for Jeff Daly Content-Disposition: attachment; filename="vcard.vcf" begin: vcard fn: Jeff Daly n: Daly;Jeff org: Sanders, A Lockheed Martin Co. email;internet: jeffrey.j.daly@lmco.com title: SW Guy note: B-sharps x-mozilla-cpt: ;0 x-mozilla-html: FALSE version: 2.1 end: vcard - --------------8725BF077EA21B71DBD80CFE-- --------------------------- Newsgroups: comp.arch.embedded,comp.os.vxworks,comp.realtime Subject: Re: compression source code Date: Tue, 2 Mar 1999 14:41:23 -0600 From: "Pat Rogers" Organization: Software Arts & Sciences Message-ID: <7bhicu$5bc$1@remarQ.com> References: <79s8hi$3r2$1@news.netway.at> <36D975BD.B69BADD1@lucent.com> <36DBA39E.598491AB@lucent.com> Lord Apollyon wrote in message ... >In article <36DBA39E.598491AB@lucent.com>, Zvika Perry > wrote: > >> Any ideas where can I find a code that implements Markov coders, a la PPMZ ? >> Modification of bzip is really tedious (but I desperately need it) > >Give Charlie Bloom's site a look-see. > >http://lonester.texas.net/~cbloom/src/ppmz.html That's "lonestar", rather than "lonester". You're not from Texas, are you? :-) --------------------------- Newsgroups: comp.os.vxworks Subject: Re: VxWorks memory partitions Date: Sun, 28 Feb 1999 10:41:13 +0300 From: Leonid Rosenboim Organization: BitBand Technologies Ltd. Message-ID: <36D8F319.DCA2B5CB@bitband.com> References: <199902190926.JAA08729@dsl-2.tadpole.co.uk> <36CD5189.28EB49ED@bitband.com> <36CDB722.1E7B6549@pso.com> Just a brief follow-up: Hwa Jin Bae wrote: > [snip] > > The partitioning scheme, as Leonid suggests, solves some problems and is > quite useful as well. Unfortunately, VxWorks memory partitions, once > created cannot be destroyed. This makes it less desirable. Plus it > requires a lot of changes in ported Unix code base and is not as easy to > do. Indeed being unable to destroy or even reduce the size of a partition is a drawback. However, there is a way to avoid modifying a large code base (e.g. when porting from UNIX) by means of creating a replacement for the ANSi functions: malloc(), calloc(), realloc() and free(), such that these function call mamPartAlloc() with a particular partition ID deduced from the SIZE of the block requested, and can also automatically increase the size of the selected partition with memPartAddToPool(), borrowing more memory from the main partition. The replacement free() has to peek into the blocks overhead header to see its size and return it to the appropriate partition. This has been already done by at least one customer without access to WRS sources. Leonid --------------------------- Newsgroups: comp.os.vxworks Subject: Re: dosFsLib Y2K compliance Date: Wed, 03 Mar 1999 10:47:39 -0800 From: Scott Kamara Organization: AP Labs Message-ID: <36DD83CB.4AE85E90@pop.sttl.uswest.net> References: <7bhrdo$dn8@overload.lbl.gov> Reply-To: spk@uswest.net When using mv167 under vxWorks 5.3.1, I do not see the result you see. The resulting file timestamp gives 'MAR-31-2000' after I do dosFsDateSet(2000,3,31). You did not mention the user supplied function you used when using dosFsDateTimeInstall(). In this case, the user supplied function gives 100% of the time-date information. Legoas, Marco A (FL51) wrote: > > VxWorks > > Does anyone know of a Y2K limitation on VxWorks 5.1(mvme167 BSP) dosFsLib ? > > When using: > > ->dosFsDateSet(2000,03,31) > > I get a file time stamp of APR-00-2000 > > I get the same behavior for dosFsDateTimeInstall() > > WRS Y2K white paper (page 8) > , says that dosFsLib can > handle dates through 12/31/2099?. > > Any ideas? > > Marco A. Legoas > malegoas@space.honeywell.com - -- - ----------------------- Scott Kamara AP Data Systems Seattle Engr. Office (206) 525-8592 scott@aplabs.com --------------------------- End of New-News digest ********************** From owner-vxwexplo-process Mon Mar 15 04:00:43 1999 From: daemon@csg.lbl.gov Date: Mon Mar 15 04:00:46 PST 1999 Subject: comp.os.vxworks newsdigest Status: O Comp.Os.Vxworks Daily Digest Mon Mar 15 04:00:40 PST 1999 Subject: ftpd and multihomed vxWorks system Subject: Re: Help! - Subject: RARP support? Subject: Re: Copy from host to target FS over serial WDB? Subject: Re: MAC query Subject: Tornado Shell for Linux Pre-Release avaliable. Subject: 100 BaseT PCI card Subject: Ampro Pentium CPU board Subject: A bus error problem..please help.. ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: ftpd and multihomed vxWorks system Date: Thu, 4 Mar 1999 16:26:41 +0200 From: Message-ID: <36de9755.0@news.barak.net.il> Hi, How do I tell tftpd which port and which interface to bind to on a multihomed system. Thanks, Rami --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Help! - Date: Wed, 10 Mar 1999 22:05:53 +0000 From: Rachel-Louise Koktava Organization: KittyCrap Message-ID: References: <36E59F35.A43CE7A8@metrolink.net> > The error that >we have come across occurs with a use of memcpy. In once instance of >using the memcpy we receive a data access error. When this is >substituted with a for loop to do the copy, the error does not occur. > >Can anyone provide an explanation of this or give me an idea of what to >look into? > hello Peter, you will often get a data access error with a stray pointer, or a NULL pointer. This would probably tally if you are having problems with memcpy, so check your pointers to see that they are pointing/addressing correctly. hope this helps RL - -- Rachel-Louise Koktava --------------------------- Newsgroups: comp.os.vxworks Subject: RARP support? Date: Sun, 14 Mar 1999 14:49:52 -0700 From: Jay Hall Organization: AT&T WorldNet Services Message-ID: <7chav5$7vs$1@mtinsc01.worldnet.att.net> Reply-To: jaydhall@worldnet.att.net Hi vxworkers, I need to use RARP for my project. Any idea how to go about it? Seems VxWorks does not support it. Jay Hall --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Copy from host to target FS over serial WDB? Date: Thu, 11 Mar 1999 09:41:46 -0800 From: "John Finley" Organization: Kivala Systems, Inc. Message-ID: <7c8v33$om7@dfw-ixnews3.ix.netcom.com> References: <7c8qf6$d0l@post.gsfc.nasa.gov> nobody@nowhere wrote in message <7c8qf6$d0l@post.gsfc.nasa.gov>... >Is it possible to copy files from a host operating system disk to a vxWorks >target disk over a serial WDB link? > >Usually when I am using the networked version of tornado, the host name shows >up in the 'devs' list and I can copy files from the host to the target using >the tornado shell. > >Now I have stripped down the kernel without networking and I would like to >copy a file from the host to the target using the host tornado shell over a >serial WDB link. > >Is this possible? Do I need networking code to do this anyway? Tornado 2's target server has a thing called the "Target Server File System" (TSFS). TSFS does exactly what you want to do: the target server serves host files to the target without the target having networking installed. You can think of it sort of like a little NFS server. You could ask your local Wind River folks if the later Tornado 1 target servers include it. The last component I installed (TFFS for PPC) "upgraded" a whole lot of stuff, including the target server exe, but I didn't notice if TSFS was in there. Otherwise, I think you have to write a little code in TCL, C, or, Visual Basic (see link below) to do it. I was writing just such a thing in VB as a VxTool demo program when I found out TSFS was coming, and aborted the effort. Maybe there are enough people stuck with T1 for a while so I should have kept it up (you don't have to buy VxTool to run the demos) . The complexity of the program (any language) would depend on how featureful you need it. Doing it like NFS, e.g. arbitrary reads and writes, and directory ops, is the full featured end. There is nothing tricky going on, but it is nontrivial. If you just want to occasionally spew a file from host to target through a VIO channel while your target is "offline" then it becomes really easy. John - -------------------------------------------------- John Finley Kivala Systems, Inc. Project Manager (619) 689-0032 john@kivala.com http://www.kivala.com - -------------------------------------------------- VxTool - Build Tornado Apps with MS Visual Basic - -------------------------------------------------- --------------------------- Newsgroups: comp.os.vxworks Subject: Re: MAC query Date: Sun, 14 Mar 1999 23:15:26 -0600 From: petekz@cyberramp.net (Pete Kockritz) Organization: Airnews.net! at Internet America Message-ID: References: <7ccgf0$gl7$1@nnrp1.dejanews.com> <5914149BDA7E7622.2718831C80A97E46.7849779FAA73C516@library-proxy.airnews.net> <7cgq6v$slm$1@nnrp1.dejanews.com> > >The problem I had with sysEnetAddrGet is that each time I run it the addr >increases by 1. The first time I run sysEnetAddrGet the address the 2nd >parameter points to is x+1, the second time I run it it's x+2 and so on (x = >real MAC). >By the way, is this function - sysEnetAddrGet - documented anywhere, becuase I >found it digging into the source I have. I'd say your BSP is broken. Or perhaps someone was trying to do something clever with the MAC address. Here is what we have in our BSP for that function: STATUS sysEnetAddrGet ( UINT8 unit, UINT8 * addr ) { sysCpmEnetAddrGet (unit, addr); return (OK); } STATUS sysCpmEnetAddrGet ( int unit, /* not used - only slave SCC1 is wired to port */ UINT8 * addr ) { sysNvRamGet (addr, ENET_SIZE, ENET_OFFSET); /* * The MS 3 bytes are fixed. */ addr[0] = (ENET_DEFAULT_MS >> 16) & 0xff; addr[1] = (ENET_DEFAULT_MS >> 8) & 0xff; addr[2] = ENET_DEFAULT_MS & 0xff; return (OK); } ENET_DEFAULT_MS, ENET_SIZE and ENET_OFFSET are not WRS standard defines. So all it does is retrieve the MAC address from the NV memory, a serial EEPROM in our case. You should also be able to read the MAC address out of the ethernet chip, but it shouldn't be necessary. Are you using an off-the-shelf BSP or a custom one? I think that sysEnetAddrGet() may be documented in the BSP developer's manual. Regards, Pete - -- +-----------------------------------------------------------------+ | Pete Kockritz mailto:petekz@cyberramp.net | | This space for rent. | +-----------------------------------------------------------------+ --------------------------- Newsgroups: comp.os.vxworks Subject: Tornado Shell for Linux Pre-Release avaliable. Date: Mon, 15 Mar 1999 07:29:25 +0100 From: Philippe Le Foll Organization: FRIDU Message-ID: <36ECA8C5.D371C5C1@fridu.com> I prerelease a free version of a Tornado shell for Linux. This shell is a merge of WinSh and WtxTcl, it uses tclsh as backbone and readline as user interface. I develop on RH-5.2 but binaries run on Debian after some symbolic link in order making share libraries names equivalent. I also shipped a Solaris-Sparc version, in this case user should take care using a tclsh that as been compiled with --enable-share. Outside of those share libraries name, system should run almost out of the box. % TornadoSh --registery=yyyy --target=xxxx - --> set A [malloc 12] - --> free $A - --> printf "int=%d string=%s\n" 123 {--string 123} - --> exit ftp://ftp.neosoft.com/pub/tcl/sorted/packages-8.0/devel/jWrap-1.30-linux86.tgz TornadoSh is a module of jWrap. Jwrap provides a very generic mechanism to map C routines from Host or Target into TCLSH. Modules can be load statically or dynamically in any tcl8/wish, jWrap support both C and C++ nevertheless current version of WTX limit Tornado interface to C. This is a first version, any comments are welcome. Bruce O.Benson is looking to start an anonymous CVS server for source distribution. TornadoSh is normally fully free and uses TCL license, it nevertheless rely on some files shipped with standard Tornado distribution that are copywriter by WRS. TornadoSh is a production of Fridu in collaboration with South Brittany University Valor laboratory. Have fun. - -- La vitesse peut tuer: Utilisez Windows (o^o) ======================================oOO==(~)==OOo======== phillf@fridu.com, Philippe Le Foll Fridu (+33/0)609.794.781 --------------------------- Newsgroups: comp.os.vxworks Subject: 100 BaseT PCI card Date: Thu, 11 Mar 1999 19:42:29 GMT From: bijtaj@my-dejanews.com Organization: Deja News - The Leader in Internet Discussion Message-ID: <7c96b3$j21$1@nnrp1.dejanews.com> Hi, I am looking for 100 BaseT PCI cards with VxWorks drivers. This is for a Intel- Pentium architecture using the pc486 BSP. I understand the if_fei driver may support the Intel EtherExpress Pro 100+ card. Is that true and if so what is your experience with this card in terms of reliability under VxWorks? Are there any other PCI 100 BaseT boards that have VxWorks drivers and what kind of experience,good or bad, have people had with these boards? Preferably we would like at least 3 or 4 sources for ethernet boards. Popular NICs like the 3COM Fast Etherlink would be nice if people have used it with VxWorks. We don't want to spend time developing drivers so any that come with driver support from VxWorks or third-party would be preferable. Thanks BT - -----------== Posted via Deja News, The Discussion Network ==---------- http://www.dejanews.com/ Search, Read, Discuss, or Start Your Own --------------------------- Newsgroups: comp.os.vxworks Subject: Ampro Pentium CPU board Date: Mon, 15 Mar 1999 08:12:30 +0100 From: Ole Asbjorn Fadum Organization: Telenor Online Public Access Message-ID: <36ECB2DE.E4B5BF84@scanmar.no> Hi, For a long time I've tried to get a Ampro P5x CPU board to work with VxWorks, and Zinc graphic (Delivered by WRS). This CPU board has a 79c972 ethernet controller from AMD, and a C&T 69000 VGA controller, sounds nice but.... I'm not able to get the network work on this CPU board. 1. Is there anyony in the world who has made the P5x to work with network and standard WRS 486 BSP??? 2. Anyone which has experience with some other Ampro CPU boards and VxWorks? --------------------------- Newsgroups: comp.os.vxworks Subject: A bus error problem..please help.. Date: Mon, 15 Mar 1999 11:15:16 +0000 From: lihb Organization: Computing Center, Academia Sinica Message-ID: <36ECEBC4.BD2FFC7C@phys.sinica.edu.tw> I run this programme on my vxWorks5.1.1(MVME167) : #include "stdio.h" void xxxx() { unsigned int i; unsigned int xxdata[6]; for(i=1;i<=6;i++) { xxdata[i] = i; printf("%d\n",i); } printf("Hello \n"); } and i found that it's ok when i spawn the task with: - ->sp xxxx but if i do the below, i get into troble - ->xxxx 1 2 3 4 5 6 Hello Bus Error Program Counter: 0x0001b622 Status Register: 0x3000 Access Address : 0xffffffce Special Status : 0x0085 Bus Error Program Counter: 0x0004bf58 Status Register: 0x3008 Access Address : 0xb3b00004 Special Status : 0x0545 Task: 0x787500 "tShell" both the 0xffffffce and 0xb3b00004 are not existed, who access them?? if i define a larger array: unsigned int xxdata[4096] it won't even show me any message the system halt right after - ->xxxx anyone can show me the answer and the right way? thanks... lihb@phys.sinica.edu.tw --------------------------- End of New-News digest ********************** From owner-vxwexplo-process Mon Mar 15 06:14:09 1999 From: "Tim Allen" Date: Mon Mar 15 06:14:13 PST 1999 Subject: Re: Ampro Pentium CPU board & VxWorks Status: O >1. Is there anyony in the world who has made the P5x to work with >network and standard WRS 486 BSP??? >2. Anyone which has experience with some other Ampro CPU boards and >VxWorks? I haven't experienced the P5x yet (though I'm planning to) but I'm currently evaluating a P5i which has the SMC9000 series ethernet controller and C&T65550 graphics controller. I changed the ESMC driver parameters in config,h to match my BIOS defaults and then the pc486 vxWorks booted without any trouble. I was also able to run the Zinc/UGL demos quickly too. I've had no problems with the VxWorks/Ampro side of things, my problem is with LCD panels... The AMD device on the P5x is on PCI (as opposed to ISA on the P5i), so there will be some PCI configuration required, and the if_lnPci (?) driver parameters have to match the PCI configuration. Apologies if that's obvious to you, I'd be interested to hear how you get on, or if you think I can help, because I'll be doing this shortly! Regards, Tim. From owner-vxwexplo-process Mon Mar 15 06:57:27 1999 From: Roger Holmquist Date: Mon Mar 15 06:57:31 PST 1999 Subject: Lost in the RTX/X11-jungle..... HELP!!! Status: O Hello vxWorkwers! I have been trying to install/run RTX version 1.0 on a MVME2700 (PPC750) using Tornado 1.01 on our Sparc but I still haven't managed to run the demo-code... :( I'm only running the client-part using the options suggested in the Programmes Guide, see the file rtxcnfg.h below I have tried to run 2 different demoProggs: oclock and xlogo, both aborting on the same function, XrmParseCommand() - I'm loading the following to target from windshell: First, virtual console: vf0=open("/vio/0",2,0) ioGlobalStdSet(0,vf0) ioGlobalStdSet(1,vf0) ioGlobalStdSet(2,vf0) 2) Loading rtx: ld < /opt/wind101/ppc/target/config/rtx/rtx.o 3) Initializing the X-environment: > rtx 4) Downloading & running the democode: taskSpawn("oclock",100,0,100000,oclock) ...and so it crashes.. Here is the stack-content: 1733cc vxTaskEntry +60 : oclock (5a1be4) 5a1be4 oclock +f8 : oclockmain (5) 5a1c8c oclockmain +8c : XtOpenApplication ([]) 26d634 XtOpenApplication+88 : _XtAppInit ([]) 28d004 _XtAppInit +d8 : XtOpenDisplay ([]) 28cdd4 XtOpenDisplay +a0 : _XtPreparseCommandLine ([]) 26ce8c _XtPreparseCommandLine+d0 : XrmParseCommand ([]) Questions: - Is the environment variables correcly defined? (rtxContextGet returns 0) - Is the MVME2700-card supported? - In that case, is it correct to use the PPC604-RTX-libraries? (The alternative is the PPC603-code) - Is it necessary to set the default directory on the target by using @cd "" in windshell ? - Or is it any other error?? Thanks for any help! Greetings / Roger *********************************************** ---------------------------------------- About the configuration: /opt/wind101/ppc/target/config/rtx/: - I have included the demoCode oclock in the rtx-codeblock by adding CLIENTS = /opt/wind101/ppc/target/src/demo/rtx/oclock/oclock in the Makefile ---------------------------------------------------- rtxcnfg.h ------------------------------------------------ /* rtxcnfg.h - RtX-Windows configuration header file */ /* * This file is automatically generated from the configuration * program. Modification of the file will result in various * inconsistancies between the configuration program and this * file. Modify at your own risk. */ /* File created Mon Mar 1 14:51:45 MET 1999 from RogersRtxClientConfig_2 */ #define WIND_BASE "/opt/wind101/ppc" /***********************************************************/ /* Included Options */ /***********************************************************/ #define INCLUDE_RTX_BIGREQ_EXT /* Include X Big Request server extension */ #define INCLUDE_RTX_CLIENTS /* Include X Client startup code */ #define INCLUDE_RTX_ERROR_DATABASE /* Include text strings from error data base */ #define INCLUDE_RTX_FILE_SYSTEM /* Include file system support */ #define INCLUDE_RTX_I18N /* Include international font support */ #define INCLUDE_RTX_MIX_ENDIAN /* Include different byte order client/server connections */ #define INCLUDE_RTX_SHAPE_EXT /* Include X Shape server extension */ #define INCLUDE_RTX_TCP_CONN /* Include TCP/IP server/client connections */ #define INCLUDE_RTX_X11_LIBS /* Include X11 client libraries */ #define INCLUDE_RTX_XAW_LIBS /* Include Xaw client libraries */ #define INCLUDE_RTX_XCMISC_EXT /* Include XC misc. extension */ #define INCLUDE_RTX_XEXT_LIBS /* Include Xext client libraries */ #define INCLUDE_RTX_XKEYDB /* Include the default key symbol database */ #define INCLUDE_RTX_XMU_LIBS /* Include Xmu client libraries */ #define INCLUDE_RTX_XM_LIBS /* Include Xm client libraries */ #define INCLUDE_RTX_XT_LIBS /* Include Xt client libraries */ /***********************************************************/ /* Standard Options */ /***********************************************************/ #define LOC_CON_RCV_SIZE 4096 /* Server recv queue size */ #define LOC_CON_SND_SIZE 4096 /* Server send queue size */ #define LOC_CON_CNTXT_SWTCH TRUE /* Force client context switch after a write to server */ #define RTX_SERVER_PRIO 75 /* Server task priority */ #define RTX_SERVER_STACK 50000 /* Server stack size */ #define RTX_SERVER_FONT_PATH "" -------------------------------------------------------------------------------- ------------------------------- - # -----------------------------------------------------------------# # Roger Holmquist, FOA, National Defence Research Establishment # # sect 750, Olaus Magnus vag 42, Box 1165, 581 11 LINKOPING ------# # --------------------------- SWEDEN ------------------------------# # Phone: +46 13-378382 --------------------------------------------# # Fax: +46 13-378009 ----------------------------------------------# # Home: +46-13-214733 Mobile +46-706-250123 ----------------------# # Email: roghol@lin.foa.se ----------------------------------------# # Home address: Hovslagargrand 590 18 MANTORP ---------------------# # -----------------------------------------------------------------# xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx | This message is MIME-coded in order to avoid | | problems with national characters. MIME coding | | should be enabled when replying in Swedish. | xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx From owner-vxwexplo-process Mon Mar 15 07:55:07 1999 From: Fred Roeber Date: Mon Mar 15 07:55:10 PST 1999 Subject: Re: A bus error problem..please help.. Status: O lihb@phys.sinica.edu.tw asked: > I run this programme on my vxWorks5.1.1(MVME167) : > > #include "stdio.h" > void xxxx() > { > unsigned int i; > unsigned int xxdata[6]; > > for(i=1;i<=6;i++) > { > xxdata[i] = i; > printf("%d\n",i); > } > printf("Hello \n"); > } < SNIP > > Bus Error > Program Counter: 0x0004bf58 > Status Register: 0x3008 > Access Address : 0xb3b00004 > Special Status : 0x0545 > Task: 0x787500 "tShell" > > both the 0xffffffce and 0xb3b00004 are not existed, who access them?? > if i define a larger array: unsigned int xxdata[4096] > it won't even show me any message the system halt right after > - ->xxxx I think your problem is stack corruption. You are setting elements 1 to 6 of the xxdata array. This array is only declared as having 6 elements though. Thus only indices 0 - 5 are valid. When you set xxdata[6] you are clobbering whatever is on the stack after xxdata. What that is depends on many factors including target architecture and how you ran the program. Declare xxdata to have 7 elements and your problem should go away. Fred | Fred J Roeber, BBN Systems & Technologies | | 4 John Clarke Road Middletown, RI 02842-5202 | | froeber@bbn.com 401-848-3548 | | TraceMaker product manager -> www.tracemaker.bbn.com | From owner-vxwexplo-process Mon Mar 15 16:28:02 1999 From: "Kasse, Mary K." Date: Mon Mar 15 16:28:06 PST 1999 Subject: Problem in transferring control to the vxWorks image on MVME1603 Status: O I'm newly assigned to a project using Tornado 1.0.1/VxWorks 5.3 with Sun/Solaris 2.5.1 host and MVME1603 target using an Ethernet connection. I've programmed the boot ROM and am able to download the default kernel. After getting the following messages nothing further prints to the terminal attached to the target's serial port: Attaching network interface dc0... done. Attaching network interface lo0...done. Loading... 466880 Starting at 0x100000... I've installed the cumulative and Y2k patches for Solaris 2.5.1. I've installed the cumulative patch for Tornado 1.0.1 from WRS I've reviewed and implemented changes given in WTN 32: size of vxWorks system image < 570K, target is properly jumpered per mv1603(1), changed LOCAL_MEM_SIZE to 32MB to reflect amount of memory on target (0x02000000), turned off the cache in config.h (per note) What else can I check? Should I be adding any includes via WindConfig? Please respond via the email address below. I will compile the suggestions and post the solution. Thanks in advance, Mary Kasse Senior Design Engineer Northrop Grumman 401 E. Hendy, MS 51/2B Sunnyvale, CA 94088 (408) 735-4250 Mary_Kasse@mail.northgrum.com From owner-vxwexplo-process Mon Mar 15 17:02:56 1999 From: "Bill W.G. Shi" Date: Mon Mar 15 17:03:00 PST 1999 Subject: vxWorks SNMP Status: O Hi, vxWorks gurus: I have been instructed to add mibs to vxWorks on a platform where SENS is still not available. So we are still waiting. But can anyone out there tell me if WRS uses the SNMP implementation of that of CMU's, or their private version in SENS? Thanks. Bill W.G. Shi at BBT Inc. E-Mail Address: wgshi@bbti.com.cn From owner-vxwexplo-process Mon Mar 15 21:22:21 1999 From: "Anurag Srivastava" Date: Mon Mar 15 21:22:25 PST 1999 Subject: Master-Subagent architecture Status: O This is a multi-part message in MIME format. ------=_NextPart_000_0019_01BE6F9B.1FDF6260 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable Hi Everybody, I am using SNMP in my NMS design and using Master-Subagent = architecture (RFC 2257). I would like to know from those people who have = used WindNet code on VxWorks that whether that code supports = Master-Subagent architeecture or not ? How much effort it will take to = use the WindNet SNMP code to make it supports Master-Subagent = Architecture ? Thanks in advance. Regards, Anurag. mail to : anurags@nmi.stpn.soft.net ------=_NextPart_000_0019_01BE6F9B.1FDF6260 Content-Type: text/html; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable
Hi Everybody,
 
    I am using SNMP = in my NMS=20 design and using Master-Subagent architecture (RFC 2257). I would like = to know=20 from those people who have used WindNet code on VxWorks that whether = that code=20 supports Master-Subagent architeecture or not ? How much effort it will = take to=20 use the WindNet SNMP code to make it supports Master-Subagent = Architecture=20 ?
 
Thanks in advance.
 
Regards,
 
Anurag.
 
mail to :=20 anurags@nmi.stpn.soft.net
------=_NextPart_000_0019_01BE6F9B.1FDF6260-- From owner-vxwexplo-process Tue Mar 16 04:00:22 1999 From: daemon@csg.lbl.gov Date: Tue Mar 16 04:00:27 PST 1999 Subject: comp.os.vxworks newsdigest Status: O Comp.Os.Vxworks Daily Digest Tue Mar 16 04:00:18 PST 1999 Subject: Re: A bus error problem..please help.. Subject: About writing a driver. Subject: Re: How to get system time and date with ms resolution Subject: Re: TrueFFS Subject: Re: How to read CMOS into VxWorks clock? Subject: Re: How to get system time and date with ms resolution Subject: FS: CodeTAP MPC8xx, MPC860FADS, MT32000T Subject: Question about ethernet interface Subject: Do I need to use munch? Subject: GNU objcopy problem Subject: Re: Do I need to use munch? Subject: Re: wlmd error message "System clock has been set back too far" Subject: Re: Ampro Pentium CPU board & VxWorks Subject: Radstone ATM PMC module, Com-Path ATM vxworks driver, boot problems, performance, etc. ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: A bus error problem..please help.. Date: Mon, 15 Mar 1999 15:24:35 +0300 From: Leonid Rosenboim Organization: BitBand Technologies Ltd. Message-ID: <36ECFC03.E7E4D87D@bitband.com> References: <36ECEBC4.BD2FFC7C@phys.sinica.edu.tw> This one is really simple, and it has little to do with VxWorks, just good old "C": in "C" when you declare an array, its first element is indexed with a 0, not a 1 as in Pascal, so essentially your for loop goes from 1 to 6 inclusively, meaning that the last iteration it will write the 7th element of the array, which has only space for 6 elements, thereby overwriting a neighbouring stack location. It is worth mentioning that VxWorks was designed for experienced programmers, and somewhat lacks tools for novice programmers, it won't elegantly deal with either memory curroptions or heap memory leaks, etc. Leonid lihb wrote: > I run this programme on my vxWorks5.1.1(MVME167) : > > #include "stdio.h" > void xxxx() > { > unsigned int i; > unsigned int xxdata[6]; > > for(i=1;i<=6;i++) > { > xxdata[i] = i; > printf("%d\n",i); > } > printf("Hello \n"); > } > > and i found that it's ok when i spawn the task with: > ->sp xxxx > > but if i do the below, i get into troble > ->xxxx > 1 > 2 > 3 > 4 > 5 > 6 > Hello > > Bus Error > Program Counter: 0x0001b622 > Status Register: 0x3000 > Access Address : 0xffffffce > Special Status : 0x0085 > > Bus Error > Program Counter: 0x0004bf58 > Status Register: 0x3008 > Access Address : 0xb3b00004 > Special Status : 0x0545 > Task: 0x787500 "tShell" > > both the 0xffffffce and 0xb3b00004 are not existed, who access them?? > if i define a larger array: unsigned int xxdata[4096] > it won't even show me any message the system halt right after > ->xxxx > > anyone can show me the answer and the right way? > thanks... > > lihb@phys.sinica.edu.tw --------------------------- Newsgroups: comp.os.vxworks Subject: About writing a driver. Date: Mon, 15 Mar 1999 23:04:27 +0900 From: "ÀÌÇص¿" Organization: Set your organization ($ORGANIZATION) Message-ID: <7cj4f1$4fo$1@baikdu.kaist.ac.kr> I am new in a VxWorks. In this term, I have a term project at real-time system course. I have to do RTOS porting, make interface board, write a device driver and an application. I will implement that on VxWorks. Can anyone recommend and introduce related resources & Web sites & Books & any other material on writing a driver on ISA bus. Thanks in advance....... - Haedong Lee ( haedong@cnet.kyungpook.ac.kr ) --------------------------- Newsgroups: comp.os.vxworks Subject: Re: How to get system time and date with ms resolution Date: Mon, 15 Mar 1999 09:09:13 -0500 From: Van_D_Nguyen Organization: Raytheon Company Message-ID: <36ED1489.211376D2@res.raytheon.com> References: <7c7n0s$l2b@as1fw.mannesmann.de> <7c8321$t6g$1@menelao.polito.it> I took a look at the manual pages in the clockLib library and it seems as though the routine clock_gettime() returns time in ns, which is the unit of the returned value, but is the resolution is ns ? The routine clock_getres() in the same library returns the clock resolution and the manual page said it bases on the sysClkRateGet( ), which returns the system clock rate and it's not very high. Does this mean if your system clock is set to 100Hz (ticks) and the first call to clock_gettime() returns (after converted to nanosecond) 10000, the next calls would return something like 20000 or 30000 and never 10001 or 10002 ? Van M.Novaro wrote: > Torsten Trzeciak ha scritto nel messaggio > <7c7n0s$l2b@as1fw.mannesmann.de>... > >Hello, > >I like to get the system time with a ms resolution. Is there a kind of > >getTime function? > > > >Torsten > > Try the clock_gettime() procedure: it should return the time till ns (!) > Regards, > Marco --------------------------- Newsgroups: comp.os.vxworks Subject: Re: TrueFFS Date: Wed, 3 Mar 1999 12:16:24 -0500 From: "Rich Duszczak" Organization: Monmouth Internet Message-ID: <7bjpd9$al1$1@news.monmouth.com> References: <7be6rt$f45$1@nnrp1.dejanews.com> Stephen, You are in luck. I recently got a set of LV116BB's to work on an 860DE board using TrueFFS. It was a real pain, but here is what you have to do: Take this copy of amdmtd.c and modify the defines and the erase routine to support the JEDEC(?) code for your flash chips. Regardless of what WRS says, they will work in 32-bit mode. TrueFFS tries 4 to 1 interleaving, then 2 to 1, then it tries 1 to 1 if all else fails. You will also need to modify amdMTDIdentify to correctly detect the chips and report the proper chip size. Look at the changes I made and modify them for your needs. Also, make sure you change the switch statement in the eraseFirstBlockLV008 routine. You will see what I mean when you look at the code. If you have any other problems feel free to post more questions. Good luck, Rich Duszczak smooney@my-dejanews.com wrote in message <7be6rt$f45$1@nnrp1.dejanews.com>... >We are evaluating WindRivers flash file system TrueFFS and are >running into a problem getting it running. We are using a PPC >860T processor and 4M of AMD29LV160B flash on a 32-bit bus, so >thats two 16-bit flash chips interleaved to get the 32-bit access. > >WindRivers doesn't supply any 32bit MTD's and the first level of >customer support says they have never run in this configuration, >(although the sales support feel they have.) > >My attempts to modify the existing MTD's have been unsuccessful, has >anyone ever used TrueFFS with 32-bit access? Any pointers would be >appreciated. > >-Stephen Mooney > >-----------== Posted via Deja News, The Discussion Network ==---------- >http://www.dejanews.com/ Search, Read, Discuss, or Start Your Own begin 666 amdmtd.c M+RH-"B J("1,;V2!D:69F97)E;G0@:6YT97)L96%V:6YG2@I M#0H@*@T*("H@(" @4F5V(#$N,38@(" R-"!397 @,3DY-R Q-SHT-3HU,B @ M(&1A;FEG#0H@*B!$969A=6QT(&EN=&5R;&5A=FEN9R!V86QU92!I#DP#0HC9&5F:6YE(%-54U!%3D1?15)!4T4) M,'AB, T*(V1E9FEN92!314-43U)?15)!4T4),'@S, T*(V1E9FEN92!215-5 M345?15)!4T4),'@S, T*(V1E9FEN92!214%$7T%24D%9"3!X9C -"@T*(V1E M9FEN92!53DQ/0TM?,0DP>&%A#0HC9&5F:6YE"55.3$]#2U\R"3!X-34-"@T* M(V1E9FEN92!53DQ/0TM?041$4C$),'@U-34U=0T*(V1E9FEN90E53DQ/0TM? M041$4C(),'@R86%A=0T*#0HC9&5F:6YE"40R"0DT"2\J(%1O9V=L97,@=VAE M;B!E# Q830-"B-D969I;F4)06TR.48P.#!?1DQ!4T@)"3!X,#%D M-0T*(V1E9FEN90E!;3(Y3%8P.#!?1DQ!4T@)"3!X,#$S. T*(V1E9FEN92!! M;3(Y3%8P,#A?1DQ!4T@@(" @(" @(" P># Q,S<-"B-D969I;F4@06TR.48P M,39?1DQ!4T@@(" @(" @(" @,'@P,6%D#0HC9&5F:6YE"4%M,CE&,#$V0U]& M3$%32" @"3!X,#$S9 T*(V1E9FEN92!!;3(Y3%8P,3=?1DQ!4T@@(" @(" @ M(" P># Q8S@-"@T*+RH@*BHJ*BHJ*BHJ*BHJ*BHJ*BHJ*BHJ*BHJ*BHJ*BHJ M*BHJ*BHJ*BHJ*BHJ*B J+PT*+RH@*BHJ*BHJ*BHJ*B!!1$1%1"!"62!224-( M("HJ*BHJ*BHJ*BHJ*BHJ*BHJ*B J+PT*+RH@*BHJ*BHJ*BHJ*BHJ*BHJ*BHJ M*BHJ*BHJ*BHJ*BHJ*BHJ*BHJ*BHJ*BHJ*B J+PT*#0HC9&5F:6YE($%M,CE, M5C$Q-D)"7T9,05-("3!X,#$T8PT*#0H-"B-D969I;F4@1G5J,CE&,#0P7T9, M05-((" @(" @(" @,'@P-&$T#0HC9&5F:6YE"49U:C(Y1C X,%]&3$%32 D) M,'@P-&0U#0HC9&5F:6YE"49U:C(Y3%8P.#!?1DQ!4T@),'@P-#,X#0HC9&5F M:6YE($9U:C(Y3%8P,#A?1DQ!4T@@(" @(" @(#!X,#0S-PT*(V1E9FEN92!& M=6HR.48P,39?1DQ!4T@@(" @(" @(" P># T860-"B-D969I;F4)1G5J,CE& M,#$V0U]&3$%32" @"3!X,#0S9 T*(V1E9FEN92!&=6HR.4Q6,#$W7T9,05-( M(" @(" @(" P># T8S@-"@T*#0HO*BTM+2TM+2TM+2TM+2TM+2TM+2TM+2TM M+2TM+2TM+2TM+2TM+2TM+2TM+2TM+2TM+2TM+2TM+2TM+2TM+2TM+2TM+2TM M+2TJ+PT*+RH@(" @(" @(" @(" @(" @(" @(" @(" @;2!A(' @0B!A(',@ M90D)"0DJ+PT*+RH)"0D)"0D)"0DJ+PT*+RH@36%P('1H92!W:6YD;W<@=&\@ M82!P86=E(&)A6EN9R!DPT*#0H-"B @:68@*'1H:7-6 M87)S+3YU;FQO8VM!9&1R,2 A/2!.3U]53DQ/0TM?041$4BD-"B @>PT*(" @ M("HH*$9L87-H4%12*2!F;$%D9$QO;F=4;T9A#$P,# P;"!\("AV;VPN:6YT97)L96%V:6YG("T@,2DI.PT* M#0H@(" @*BA&;&%S:%!44BD@9FQ-87 H=F]L+G-O8VME="QB87-E061D&8P(#L-"B @(" J*$9L87-H4%12*2!F;$UA<"AV;VPN71E" @;&5N9W1H/25D("!O=F5R=W)I=&4] M)61<;B(L8G5F9F5R+&QE;F=T:"QO=F5R=W)I=&4I(#L-"@ED*&)U9F9EPT*;&%S=$)Y M=&4Z#0HC:69D968@7U]C<&QU#$P,3L-"B @(" @("IW M56YL;V-K061DPT*(" @(" @("!I9B H*"AW1FQA M('="=69F97); M+3%=*2 F(#!X9F8I*2!\? T*(" @(" @(" @(" @*"AW1FQA&9F,# I*2D-"@D)>PT*"0D)=T9L87-H4'1R6RTQ72 ](%)% M041?05)205D@*B P>#$P,3L-"B-I9F1E9B!$14)51U]04DE.5 T*"0D)1$5" M54=?4%))3E0H(E)E='5R;FEN9R!F;%=R:71E1F%U;'0N(BQF;%=R:71E1F%U M;'0I.PT*(V5N9&EF#0H)"0ER971U#$P,3 Q,#%L=3L-"B @(" @("ID56YL;V-K061DPT*"6EN="!I.PT*"69O&9F*2D@?'P-"B @(" @(" @(" @("@H9$9L87-H4'1R6RTQ M72 F("A$-2 J(#!X,3 P*2D@)B8@*"AD1FQA M(&1"=69F97);+3%=*2 F(#!X9F8P,# P;'4I*2!\? T*(" @(" @(" @(" @ M*"AD1FQA&9F,# P,# P;'4I*2D- M"@D)>PT*"0D)9$9L87-H4'1R6RTQ72 ](%)%041?05)205D@*B P>#$P,3 Q M,#%L=3L-"B-I9F1E9B!$14)51U]04DE.5 T*"0D)1$5"54=?4%))3E0H(E)E M='5R;FEN9R!F;%=R:71E1F%U;'0N(BQF;%=R:71E1F%U;'0I.PT*(V5N9&EF M#0H)"0ER971U71EPT* M("!I;G0@:5-U8D)L;V-K.PT*("!L;VYG('-U8D)L;V-K4VEZ92 ](# [#0H- M"B\J1$5"54=?4%))3E0H(E-T87)T:6YG+BXN(BPP*2 [*B\-"@T*("!F;W(@ M*&E3=6)";&]C:R ](# [(&E3=6)";&]C:R \(#0[(&E3=6)";&]C:RLK*0T* M("![#0H@(" @:6YT(&D[#0H@(" @1FQAPT*(" @(" @8V%S92 Q.@T*"0D)F4@*ST@,'@R,# P.PT*"0D)8G)E86L[#0H@(" @ M?0T*#0H@(" @9FQAF4@*@T*"0D)"79O;"YI M;G1EPT*"0D)9F]R("AI(#T@,#L@:2 \('9O;"YI;G1E(&9L87-H4'1R6VE=*2 F($0V*2 [#0H)"0E] M#0H)"7T-"B-E;F1I9@T*"0EF:6YIPT*"2!I M;G0@:2 [#0H)(&9O&9F*0T*"0D)>PT*"0D)"6EF("@H9FQA6EN9R!D" @9FER#0P,# P+"!F:7)S M=$5R87-A8FQE0FQO8VLL(&YU;4]F17)A'1R82!C;VYD M:71I;VX@9F]R($Q6,3$V(&EN(&EF('-T871E;65N=" J+PT*+RH@*BHJ*BHJ M*BHJ*BHJ*BHJ*BHJ*BHJ*BHJ*BHJ*BHJ*BHJ*BHJ*BHJ*BHJ*B J+PT*#0H@ M(" @+RH@5&AE(&9IPT*"0EC:&5C:U-T871U2 J+PT*(" @(&9L87-H4'1R(#T@*$9L87-H M4%12*79O;"YM87 H)G9O;"P@*&9IPT*"0EEPT*"0EI9B H9FQAPT*"0D):68@*"AF;&%S:%!T&9F*0T*"0D)>PT*"0D)"69L87-H4'1R6VE= M(#T@4D5!1%]!4E)!63L-"B-I9F1E9B!$14)51U]04DE.5 T*"0D)"41%0E5' M7U!224Y4*")2971UF4H=F]L+G-O8VME="PR*3L)+RH@."!+0GET97,@*B\- M"@T*("!V;VPN;71D5F%R2!D:69F97)E;G0@:6YT97)L96%V:6YGPT*(" @(&EF("AI;FQV(#T](#$I#0H@(" @("!F;%-E=%=I;F1O=T)U7!E(#T@(BQV;VPN='EP92D@.PT*(V5N9&EF#0H- M"B\J("HJ*BHJ*BHJ*BHJ*BHJ*BHJ*BHJ*BHJ*BHJ*BHJ*BHJ*BHJ*BHJ*BHJ M*BH@*B\-"B\J("HJ*BHJ*BHJ*BH@041$140@0ED@4DE#2" J*BHJ*BHJ*BHJ M*BHJ*BHJ*BH@*B\-"B\J($%D9&5D(&5X=')A(&-O;F1I=&EO;B!F;W(@3%8Q M,38@:6X@:68@7!E(#T]($%M,CE&,#$V0U]&3$%32"!\?"!V M;VPN='EP92 ]/2!&=6HR.48P,39#7T9,05-(('Q\#0H)"79O;"YT>7!E(#T] M($%M,CE&,#@P7T9,05-(('Q\('9O;"YT>7!E(#T]($9U:C(Y1C X,%]&3$%3 M2"!\? T*"0EV;VPN='EP92 ]/2!!;3(Y3%8P.#!?1DQ!4T@@?'P@=F]L+G1Y M<&4@/3T@1G5J,CE,5C X,%]&3$%32"!\? T*"0EV;VPN='EP92 ]/2!!;3(Y M3%8P,#A?1DQ!4T@@?'P@=F]L+G1Y<&4@/3T@1G5J,CE,5C P.%]&3$%32"!\ M? T*"0EV;VPN='EP92 ]/2!!;3(Y1C T,%]&3$%32"!\?"!V;VPN='EP92 ] M/2!&=6HR.48P-#!?1DQ!4T@@?'P-"@D)=F]L+G1Y<&4@/3T@06TR.4Q6,#$W M7T9,05-(('Q\('9O;"YT>7!E(#T]($9U:C(Y3%8P,3=?1DQ!4T@@?'P-"@D) M=F]L+G1Y<&4@/3T@06TR.4Q6,3$V0D)?1DQ!4T@I#0H)"0EB'1R82!C;VYD:71I;VX@9F]R M($Q6,3$V(&EN(&EF('-T871E;65N=" J+PT*+RH@*BHJ*BHJ*BHJ*BHJ*BHJ M*BHJ*BHJ*BHJ*BHJ*BHJ*BHJ*BHJ*BHJ*BHJ*B J+PT*#0H@(&EF("AV;VPN M='EP92 ]/2!!;3(Y1C Q-E]&3$%32"!\?"!V;VPN='EP92 ]/2!&=6HR.48P M,39?1DQ!4T@@?'P-"B @(" @('9O;"YT>7!E(#T]($%M,CE&,#$V0U]&3$%3 M2"!\?"!V;VPN='EP92 ]/2!&=6HR.48P,39#7T9,05-(('Q\#0H@(" @("!V M;VPN='EP92 ]/2!!;3(Y3%8P,3=?1DQ!4T@@?'P@=F]L+G1Y<&4@/3T@1G5J M,CE,5C Q-U]&3$%32"!\? T*"2 @=F]L+G1Y<&4@/3T@06TR.4Q6,3$V0D)? M1DQ!4T@I#0H)"79O;"YC:&EP4VEZ92 ](#!X,C P,# P;#L-"B @96QS92!I M9B H=F]L+G1Y<&4@/3T@1G5J,CE&,#@P7T9,05-(('Q\('9O;"YT>7!E(#T] M($%M,CE&,#@P7T9,05-(('Q\#0H)"79O;"YT>7!E(#T]($9U:C(Y3%8P.#!? M1DQ!4T@@?'P@=F]L+G1Y<&4@/3T@06TR.4Q6,#@P7T9,05-(('Q\#0H)"79O M;"YT>7!E(#T]($9U:C(Y3%8P,#A?1DQ!4T@@?'P@=F]L+G1Y<&4@/3T@06TR M.4Q6,# X7T9,05-(*0T*"0D)=F]L+F-H:7!3:7IE(#T@,'@Q,# P,#!L.PT* M("!E;'-E(&EF("AV;VPN='EP92 ]/2!&=6HR.48P-#!?1DQ!4T@@?'P@=F]L M+G1Y<&4@/3T@06TR.48P-#!?1DQ!4T@I#0H)"0EV;VPN8VAI<%-I>F4@/2 P M>#@P,# P;#L-"B @96QS90T*("![#0HC:69D968@1$5"54=?4%))3E0-"B @ M("!$14)51U]04DE.5"@B4F5T=7)N:6YG(&9L56YK;F]W;DUE9&EA+B(L9FQ5 M;FMN;W=N365D:6$I.PT*(V5N9&EF#0H@(" @#4U-2 J('9O;"YI;G1EPT*("!I9B H;F]/9DU41',@/CT@351$4RD-"B @ M("!R971U then set VxWorks's system time/date with this data? I didn't see any > thing to do this built in to the PC486 BSP or otherwise? > > Thanks for any ideas you may have that will help me, > Marc Try this peace of code to access the PC RTC and do what you want with..... I use it and it work fine. Olivier STOLZENBERG PROSYS S.A. unsigned char uch_Temp[7]; /* Sauvergarde de l'heure et de la date */ sysOutByte(0x70, 4); /* Hours */ uch_Temp[0] = sysInByte(0x71); sysOutByte(0x70, 2); /* Minutes */ uch_Temp[1] = sysInByte(0x71); sysOutByte(0x70, 0); /* Seconds */ uch_Temp[2] = sysInByte(0x71); sysOutByte(0x70, 7); /* Month day */ uch_Temp[3] = sysInByte(0x71); sysOutByte(0x70, 8); /* Month */ uch_Temp[4] = sysInByte(0x71); sysOutByte(0x70, 9); /* Year */ uch_Temp[5] = sysInByte(0x71); sysOutByte(0x70, 0x32); /* Century */ uch_Temp[6] = sysInByte(0x71); --------------------------- Newsgroups: comp.os.vxworks Subject: Re: How to get system time and date with ms resolution Date: Mon, 15 Mar 1999 08:51:15 -0800 From: Scott Kamara Organization: AP Labs Message-ID: <36ED3A83.F49685F8@pop.sttl.uswest.net> References: <7c7n0s$l2b@as1fw.mannesmann.de> <7c8321$t6g$1@menelao.polito.it> <36ED1489.211376D2@res.raytheon.com> Reply-To: spk@uswest.net > Does this mean if your system clock is > set to 100Hz (ticks) and the first call to clock_gettime() returns (after > converted to nanosecond) 10000, the next calls would return something like > 20000 or 30000 and never 10001 or 10002 ? That is right. Again, clock_gettime() returns 'secs:ns' based on the posix clock of type CLOCK_REALTIME which, is the same as the system clock (sysClk) resolution, whatever that may be for your system. - ----------------------- Scott Kamara AP Data Systems Seattle Engr. Office (206) 525-8592 scott@aplabs.com --------------------------- Newsgroups: comp.os.vxworks Subject: FS: CodeTAP MPC8xx, MPC860FADS, MT32000T Date: Fri, 5 Mar 1999 12:15:48 -0500 From: "David Ebinger" Organization: Together Networks - Burlington, VT. Message-ID: <36e010d9.0@news.together.net> We have the following equipment for sale: (4) Applied Microsystems MPC8xx CodeTAP w/ MWX-ICE MPC8xx Ver. 3.50.1 for SunOS 4.1.x - Unused (1) Applied Microsystems MPC8xx CodeTAP - used, no software (4) Motorola MPC860FADS boards (1) Mitsubishi M32000T-2MB-E Unused - no software If you are interested in any of this equipment, please make us an offer. Pictures available here: http://www.ebinger.com/MiscTest.htm Thanks, David Ebinger Ebinger Equipment --------------------------- Newsgroups: comp.os.vxworks Subject: Question about ethernet interface Date: Tue, 16 Mar 1999 09:26:14 +0800 From: "SRD Zha Junhai" Organization: SBELL Message-ID: <36edb43a.1@dnews.sbell.com.cn> Hi, everybody I have a board use 860. There are 2 ethernet interface on it. Now I want to send a package from No.1 ethernet at first. I can get a package from which I will know the interface structure,then I can send the package to it. But I want to send the package at first. So please tell How I can do it? Hope for your answer, thanks. Best Regards Zha Junhai --------------------------- Newsgroups: comp.os.vxworks Subject: Do I need to use munch? Date: Mon, 15 Mar 1999 20:32:56 -0500 From: Keith A Arner Organization: Carnegie Mellon, Pittsburgh, PA Message-ID: For VxWorks, does munch do anything for C++ code, other than set up the static constructors? That is, is there any reason to munch my .o files if I'm not actually using any static objects? My impression from the documentation is that munch in this case would just be overhead, but I'd like to make sure. Thanks, Keith --------------------------- Newsgroups: comp.os.vxworks Subject: GNU objcopy problem Date: Mon, 15 Mar 1999 22:12:29 -0600 From: defaultuser@domain.com Organization: http://www.msfc.nasa.gov/ Message-ID: <36EDDA2C.1B2369A0@domain.com> Greetings, I am attempting to use objcopymips from the GNU toolkit to generate a binary file using the following command: objcopymips -Obinary monitor monior.bin When I run the command I get an error that the output device is full even though I have >200MB free diskspace (the binary file should be less than 256kbytes). I also had problems using the --byte command line switch. I am using objcopymips version cygnus-2.6 under windows NT4. Any help would be greatly appreciated David --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Do I need to use munch? Date: Tue, 16 Mar 1999 07:51:08 GMT From: bobarmstrong@home.net (Robert Armstrong) Organization: @Home Network Message-ID: References: It's constructors and destructors actually. If you don't have them then it won't waste execution time and the compilation time hit is small (perhaps I'm spoilt for host machines). Bob In article , Keith A Arner wrote: > >For VxWorks, does munch do anything for C++ code, other than set up the >static constructors? That is, is there any reason to munch my .o files if >I'm not actually using any static objects? > >My impression from the documentation is that munch in this case would just >be overhead, but I'd like to make sure. > >Thanks, >Keith > > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: wlmd error message "System clock has been set back too far" Date: Tue, 16 Mar 1999 07:57:06 GMT From: bobarmstrong@home.net (Robert Armstrong) Organization: @Home Network Keywords: License Server, Tornado 1.0.1, Wind River, Y2K Message-ID: References: <1999Mar15.105252@sun18> If you have support go to the web site www.wrs.com, go to support and look for the patch to the license manager which is the third entry on the list. This is not a Y2K problems, it's just a daylight saving time problem and has been fixed for at least a year. Bob In article <1999Mar15.105252@sun18>, mathar@mpia-hd.mpg.de wrote: >The execution of >wlmd -e $WIND_BASE/.wind/license -l /tmp/wlmd.log -v 4 >on our Sun 5.5.1 host is terminated with the error message >99-Mar-15 10:46:48: System clock has been set back too far!. WLMD Terminated. > >What does this mean? Is there a Y2K interferences? >date >on the host responds as it ought to: >Mon Mar 15 10:55:41 MET 1999 > >=============================================================================== >e-mail: mathar@mpia-hd.mpg.de >Richard J Mathar, Max-Planck-Institut fur Astronomie, >Konigstuhl 17, 69117 Heidelberg, Germany --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Ampro Pentium CPU board & VxWorks Date: Tue, 16 Mar 1999 08:19:42 +0100 From: Ole Asbjorn Fadum Organization: Telenor Online Public Access Message-ID: <36EE060E.413BFF60@scanmar.no> References: Hi, Yes, and you also have to install SENS, because the 79c972 devicedriver is a SENS driver. But my problem is that I got various errors. When I try to boot from the network, it doesn't work. Just hung. If I boot from a floppy, I could even get ping response, but when I try to connect to the target from Tornado, BANG connection timeout. I'm really fed up with Ampro, and it seems that the Ampro guys don't want to comment my problems or tell what they have done to get the VxWorks kernel to work. Tim Allen wrote: > >1. Is there anyony in the world who has made the P5x to work with > >network and standard WRS 486 BSP??? > >2. Anyone which has experience with some other Ampro CPU boards and > >VxWorks? > > I haven't experienced the P5x yet (though I'm planning to) but I'm currently evaluating a P5i which has the SMC9000 series ethernet controller and C&T65550 graphics controller. I changed the ESMC driver parameters in config,h to match my BIOS defaults and then the pc486 vxWorks booted without any trouble. > I was also able to run the Zinc/UGL demos quickly too. I've had no problems with the VxWorks/Ampro side of things, my problem is with LCD panels... > > The AMD device on the P5x is on PCI (as opposed to ISA on the P5i), so there will be some PCI configuration required, and the if_lnPci (?) driver parameters have to match the PCI configuration. > > Apologies if that's obvious to you, I'd be interested to hear how you get on, or if you think I can help, because I'll be doing this shortly! > > Regards, > Tim. --------------------------- Newsgroups: comp.os.vxworks Subject: Radstone ATM PMC module, Com-Path ATM vxworks driver, boot problems, performance, etc. Date: 12 Mar 1999 17:19:27 GMT From: jcl@apl.washington.edu (Jim Luby) Organization: APL-UW Message-ID: <7cbiav$149m$1@nntp6.u.washington.edu> Hi, I received a request for information about vxworks and ATM issues I'm working on. I decided to post my reply in case it helps someone else and in case someone is able to help solve a problem I'm having. Cheers, Jim >Jim, >I just read a message you posted to the comp.os.vxworks newsgroup which was >discussing ethernet/ATM performance. I am working with a setup similiar to >yours - MVME2604 SBC, VxWorks 5.3.1. As part of our project, we will be >integrating with another lab using ATM. We are currently looking for ATM PMC >cards for the MVME2604. I was wondering if you could provide some feedback >as to what type of ATM card you are using, how well the software drivers >work, quality of support you've received from the supplier, how much time >was required to set up the ATM card, etc. We have not worked with ATM >before, so it is new ground for us. Currently we are looking at two cards - >one from Radstone, the other from Cetia. >Thanks, > > "John Doe" > Hi "John," Yes I would be glad to share my experiences with you though I can only speak to the Radstone PMC module and associated Com-Path driver (which camewith the Radstone boards in my case). We ordered 7 ATM modules from Radstone of which we had to return 2 due to hardware problems. Apparently they had some quality control problems last year and cards made before a certain date are problematic. One of the 2 cards we returned was of the older vintage while the second was of a newer batch (as best I can infer based on serial numbers). When the cards work they seem to work fine. As I think I mentioned in my newsgroup posting I am able to get transfer rates (between my PowerPC 2304 board and a Sun) of 9-10 MB/sec when transferring 1 MByte buffers using TCP/IP socket transfers. Our ATM network is configured as a single emulated local area network (ELAN). This performance was achieved only after the following two steps were taken: A. Using setsockopt() specify the TCP_NODELAY option; B. Using setsockopt() increase the size of the SO_SNDBUF from the default value of 4096 to 32768. I did this on the Sun end of things also although the effect was not as dramatic as doing it on the VxWorks side. As to the driver there are a few issues of import as follows: 1. Unless you make some minor code modications to the ravenMpic.c routine in the BSP target directory you will get a fairly high rate of MPIC spurious interrupts. I''ve made Com-Path aware of the problem though it's not in their driver but rather in one of the board support routines from Wind River. If you decide to try this combination of hardware let me know and I'll email you the code modification for ravenMpic.c. 2. If I boot vxworks in what I'll refer to as the "normal" mode (downloads the kernel over ethernet) I can then use a script to load the ATM driver code and start the ATM interface. In this mode everything works well. However, I've been trying to modify the Wind River boot loader code so that it downloads the kernel over ATM (i.e. we're trying to get rid of the ethernet altogether) and I've run into problems. I'm tantalizingly close to having this working but for some unknown reason (to me, to Compath and to WRS) when the ATM driver code tries to set up the connection to the ATM switch it fails. The Radstone and Fore Systems ATM switch are exchanging data packets but the packets arriving at the switch are garbage. This is happening in the ILMI phase where the Radstone and switch attempt to exchange hardware addresses with each other (to the best of my understanding). Since I don't have source code for the driver I'm kind of stuck although the main person at Compath (Jim Gleba) is very helpful. He has spent many hours on the phone trying to understand this problem as well as helping me get things working originally. Jim has offered to add some debugging output to the driver code to see if it helps track down the problem I'm having. There is apparently some part of the operating system that is not being initialized (or who knows what) in the bootloader (reduced vxworks kernel [see tornado/target/config/all/bootconfig.c if you are a real glutton for punishment!] ) but that is there as part of the "normal" vxworks kernel. If I find a solution to this problem I'll also let you know. That's all I can think of right now. I think I'll also post this on the newsgroup in case anyone has suggestions for my problem. Best regards, Jim Luby Dr. James C. Luby Applied Physics Laboratory University of Washington 1013 NE 40 Street Seattle, WA 98105-6698 Tel: 206-543-6854 Fax: 206-543-6785 --------------------------- End of New-News digest ********************** From owner-vxwexplo-process Tue Mar 16 06:30:13 1999 From: Knower.Cat@Nowhere.Net Date: Tue Mar 16 06:30:17 PST 1999 Subject: Re: advantages of SENS Status: O Markus Mitterer wrote: > > Hi all, > when I look at this newsgroup, I see a lot of people using SENS. I > don't have this optional product but I must be able to transmit much > data via ethernet. So my questions are: Is SENS much faster (and how > much) than the default WRS network stack? Are there any other > advantages of SENS? Markus, the answer is not as simple as one may think: SENS implements an IP routing algorithm which is way more complex then the 4.3bsd, in order to achieve faster and constant time routing when hundreds of routing table entries are resent. However, in the most common case where only a handful of entries are present in the routing table, SENS performance is worse when compared to the old stack. Also, in SENS they recommend to use a newly devised END driver API, which adds a layer (MUX) between the driver and IP, and that has a price too. The bad news however, is that SENS is going to be the DEFAULT stack in all future releases of VxWorks, starting Tornado 2.0, hence you don't really have a choice. Over-and-out. From owner-vxwexplo-process Tue Mar 16 07:29:38 1999 From: "Kasse, Mary K." Date: Tue Mar 16 07:29:42 PST 1999 Subject: Problem in transferring control to the vxWorks image on MVME2603 Status: O (same problem...correct target number). > I'm newly assigned to a project using Tornado 1.0.1/VxWorks 5.3 with > Sun/Solaris 2.5.1 host and MVME2603 target using an Ethernet connection. > I've programmed the boot ROM and am able to download the default kernel. > After getting the following messages nothing further prints to the > terminal attached to the target's serial port: > > Attaching network interface dc0... done. > Attaching network interface lo0...done. > Loading... 466880 > Starting at 0x100000... > > I've installed the cumulative and Y2k patches for Solaris 2.5.1. > I've installed the cumulative patch for Tornado 1.0.1 from WRS > I've reviewed and implemented changes given in WTN 32: > size of vxWorks system image < 570K, > target is properly jumpered per mv1603(1), > changed LOCAL_MEM_SIZE to 32MB to reflect amount of memory on target > (0x02000000), > turned off the cache in config.h (per note) > > What else can I check? > Should I be adding any includes via WindConfig? > Please respond via the email address below. > I will compile the suggestions and post the solution. > > Thanks in advance, > Mary Kasse > Senior Design Engineer > Northrop Grumman > 401 E. Hendy, MS 51/2B > Sunnyvale, CA 94088 > (408) 735-4250 > Mary_Kasse@mail.northgrum.com From owner-vxwexplo-process Tue Mar 16 10:56:17 1999 From: spillai@corp.cirrus.com (Subramonia Pillai) Date: Tue Mar 16 10:56:21 PST 1999 Subject: Hi Status: O VXWORKS: Hi, This is related to Point-Point Protocol implementaion of VxWorks. The current implementaion is based on the assumption that PPP will always run above serial link. And we can configure a particular PPP link by pppInit (unit, tty driver, ...). The tty deiver will be used for sending and receiving character for PPP protcol data units. I want to use the vxworks PPP implemetaion to run above ATM. Since the current implementaion is tightly coupled with tty driver, How I can make existing PPP to use a ATM PVC connection establised by me (as per RFC 2364, PPP over AAL5) instead of tty device for sending PPP protocol data units?. Any hints will be useful for me. Thanks ssp From owner-vxwexplo-process Tue Mar 16 13:24:01 1999 From: Dennis Lewis Date: Tue Mar 16 13:24:05 PST 1999 Subject: Re: Slow TCP/IP Sockets Status: O We here at Lockheed Martin in Goodyear, AZ have also encountered slow transfer rates between the PowerPC (MVME2604, vxWorks 5.3.1) and the Sun host with Solaris 2.5.1 (SunOS 5.5.1). Steve Riffle fixed the problem as follows: To get better TCP/IP performance, set the TCP_NODELAY on the socket. This needs to be done before a connect. Need to include #include #ifdef VXWORKS int one = 1; setsockopt (Sock, IPPROTO_TCP, TCP_NODELAY, (char *) &one, sizeof(int)); #endif The setsockopt in windman talks about this in more detail. **************************************** * Dennis B. Lewis * * Lockheed Martin TDS-AZ * * MS 1322 * * 1300 S. Litchfield Rd. * * Goodyear, AZ 85338 * * * * Phone: (602) 925-7593 * * Fax : (602) 925-7873 * * Email: lewis@master.tds-az.lmco.com * * * **************************************** From owner-vxwexplo-process Tue Mar 16 14:48:11 1999 From: spillai@corp.cirrus.com (Subramonia Pillai) Date: Tue Mar 16 14:48:15 PST 1999 Subject: PPP over ATM Status: O VXWORKS: Hi, This is related to Point-Point Protocol implementaion of VxWorks. The current implementaion is based on the assumption that PPP will always run above serial link. And we can configure a particular PPP link by pppInit (unit, tty driver, ...). The tty deiver will be used for sending and receiving character for PPP protcol data units. I want to use the vxworks PPP implemetaion to run above ATM. Since the current implementaion is tightly coupled with tty driver, How I can make existing PPP to use a ATM PVC connection establised by me (as per RFC 2364, PPP over AAL5) instead of tty device for sending PPP protocol data units?. Any hints will be useful for me. Thanks ssp From owner-vxwexplo-process Tue Mar 16 23:41:22 1999 From: "Michael Lawnick" Date: Tue Mar 16 23:41:26 PST 1999 Subject: Re: Problem in transferring control to the vxWorks image on MVME Status: O > (same problem...correct target number). > > I'm newly assigned to a project using Tornado 1.0.1/VxWorks 5.3 with > > Sun/Solaris 2.5.1 host and MVME2603 target using an Ethernet connection. > > I've programmed the boot ROM and am able to download the default kernel. > > After getting the following messages nothing further prints to the > > terminal attached to the target's serial port: > > > > Attaching network interface dc0... done. > > Attaching network interface lo0...done. > > Loading... 466880 > > Starting at 0x100000... -- snip -- Hi Mary, check the kernel size (file vxWorks). If it is grater than 1 MB you should move the bootimage to a higher load address : Change the constant RAM_HIGH_ADRS in your BSP makefile and in your config.h (must be same in both). Now rebuild your bootimage and kernel and flash the bootimage. HTH Michael 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 owner-vxwexplo-process Wed Mar 17 00:52:35 1999 From: Andy Foster Date: Wed Mar 17 00:52:39 PST 1999 Subject: Tornado "round" function on PPC (MVME2700) Status: O Does anyone know if the "round" function is supported under Tornado on the MVME2700 PPC? Cheers, Andy =================================================================================== Andy Foster Gemini 8-m Telescopes Project Tel: 1 (808) 974 2556 (direct) Gemini-North Base Facility Fax: 1 (808) 935 9235 670 North A'ohoku Place University Park E-mail: afoster@gemini.edu or ajf@noao.edu Hilo Home Page: http://controls.gemini.edu/ajf Hawaii 96720 USA From owner-vxwexplo-process Wed Mar 17 04:00:53 1999 From: daemon@csg.lbl.gov Date: Wed Mar 17 04:00:57 PST 1999 Subject: comp.os.vxworks newsdigest Status: O Comp.Os.Vxworks Daily Digest Wed Mar 17 04:00:50 PST 1999 Subject: Re: wlmd error message "System clock has been set back too far" Subject: Re: tornado utilities on linux under dosemu? ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: wlmd error message "System clock has been set back too far" Date: 16 Mar 1999 18:14:16 GMT From: mathar@mpia-hd.mpg.de (Richard Mathar) Organization: Max-Planck-Institut fuer Astronomie Keywords: License Server, Tornado 1.0.1, Wind River, Y2K Message-ID: <7cm71o$cdm@sun0.urz.uni-heidelberg.de> References: <1999Mar15.105252@sun18> Reply-To: mathar@mpia-hd.mpg.de In article <1999Mar15.105252@sun18>, mathar@mpia-hd.mpg.de (Richard Mathar) writes: |> The execution of |> wlmd -e $WIND_BASE/.wind/license -l /tmp/wlmd.log -v 4 |> on our Sun 5.5.1 host is terminated with the error message |> 99-Mar-15 10:46:48: System clock has been set back too far!. WLMD Terminated. |> The problem has been solved now. The Elan license manager checks /etc/wtmp (yes, that's where the user's are logged if they connect to the system) in a way and for a reason that is not obvious to us, and for an unknown reason there were entries dated in the future listed there. Removing this file was our way to get wlmd running.. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: tornado utilities on linux under dosemu? Date: 3 Mar 1999 15:27:47 GMT From: hat@se-46.wpa.wtb.tue.nl (Albert Hofkamp) Organization: Eindhoven University of Technology, The Netherlands Message-ID: References: <36D44CF5.71F167CA@yahoo.com> Reply-To: a.hofkamp@wtb.tue.nl On 24 Feb 1999 11:05:19 PST, Jim Wiggins wrote: >has anyone tried doing vxworks builds on linux by running the >tornado utilities like ccppc, ldppc, nmppc, etc. under dosemu? >as you can see, i'm interested in the power-pc target. I have been using vnc to have control of a remote w95 box on my linux box. Works great ! Albert - --- Look ma, windows without Windows !! --------------------------- End of New-News digest ********************** From owner-vxwexplo-process Wed Mar 17 07:50:36 1999 From: "Paul Myers" Date: Wed Mar 17 07:50:39 PST 1999 Subject: RPC Timeouts Status: O Does anybody have any suggestions as to what might be causing an RPC timeout when trying to connect to the Wind River system target agent on a Coldfire core? I am using a VisionICE EST box to connect via EtherNet to connect to a board running a ColdFire core. The Target Server indicates the back end is successfully connected to and attempts to connect to the target agent. It then after a second or two generates the error: Wind River Systems Target Server: NT/Win95 version Attaching backend... succeeded. Connecting to target agent... Error: rpccore backend client Timed out failed. I am working in a separate development environment from the main group at another site. The other site does not experience this problem so I don't believe the VxWorks Kernel software is the problem since every tool, the Kernel build and the EST box are identical. I have spent considerable time examining my PC and the EST box configurations which I believe to be the cause. Paul Myers Questra Corporation From owner-vxwexplo-process Wed Mar 17 09:31:52 1999 From: wardd@ttc.com Date: Wed Mar 17 09:31:55 PST 1999 Subject: VxWorks - I18n Status: O We're building an embedded web application that uses Wind web server and VxWorks. Any thoughts/experiences with peforming internationalization on the HTML pages to be sent back. Some of the pages are static (we can just translate them); however, some of the pages are created dynamically. How do we internationalize them? I don't see any direct support for i18n in VxWorks as there is in other operating systems. Thanks, David Ward From owner-vxwexplo-process Wed Mar 17 16:33:15 1999 From: "Murphy, John" Date: Wed Mar 17 16:33:18 PST 1999 Subject: opening files over wdbSerial Status: O ob vxWorks tornado et al. I've got some hardware with a single serial port running vxWorks 5.3.1. Because of other hardware restrictions, I cannot use my Ethernet interface with tornado. I've successfully compiled in WDB serial support and have the wdb serial backend working with my system. The tornado tools all seem to work well (browser, windShell, etc). My problem is that I'm unable to open files. I can load/unload module files inside the windShell (ld< , unld "", etc). I need to open some small'ish binary files for short term development. Obviously, NFS (which worked great with the Ethernet interface) is no longer an option, unless I reconfigure both the host and the target and get PPP or SLIP running. Given the temporary nature of the development, if I can avoid reconfiguring everything for PPP I'd like to avoid it. I thought netDriver might be a solution, but it's also based on an IP layer network (rsh, ftp). Playing around with netDriver, netDrv and netDevCreate execute successfully (after a hostAdd), but I get an errno indicating an invalid file descriptor when I try an open, i.e. fd = open("star:/home/fred/foo.txt",2) fails with ERROR. Any suggestions? I'm looking for a quick way to read bytes from a binary file and write them to some custom VME hardware. If I can get the data into my target, I'd be set. I'd rather not compile my binary data into the object module I load. I've started researching writing a wtx based tool, but would rather avoid that also. Any recommendations? Is PPP the answer? I know it can be, but would rather not invest a significant amount of time implementing PPP for 3 or 4 day development effort. Any ideas are greatly appreciated. John Murphy From owner-vxwexplo-process Thu Mar 18 00:13:01 1999 From: Rainer Stelzer Date: Thu Mar 18 00:13:05 PST 1999 Subject: Own flash driver with vxworks filesystem Status: O vxworks tornado Hi, we want to use a flash filesystem on the MBX860 board. We want to do this by providing the low-level block I/O driver routines to the integrated filesystem e.g. DOSFS. Here some questions: # because we are very familar with writing hardware-depending code we think it is quite easy to solve the problem, but we wondered that there is a product (TrueFFS from M-Systems) which costs more than 4000US$ to do the same thing. Is there a trap solving the problem like we want it to do ? (e.g. will the FAT be written to the Flash after every write access ?) # to increase the life cycle of the flash we want the driver to buffer special files in memory. The buffer will be written after a powerfail interrupt ( the final system will be a own hardware wich keeps on running for about 30 secondes after powerfail ) To distinguish between buffered an unbuffered files i would like to get the filehandle on every block access? # in a later version we want the driver to store special files compressed in flash. To do this, we need a filehandle too. And we are looking for a fast and deterministic compressing algorithm? any hint is welcome. thanks in advance. regards rainer ********************************************************************** fraunhofer institute of microelectronic circuits and systems department SAT (systems and applications) name rainer stelzer address finkenstrasse 61 47057 duisburg germany phone +0049/203/3783-238 fax +0049/203/3783-266 email rstelzer@ims.fhg.de web www.fhg.de/depts/ims-de-e.html ********************************************************************** From owner-vxwexplo-process Thu Mar 18 00:58:45 1999 From: Zoltan Kakucs Date: Thu Mar 18 00:58:49 PST 1999 Subject: Re: RPC Timeouts Status: O This is a multi-part message in MIME format. --------------4330DEABD21D5340A8A41FF8 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit > Submitted-by owner-vxwexplo-process Wed Mar 17 07:50:36 1999 > Submitted-by: "Paul Myers" > > Does anybody have any suggestions as to what might be causing an RPC timeout > when trying to connect to the Wind River system target agent on a Coldfire > core? > > I am using a VisionICE EST box to connect via EtherNet to connect to a board > running a ColdFire core. > > The Target Server indicates the back end is successfully connected to and > attempts to connect to the target agent. It then after a second or two > generates the error: > > Wind River Systems Target Server: NT/Win95 version > Attaching backend... succeeded. > Connecting to target agent... > Error: rpccore backend client Timed out failed. > > I am working in a separate development environment from the main group at > another site. The other site does not experience this problem so I don't > believe the VxWorks Kernel software is the problem since every tool, the > Kernel build and the EST box are identical. I have spent considerable time > examining my PC and the EST box configurations which I believe to be the > cause. > > Paul Myers > Questra Corporation > Hi Paul, I've posted the same message one year before, without getting any answer on it. In the meantime I found a solution. After I made a new vxWorks image including two additional options and , I've got the problem resolved. Hope this helps. I am using Motorola MVME 162, 167, 177 Boards. Regards, Zoltan Kakucs --------------4330DEABD21D5340A8A41FF8 Content-Type: text/x-vcard; charset=us-ascii; name="kakucs.vcf" Content-Transfer-Encoding: 7bit Content-Description: Card for Zoltan Kakucs Content-Disposition: attachment; filename="kakucs.vcf" begin:vcard n:Kakucs;Zoltan tel;fax:040 - 8998 4388 tel;work:040 - 8998 3283 x-mozilla-html:FALSE url:www-kryo.desy.de org:Deutsches Elektronen-Synchrotron;KRYK/MKS2 adr:;;Notkestr. 85;Hamburg;;22607;Germany version:2.1 email;internet:kakucs@herant.desy.de title:Dipl.-Ing. fn:Zoltan Kakucs end:vcard --------------4330DEABD21D5340A8A41FF8-- From owner-vxwexplo-process Thu Mar 18 01:24:05 1999 From: David Laight Date: Thu Mar 18 01:24:08 PST 1999 Subject: Re: opening files over wdbSerial Status: O > Subject: opening files over wdbSerial > Submitted-by: "Murphy, John" > > My problem is that I'm unable to open files. I can load/unload module files > inside the windShell (ld< , unld "", etc). I need to open some small'ish > binary files for short term development. > > Any ideas are greatly appreciated. > > John Murphy > How about using the data from the file to initialise a C structure in a dummy module. Load the module, define the data area as a raw file, read the file, unload the module. That might do the trick! David - I need a vxworks in here somewhere... ---------------------------------------------------------------- David Laight email: dsl@tadpole.co.uk Tadpole Technology plc phone: +44 1223 278 256 Cambridge, UK fax: +44 1223 278 201 From owner-vxwexplo-process Thu Mar 18 04:02:36 1999 From: daemon@csg.lbl.gov Date: Thu Mar 18 04:02:41 PST 1999 Subject: comp.os.vxworks newsdigest Status: O Comp.Os.Vxworks Daily Digest Thu Mar 18 04:02:32 PST 1999 Subject: 32-bit RISC Development Environment Subject: Port of Perl to Vxworks? Subject: Re: Slow TCP/IP Sockets Subject: TCP buffer corruption with SENS 1.1 Subject: Realtime clock on MPC8xx Subject: Ethernet Promiscuous mode on ADS860 Subject: Re: do_protocol_with_type() question. Subject: Re: Debugger Support For Ethernet & Kernel drivers ------------------------------------------------------- Newsgroups: comp.os.ms-windows.ce,comp.os.vxworks Subject: 32-bit RISC Development Environment Date: Wed, 17 Mar 1999 07:08:00 -0500 From: Jeff Hovis <71370.1527@CompuServe.COM> Organization: HighGround Message-ID: Has anyone heard about or used the new embedded architecture from a company Navic Systems (www.navicsys.com)? It is supposed to provide a really rapid development environment for MIPS and ARM core processors running Win CE or VxWorks. Apparently it has an embedded soft core that takes all the hassle out of low level device driver development for networking and I/O.  Apparently it has been used by Toshiba and Polaroid in some new products.  Any opinions or knowledge about the enviornment or products would be appreciated. I am working on a new product architecture that needs to be heavily network enabled (lots of I/O).  Thanks - -- Jeff Hovis --------------------------- Newsgroups: comp.lang.perl.misc,comp.os.vxworks Subject: Port of Perl to Vxworks? Date: Wed, 17 Mar 1999 10:28:09 -0500 From: bernard@ultranet.com (Bernard Farrell) Organization: Farrell Family Message-ID: Sender: bernard@hfa2c.s84f5.baynetworks.com Has anyone ported Perl (5.x or 4.x) to VxWorks? I'm intrigued that I can find a port of Tcl and Python for VxWorks, but apparently no one has done the same for Perl. Please email direct at bernard@bernardfarrell.com if you can help as I don't read these newsgroups very often. Thanks Bernard bernard@bernardfarrell.com --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Slow TCP/IP Sockets Date: Wed, 17 Mar 1999 18:35:53 GMT From: NO-SPAM-xad@NO-SPAM-home.com (Ron Dempster) Organization: @Home Network Message-ID: <36eff3ec.77307071@news> References: <7cml5a$h7p3@overload.lbl.gov> I also experienced very slow performance between Solaris 2.5.1 and 2.6. I was doing ftp from the PPC to the Sun with terrible performance. At the same time ftp initiated from the Sun and getting files from the PPC performed as expected (good performance). To fix this I used the setsockopt on the dataSocket to change the send buffer size. I am using 100 MBit ethernet and found the following results: 4*1460 Poor 5*1460 Worse 6*1460 Good (Actually the best numbers, but not always the best) 7*1460 - 11*1460 Fair 12*1460 and above Good and stable (not as good as the 6 peak, but overall as expected) Notice that the 1460 is a multiple of the ethernet payload (1500-40 for header). I also tried multiples of 1024, but got inconsistent results. So, I settled on: #define SO_BUFFERS (16 * 1460) val = SO_BUFFERS; status = setsockopt(dataSocket, SOL_SOCKET, SO_SNDBUF, &val, sizeof(val)); Hope this helps, Ron On Tue, 16 Mar 1999 14:22:23 -0700 (MST), Dennis Lewis wrote: > >We here at Lockheed Martin in Goodyear, AZ have also encountered slow transfer rates >between the PowerPC (MVME2604, vxWorks 5.3.1) and the Sun host with Solaris 2.5.1 >(SunOS 5.5.1). Steve Riffle fixed the problem as follows: > > >To get better TCP/IP performance, set the TCP_NODELAY on the socket. >This needs to be done before a connect. > >Need to include >#include > > >#ifdef VXWORKS >int one = 1; >setsockopt (Sock, IPPROTO_TCP, TCP_NODELAY, (char *) &one, sizeof(int)); >#endif > >The setsockopt in windman talks about this in more detail. > > > > >**************************************** >* Dennis B. Lewis * >* Lockheed Martin TDS-AZ * >* MS 1322 * >* 1300 S. Litchfield Rd. * >* Goodyear, AZ 85338 * >* * >* Phone: (602) 925-7593 * >* Fax : (602) 925-7873 * >* Email: lewis@master.tds-az.lmco.com * >* * >**************************************** > --------------------------- Newsgroups: comp.os.vxworks Subject: TCP buffer corruption with SENS 1.1 Date: Mon, 15 Mar 1999 15:59:07 GMT From: rbose@my-dejanews.com Organization: Deja News - The Leader in Internet Discussion Message-ID: <7cjao4$vhu$1@nnrp1.dejanews.com> Hi, We are using SENS 1.1 and VXWorks 5.3.1 and are seeing occasional corruptions in TCP packets being sent out from the target. I understand that the patch for SPR 20948 fixed a TCP buffer corruption problem and that the patch was obsoleted by the SENS 1.1 maintenance release. Has anyone else seen this type of corruption? Rahul Bose NBX Corporation 100 Brickstone Square Andover, MA 01810 - -----------== Posted via Deja News, The Discussion Network ==---------- http://www.dejanews.com/ Search, Read, Discuss, or Start Your Own --------------------------- Newsgroups: comp.os.vxworks Subject: Realtime clock on MPC8xx Date: Wed, 17 Mar 1999 21:19:06 -0600 From: Kevin Agnew Organization: Digital Highway (using Airnews.net!) Message-ID: <433F274FA1EDBCD0.B9B0C5670A0B84D0.F5D0E0B6EAAB87B5@library-proxy.airnews.net> Reply-To: agnew@acm.org I need to implement a date/time clock on the MPC8xx processor. The MBX860 and the FADS860 BSPs do not supply drivers for the RTC in the 8xx processor. Does anyone know if a device driver has been written for the RTC (to convert ticks to/from time/day/month/year)? Or can someone point me to a good (and Y2K compliant) algorithm so I can implement one myself? Thank you!! -- Kevin Agnew --------------------------- Newsgroups: comp.os.vxworks Subject: Ethernet Promiscuous mode on ADS860 Date: Mon, 15 Mar 1999 19:32:40 GMT From: dejadeep@my-dejanews.com Organization: Deja News - The Leader in Internet Discussion Message-ID: <7cjn8n$c2d$1@nnrp1.dejanews.com> Hi, Has anybody managed to get Ethernet in promiscuous mode working on the SCC of the PPC860 on the ADS860 boards? From the driver source, it appears as though frames are discarded while in promiscuous mode, unless the destination address of the frame is the address of the interface or the broadcast address. The other thing I can think of is that the REJECT pin is somehow asserted but I suspect the driver is the culprit. Thanks - -- Deep - -----------== Posted via Deja News, The Discussion Network ==---------- http://www.dejanews.com/ Search, Read, Discuss, or Start Your Own --------------------------- Newsgroups: comp.os.vxworks Subject: Re: do_protocol_with_type() question. Date: Thu, 04 Mar 1999 13:28:09 -0800 From: Stephen Macmanus Organization: Wind River Systems Message-ID: <36DEFAE9.5F76@wrs.com> References: <7bkdk6$sl1@news.or.intel.com> rvsaripa wrote: > > In the type field for do_protocol_with_type() function , when calling this > function, > do I pass the type after converting it to host order ( by calling NTOHS) or > do I pass it as it is ?. > > The reason I ask is because I have 2 ENET drivers with me for which I have > sources. > One driver appears to do the conversion before calling this function and > another does not appear to do that. The type parameter should be in host byte order. One reason your drivers might differ is that the BSD 4.3 compliant drivers call do_protocol_with_type directly while drivers using the BSD 4.4 model typically call a higher-level routine which calls ntohs() internally. > > On a different note, > how do I compile ping utility to be part of the vxworks image?. I tried > "Configure BSP", chose INCLUDE_PING from the excluded options list ( it was > part of the list ) > and then chose build to build the boot ROM ( bootrom_uncmp ) and the vxworks > image ( vxworks ). But when I try lkup "ping", it says no ping found. > > How do I enable ping?. The INCLUDE_PING definition is sufficient to link that module into the image. The lkup call might not display it if the symbol table on the target is not updated. There are various definitions to link the symbol table depending on the type of runtime image used (i.e. standalone, network, etc). Try defining INCLUDE_SYM_TBL for starters. - -- - ------------------ Stephen Macmanus #include Software Engineer Wind River Systems --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Debugger Support For Ethernet & Kernel drivers Date: Mon, 15 Mar 1999 15:26:51 -0600 From: John Organization: Rockwell International Message-ID: <36ED7B1B.352738E0@cacd.rockwell.com> References: <36E70D71.7F8DDC55@NOSPAM.metabyte.com> Ganesh, Agreed, system mode debugging (for ethernet and sysHwInit work) via serial backend has not worked for us. Also, for a Solaris host, ours is Solaris 2.6, SLIP is no longer included. Look for PPP implementation - Morning Star PPP from Progressive Systems. John Robert Armstrong wrote: > You can do this using the serial backend. If you need to debug the interrupts > you can do that in system mode, again via the serial backend. Your chances of > debugging ethernet whilst using it as the debug link are pretty slim :) > > Bob > > In article <36E70D71.7F8DDC55@NOSPAM.metabyte.com>, Ganesh Sundaram > wrote: > >Hi, > > > >I am new to the VxWorks operating system, so excuse me if this is a > >repeat/naive question. I wanted to know if Tornado supports "source > >level debugging for ethernet drivers" ? I am looking for debugging > >support similar to what SoftIce provides for Windows NT driver > >debugging. I will be developing an ethernet driver under vxWorks and > >would be very much interested to know if this is possible. > > > >On what kind of hardware platform is the source level debugging > >supported? What is the setup required to acheive this? Does the basic > >Tornado environment support source level debugging of ISR's? > > > >Any information or direction to obtain the setup required for debugging > >is greatly appreciated. > > > >Thanks in advance. > >-Ganesh > > --------------------------- End of New-News digest ********************** From owner-vxwexplo-process Thu Mar 18 05:46:14 1999 From: David Laight Date: Thu Mar 18 05:46:17 PST 1999 Subject: Re: Own flash driver with vxworks filesystem Status: O > Subject: Own flash driver with vxworks filesystem > Submitted-by: Rainer Stelzer > > Hi, > we want to use a flash filesystem on the MBX860 board. > We want to do this by providing the low-level block I/O driver routines > to the integrated filesystem e.g. DOSFS. > Here some questions: > # because we are very familar with writing hardware-depending code > we think it is quite easy to solve the problem, but we wondered that > there is a product (TrueFFS from M-Systems) which costs more > than 4000US$ to do the same thing. Is there a trap solving the > problem like we want it to do ? (e.g. will the FAT be written to the > Flash after every write access ?) I know of no trap:-) The block read/write routines and DOSFS will work (3.5k ARM object for the routines I wrote). There are some caveats. - On the chips we have erase is SLOW, about 1/2 second per sector. Multi-sector erase is no faster, parallel erase (of multiple chips) will help. - Avoid erasing flash if the write only requires changing ones to zeros. - Cache at least one flash sector (128kb in my case) - Erasing the flash prior to creating the filesystem will help - if you can afford to wait the 30 seconds or so. - The FAT is only updated on close, creating files isn't too bad (about 1 second per file for small files) The lifetime of modern flash chips is around 10^6 writes, so you have to do a lot of writes (1 a minute for two years) to exceed it. The effect of writes is to reduce the time that the data is guarantted to be valid (maybe 10 years?), it is unlikely that you need long guaranteed retention on the sectors you keep writing... I suspect that TrueFFS (it's what I would do) uses a block map so that it can allocate space in an erased sector for each write instead of having to do a read-modify-write cycle. Maybe that is worth some money. > # to increase the life cycle of the flash we want the driver to buffer > special files in memory. The buffer will be written after a > powerfail interrupt > ( the final system will be a own hardware wich keeps on running for > about 30 > secondes after powerfail ) > To distinguish between buffered an unbuffered files i would like to > get the filehandle on every block access? Why not use two filesystem? Copy from fLASH to DRAM during boot, and DRAM to flash on shutdown? Actually just caching a few flash sectors will probably have the same effect. (The cache rules are odd - write-back, but read misses don't allocate) 30 seconds isn't long enough to erase some 2Mb flash parts. > # in a later version we want the driver to store special files > compressed in flash. > To do this, we need a filehandle too. And we are looking for a fast > and deterministic compressing algorithm? You don't need a fast compression, do it while erasing the flash! > > any hint is welcome. > > thanks in advance. > > regards > > rainer > > ---------------------------------------------------------------- David Laight email: dsl@tadpole.co.uk Tadpole Technology plc phone: +44 1223 278 256 Cambridge, UK fax: +44 1223 278 201 From owner-vxwexplo-process Thu Mar 18 11:04:12 1999 From: sucro@eng.kns.com (Paul Sucro) Date: Thu Mar 18 11:04:16 PST 1999 Subject: Vx Exploder Digest: Memory allocation tracking routines. Status: O Anyone interested in memory tracking routine for VxWorks. Allow you to track all allocated memory, through malloc, calloc, valloc, memPartAlloc, etc. These routine will determine if application code overwrites any of these allocated buffers by adding padding bytes to the beginning and end of every allocated data block. I've found a number of bugs of memory being allocated for string with the following code which is in error. str2 = malloc(strlen(str)); strcpy(str2, str); This results in one byte overwrite. With only a 16 megabyte memory board this causes no problems since the first byte of allocated data block is a pointer and with 16 Megabytes is 0. But when above 16 Megabytes the first byte is non-zero and result in memPartFree errors. The code I've written will also allow you to validate all system allocated memory at anytime during running. In addition it allows many ways to display this allocated memory. Found to be very useful for memory leaks to interrogate the extra allocated blocks to determine the owner, based on contents. Tracked with the memory is size, file, line, user_id who create the file. Also part of this is freed memory tracking. This is very memory wasteful but can find the bug of code which writes to memory already freed. Thanks, Paul Sucro From owner-vxwexplo-process Thu Mar 18 11:12:59 1999 From: David Laight Date: Thu Mar 18 11:13:03 PST 1999 Subject: starting non-boot network devices using DHCP Status: O Does anyone know the required incantations to get vxworks to initialise: - An ethernet interface using DHCP - A dial up PPP link using DHCP AFTER the system has booted from ROM. I'm using the compressed bootrom loader to load a full vxworks image (exactly the one you would normally network load). At the moment it thinks it has been network loaded (although the boot string isn't in low memory) and initialises the ethernet with fixed parameters. Our production systems will not normally be directly connected to the network. We intend using dial-up PPP, dynamically establishing and dropping the connection. There is also the possibility that the ethernet card may be connected - but only for short times and not associated with machine reloads (which never happen). It would be nice to use DHCP for both the ethernet and PPP links. src/config/usrNetwork.c and config/all/bootConfig.c between them more or less explain how to get an ethernet interface up. But PPP and DHCP seem to involve a lot of magic undocumented routines. At the moment I'm a little stuffed because endFindbyName() is broken:-) David ---------------------------------------------------------------- David Laight email: dsl@tadpole.co.uk Tadpole Technology plc phone: +44 1223 278 256 Cambridge, UK fax: +44 1223 278 201 From owner-vxwexplo-process Thu Mar 18 12:53:49 1999 From: piyush agarwal Date: Thu Mar 18 12:53:52 PST 1999 Subject: C++ template class on VxWorks Status: O We are doing development on MIPS processor with VxWorks 5.3 OS. We have defined some template classes. We follow the procedure described in the programmer's guide. We have defined a template class in file.h file. Then we have a file1.cpp file in which we include file.h and define methods for this template class. We have another file file2.cpp in which we include file1.cpp and instantiate the template class. We have other files file3.cpp etc which use these. Now when we link these files we get an error in file3.cpp as "undefined reference to list:..." where list is our template class. We then linked with -r option and did nmmips on the object module. Portion of the output is given below: 000080c8 t appendElem__t4list1Z11ComponentIdR11ComponentId U appendElem__t4list1Z11ComponentIdR11ComponentId U appendElem__t4list1Z11RegListElemR11RegListElem 00003238 t appendElem__t4list1Z11RegListElemR11RegListElem 000038b0 t appendElem__t4list1Z13RemoteRegInfoR13RemoteRegInfo U appendElem__t4list1Z13RemoteRegInfoR13RemoteRegInfo U appendElem__t4list1Z15GatewayCompInfoR15GatewayCompInfo 00003574 t appendElem__t4list1Z15GatewayCompInfoR15GatewayCompInfo 00003bec t appendElem__t4list1Z9EventInfoR9EventInfo Thanks, Piyush From owner-vxwexplo-process Thu Mar 18 15:45:07 1999 From: Charlie Grames Date: Thu Mar 18 15:45:10 PST 1999 Subject: System Clock Status: O VxWorks 5.3.1 MVME2700 1.1/4 I have seen a lot of discussion about running the system clock at high rates and the negative effects it can have on system performance. I am a little concerned about the blanket "don't do it" advice that has been given in this forum, and I thought I'd add my two cents worth. Whether or not you can afford to run the system clock at a high rate (e.g., 1000 Hz) depends a lot on your processor. If you are using older 680x0-based boards, I would agree that high system clock rates are best avoided, because you will spend a significant amount of processor time handling interrupts. If, however, you are running on a newer, faster architecure such as the PowerPC, the amount of time spent in interrupt handling with a 1000 Hz system clock is relatively low (I measured about 1%). There are good reasons for running the system clock faster. The most important reason for us was being able to provide high timeout resolution for the select() system call. As far as I know, uping the system clock rate is the ONLY way to improve select() timeout resolution. Anyway, I agree that caution should be used when increasing the system clock rate, but I believe users should determine the impact on an individual basis rather than avoiding it because it's "just a bad idea." Charlie Grames The Boeing Company (314) 233-1956 Charles.R.Grames@boeing.com From owner-vxwexplo-process Thu Mar 18 15:51:16 1999 From: Charlie Grames Date: Thu Mar 18 15:51:21 PST 1999 Subject: File I/O Status: O Joe, Did you ever get a useful answer to your recent post? I have been out of circulation lately and just read it. The important thing to know about VxWorks is that there are two different network file system drivers. The default one is netDrv, which uses ftp or rsh (depending on how you configure your target). The other is nfsDrv, which uses nfs. Both are discussed in the Reference Manual. Based on your results, I am guessing that you are using the default netDrv via ftp. The way this works is that when you open a file, the entire file is ftp'ed into memory. All writes and reads occur to and from memory. When you close the file, it is ftp'ed back to the remote system. You will always get the "The system cannot find the file specified" when you open a file that doesn't exist, even if it isn't really an error. The real question is whether you get a valid file descriptor back from open() or -1. If you don't get -1, the file was opened, so you can write your data and close the file. If you do get -1, make sure that the target's view of the file system is valid (i.e., use cd() programatically or @cd from WindSh) to change to the proper directory before opening the file. Note, also, that many ftp servers (notably Microsoft's Peer Web Services and Internet Information Server) do NOT enable write access by default. HTH Charlie Grames The Boeing Company (314) 233-1956 Charles.R.Grames@boeing.com From owner-vxwexplo-process Fri Mar 19 00:22:45 1999 From: stevens Date: Fri Mar 19 00:22:49 PST 1999 Subject: missing symbols (CORBA with vxworks) Status: O To the VxWorks World ! We use a Motorola 68332 (4MB RAM) with Tornado 1.01 and Visibroker's CORBA 3.2.1. From Visigenic we got four objects liborb.o, osagent.o, liborb_munched.o and osagent_munched.o. We tried to link the non-munched to the vxworks objekt (makefile). After booting and loading some excamples from Visigenic a lot of corba specific symbols are missing. The missing symbols can be found in the browser's symbol list. Is there someone out there who had a similar problem (It's not CORBA specific) ? Who is using Visigenic's CORBA 3.2.1 implementation and are the examples OK ? Thanks, Torsten -- ************************************************************************** fraunhofer institute of microelectronic circuits and systems department sat name torsten stevens mailto:stevens@ims.fhg.de address finkenstrasse 61 phone +0049/203/3783-238 47057 duisburg fax +0049/203/3783-266 germany http://www.fhg.de/depts/ims-du-e.html ************************************************************************** From owner-vxwexplo-process Fri Mar 19 04:02:26 1999 From: daemon@csg.lbl.gov Date: Fri Mar 19 04:02:29 PST 1999 Subject: comp.os.vxworks newsdigest Status: O Comp.Os.Vxworks Daily Digest Fri Mar 19 04:02:21 PST 1999 Subject: Re: Performance impact of C++ Subject: serial port driver Subject: VxWorks SENS and IEEE803.3 Ethernet Frames Subject: VxWorks Job Opportunity, Sydney, Australia Subject: Re: C++ wrapper for system calls ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: Performance impact of C++ Date: Mon, 08 Mar 1999 21:12:34 -0500 From: Mark Clayton Organization: Shore.Net/Eco Software, Inc; (info@shore.net) Message-ID: <36E48392.DCE2B62F@shore.net> References: Bill Pringlemeir wrote: > Many CPUs support the Green Hills compiler. Green Hills provide PJ Plaugers I would recommend you NOT use green hills unless you're working on a PowerPC. I use gh on an SH-1 and it's terrible. Tech support is no help as well. Mark --------------------------- Newsgroups: comp.os.vxworks Subject: serial port driver Date: Thu, 18 Mar 1999 13:42:57 +0100 From: Entwicklung ELAU Organization: Citylink Internet Service Center Message-ID: <36F0F4CF.14D7B2D2@elau.de> How to set up a serial port driver (/tyCo/0) for use with a modem and the ppp-Protocol as a dial-in server ? --------------------------- Newsgroups: comp.os.vxworks Subject: VxWorks SENS and IEEE803.3 Ethernet Frames Date: Thu, 18 Mar 1999 18:31:45 GMT From: "Peter Smith" Message-ID: Dear All, I have a number of questions about SENS and IEEE802.3/Ethernet Frames. 1) What type of frame does the SENS TCP/IP stack transmit by default; Ethernet or IEEE802.3 ? 2) Can the SENS TCP/IP stack be forced to transmit IEEE802.3 frames, if so how is this done ? 3) What value does the SENS TCP/IP stack use for the "protocol" argument when it calls muxBind() ? Thanks in anticipation of lots of interersting answers Peter Smith ALSTOM T&D P&C Ltd --------------------------- Newsgroups: comp.os.vxworks Subject: VxWorks Job Opportunity, Sydney, Australia Date: Tue, 16 Mar 1999 03:39:19 GMT From: nick@interpro.com.au Organization: Deja News - The Leader in Internet Discussion Message-ID: <7ckjp6$5vh$1@nnrp1.dejanews.com> I urgently need two Technical Software Engineers and am offering the chance to work on Australia's most exciting defence project (due to security clearance requirements, you must be an Australian CItizen or Permanent Resident) You will have 1) Experience in developing low level software on custom hardware 2) Knowledge of VME hardware 3) Preferably experience of VxWorks operating system 4) Background as a software engineer using eg C/C++ or ADA This position can be either permanent or contract Best Regards, Nick Blakesley Senior Consultant Interpro CRI Pty Ltd Tel: +61 (0)2 9283 3555 Fax: +61 (0)2 9283 3634 email: nick@interpro.com.au Please note that by using any information in, or attached to, this e-mail, you agree to be bound by the standard terms and conditions of InterPro CRI Pty Limited. You agree not to employ or arrange employment for any candidates supplied in this document without first entering into a contractual agreement with InterPro CRI Pty Limited. You further agree not to divulge any information contained in this document to any persons or entities without the express permission of InterPro CRI Pty Limited - -----------== Posted via Deja News, The Discussion Network ==---------- http://www.dejanews.com/ Search, Read, Discuss, or Start Your Own --------------------------- Newsgroups: comp.os.vxworks Subject: Re: C++ wrapper for system calls Date: Thu, 18 Mar 1999 18:43:24 -0500 From: Tim Sohacki Organization: Nortel (Northern Telecom) Message-ID: <36F18F9C.AE7EC8EF@nortelnetworks.com> References: <36F185EE.50B3173@virgin.net> Darren, Responses embedded below. Darren Long wrote: > > I'm wondering if the wrapper classes that are apparently supplied with > the OS fully implement C++ exception handling, i.e. throwing exceptions > in response to errors detected by the underlying C code, and with > exception specifications in the member function declarations. Not in vxWorks 5.3.1 (current release), but some (complete?) support in the Tornado II version, available May (?). See http://www.wrs.com for more information. > Under VxWorks, do all tasks exist in a single data space i.e. all tasks > could see a (shock-horror) global variable? Yes. And all tasks run in supervisor mode (does that apply to all targets? The PPC and 68K, for sure. Right?) > Does each task have it's own stack? Yes. And most target processors have a single but separate interrupt stack. > Could an exception thrown by a task be caught by the entity > that spawned it, or should they definitely be caught within the > thread/task itself? Exceptions ... the C++ exceptions probably need to be addressed by the Tornado II material. (Any beta testers out there?) In general, when an exception (divide by zero, for example) is generated, vxWorks handles it by logging some relevant info to standard output and then suspending the task that generated the exception. (Exceptions at interrupt level?) VxWorks provides an API to add a hook called during this processing, so that you can do your own cleanup. In this case, you'd be handling exceptions outside your task. You can use ANSI setjmp, longjmp, and signal handlers too. If you do this, you'll be handling the exception from within your task. Both may be appropriate in the same system for different tasks or different exceptions. > Does the round-robin scheduling mode of VxWorks work on a priority > basis? Pre-emptive priority based scheduling. Higher priorities correspond to lower absolute numbers (i.e., 0 is the highest, 254 is the lowest). Tasks at the same priority schedule unfairly by default. A kernel call allows round-robin at the same priority with a number of system clock ticks per slice. For example, if 5 tasks at priority 100 are ready, the first will run until it pends. With time slicing enabled, it will run until 1, 2, or x ticks pass (or it pends), at which time the next task will be scheduled. Of course a higher priority (with a lower absolute number) can preempt the priority 100 tasks at any time. > > How does pre-emptive scheduling work in conjunction with clock ticks > and context switching? > > Is there an ageing mechanism akin to that in OS9? OS what? I'll leave that to someone else. Hope that helps, Tim - -- Tim Sohacki, Nortel Networks email: sohacki@nortelnetworks.com --------------------------- End of New-News digest ********************** From owner-vxwexplo-process Fri Mar 19 04:50:33 1999 From: "Wu, Huang" Date: Fri Mar 19 04:50:36 PST 1999 Subject: how to use excHookAdd() Status: O Hello vxWorks world, Does anyone know how to use excHookAdd() in a task? I have written a test code, but the exception handler is not called when an exception occurs. Is there anyone who likes to share their successful story with me? Thanks in advance Huang #include "vxWorks.h" #include "taskLib.h" #include "stdio.h" #include "arch/mc68k/archMc68k.h" #include "arch/mc68k/regsMc68k.h" #include "arch/mc68k/esfMc68k.h" #include "semLib.h" #include "excLib.h" /* Globals */ SEM_ID syncSem; /* Forward declarations */ void myHandler (int, int, ESF68K_BA *); void testExc (void); void fuBar (void); /* Function: myHandler * Purpose : User-defined handler executes after a hardware exception oc curs. * Returns : N/A */ void myHandler (int task, int vector, ESF68K_BA * pEsf) { printf ("\n\n"); printf ("USER-DEFINED HANDLER...\n"); printf (" Offending task ID: 0x%x\n", task); printf (" Exception vector: %d\n", vector); printf (" Status reg: %d\n", pEsf->sr); semGive (syncSem); /* activate fuBar() */ } /* function: testExc * Purpose : Test the execution of a user-defined exception handler insta lled * vio excHookAdd(). This routine will generate an exception wh ich, * in turn, will invoke the user-defined exception handler. The * user-defined handler will give a semaphore which unblocks an other * task - fuBar(). fuBar() will then generate another exceptio n. * Returns : N/A */ void testExc (void) { unsigned long goomBah; /* add the exception hook routine in this task */ excHookAdd ((FUNCPTR) myHandler); /* create semaphore */ syncSem = semBCreate (SEM_Q_FIFO, SEM_EMPTY); /* spawn a task which will be called after the hook activates */ taskSpawn ("tFuBar", 100, 0, 2000, (FUNCPTR)fuBar, 0,0,0,0,0,0,0,0,0, 0); /* this is a NON-EXISTANT ADDR*/ goomBah = *(unsigned long *)0xfffff98; } /* function: fuBar * Purpose : Pend on a semaphore given by a user-defined exception handle r. * After getting the semaphore, force another exception. * Returns : N/A */ void fuBar (void) { unsigned long goomBah = 0x0000ffff; /* wait for exception event to occur */ semTake (syncSem, WAIT_FOREVER); printf ("\nfuBar() got the semaphore\n"); /* a brief delay to give the system a chance to recover before */ /* we generate the next exception. */ taskDelay (10); /* assign to a NON-EXISTANT ADDRESS */ *(unsigned long *)0xfffff98 = goomBah; } I From owner-vxwexplo-process Fri Mar 19 06:59:47 1999 From: jmk@alum.mit.edu (Joel Katz) Date: Fri Mar 19 06:59:51 PST 1999 Subject: Re: System Clock Status: O >>----snip >VxWorks 5.3.1 >MVME2700 1.1/4 > >I have seen a lot of discussion about running the system clock at high rates >and the negative effects it can have on system performance. I am a little >concerned about the blanket "don't do it" advice that has been given in this >forum, and I thought I'd add my two cents worth. > > >There are good reasons for running the system clock faster. The most >important reason for us was being able to provide high timeout resolution >for the select() system call. As far as I know, uping the system clock rate >is the ONLY way to improve select() timeout resolution. > >Anyway, I agree that caution should be used when increasing the system clock >rate, but I believe users should determine the impact on an individual basis >rather than avoiding it because it's "just a bad idea." I just wanted to second Charlie's thoughts on the system clock issue. I'd be willing to bet money that the 60 hz rate is based on using a processor along the lines of a 30 Mhz 68030. I did some quick, non-empirical experiments running our board with a 1Khz system clock rate and detected no adverse effects on normal functioning. Joel Katz, Principal Engineer --- (603)-577-5895 Acuity Imaging, 9 Townsend West, Nashua, NH 03063 --- (603)-577-5965 (FAX) visit our Web Site at >>> www.acuityimaging.com <<< From owner-vxwexplo-process Fri Mar 19 07:11:11 1999 From: "Thomas Keith Buchanan" Date: Fri Mar 19 07:11:14 PST 1999 Subject: Re: C++ wrapper for system calls Status: O > > How does pre-emptive scheduling work in conjunction with clock ticks > > and context switching? Most preemptive operating systems operate in a similar manner. A periodic time interrupt increments the "clock ticks" and calls the scheduler. A context switch only takes place when the scheduling algorithm determines that the current context must changed. VxWorks is not different in this respect. VxWorks always uses a priority-based preemptive scheduler. In addition, it implements a simple priority inheritance protocol algorithm and can do fair round-robin scheduling within the SAME priority level. > > Is there an ageing mechanism akin to that in OS9? No, that is a sign of OS-9's Unix-like heritage... not so useful in hard real-time apps adios ------------------------------------ /\ Thomas Keith Buchanan /**\ Principal Engineer /****\ SPARTA, Inc. /****/.. 205 East IH 30 /****/.... Rockwall, TX 75087 .\****\__... ...\**\ /.... Voice: 972 772 4487 ....\**\/\.... FAX: 972 722 3201 ....\****\.. PCS: 972 672 9657 ....\****/ WWW: www.sparta.com .../***/ ./***/ SPARTA \**/ ====== \/ Pride In Performance From owner-vxwexplo-process Fri Mar 19 07:37:13 1999 From: Andreas Wolf Date: Fri Mar 19 07:37:16 PST 1999 Subject: RE: missing symbols (CORBA with vxworks) Status: O Hi Torsten, the vxWorks Users Group Exploder writes: > We use a Motorola 68332 (4MB RAM) with Tornado 1.01 and Visibroker's CORBA 3.2.1. > >From Visigenic we got four objects liborb.o, osagent.o, liborb_munched.o and osagent_munched.o. > We tried to link the non-munched to the vxworks objekt (makefile). After booting and loading some excamples from > Visigenic > a lot of corba specific symbols are missing. The missing symbols can be found in the browser's symbol list. > > Is there someone out there who had a similar problem (It's not CORBA specific) ? > Who is using Visigenic's CORBA 3.2.1 implementation and are the examples OK ? We use the same ORB on MC68040 and PPC604 machines and it works fine for us. However I did play only once with the demos... Here is what we do: 1) We link the liborb.o to the VxWorks kernel, so it gets munched during the creation of the boot file. Remember to follow the description in the installation manual, how to get the code to be linked correctly. 2) We compile the code (demo or real) and link all local code excluding 'liborb.o'. Then we munch it. 3) We then download it into the target, using the shell (target based in our case). As I said, this works fine for us. I think, you do the same steps. Are you using the target shell or the host shell? Only the host shell is setup to manage symbols correctly, when you use the 'out-of-the-box' VxWorks configuration. You need to get the symbols into the target, if you want to use the target shell. I can check the examples once more, but I doubt that the problem is there... Hope this helps, Andreas =:-) -- ---------------------------------------------------------------------- ** Andreas H. Wolf ** Cornell University ** Wilson Lab ** Ithaca, NY 14853 ** Phn: (607) 255-4882 ** E-Mail: aw@mail.lns.cornell.edu ** WWW: http://www.lns.cornell.edu/~aw ** ** 'Help Gromit! The techno trousers are going haywire!!' - Wallace ** ** 'This is Linux Country. On a quiet night, you can hear Windows reboot.' ---------------------------------------------------------------------- From owner-vxwexplo-process Fri Mar 19 09:22:41 1999 From: "Thomas Keith Buchanan" Date: Fri Mar 19 09:22:45 PST 1999 Subject: RE: System Clock Status: O > >>----snip > >VxWorks 5.3.1 > >MVME2700 1.1/4 > > > >I have seen a lot of discussion about running the system clock > at high rates > >and the negative effects it can have on system performance. I > am a little > >concerned about the blanket "don't do it" advice that has been > given in this > >forum, and I thought I'd add my two cents worth. > > > > > >There are good reasons for running the system clock faster. The most > >important reason for us was being able to provide high timeout resolution > >for the select() system call. As far as I know, uping the > system clock rate > >is the ONLY way to improve select() timeout resolution. > > > >Anyway, I agree that caution should be used when increasing the > system clock > >rate, but I believe users should determine the impact on an > individual basis > >rather than avoiding it because it's "just a bad idea." > > > I just wanted to second Charlie's thoughts on the system clock > issue. I'd be willing to bet money that the 60 hz rate is based > on using a processor along the lines of a 30 Mhz 68030. I did > some quick, non-empirical experiments running our board with a > 1Khz system clock rate and detected no adverse effects on normal > functioning. 60 hz is a historical choice dating back to the days of the first PDPs which used the zero crossing of the AC power signal to trigger the timer interrupts. It is still kinda silly to increase the system clock rate by an order of magnitude just to increase the resolution of the tick counter. That's what timers are for. That's what the sysAuxClock support code is for. How many consider the addition of a 1Khz interrupt source of little or no concern? Not me... Sure, a faster processor can handle more interrupts and more calls to the scheduler but nothing is free. The kernel is spending more time with interrupts disabled and more time in an interrupt context. The application gains very little (maybe select() is an exception) from this behaviour. Most context changes in a good design are prompted by calls to the OS, not by preemption. In summary, I still say is a questionable design decision. adios ------------------------------------ /\ Thomas Keith Buchanan /**\ Principal Engineer /****\ SPARTA, Inc. /****/.. 205 East IH 30 /****/.... Rockwall, TX 75087 .\****\__... ...\**\ /.... Voice: 972 772 4487 ....\**\/\.... FAX: 972 722 3201 ....\****\.. PCS: 972 672 9657 ....\****/ WWW: www.sparta.com .../***/ ./***/ SPARTA \**/ ====== \/ Pride In Performance From owner-vxwexplo-process Fri Mar 19 13:41:50 1999 From: Charlie Grames Date: Fri Mar 19 13:41:53 PST 1999 Subject: RE: System Clock -Reply Status: O > Submitted-by: "Thomas Keith Buchanan" > Sure, a faster processor can handle more interrupts and more calls to the > scheduler but nothing is free. The kernel is spending more time with > interrupts disabled and more time in an interrupt context. The > application gains very little (maybe select() is an exception) from this > behaviour. Most context changes in a good design are prompted by calls to > the OS, not by preemption. In summary, I still say is a questionable > design decision. I agree that it is "questionable," in the sense that it is something that should be seriously considered. When you are wanting high frequency for something that can be easily driven by an alternate timer, use it. When you want high resolution for a service provided only by the system, don't be afraid to at least consider the possibility of increasing the system clock rate. If select() could be disconnected from the system clock, I would do so. Unfortunately, that is not an option unless you buy VxWorks source. Raising the system clock rate to 1000 Hz solves our problem and does not have adverse effects on our application. I cannot guarantee that would be true for all users. As I stated originally, users need to make this determination for themselves. Charlie Grames The Boeing Company (314) 233-1956 Charles.R.Grames@boeing.com From owner-vxwexplo-process Sat Mar 20 04:00:32 1999 From: daemon@csg.lbl.gov Date: Sat Mar 20 04:00:36 PST 1999 Subject: comp.os.vxworks newsdigest Status: O Comp.Os.Vxworks Daily Digest Sat Mar 20 04:00:29 PST 1999 Subject: Re: compression source code Subject: PPPOE Subject: Re: WindFoundationClass (WFC) Subject: Re: Wind River' Tornado 1.0.1 for Windows NT is bugged Subject: AZ:Tempe - VxWorks OS/Driver Contract - Agency Subject: Re: purposefully crashing VxWorks Subject: Re: VxWorks Job Opportunity, Sydney, Australia Subject: How to get system time and date with ms resolution ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: compression source code Date: Fri, 19 Mar 1999 14:35:30 +0000 From: "r.pettipher" Organization: UUNET WorldCom server (post doesn't reflect views of UUNET WorldCom) Message-ID: <36F260B1.927D9B18@dial.pipex.com> References: <79s8hi$3r2$1@news.netway.at> <36cb0ee6.907923124@news.pacificnet.net> <7a44pa$1rja$1@node17.cwnet.frontiernet.net> <36d83522.983308052@news.pacificnet.net> Obviously the question I want to ask has recently been answered, but unfortunately I missed the start of this thread, so please bear with me. I am looking for a way of compressing files from within an vxworks application, so is there source code out there for gzip or something similar, or alternatively is there a version of gzip that I can execute on my vxworks machine from my application? And is there a better way of finding old messages other than asking the questions again? Thanks Rodney Pettipher --------------------------- Newsgroups: comp.os.vxworks Subject: PPPOE Date: Fri, 19 Mar 1999 12:10:44 -0500 From: "Bob Beers" Organization: CyberGate, Inc. Message-ID: <7cu0f7$2p34$1@news.gate.net> Hello all, It's on my schedule to start looking into this (PPP over Ethernet), so I am asking for some pointers to necessary reading, experiences (successful or not), or opinions on the feasability of the idea that some of you may be willing to share on the subject. My target is x86, but experiences from any target environment will be appreciated. Thanks, Bob Beers bbeers(_at_)ipaxs(_dot_)com --------------------------- Newsgroups: comp.os.vxworks Subject: Re: WindFoundationClass (WFC) Date: 6 Mar 1999 18:23:20 GMT From: fnordz@aol.com (Fnordz) Organization: AOL http://www.aol.com Message-ID: <19990306132320.07559.00004201@ng-fa1.aol.com> References: <7arfne$d5p$1@fleetstreet.Austria.EU.net> Hello Markus, The document you are referring to is actually the library documentation for AT&T C++ 3.0. It seems that Wind River supplies this to WFC/VxWorks users because the iostreams implementation (and one or two other C++ libraries in vxWorks) are actually from the AT&T C++ 3.0 libs, and perhaps also because Wind River used to claim support for the AT&T cfront C++ translator at one time. Unfortunately, the document you are referring to is also confusing because Wind River does _not_ ship all of the classes and libraries documented in that manual. In fact, given the VxWorks RTOS, it would likely be absurd to ship the Task Library you are reading about. So, read the "C++ Language System" with the understanding that ther are only a few sections which actually apply to the C++ libraries you will find in Tornado 1.0.1/VxWorks 5.3.1. As an aside, the "C++ Language System" document is fully useless for Tornado 2.0 / VxWorks 5.4 because Wind River is not shipping any remnant of AT&T C++ 3.0 in that package; the iostreams implementation has been replaced with a more current cygnus/GNU version. Regards, paul --------------------------- Newsgroups: comp.arch.embedded,comp.os.vxworks Subject: Re: Wind River' Tornado 1.0.1 for Windows NT is bugged Date: Sat, 6 Mar 1999 20:07:16 -0000 From: "Darren J Longhorn" Organization: Customer of Planet Online Message-ID: <7bs1vp$52u$1@newsreader1.core.theplanet.net> References: <7bn25a$jeu$1@nslave1.tin.it> <7sC$fQAoq+32Ew9$@noco.demon.co.uk> Hmm, I don't see that on mine. Where is it exactly? Mike Davies wrote in message <7sC$fQAoq+32Ew9$@noco.demon.co.uk>... >Make sure that you have the patch installed : when you >start the Tornado environment pull down the help dialog >box and look at the version details. There should be an >entry there for something like "cumulative patch x86". >Tornado is not really usable without it so be warned. > > > >-- >Mike Davies --------------------------- Newsgroups: comp.os.vxworks,comp.realtime,misc.jobs.contract Subject: AZ:Tempe - VxWorks OS/Driver Contract - Agency Date: 17 Mar 1999 13:07:53 GMT From: Tate Consulting Organization: Tate Consulting, Inc. Message-ID: <7co9f9$co7$7@remarQ.com> Fortune 100 Company needs contractor to work on VxWorks Drivers and OS tuning. Refer to Job Code: JC-A10078-01. Please reference the JOB CODE listed above in any correspondence to us. Send your resume in ASCII or MSWord format to: Email: hotjobs@tateinc.com Fax: (561) 852-2385 Snail: Tate Consulting, Inc. 10647 Avenida Santa Ana Boca Raton, FL 33498 Please visit our Hi-Tech Job Database at http://www.tateinc.com Tate Consulting is a Computer Consulting and Placement Company in business since 1991. Andrew Tate, President/Owner Tate Consulting, Inc. atate@tateinc.com (561) 852-TATE (8283) We are an EOE. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: purposefully crashing VxWorks Date: Wed, 17 Mar 1999 12:50:21 GMT From: "Don Bockenfeld" Organization: FlashNet Communications, http://www.flash.net Message-ID: References: <7ceuin$k3t$1@goldenapple.srv.cs.cmu.edu> <36EB0F18.6A70@nospam.aracnet.com> <36ED3BB7.41C6@utsci.com> Mike Stimpson wrote in message <36ED3BB7.41C6@utsci.com>... >Perhaps a cleaner way to do it is disable interrupts and then execute >the HALT instruction. Depends on if you need to prevent NMI from reviving your system. I don't have the details in front of me; but I convinced myself a couple of years ago that the only definitive way for a program to shut down the 68040 was by provoking a double bus fault. Speaking of unclean ways, the technique I use is pretty ugly. Notice that the double bus fault state can be entered by a double address fault too. I load the appropriate stack pointer (supervisor or master) with an odd number and execute a trap instruction. This approach can be used by a library routine that has no knowledge of how a particular application has loaded the exception vector table. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: VxWorks Job Opportunity, Sydney, Australia Date: 17 Mar 1999 19:49:15 GMT From: mfb@mbunix.mitre.org (Michael F Brenner) Organization: none Message-ID: <7cp0vr$7fg@top.mitre.org> References: <7ckjp6$5vh$1@nnrp1.dejanews.com> Nick > you agree to be bound by the standard terms and conditions of InterPro CRI Pty Limited. You agree not to employ or arrange employment for any candidates supplied in this document without first entering into a contractual agreement with InterPro CRI Pty Limited. You further agree not to divulge any information contained in this document to any persons or entities without the express permission of InterPro CRI Pty Limited Nick, I do not agree to be bound by those terms or conditions, which are not even listed in your message. Please understand that posting a message on the Net can never obligate anyone else to anything, you can only obligate yourself. Good luck in finding what your message was seeking, though! --------------------------- Newsgroups: comp.os.vxworks Subject: How to get system time and date with ms resolution Date: Thu, 11 Mar 1999 07:14:43 +0100 From: "Torsten Trzeciak" Organization: mannesmann.de Message-ID: <7c7n0s$l2b@as1fw.mannesmann.de> Hello, I like to get the system time with a ms resolution. Is there a kind of getTime function? Torsten --------------------------- End of New-News digest ********************** From owner-vxwexplo-process Sun Mar 21 04:01:27 1999 From: daemon@csg.lbl.gov Date: Sun Mar 21 04:01:31 PST 1999 Subject: comp.os.vxworks newsdigest Status: O Comp.Os.Vxworks Daily Digest Sun Mar 21 04:01:21 PST 1999 Subject: problem with vxworks - tcp/ip stack Subject: Task Scheduling using VxWorks Subject: Re: MAC query Subject: nfs problem...can't ls and ld Subject: Re: purposefully crashing VxWorks ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: problem with vxworks - tcp/ip stack Date: Thu, 18 Mar 1999 11:10:11 +0200 From: "RonenO" Organization: Unspecified Organization Message-ID: <7cqg4p$eu4$1@news.ibm.net.il> I am using the new tcp/ip stack - SENS 1.1 with the END driver on PPC860. It seems that the tcp/ip stack stops working when a socket is full. I open a UDP socket & transmiting packets to this socket. In a few seconds, the target loses contact, & does not answer to pings anymore. there is no printings in the shell/debugger. there is no recovery & a reboot is needed. Does anyone having the same problem?? Or know a possible solution for it?? Ronen. --------------------------- Newsgroups: comp.os.vxworks Subject: Task Scheduling using VxWorks Date: 20 Mar 1999 17:55:38 GMT From: seshaga@email.uah.edu (Aravind T. Seshagiri) Organization: The University of Alabama in Huntsville Message-ID: <7d0neq$k8b$1@info2.uah.edu> We are looking for information on Task scheduling using VxWorks. If any of you can provide us help regarding the algorithms that would help us in coding a rate monotonic priority based preemptive scheduling module, we would greatly appreciate your help. Thanks - you can email responses to seshaga@email.uah.edu --------------------------- Newsgroups: comp.os.vxworks Subject: Re: MAC query Date: Sun, 14 Mar 1999 17:04:36 GMT From: reperjesi@my-dejanews.com Organization: Deja News - The Leader in Internet Discussion Message-ID: <7cgq6v$slm$1@nnrp1.dejanews.com> References: <7ccgf0$gl7$1@nnrp1.dejanews.com> <5914149BDA7E7622.2718831C80A97E46.7849779FAA73C516@library-proxy.airnews.net> In article <5914149BDA7E7622.2718831C80A97E46.7849779FAA73C516@library- proxy.airnews.net>, petekz@cyberramp.net (Pete Kockritz) wrote: > In article <7ccgf0$gl7$1@nnrp1.dejanews.com>, reperjesi@my-dejanews.com wrote: > > > Hi vxworkers, > > > > > > I'd like to get the ethernet controller's Ethernet Address > > (6 byte unique ID). Any hints? > > Assuming your BSP supports it > > STATUS sysEnetAddrGet > ( > UINT8 unit, > UINT8 * addr > ) > > will return it. > > Regards, > Pete > -- > +-----------------------------------------------------------------+ > | Pete Kockritz mailto:petekz@cyberramp.net | > | This space for rent. | > +-----------------------------------------------------------------+ > The problem I had with sysEnetAddrGet is that each time I run it the addr increases by 1. The first time I run sysEnetAddrGet the address the 2nd parameter points to is x+1, the second time I run it it's x+2 and so on (x = real MAC). By the way, is this function - sysEnetAddrGet - documented anywhere, becuase I found it digging into the source I have. Thanks, Richard - -----------== Posted via Deja News, The Discussion Network ==---------- http://www.dejanews.com/ Search, Read, Discuss, or Start Your Own --------------------------- Newsgroups: comp.os.vxworks Subject: nfs problem...can't ls and ld Date: Mon, 15 Mar 1999 11:44:12 +0000 From: lihb Organization: Computing Center, Academia Sinica Message-ID: <36ECF28C.178A9AF6@phys.sinica.edu.tw> i'm using vxWorks(MVEM167) to mount an UNIX(SGI) directory. on the UNIX, i had add the IP of target into /etc/host.equiv and /etc/exports, chmod 777 the directory and every sub-directories and every files and exportfs it, on the vxWorks, i had hostAdd the UNIX and nfsMount the directory. everythings seem ok, nfsDevShow have the correct things on it. i cd to that directory by cd "/directory". when i ls or ld < file, it seem halt untill i ctrl-c. and if i cd "host:/directory", both ls and ld < file work. what is the possible mistake i made? please help. thanks......... lihb@phys.sinica.edu.tw --------------------------- Newsgroups: comp.os.vxworks Subject: Re: purposefully crashing VxWorks Date: Mon, 15 Mar 1999 12:36:06 GMT From: "Don Bockenfeld" Organization: FlashNet Communications, http://www.flash.net Message-ID: References: <7ceuin$k3t$1@goldenapple.srv.cs.cmu.edu> <36EB0F18.6A70@nospam.aracnet.com> I don't know about Michael, but we have a customer requirement that our system "halt" under certain error conditions. We accomplish this by coercing a double-bus fault, as suggested by Scott. Scott Johnson wrote in message <36EB0F18.6A70@nospam.aracnet.com>... >Michael J. Abowd wrote: >> >> Hi, >> >> I am working on a project that needs to consistently create a type of > OS failures in VxWorks running on a Motorola 68040. Create a process > that cause the OS to hang altogether. At this point, the machine >> would need to rebooted. > > Just curious--what are you doing this for? The fact that vxWorks, > a striclty supervisor-mode OS with no memory protection whatsoever, > is easy to crash has already been proven... --------------------------- End of New-News digest ********************** From owner-vxwexplo-process Sun Mar 21 07:46:06 1999 From: Eric.vanderSanden@EMT-EHV.BE.PHILIPS.COM Date: Sun Mar 21 07:46:10 PST 1999 Subject: missing symbols (CORBA with vxworks) Status: O Status Distribution March 19, 1999 10:43:57 The message regarding "missing symbols (CORBA with vxworks)" sent on March 19, 1999 10:43:57 was sent by Status Recipient Type To Native Name vxwexplo@lbl.gov Foreign Native Name vxwexplo@lbl.gov\n\n\nSMTP Recipients Status Reporters Type From Name Domain NOTES Native Name CN=Eric van der Sanden/OU=EHV/OU=EMT/OU=BE/O=PHILIPS@PHILIPS07 Foreign Native Name CN=Eric van der Sanden/OU=EHV/OU=EMT/OU=BE/O=PHILIPS\nPHILIPS07\n\n Organization PHILIPS Org Unit 1 BE Org Unit 2 EMT Org Unit 3 EHV Last Name Sanden First Name Eric Status 769 Explanation Invalid recipient X.400 Status 769 Explanation Delivery time expired From owner-vxwexplo-process Sun Mar 21 07:49:11 1999 From: Eric.vanderSanden@EMT-EHV.BE.PHILIPS.COM Date: Sun Mar 21 07:49:14 PST 1999 Subject: comp.os.vxworks newsdigest Status: O Status Distribution March 19, 1999 17:49:46 The message regarding "comp.os.vxworks newsdigest" sent on March 19, 1999 17:49:46 was sent by Status Recipient Type To Native Name vxwexplo@lbl.gov Foreign Native Name vxwexplo@lbl.gov\n\n\nSMTP Recipients Status Reporters Type From Name Domain NOTES Native Name CN=Eric van der Sanden/OU=EHV/OU=EMT/OU=BE/O=PHILIPS@PHILIPS07 Foreign Native Name CN=Eric van der Sanden/OU=EHV/OU=EMT/OU=BE/O=PHILIPS\nPHILIPS07\n\n Organization PHILIPS Org Unit 1 BE Org Unit 2 EMT Org Unit 3 EHV Last Name Sanden First Name Eric Status 769 Explanation Invalid recipient X.400 Status 769 Explanation Delivery time expired From owner-vxwexplo-process Sun Mar 21 08:05:36 1999 From: Eric.vanderSanden@EMT-EHV.BE.PHILIPS.COM Date: Sun Mar 21 08:05:40 PST 1999 Subject: comp.os.vxworks newsdigest Status: O Status Distribution March 20, 1999 13:46:29 The message regarding "comp.os.vxworks newsdigest" sent on March 20, 1999 13:46:29 was sent by Status Recipient Type To Native Name vxwexplo@lbl.gov Foreign Native Name vxwexplo@lbl.gov\n\n\nSMTP Recipients Status Reporters Type From Name Domain NOTES Native Name CN=Eric van der Sanden/OU=EHV/OU=EMT/OU=BE/O=PHILIPS@PHILIPS07 Foreign Native Name CN=Eric van der Sanden/OU=EHV/OU=EMT/OU=BE/O=PHILIPS\nPHILIPS07\n\n Organization PHILIPS Org Unit 1 BE Org Unit 2 EMT Org Unit 3 EHV Last Name Sanden First Name Eric Status 769 Explanation Invalid recipient X.400 Status 769 Explanation Delivery time expired From owner-vxwexplo-process Sun Mar 21 08:21:27 1999 From: Eric.vanderSanden@EMT-EHV.BE.PHILIPS.COM Date: Sun Mar 21 08:21:31 PST 1999 Subject: RE: missing symbols (CORBA with vxworks) Status: O Status Distribution March 19, 1999 18:59:50 The message regarding "RE: missing symbols (CORBA with vxworks)" sent on March 19, 1999 18:59:50 was sent by Status Recipient Type To Native Name vxwexplo@lbl.gov Foreign Native Name vxwexplo@lbl.gov\n\n\nSMTP Recipients Status Reporters Type From Name Domain NOTES Native Name CN=Eric van der Sanden/OU=EHV/OU=EMT/OU=BE/O=PHILIPS@PHILIPS07 Foreign Native Name CN=Eric van der Sanden/OU=EHV/OU=EMT/OU=BE/O=PHILIPS\nPHILIPS07\n\n Organization PHILIPS Org Unit 1 BE Org Unit 2 EMT Org Unit 3 EHV Last Name Sanden First Name Eric Status 769 Explanation Invalid recipient X.400 Status 769 Explanation Delivery time expired From owner-vxwexplo-process Mon Mar 22 00:01:08 1999 From: Arand Thomas Date: Mon Mar 22 00:01:12 PST 1999 Subject: AW: C++ template class on VxWorks Status: O You will have these kind of problems very often with C++, at least when you use the GNU compiler. This is not a specific vxWorks problem. What the compiler does is instantiating the methods of a template class only then when you use it. That means, if file2.cpp uses the list, but not all methods, the methods used in file3.cpp are NOT instantiated, what results in your error. What you have to do (if that really is the problem): Put the instantiation in a .h file (what produces a lot of unnecessary inline code eventually) or instantiate the missing functions explicitely. Hope that helps. Ciao, Thomas ---------------------------------------------------------------------- Thomas Arand Siemens AG Tel. +49 - 89 - 722 42296 Fax. +49 - 89 - 722 26572 thomas.arand@icn.siemens.de > -----Ursprüngliche Nachricht----- > Von: vxwexplo@lbl.gov [SMTP:vxwexplo@lbl.gov] > Gesendet am: Donnerstag, 18. März 1999 21:54 > An: vxworks_users@csg.lbl.gov > Betreff: C++ template class on VxWorks > > Submitted-by owner-vxwexplo-process Thu Mar 18 12:53:49 1999 > Submitted-by: piyush agarwal > > > We are doing development on MIPS processor with VxWorks 5.3 OS. We have > defined some template classes. We follow the procedure described in the > programmer's guide. We have defined a template class in file.h file. Then > we have a file1.cpp file in which we include file.h and define methods for > this template class. We have another file file2.cpp in which we include > file1.cpp and instantiate the template class. We have other files > file3.cpp etc which use these. Now when we link these files we get an > error in file3.cpp as "undefined reference to list:..." where list is > our template class. > > We then linked with -r option and did nmmips on the object module. Portion > of the output is given below: > > 000080c8 t appendElem__t4list1Z11ComponentIdR11ComponentId > U appendElem__t4list1Z11ComponentIdR11ComponentId > U appendElem__t4list1Z11RegListElemR11RegListElem > 00003238 t appendElem__t4list1Z11RegListElemR11RegListElem > 000038b0 t appendElem__t4list1Z13RemoteRegInfoR13RemoteRegInfo > U appendElem__t4list1Z13RemoteRegInfoR13RemoteRegInfo > U appendElem__t4list1Z15GatewayCompInfoR15GatewayCompInfo > 00003574 t appendElem__t4list1Z15GatewayCompInfoR15GatewayCompInfo > 00003bec t appendElem__t4list1Z9EventInfoR9EventInfo > > Thanks, > Piyush > > > From owner-vxwexplo-process Thu Mar 11 10:16:58 1999 From: "Bruce A. Wilson" Date: Mon Mar 22 03:19:20 PST 1999 Subject: RE: question about load vxworks Status: O Use the printErrno function from the shell to get a description as shown below: -> printErrno 0xd0003 0xd0003 = S_iosLib_INVALID_FILE_DESCRIPTOR >Submitted-by owner-vxwexplo-process Wed Mar 10 07:48:22 1999 >Submitted-by: Bill Mills < > >... I don't know where to find the meaning of >the error code, I did look one time. If anyone else knows, please let us >know. > >> >> Loading... >> Error loading file: errno = 0xd0003 >> >> The procedure is very short. Could you tell me what caused this >> problem and the means of errno( 0xd0003 ), by the way, please tell >> me how can I find the error's mean. >> Bruce A. Wilson U.C. Lawrence Livermore National Lab P.O. Box 808 (L-054) Livermore, CA 94550 voice: (925) 422-4145 fax: (925) 422-9905 From owner-vxwexplo-process Mon Mar 22 03:19:14 1999 Received: from postal1.lbl.gov (root@postal1.lbl.gov [128.3.7.82]) by csg.lbl.gov (8.8.7/8.8.7) with ESMTP id DAA04614 for ; Mon, 22 Mar 1999 03:19:13 -0800 (PST) From: Roy.Wilmhoff@vickers-systems.com Received: from SpamWall.lbl.gov (root@localhost [127.0.0.1]) by postal1.lbl.gov (8.9.0/8.9.0) with ESMTP id DAA00392 for ; Mon, 22 Mar 1999 03:19:12 -0800 (PST) Received: from ctlas08.tnv.com (ctlas08.aeroquip-vickers.com [148.179.192.231]) by SpamWall.lbl.gov (8.9.0/8.9.0) with ESMTP id DAA00388 for ; Mon, 22 Mar 1999 03:19:11 -0800 (PST) Received: by ctlas08.tnv.com (NPlex 1.3.167); 22 Mar 1999 06:28:01 -0500 Message-ID: <36F3E54100001676@ctlas08.tnv.com> (added by MTA ctlas08.tnv.com) To: vxwexplo@lbl.gov Subject: X.400 Inter-Personal Notification Message-Type: InterPersonal Notification References: <199903191722.JAA07374acsg.lbl.gov@MHS> Date: 22 Mar 1999 06:22:35 -0500 Status: O Your message to: was discarded for the following reason:Obsoleted The Original Message is not available From owner-vxwexplo-process Fri Mar 19 06:59:47 1999 From: jmk@alum.mit.edu (Joel Katz) Date: Mon Mar 22 03:19:21 PST 1999 Subject: Re: System Clock Status: O >>----snip >VxWorks 5.3.1 >MVME2700 1.1/4 > >I have seen a lot of discussion about running the system clock at high rates >and the negative effects it can have on system performance. I am a little >concerned about the blanket "don't do it" advice that has been given in this >forum, and I thought I'd add my two cents worth. > > >There are good reasons for running the system clock faster. The most >important reason for us was being able to provide high timeout resolution >for the select() system call. As far as I know, uping the system clock rate >is the ONLY way to improve select() timeout resolution. > >Anyway, I agree that caution should be used when increasing the system clock >rate, but I believe users should determine the impact on an individual basis >rather than avoiding it because it's "just a bad idea." I just wanted to second Charlie's thoughts on the system clock issue. I'd be willing to bet money that the 60 hz rate is based on using a processor along the lines of a 30 Mhz 68030. I did some quick, non-empirical experiments running our board with a 1Khz system clock rate and detected no adverse effects on normal functioning. Joel Katz, Principal Engineer --- (603)-577-5895 Acuity Imaging, 9 Townsend West, Nashua, NH 03063 --- (603)-577-5965 (FAX) visit our Web Site at >>> www.acuityimaging.com <<< From owner-vxwexplo-process Mon Mar 22 03:19:15 1999 Received: from postal1.lbl.gov (root@postal1.lbl.gov [128.3.7.82]) by csg.lbl.gov (8.8.7/8.8.7) with ESMTP id DAA04620 for ; Mon, 22 Mar 1999 03:19:14 -0800 (PST) From: Roy.Wilmhoff@vickers-systems.com Received: from SpamWall.lbl.gov (root@localhost [127.0.0.1]) by postal1.lbl.gov (8.9.0/8.9.0) with ESMTP id DAA00400 for ; Mon, 22 Mar 1999 03:19:13 -0800 (PST) Received: from ctlas08.tnv.com (ctlas08.aeroquip-vickers.com [148.179.192.231]) by SpamWall.lbl.gov (8.9.0/8.9.0) with ESMTP id DAA00393 for ; Mon, 22 Mar 1999 03:19:12 -0800 (PST) Received: by ctlas08.tnv.com (NPlex 1.3.167); 22 Mar 1999 06:28:02 -0500 Message-ID: <36F3E5410000167B@ctlas08.tnv.com> (added by MTA ctlas08.tnv.com) To: vxwexplo@lbl.gov Subject: X.400 Inter-Personal Notification Message-Type: InterPersonal Notification References: <199903192141.NAA12084acsg.lbl.gov@MHS> Date: 22 Mar 1999 06:22:34 -0500 Status: O Your message to: was discarded for the following reason:Obsoleted The Original Message is not available From owner-vxwexplo-process Mon Mar 22 04:03:51 1999 From: daemon@csg.lbl.gov Date: Mon Mar 22 04:03:54 PST 1999 Subject: comp.os.vxworks newsdigest Status: O Comp.Os.Vxworks Daily Digest Mon Mar 22 04:03:48 PST 1999 Subject: hdlc driver for MC68360 (QUICC) Subject: Re: Realtime clock on MPC8xx ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: hdlc driver for MC68360 (QUICC) Date: Thu, 18 Mar 1999 14:59:17 GMT From: shad.ansari@premisys.com Organization: Deja News - The Leader in Internet Discussion Message-ID: <7cr4c0$mog$1@nnrp1.dejanews.com> Hi, I am working on a project that requires to implement ip over hdlc on a MC68360 (QUICC) platform running vxWorks. Does anybody know anything about commercial/freeware availability of a vxworks hdlc driver for MC68360 that can run on any SCC? Thanks in advance, Shad Ansari Software Engineer, Premisys Communications shad.ansari@premisys.com 613.722.4332x2269 Wind River License# 15581 - -----------== Posted via Deja News, The Discussion Network ==---------- http://www.dejanews.com/ Search, Read, Discuss, or Start Your Own --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Realtime clock on MPC8xx Date: Thu, 18 Mar 1999 10:58:34 -0800 From: "John Finley" Organization: Kivala Systems, Inc. Message-ID: <7cri72$c80@dfw-ixnews4.ix.netcom.com> References: <433F274FA1EDBCD0.B9B0C5670A0B84D0.F5D0E0B6EAAB87B5@library-proxy.airnews.net> Kevin Agnew wrote in message <433F274FA1EDBCD0.B9B0C5670A0B84D0.F5D0E0B6EAAB87B5@library-proxy.airnews.ne t>... >I need to implement a date/time clock on the MPC8xx >processor. The MBX860 and the FADS860 BSPs do not supply >drivers for the RTC in the 8xx processor. > >Does anyone know if a device driver has been written for >the RTC (to convert ticks to/from time/day/month/year)? > >Or can someone point me to a good (and Y2K compliant) >algorithm so I can implement one myself? To convert seconds to date/time, look at Dallas Semiconductor App Note 31, "Algorithm for Converting Binary Seconds into Date/Time", at http://www.dalsemi.com/DocControl/PDFs/appindex.html I have not used it, but it's at least a starting place. John - -------------------------------------------------- John Finley Kivala Systems, Inc. Project Manager (619) 689-0032 john@kivala.com http://www.kivala.com - -------------------------------------------------- VxTool - Build Tornado Apps with MS Visual Basic - -------------------------------------------------- --------------------------- End of New-News digest ********************** From owner-vxwexplo-process Mon Mar 22 07:55:03 1999 From: Bhavin Parikh Date: Mon Mar 22 07:55:06 PST 1999 Subject: Porting Web Server to VxWorks Status: O Hi Everybody, Currently I am working hard with windweb server and VxWorks. Actually WindWeb server doesn't support everything whatever I need like java servlet and some more. So now I am looking for other web server, who are available on VxWorks. I need your help in following issues. It can save my lots of time. 1. Does anybody has some recommedations and solutions so WindWeb server can be extended for achieving servlet support? 2. Is there a freeware HTTP server with servlet support already ported to VxWorks available? 3. I found Apache is really good for servlet support. What is porting effect required for Apache Web Server porting to VxWorks. I would appreciate your reply and help. Thanks Bhavin Parikh Microsoft Certified Solution Developer, From owner-vxwexplo-process Mon Mar 22 08:38:38 1999 From: Marcus Hardy Date: Mon Mar 22 08:38:42 PST 1999 Subject: HDLC driver for MC68360 (QUICC) Status: O Dear Shad Having seen your question to the VxWorks user group, it may be interesting to note that RTP has an off-the-shelf generic HDLC VxWorks driver available. The device driver provides a generic programming interface to VxWorks applications which is independent of the underlying hardware. This allows applications requiring HDLC services to be easily ported to platforms with differing HDLC services and even provides support for different devices on a single platform. The HDLC driver is structured as a well-defined set of library functions analogous to the existing VxWorks serial device driver library. The HDLC library handles common functions such as buffer management, HDLC programming and multiplexing and has several easily configurable modes of operation to support different applications requirements. Support is provided for PPP (RFC1662) and SLIP as well as for framed or unframed raw data channels. There will in fact be a version available soon that we will have already ported to the MC68360 (QUICC). RTP has a number of VxWorks drivers available. You may wish to e-mail enquiries@rtp.co.uk, or visit our website : http://www.rtp.co.uk or indeed contact me direct for more information. Best Regards Marcus ------------------------------------------------------- Marcus Hardy E-mail: mailto:mjh@rtp.co.uk Real Time Products Ltd. Direct: +44 (0) 121 234 6626 Chancery House, Tel: +44 (0) 121 234 6600 8 Edward Street, Fax: +44 (0) 121 234 6611 Birmingham. Web: http://www.rtp.co.uk B1 2RX. England. ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: hdlc driver for MC68360 (QUICC) Date: Thu, 18 Mar 1999 14:59:17 GMT From: shad.ansari@premisys.com Organization: Deja News - The Leader in Internet Discussion Message-ID: <7cr4c0$mog$1@nnrp1.dejanews.com> Hi, I am working on a project that requires to implement ip over hdlc on a MC68360 (QUICC) platform running vxWorks. Does anybody know anything about commercial/freeware availability of a vxworks hdlc driver for MC68360 that can run on any SCC? Thanks in advance, Shad Ansari Software Engineer, Premisys Communications shad.ansari@premisys.com 613.722.4332x2269 Wind River License# 15581 From owner-vxwexplo-process Mon Mar 22 11:02:31 1999 From: "Blake, Don" Date: Mon Mar 22 11:02:35 PST 1999 Subject: RE: how to use excHookAdd() Status: O I don't have any experience with excLib. However, I doubt that you can do printf's from an exception handler. The vxWorks manuals seem to be silient on any excHookAdd function limitations but I would guess they would be similar to those for intConnect - I/O is not permitted and don't do anything that could potentially block. Don Lockheed Martin Federal Systems Owego, NY From owner-vxwexplo-process Mon Mar 22 13:09:05 1999 From: mmenge@dspt.com Date: Mon Mar 22 13:09:09 PST 1999 Subject: Cheap VME Controller with Ethernet? Status: O From: Mark Menge@DSPT on 03/22/99 04:08 PM Does anyone know what the most inexpensive VME controller that supports ethernet is? I would like it to support tornado and vxWorks, but that is not required. From owner-vxwexplo-process Tue Mar 23 02:38:57 1999 From: Ilia Bosis Date: Tue Mar 23 02:39:01 PST 1999 Subject: Re: HDLC driver for MC68360 (QUICC) Status: O Take a look at the Aisys inc. http://www.aisysinc.com They have a driver generator for the popular uC and RTOS'es. AFAIK, QUICC & VxWorks included. Hope, this helps. Ilia. Disclaimer: The opinions here are mine and not of my employer Not I, nor my employer haven't linked with Aisys in any way. +=====================================================================+ | Bosis Ilia, Software engineer, Telegate ltd. | | E-mail: iliab 'at' telegate.co.il | WWW: http://www.telegate.co.il | +=====================================================================+ >Hi, > >I am working on a project that requires to implement ip over hdlc >on a MC68360 (QUICC) platform running vxWorks. >Does anybody know anything about commercial/freeware >availability of a vxworks hdlc driver >for MC68360 that can run on any SCC? > >Thanks in advance, >Shad Ansari > >Software Engineer, Premisys Communications >shad.ansari@premisys.com >613.722.4332x2269 -- +=====================================================================+ | Bosis Ilia | Work phone: 972-3-5384685 | | Software engineer | Fax: 972-3-5335877 | | Telegate ltd. | E-mail: iliab at telegate.co.il | | 60218 Haplada7 Or-Yehuda, Israel | WWW: http://www.telegate.co.il | +=====================================================================+ From owner-vxwexplo-process Tue Mar 23 04:02:28 1999 From: daemon@csg.lbl.gov Date: Tue Mar 23 04:02:32 PST 1999 Subject: comp.os.vxworks newsdigest Status: O Comp.Os.Vxworks Daily Digest Tue Mar 23 04:02:20 PST 1999 Subject: SENS Problem Subject: Re: GNU objcopy problem Subject: purposefully crashing VxWorks Subject: Re: Sockets in serial interrupt handler ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: SENS Problem Date: Mon, 22 Mar 1999 19:24:22 +0200 From: "RonenO" Organization: Unspecified Organization Message-ID: <7d5uip$d0s$1@news.ibm.net.il> I am using SENS 1.1 on PPC860. It seems that the tcp/ip stack stops working when a socket is full (out of buffers) & does not recover. Does anyone have more information or a patch for this problem? Ronen. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: GNU objcopy problem Date: Mon, 22 Mar 1999 21:51:23 GMT From: rjdejoie@my-dejanews.com Organization: Deja News - The Leader in Internet Discussion Message-ID: <7d6e0n$bsa$1@nnrp1.dejanews.com> References: <36EDDA2C.1B2369A0@domain.com> In article <36EDDA2C.1B2369A0@domain.com>, defaultuser@domain.com wrote: > Greetings, > > I am attempting to use objcopymips from the GNU toolkit to generate a > binary file using the following command: > objcopymips -Obinary monitor monior.bin > When I run the command I get an error that the output device is full > even though I have >200MB free diskspace (the binary file should be less > than 256kbytes). I also had problems using the --byte command line > switch. > I am using objcopymips version cygnus-2.6 under windows NT4. > > Any help would be greatly appreciated > David > wrote in message >news:36EDDA2C.1B2369A0@domain.com... > Greetings, > > I am attempting to use objcopymips from the GNU toolkit to generate a > binary file using the following command: > objcopymips -Obinary monitor monior.bin > When I run the command I get an error that the output device is full > even though I have >200MB free diskspace (the binary file should be less > than 256kbytes). I also had problems using the --byte command line > switch. > I am using objcopymips version cygnus-2.6 under windows NT4. > > Any help would be greatly appreciated > David I also had a problem with the -b option of objcopy68k delivered with Tornado from Wind River (cygnus 2.6 running under NT4). I downloaded and built latest binutils which produced objcopy 2.9.4, it cleaned up the problem with -b. - -----------== Posted via Deja News, The Discussion Network ==---------- http://www.dejanews.com/ Search, Read, Discuss, or Start Your Own --------------------------- Newsgroups: comp.os.vxworks Subject: purposefully crashing VxWorks Date: Sat, 13 Mar 1999 19:06:51 -0500 From: "Michael J. Abowd" Organization: CMU Software Engineering Message-ID: <7ceuin$k3t$1@goldenapple.srv.cs.cmu.edu> Hi, I am working on a project that needs to consistently create a type of OS failures in VxWorks running on a Motorola 68040. Create a process that cause the OS to hang altogether. At this point, the machine would need to rebooted. If this processes could be created by calling a POSIX call, that would be very handy, but it's not required. Any suggestions on how to create this failure? I have noticed that when I execute a while(true) loop, the OS hangs, but don't think it should and was wondering if there were other ways to do this. Thanks Much, Michael --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Sockets in serial interrupt handler Date: Tue, 23 Mar 1999 08:35:06 GMT From: gabry@ian.ge.cnr.it (Gabriele Bruzzone) Organization: C.N.R. - Istituto per l'Automazione Navale Keywords: socket, serial, interrupt Message-ID: <7d7ngu$t80$1@news.csita.unige.it> References: <7d6095$gar$1@news.csita.unige.it> Followup-To: comp.os.vxworks Reply-To: gabry@ian.ge.cnr.it In article , chc@nasa2.ksc.nasa.gov (Charles H. Chapman) wrote: >On Mon, 22 Mar 1999 16:52:23 GMT, Gabriele Bruzzone > wrote: >> I'm writing a serial "repeater", i.e. a program having the task to >>read a serial input and send it through a socket. I though to call the >>socket's write(...) function directly in the serial interrupt handler every >>time I receive the received char's interrupt. In your opinion, could this be a > >>dangerous approach? > >Dangerous, and not allowed. No sockLib functions are allowed from an >interrupt handler -- see Table 2-22 in the VxWorks 5.3 Programmers >Guide. I'm sorry I haven't read the manual very carefully... > >>Another approach could be the following: the interrupt handler when it >>receives a received char's interrupt it before fills a queue and then it >>unblocks a semaphore which signals that there are characters availables in the > >>buffer. Then, there is another task blocked on the same semaphore which, when >>it unblocks, it reads the characters in the queue and sends them through the >>socket. But I think this second solution could be very inefficient. Do you >>have any suggestions? > >The second approach would work but it may be a lot more straightforward >to have the ISR do a msgQSend (with the NO_WAIT option) with the character >received in the message and have another task pend on that message queue >and write the character to the socket. > >Chuck > Surprise, surprise... as told me Stephen (Steve) L. Egber some years ago: A potential workQPanic crash is possible if multiple msgQSend is done within the same interrupt or series of successive interrupts. You would be limited to the fixed workQSize of VxWorks kernel (so difficult but not impossible to increase.) It is better to home-brew your own message queueing mechanism and not having to call semGive() or msgQSend() from ISR which contributes to the expanding workQ size and potentially leading to workQPanic condition. Most in-house created and specialized message/semaphore libraries are faster than the VxWorks's msgQLib/semLib which is provided for maximum flexibility. I'm afraid the solution to my problem could be not so easy :( Many thanks for the attention you paid to my question. Kindest regards Gabriele +---------------------------------------------------------------+ | Gabriele BRUZZONE | | Research Scientist of Robotics Department | | C.N.R. - Istituto per l'Automazione Navale | | Via De Marini, 6 Direct +(39)-010-6475657 | | 16149-GENOVA-ITALY Fax +(39)-010-6475600 | | mailto:gabry@ian.ge.cnr.it http://www.ian.ge.cnr.it/robotics | +---------------------------------------------------------------+ --------------------------- End of New-News digest ********************** From owner-vxwexplo-process Tue Mar 23 05:03:26 1999 From: Bob Anderson Date: Tue Mar 23 05:03:29 PST 1999 Subject: RE: Cheap VME Controller with Ethernet? Status: O Depends what you mean by inexpensive. A bare-bones MVME162-212 from Motorola lists for $2045. This is probably overkill for your needs but it is a proven reliable solution (also comes with a five year warranty). You can get a data sheet for this on the Motorola Computer Group web site at http://www.mcg.mot.com and then follow the URL for product portfolio. Select 68K VME boards in the table. Thanks, Bob Anderson > -----Original Message----- > From: vxwexplo@lbl.gov [SMTP:vxwexplo@lbl.gov] > Sent: Monday, March 22, 1999 4:09 PM > To: vxworks_users@csg.lbl.gov > Subject: Cheap VME Controller with Ethernet? > > Submitted-by owner-vxwexplo-process Mon Mar 22 13:09:05 1999 > Submitted-by: mmenge@dspt.com > > > From: Mark Menge@DSPT on 03/22/99 04:08 PM > Does anyone know what the most inexpensive VME controller that supports > ethernet is? > I would like it to support tornado and vxWorks, but that is not required. > From owner-vxwexplo-process Tue Mar 23 07:46:31 1999 From: "Jonas Boegård" Date: Tue Mar 23 07:46:35 PST 1999 Subject: Tornado and ClearCase Status: O Hi, I am participating in a project where we would like to use ClearCase for source code version control. I would like to know if anyone has had any success in using Tornado and ClearCase together. We are using Tornado version 1.0.1 for solaris. Thanks in advance Jonas -- _______________________________________________________________ Jonas Boegård Ericsson Radio Access RSA/B/UD email: jonas.boegard@rsa.ericsson.se +46(0)8-585 32139 memo: rsa.rsajboe Borgarfjordsg 14 Kista, Sweden From owner-vxwexplo-process Tue Mar 23 08:03:54 1999 From: Bill Brown Date: Tue Mar 23 08:03:58 PST 1999 Subject: cPCI under vxWorks Status: O In article <7d83e4$i7a$1@client2.news.psi.net>, Bill Rogers wrote: >Hi, >I have been struggling with reading compact PCI I/O bus addresses from my >Force 603 Power PC board. My goal is to read a Mailbox register of a I/O >board(for which no device drive exists). However, it seems that the Plug and >Play nature of the PCI bus requires that first the I/O address must be >assigned by the 603PPC. >I have received little support from Force regarding the BSP configuration to >assign PCI I/O addresses. >I would appreciate any information anyone has in this area (e.g. recommended >reading or prior experience). We, also, are sticking our noses into the PPC/cPCI world; it's sure a whole different universe than vme! WRS "supports" the Motorola BSP; so far I've found that "support" to be somewhat underwhelming. Essentially, what you need to do is write some stuff that probes the devices on the cPCI bus(es) - they can be layered! Bridges (header-type 1) are initialized differently than are other (type 0) devices on the bus. A good reference is "PCI System Architecture (third edition)" by Shanley and Anderson, ISBN 0-201-40993-3. There is another book "available," but we've had it on backorder for many weeks now. IMHO, learning the cPCI bus is _at_least_ a "3-beer problem." Even if I understood it well, I wouldn't try to walk anyone thru a problem like this on email. Try going to someone like amazon.com or one of the other booksellers and do a search on "P.C.I." or some variant. ("PCI Bus"?) So far, we've tinkered with a cpu from Motorola (MCP-750) and V.I. Computer (power4ecpci, I think). Motorola has a much more complete bsp, but there are a couple of "gotchas" in the bsp which are apparently caused by some strange features in the hardware, that had me going for a couple of weeks. However, it is a good example of how to configure the cPCI stuff for the general case. I have to admit that I really didn't understand what was going on until our cPCI analizer board arrived from VMETRO. The VI Comp. bsp is much less complete, but so far I haven't found any real "gotchas" in it. It does not include the "automatic configuration" code that the Motorola package does, but what _is_ there seems somewhat less quirkey. It sounds like it's probably similar to the code you got from Force. Unfortunately, I have no experience with the Force board. I suspect that it is similar, as everyone seems to be using the same PCI-toPCI bridge chips, designed by Digital and now sold by Intel, I believe. I can send you the stuff I wrote to get the V.I. Comp. board "talking to" the "first level" cPCI bus, if you'd like it. Perhaps it will be of some small help in understanding what needs to be done, but you're really going to need a _good_ reference text on PCI to move forward. Hope this is helpful. Disclaimer: Any opinions are my own and have | -bill nothing to do with the official policy or the | wlbrown@lbl.gov management of L.B.N.L, who probably couldn't | Berkeley, CA care less about employees who play with trains. | aka wlbrown@netcom.com From owner-vxwexplo-process Tue Mar 23 08:39:32 1999 From: "Sowko, Nick" Date: Tue Mar 23 08:39:36 PST 1999 Subject: RE: Slow TCP/IP Sockets Status: O TSR #116602 addresses the slow boot problem on Solaris 2.6 (expecially noticed going from 2.5.1 to 2.6) TITLE: Why does Solaris 2.6 take so long to boot vxworks? AUTHOR: David Holloway SCOPE DISCOVERED ON HOST: Solaris 2.6 / All Solaris 2.6 ------------------- ------------------- ARCHITECTURE: All PowerPC ------------------- ------------------- BSP: All mtx603 ------------------- ------------------- PRODUCT / REVISION: N/A vxWorks 5.3.1 ------------------- ------------------- DESCRIPTION: It takes far too long to boot vxworks using a solaris 2.6 ftp server under the following conditions: 1. when it is rebooted in quick succession. 2. when symbol tables loaded via the network take too long to load. RESEARCH: The bootrom uses a local port number to download vxWorks. VxWorks will attempt to use the same port number to download vxWorks.sym. In accordance with TCP/IP's proper implementation, after the bootrom's transaction, Solaris 2.6 still has a connection in a TIME_WAIT state as vxWorks attempts to download vxWorks.sym. VxWorks must then wait for Solaris 2.6 to time out before proceeding with its connection. That is the cause for the delay. Sun Microsystems probably fixed TCP/IP in version 2.6 of Solaris. The device /dev/tcp has a "quality" modifiable by /usr/sbin/ndd called tcp_time_wait_interval. Solaris 2.5.1 doesn't seem to pay attention to this variable, if one changes it to a larger or smaller value then vxWorks/vxWorks.sym still load without interruption. RESOLUTION: If you do not have root access to the solaris ftp server, then add this to your usrConfig.c. It increments the local port number so that vxWorks.sym will load more quickly. /* Right before the call to netLoadSymTbl() */ #ifdef SOLARIS26FTPD /* increment the last used port number by one. */ #include "netinet/in.h" #include "sockLib.h" { int mybsock; struct sockaddr_in mybsockaddr; mybsockaddr.sin_addr.s_addr = htonl(INADDR_ANY); mybsockaddr.sin_port = 0; mybsockaddr.sin_family = AF_INET; mybsock = socket(AF_INET, SOCK_STREAM, 0); bind (mybsock, (struct sockaddr *) &mybsockaddr, sizeof(mybsockaddr)); close(mybsock); } #endif /* SOLARIS26 ftp server*/ Note, this would also apply when using a vxworks ftp server to boot other vxworks targets. If you do have root access to the solaris ftp server then decrease the value of tcp_time_wait_interval using /usr/sbin/ndd example usage: /usr/sbin/ndd -set /dev/tcp tcp_time_wait_interval 5000 SPR / PATCH: KEYWORDS: Solaris 2.6 symbol table boot INCLUDE_NET_SYM_TBL From owner-vxwexplo-process Tue Mar 23 15:15:11 1999 From: "Minbo Shim" Date: Tue Mar 23 15:15:15 PST 1999 Subject: SCSI hard disk configuration on mvme2700 Status: O Dear vxWorks gurus: I am trying to configure a one-slot SCSI hard drive module on mvme2700 PowerPC board. The disk is ATA (IDE) hard disk. I defined the followings in 'config.h': INCLUDE_SCSI INCLUDE_SCSI2 INCLUDE_SCSI_DMA INCLUDE_SCSI_BOOT SCSI_AUTO_CONFIG INCLUDE_DOSFS Based on the manual, it suppose to display the SCSI device at the booting time, but it didn't. The 'scsiAutoConfig()' seemed to reach the device (LED on the hard disk was blinking), but failed to identify the device (scsiShow() right after the scsiAutoConfig() showed nothing). In order to make sure the Hard disk is fine, I attached the device into the end of the SCSI chain coming from one of our SUNs. The 'probe-scsi-all' identified the hard disk, and displayed all the logical units. What else I did was that from the Motorola bug prompt, I issued 'ioi' and got the right result as follows: PPC1-Bug>ioi I/O Inquiry Status: CLUN DLUN CNTRL-TYPE DADDR DTYPE RM Inquiry-Data 0 20 NCR53C825 2 $00 N ADTRON SDV6 N19013 SDV6 2f01 1 0 PC8477 0 $00 Y Obviously, DADDR 2 is the right number that I assigned to the hard disk. Using the 'iop','mm' and 'md', I was able to verify that the system board can read and write to the device. So, I came up with that the problem exists on vxWorks. Do you have any clue about why the vxWorks does not identify the SCSI hard disk? Thanks for you help. ----------------- Minbo Shim; mshim@rst.com Robotic Systems Technology Westminster, MD 21157 410.876-9200 From owner-vxwexplo-process Tue Mar 23 23:51:10 1999 From: "Bill W.G. Shi" Date: Tue Mar 23 23:51:14 PST 1999 Subject: IEEE 802.2 support in vxWorks Status: O VxWorks Hi: Can anyone indicate how IEEE 802.2 is supported in vxWorks, if it IS supported? I have learned that the traditional protocol stack in vxWorks is basically from BSD, which does not support the IEEE 802.2, but only the DIX. Is that right? Then how about SENS? Best Regards Bill W.G. Shi at BBT Inc. E-Mail Address: wgshi@bbti.com.cn From owner-vxwexplo-process Wed Mar 24 02:04:55 1999 From: Martin Schrape Date: Wed Mar 24 02:04:59 PST 1999 Subject: Re: SCSI hard disk configuration on mvme2700 Status: O Hi Minbo, I don't think that you need INCLUDE_SCSI_DMA and I suppose that you boot via TCP/IP so that you do not need INCLUDE_SCSI_BOOT. So, are you really sure that the SCSI_AUTO_CONFIG code in usrScsiConfig is executed? There are already some printf lines. Do you see them? Some SCSI drives really need more time to get ready then the 7 seconds that my VxWorks boot rom image waits. Make a warm restart (=reset) not a power-up restart. martin Minbo Shim wrote: > > Dear vxWorks gurus: > > I am trying to configure a one-slot SCSI hard drive module > on mvme2700 PowerPC board. The disk is ATA (IDE) hard disk. > > I defined the followings in 'config.h': > > INCLUDE_SCSI > INCLUDE_SCSI2 > INCLUDE_SCSI_DMA > INCLUDE_SCSI_BOOT > SCSI_AUTO_CONFIG > INCLUDE_DOSFS > > Based on the manual, it suppose to display the SCSI device > at the booting time, but it didn't. The 'scsiAutoConfig()' > seemed to reach the device (LED on the hard disk was > blinking), but failed to identify the device (scsiShow() > right after the scsiAutoConfig() showed nothing). > > In order to make sure the Hard disk is fine, I attached the > device into the end of the SCSI chain coming from one of our SUNs. > The 'probe-scsi-all' identified the hard disk, and displayed all > the logical units. > > What else I did was that from the Motorola bug prompt, I issued > 'ioi' and got the right result as follows: > > PPC1-Bug>ioi > I/O Inquiry Status: > CLUN DLUN CNTRL-TYPE DADDR DTYPE RM Inquiry-Data > 0 20 NCR53C825 2 $00 N ADTRON SDV6 N19013 SDV6 2f01 > 1 0 PC8477 0 $00 Y > > Obviously, DADDR 2 is the right number that I assigned to the > hard disk. Using the 'iop','mm' and 'md', I was able to verify > that the system board can read and write to the device. So, > I came up with that the problem exists on vxWorks. > > Do you have any clue about why the vxWorks does not identify the > SCSI hard disk? > > Thanks for you help. > ----------------- > Minbo Shim; mshim@rst.com > Robotic Systems Technology > Westminster, MD 21157 > 410.876-9200 From owner-vxwexplo-process Wed Mar 24 02:07:31 1999 From: Reyhan ERGUN Date: Wed Mar 24 02:07:35 PST 1999 Subject: Additional SMC and SCC initializations in sysSerial.c for MBX860 Status: O This is a multi-part message in MIME format. --------------AF4A57FE23A632044B807BE7 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit I use VxWorks: 5.3.1 and BSP version: 1.1/1 for MBX860. One serial driver and one ethernet driver are supported in the BSP for MBX860/821. Other SCCs (SCC2-4) and SMC2 are not initialized in the BSP, I wonder if anyone has added these initializations to the BSP or not. Are there any other BSP which supports all serial communication features for MPC860. Thanks, Reyhan --------------AF4A57FE23A632044B807BE7 Content-Type: text/x-vcard; charset=us-ascii; name="vcard.vcf" Content-Transfer-Encoding: 7bit Content-Description: Card for Reyhan ERGUN Content-Disposition: attachment; filename="vcard.vcf" begin: vcard fn: Reyhan ERGUN n: ;Reyhan ERGUN org: ASELSAN INC. adr: Microwave and System Technologies Division ;;Software Engineering Dept.;ANKARA;;;TURKEY email;internet: findik@mst.aselsan.com.tr tel;work: (90-312) 385 19 00 (Ext. 2668) tel;fax: (90-312) 354 52 05 note: http://www.aselsan.com.tr x-mozilla-cpt: ;0 x-mozilla-html: FALSE version: 2.1 end: vcard --------------AF4A57FE23A632044B807BE7-- From owner-vxwexplo-process Wed Mar 24 02:58:44 1999 From: Rajesh Gautam Date: Wed Mar 24 02:58:48 PST 1999 Subject: Telnet Management Console. Status: O Hi, I am planning to implement a (character based) telnet management console for our VxWorks target. A user should be able to do a telnet to the device and should be able to view , get / set ( by some commands) various configuration data is some format. -Will it require to implement a shell with commands to be used with telnetd. - or a seperate standalone application. Either of them will have command parser to parse the input commands and display routines to output the results in different formats( tables etc.) If anybody has implemented similar solution, I would like to know how to go about it Thanks, gautam. -- -------------------------------- Network Machines, 311, Ganga Shopping Complex, Noida, INDIA, 201301. Ph# 91-575-544783, 548517, 548518. Fax# 91-575-543721. From owner-vxwexplo-process Wed Mar 24 04:00:15 1999 From: daemon@csg.lbl.gov Date: Wed Mar 24 04:00:24 PST 1999 Subject: comp.os.vxworks newsdigest Status: O Comp.Os.Vxworks Daily Digest Wed Mar 24 04:00:12 PST 1999 Subject: Re: NE2000 support on WinSystems LBC-586Plus? Subject: Re: Tornado and ClearCase Subject: Re: Generate/Catch Interrupt on x86 board? ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: NE2000 support on WinSystems LBC-586Plus? Date: Wed, 24 Mar 1999 09:26:14 +0000 From: Ian Love Organization: BT Labs - but speaking only for myself Message-ID: <36F8AFB6.441B@wireplay.bt.co.uk> References: <7d98s3$q5f$1@nnrp1.dejanews.com> travis@deskin.com wrote: > > Hi all, > > I have a WinSystems LBC-586Plus 133Mhz single board computer with an NE2000 > ethernet interface that I'm trying set up as a target machine with vxWorks > v. 5.3 (Tornado v. 1.0). I've got a boot floppy set up and a good network > connection to the host computer (10BaseT), but can't set the boot device > to the network. It has "ene" as a boot device option, which is correct, > and says it attaches ene0 and lo0 correctly. But when it tries to boot from > the network, it times out after ~30seconds with the error: > > Error loading file: errno = 0xd0003. This errno is S_iosLib_INVALID_FILE_DESCRIPTOR (since my system works I used printErrno) Basically, it cant open the file (as specified in the bootline)..... (I write this from a UNIX host perspective, but some of it will map to other boot hosts) 1) wrong path/filename (for the kernel) 2) wrong file permissions/no access to directory specified by path 3) incorrect user details (username/password) 4) ftp not working (if password set) 5) rsh not working (if password not set) - note: host needs to 'trust' the vxWorks card to login as the specified user Of course, your driver may not be working, but the above is much more likely..... Alternatively, if you wish to get the kernel off the floppy, and just need to get the SBC connected to the network, then you need to start the networking yourself and make the floppy the boot device regards Ian Love, The views expressed above are my own and not those of my employer --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Tornado and ClearCase Date: Wed, 24 Mar 1999 09:59:41 +0000 From: Ian Love Organization: BT Labs - but speaking only for myself Message-ID: <36F8B78D.2BAC@wireplay.bt.co.uk> References: <7d8em5$m206@overload.lbl.gov> Jonas Boeg=E5rd wrote: > = > Hi, > I am participating in a project where we would like to use > ClearCase for source code version control. I would like to > know if anyone has had any success in using Tornado and > ClearCase together. > = > We are using Tornado version 1.0.1 for solaris. > = > Thanks in advance > Jonas Answer 1: Yeah - WindRiver (according to a conversation I had with some of their guys 2 days ago) Tornado 2 [not got it - demo at above mtg] appears to have features which will enable easy integration with almost ANY code control tools: from what I saw you simply specify the commands to run for stuff like check-in and check out, and then you can use menu options within tornado. (I currently have 1.0.1, but we don't use the host tools very much...) Answer 2: I was involved as a consultant to a large project using a mixture of vxworks (pre-tornado), solaris and DSP code which was about to move = the codebase into Clearcase, and I had a good look at the tool. At the = time I was concerned about the limitation on the number of files that = could be places in individual VOBS (I think that's what they were, anyway = - - Version Object Database or something like that), but otherwise the = basic concepts seemed quite sound. I moved on before the changeover actually took place, so I can't give = any direct experience of using it, but I can't think of any reason why it shouldn't work, providing it is structured correctly to keep the file-count down (of course, this was 2 years ago so things may have moved on since then)... Answer 3: Other alternatives We actually use Teamware (Solaris based) which is basically enhanced SCCS with simple baselining added, as well as developer work-areas. = Previously I have extensively used CMVC (IBM) - probably the best tool I have com across so far, and PCMS (possibly the worst) and Continuus (vying = with PCMS for last place) 3 important factors to consider when choosing code control systems: 1) number of patches/upgrades/fixes which come out PCMS - hundreds - some involving complete database rebuilds CMVC - none - and no major problems with using the tools either 2) training required PCMS - you need an expert to manage the system, keep it running, fix problems,...... + you need to know a lot to drive the tools Clearcase - need an expert, need to learn a fair bit to drive = it properly Continuus - complete mindset-transplant required, loads of complex concepts to take on, incomplete support for what = it is trying to do CMVC - an hour with another user suffices, management = requirements more-or-less non-existant Provides quite a good fix/change-tracking and integration facility as well as basic code control Teamware - simple, doesn't do an awful lot, but easy to start using 3) the developers: what is their first impression? = If they already know AND like a tool, stick with it whatever..... If they tell you it doesn't do the job they need it too - listen to them!!! As you can probably tell, I have been thru this pain several times :-) HTH regards Ian Love, The views expressed above are my own and not those of my employer --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Generate/Catch Interrupt on x86 board? Date: Wed, 24 Mar 1999 11:19:05 +0000 From: Phil Ashby Organization: BT Wireplay Message-ID: <36F8CA29.7134@wireplay.bt.co.uk> References: <7d5ojj$ntr$1@nnrp1.dejanews.com> justiceb@spawar.navy.mil wrote: > Hello there. I have a x86 VME card (Concurrent board) on which I am > just trying to generate and catch an interrupt. based on return codes, > I am successful in connecting an isr to the interrupt via intConnect(), > and generating an interrupt via sysBusIntGen(), but I *never* seem > to land in my isr. sysBusIntGen() generates a VME bus interrupt, *not* a local hardware interrupt. Are you intending to catch interrupts from the VME bus? If so then you will need to route the interrupt through the Universe, the PCI chipset and (finally) the 8259 (which is all sysIntEnable does). If you are interested in interrupts from local PCI devices, remove the Universe chip from the previous list :) I would reccommend (if you can find it) reading the source to a typical hardware device driver. This should help you a lot with interrupt handling. - -- !--------------------------- Phil "Phlash" Ashby ---------------------! ! BT Wireplay - Lead Programmer and general technical trivia supplier ! ! Phone: 01473-644348 WWW: http://www.wireplay.com/ ! ! Snail: B81/G41, BT Labs, Martlesham Heath, Ipswich IP5 3RE, England.! --------------------------- End of New-News digest ********************** From owner-vxwexplo-process Wed Mar 24 04:31:24 1999 From: Leif Thuresson Date: Wed Mar 24 04:31:28 PST 1999 Subject: Re: Telnet Management Console. Status: O the vxWorks Users Group Exploder wrote: > Submitted-by owner-vxwexplo-process Wed Mar 24 02:58:44 1999 > Submitted-by: Rajesh Gautam > > Hi, > I am planning to implement a (character based) telnet management > console for our VxWorks target. > A user should be able to do a telnet to the device and should be able to > view , get / set ( by some commands) various configuration data is some > format. > > -Will it require to implement a shell with commands to be used with > telnetd. > - or a seperate standalone application. > > Either of them will have command parser to parse the input commands and > display routines to output the results in different formats( tables etc.) > > If anybody has implemented similar solution, I would like to know how to > go about it I implemented a very basic http-server to get /set of configuration data on our vxWorks systems. In some environments character based interfaces work best, but if this isn't an absolute condition web-interface might be an option. If your systems have access to disc the pages can be stored on disk and use server side includes to get/set vxWorks variables. You can then use ordinary web-authoring tools to do most of the layout of the pages. Contact me directly if you are interested. --leif --------------------------------------------------------- Leif Thuresson Tel: +46-18-4713860 Uppsala University Fax: +46-18-4713833 The Svedberg Laboratory E-mail: leif.thuresson@tsl.uu.se P.O.Box 533 S-751 21 Uppsala Sweden From owner-vxwexplo-process Wed Mar 24 06:58:39 1999 From: "Andrew Pearce" Date: Wed Mar 24 06:58:42 PST 1999 Subject: Re: Telnet Management Console. Status: O Hi Rajesh, check out my web site http://www.mbari.org/~pean There's a link to the source for a telnet based vxWorks application I wrote a few years ago. It allows a telnet client to contact the vxWorks system on different port numbers and responds with various services and ASCII strings. Andrew Pearce Software Engineer Monterey Bay Aquarium Research Institute From owner-vxwexplo-process Wed Mar 24 07:01:09 1999 From: Kevin Sapp Date: Wed Mar 24 07:01:13 PST 1999 Subject: Shell command from an app. Status: O I would like to give a running application the ability to execute a command using the VxWorks shell. I understand that it is not reentrant... I have tried to use shellOrigStdSet to change the STDIN of the shell to a pipe and a file. And then sent a printf command to the pipe (or had it in a file) but never saw the output. I did not redirect STDOUT or STDERR. Any Ideas ? Can a new session be started and commands passed to it ? Thanks Kevin From owner-vxwexplo-process Wed Mar 24 10:21:41 1999 From: Alistair Adams Date: Wed Mar 24 10:21:45 PST 1999 Subject: Re: HDLC driver for MC68360 (QUICC) Status: O Am I reading from this that you do not get HDLC support as standard with VxWorks for the 68360? I know that for at least one other RTOS, HDLC support is provided as standard. Alistair Adams the vxWorks Users Group Exploder wrote: > Submitted-by owner-vxwexplo-process Tue Mar 23 02:38:57 1999 > Submitted-by: Ilia Bosis > > Take a look at the Aisys inc. http://www.aisysinc.com > They have a driver generator > for the popular uC and RTOS'es. > AFAIK, QUICC & VxWorks included. > > Hope, this helps. > Ilia. > Disclaimer: > The opinions here are mine and not of my employer > Not I, nor my employer haven't linked with Aisys in any way. > > +=====================================================================+ > | Bosis Ilia, Software engineer, Telegate ltd. | > | E-mail: iliab 'at' telegate.co.il | WWW: http://www.telegate.co.il | > +=====================================================================+ > >Hi, > > > >I am working on a project that requires to implement ip over hdlc > >on a MC68360 (QUICC) platform running vxWorks. > >Does anybody know anything about commercial/freeware > >availability of a vxworks hdlc driver > >for MC68360 that can run on any SCC? > > > >Thanks in advance, > >Shad Ansari > > > >Software Engineer, Premisys Communications > >shad.ansari@premisys.com > >613.722.4332x2269 > -- > +=====================================================================+ > | Bosis Ilia | Work phone: 972-3-5384685 | > | Software engineer | Fax: 972-3-5335877 | > | Telegate ltd. | E-mail: iliab at telegate.co.il | > | 60218 Haplada7 Or-Yehuda, Israel | WWW: http://www.telegate.co.il | > +=====================================================================+ > From owner-vxwexplo-process Wed Mar 24 10:31:32 1999 From: "Rasty Slutsker (STX)" Date: Wed Mar 24 10:31:36 PST 1999 Subject: Boot of VxWorks Status: O 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_01BE7624.5D6A32F6 Content-Type: text/plain; charset="ISO-8859-1" Hi, I'm trying to boot compressed VxWorks image via TCP socket. The problem is that we do not have enough RAM to uncompress the file, hence I've created a task which unzips VxWorks and feeds packages to the socket, while bootLoadModule() should read from there and it fails. Does somebody know why loadModule(fd...) and bootLoadModule(fd...) fail when fd is socket? Thanks, Rasty ------_=_NextPart_001_01BE7624.5D6A32F6 Content-Type: text/html; charset="ISO-8859-1" Content-Transfer-Encoding: quoted-printable Boot of VxWorks

Hi,

I'm trying to = boot compressed VxWorks image via TCP = socket. The problem is that we do not = have enough RAM to uncompress the file, hence I've created a task which unzips VxWorks and = feeds packages = to the socket, while bootLoadModule() should read = from there and it fails.

Does somebody know = why loadModule(fd...) and bootLoadModule(fd...) = fail when fd is socket?

Thanks,

Rasty



------_=_NextPart_001_01BE7624.5D6A32F6-- From owner-vxwexplo-process Wed Mar 24 23:34:13 1999 From: Arand Thomas Date: Wed Mar 24 23:34:17 PST 1999 Subject: AW: Tornado and ClearCase Status: O Hi, we are using the same configuration, and it works well. The only crititcal thing I see that you MUST always be aware that the target server has the same view to the file system. That means, it must be startet out of the same view as your tornado launcher. Ciao, Thomas ---------------------------------------------------------------------- Thomas Arand Siemens AG Tel. +49 - 89 - 722 42296 Fax. +49 - 89 - 722 26572 thomas.arand@icn.siemens.de ----Ursprüngliche Nachricht----- Von: vxwexplo@lbl.gov [SMTP:vxwexplo@lbl.gov] Gesendet am: Dienstag, 23. März 1999 16:47 An: vxworks_users@csg.lbl.gov Betreff: Tornado and ClearCase Submitted-by owner-vxwexplo-process Tue Mar 23 07:46:31 1999 Submitted-by: "Jonas Boegård" Hi, I am participating in a project where we would like to use ClearCase for source code version control. I would like to know if anyone has had any success in using Tornado and ClearCase together. We are using Tornado version 1.0.1 for solaris. Thanks in advance Jonas -- _______________________________________________________________ Jonas Boegård Ericsson Radio Access RSA/B/UD email: jonas.boegard@rsa.ericsson.se +46(0)8-585 32139 memo: rsa.rsajboe Borgarfjordsg 14 Kista, Sweden From owner-vxwexplo-process Thu Mar 25 01:52:53 1999 From: Martin Schrape Date: Thu Mar 25 01:52:56 PST 1999 Subject: Re: Telnet Management Console. Status: O Hallo Gautam, ask your FAE to give you the source code of the telnetd. -martin --------------------------------------------------------------- Dipl.-Phys. Martin Schrape schrape@atomika.com Senior Software Engineer Tel. +49 89 315 891 34 Atomika Instruments GmbH Fax +49 89 315 59 21 Bruckmannring 40 http://www.atomika.com 85764 Oberschleissheim/Munich, Germany --------------------------------------------------------------- Rajesh Gautam wrote: > > Hi, > I am planning to implement a (character based) telnet management > console for our VxWorks target. > A user should be able to do a telnet to the device and should be able to > view , get / set ( by some commands) various configuration data is some > format. > > -Will it require to implement a shell with commands to be used with > telnetd. > - or a seperate standalone application. > > Either of them will have command parser to parse the input commands and > display routines to output the results in different formats( tables etc.) > > If anybody has implemented similar solution, I would like to know how to > go about it > > Thanks, > gautam. > > -- > -------------------------------- > Network Machines, > 311, Ganga Shopping Complex, > Noida, > INDIA, > 201301. > > Ph# 91-575-544783, 548517, 548518. > Fax# 91-575-543721. From owner-vxwexplo-process Thu Mar 25 02:33:03 1999 From: David Laight Date: Thu Mar 25 02:33:07 PST 1999 Subject: Re: Shell command from an app. Status: O > Subject: Shell command from an app. > > I would like to give a running application the ability to execute > a command using the VxWorks shell. I understand that it is not > reentrant... > > Any Ideas ? > How about modifying the console async driver so that you can insert characters into the keyboard stream. It's probably worth starting with a ^U to remove any junk the user has typed. David > Can a new session be started and commands passed to it ? > > Thanks > Kevin > ---------------------------------------------------------------- David Laight email: dsl@tadpole.co.uk Tadpole Technology plc phone: +44 1223 278 256 Cambridge, UK fax: +44 1223 278 201 From owner-vxwexplo-process Thu Mar 25 02:49:24 1999 From: "Moy, Ian" Date: Thu Mar 25 02:49:28 PST 1999 Subject: Ethernet Driver. Status: O Hi, As part of our project we need an Ethernet driver for VxWorks. The NIC is a Profive Ethernet-100 based on Intel's 82558 controller. We are using the MBX860 BSP and would appreciate any help with developing a driver. For example, does the BSP have functions we can use to do: * reading and writing from/to the PCI bus * DMA transfers * interrupt mapping * memory mapping Any help or examples would be great, Thanks, Ian. > -------------------------------------------------- > Ian Moy, Roke Manor Research Ltd > (tel) +44 (0)1794 83 3659 > (email) ian.moy@roke.co.uk > -------------------------------------------------- > > From owner-vxwexplo-process Thu Mar 25 04:01:41 1999 From: daemon@csg.lbl.gov Date: Thu Mar 25 04:01:45 PST 1999 Subject: comp.os.vxworks newsdigest Status: O Comp.Os.Vxworks Daily Digest Thu Mar 25 04:01:38 PST 1999 Subject: Re: How to delete a device listed by devs() command Subject: Simple Database Software For VxWorks Subject: Re: about signal facility... Subject: Atmel 40400 and 'interrupt driver' Subject: Re: 100 BaseT PCI card Subject: Re: MCP750 and cPCI ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: How to delete a device listed by devs() command Date: 24 Mar 1999 13:03:46 GMT From: chc@nasa2.ksc.nasa.gov (Charles H. Chapman) Organization: NASA, Kennedy Space Center Message-ID: References: <36F7B936.45D17934@stest.ch> On Tue, 23 Mar 1999 16:54:30 +0100, Thomas Kowatsch wrote: >For some reason I cannot use another device name as "/sd0:" what may >help it to work. So my question again: how can I get rid of the >existing "/sd0:" before call to dosFsDevInit(). You can use iosDevDelete to delete a device from the I/O system. iosDevDelete requires a pointer to the device's structure. You can use iosDevFind to obtain that (if you don't have it already). Chuck --------------------------- Newsgroups: comp.os.vxworks,comp.arch.embedded Subject: Simple Database Software For VxWorks Date: Wed, 24 Mar 1999 09:41:57 -0500 From: Dave Leney Organization: Kronos Incorporated Message-ID: <36F8F9B5.287DEB53@kronos.com> We are looking for a simple, flash-friendly, inexpensive database or other type of indexed storage software which is available for VxWorks. Our project is using an MPC860T with an M-Systems disk-on-a-chip flash device for persistant storage. The VxWorks dosFS will be thr primary interface to the flash device. We need to store both fixed and variable length data records in flash, allowing for fast retrieval and for updating which includes insertion, replacement and deletion. However, we don't want soething which will cause un-necessary wear on the flash device. Thanks in advance for any responses --------------------------- Newsgroups: comp.os.vxworks Subject: Re: about signal facility... Date: Wed, 24 Mar 1999 15:07:58 +0000 From: Ian Love Organization: BT Labs - but speaking only for myself Message-ID: <36F8FFCE.2FFF@wireplay.bt.co.uk> References: <7dad1n$6v9$1@news2.kornet.net> =C0=B1=B0=AD=BD=C4 wrote: > = > Hi! I'm Kang-Sik Yoon, Korea. > = > I am programming signal facility for tasks communication below: : : > As you see above my codes, we have four tasks, two(task1 and task2 ) = >of them are always ready status, the rest (task3 and task4)are blocked. = = >If signal is generated in task1 with kill(tid3, SIGUSR1), is myISR1() = >called without including myISR1() in task3() ? or should we insert = >myISR1() to task3()? > And also, how is control flow? Not sure I understand your problem, but I'll have a go at 2 alternate versions of what you might be asking: you wish, within the context of task1, to generate a signal directed at task3. This signal should invoke, in the context of task3, the routine myISR1(). If that is what you are trying to do, then you DONT need any calls to myISR1() or myISR2() anywhere - the whole point is that when the signal is received they will run - regardless of where in the execution flow the task is..... (same as UNIX) I am wondering if what you are trying to do here is semaphores by signals: If you are expecting task3() to get to the = myISR1() call [i.e. there is an explicit call in your task3() code and = then wait for the signal to arrive before doing it (you mention the task = is blocked?), then what you might really want is a binary semaphore: in task3, when you want task1 to proceed, semGive() = in task1, when you want to wait for task3, do semTake() In either case this isn't really an ISR: an ISR runs outside the context of any task (eg device driver interrupts). A function run ass the result of a signal raised to a task is a 'signal handler routine' (VxWorks 5.3.1 Programmers Guide, P90): "The task being signaled immediately suspends its current thread = of execution and the task-specified signal handler routine is executed the next time the task is scheduled to run. Note that the signal handler gets invoked even if the task is blocked" It also goes on to say: "Signals are more approriate for error and exception handling than as a general purpose intertask communication mechanism" I have not experimented with this, but the above quote implies that each task spawned must setup it's own signal handlers (ie you cant do it = 'globally' as in your code) - each task must call signal() on startup? I hope I haven't confused things: get back to me if you need any = expansion of any of the above. regards Ian Love, The views expressed above are my own and not those of my employer --------------------------- Newsgroups: comp.os.vxworks Subject: Atmel 40400 and 'interrupt driver' Date: Wed, 24 Mar 1999 19:05:47 GMT From: Bill Pringlemeir Message-ID: Sender: bpringle@DeadDuck I am attempting to get an 'interrupt driver' working with vxWorks and the Atmel 40400 ARM processor. I am recieving control to the 'intVecChk' function. However, I do not know what vector to pass back to vxWorks. Should I hook the following function, _func_intConnectRtn - Connect the system call intConnect to our connect routine _func_intEnableRtn, _func_intDisableRtn - Connect interrupt enable and disable These hook functions are undocumented. I don't know if it is wise to use them. I used the ARM7T BSP amba interrupt controller code to try and create something that will work for the atmel 40400. I think that I should be returning the routine that was sent in intConnect(). The AMBA code just mentions that you should return a 'vector'. I don't think they mean the interrupt code @0, but the routine to handle the hardware (written in 'C'). Thanks for any illumination, Bill Pringlemeir - -- Brave New Waves, "http://www.radio.cbc.ca/programs/bnw/" --------------------------- Newsgroups: comp.os.vxworks Subject: Re: 100 BaseT PCI card Date: Mon, 22 Mar 1999 09:55:51 +0200 From: Alef0 Organization: NetVision Israel Message-ID: <36F5F787.A535F237@127.0.0.1> References: <7c96b3$j21$1@nnrp1.dejanews.com> bijtaj@my-dejanews.com wrote: > Hi, > > I am looking for 100 BaseT PCI cards with VxWorks drivers. This is for a > Intel- Pentium architecture using the pc486 BSP. > > I understand the if_fei driver may support the Intel EtherExpress Pro 100+ > card. Is that true and if so what is your experience with this card in terms > of reliability under VxWorks? We're using this card, and we're very satisfied with it. Using it under VxWorks was fairly painless. BTW, like most other 100BaseTX cards, it's auto-sensing 10Mbit / 100Mbit. - -- Alef0 - --- Replace 127.0.0.1 with usa.net when replying by e-mail C programmers never die, they are just cast into void. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: MCP750 and cPCI Date: Wed, 24 Mar 1999 21:30:03 -0500 From: "Tim Stanka" Organization: EarthLink Network, Inc. Message-ID: <7dc728$ba$1@ash.prod.itd.earthlink.net> References: <7daub7$p99$1@front1.grolier.fr> The BSP for the MCP750 is based on the 2604 BSP. I also have been having problems with the BSP. I received my BSP on 12/98 and initially cPCI autoconfig did not work. They sent me a patch which did get it working . . . somewhat. I was able to use the "show" routines to look at and configure devices on the various PCI busses. I have noticed several problems with the BSP and PCI: 1) The BSP set the subordinate bus number for each and every bridge to 0xff. This is a problem if you have two bridges off the same PCI segment. 2) When I do pciHeaderShow() for a single function device I get config data for all 8 functions !!!! 3) My BAR values do not look correct. 4) Bridges seem to be set up with limit register values less that base so only transactions from the secondary bus are forwarded to primary. 5) BSP programs RAVEN for PREP memory map which yields 960Mb of address space. This is not enough for me. I don't know why they allocate 1Gb for PCI I/O space when each device.function is supposed to request 256 bytes max. I don't know if you guys are aware of this but all devices on PCI bus 0 (local PCI) are on an exclusion list and are not configured with the autoconfig routine. I believe they did this so they could minimize changes when they ported the 2604 BSP which had #defines for those device BARS. Anyway, I'm probably in the same boat as you guys. We will be putting several custom hardware boards in our system and we need a reliable PCI autoconfig routine. I would like to use the stuff that comes with the BSP but I think it is going to take awhile to get the bugs out. Also the routines seem more complex than they should be. I printed out the source code for the various files and it was 3/4" thick!!! I'm trying to get the problems fixed through WRS and Motorola. The Motorola real time support group actually wrote the MCP750 BSP. Let me know if you have any luck in getting PCI config to work. By the way. You had mentioned that PCI_MEM_SIZE was set to 16Mb. I was told that this value could be increased but for every 128 bytes it increased you would incur 1 byte overhead for MMU. So for 1G of PCI memory space you would lose 10Mb of DRAM for MMU overhead. I haven't looked into this yet I'm only repeating what the Motorola Computer Group told me. Good Luck, Tim Stanka stanka@kodak.com --------------------------- End of New-News digest ********************** From owner-vxwexplo-process Thu Mar 25 04:03:00 1999 From: Michael Bowring Date: Thu Mar 25 04:03:04 PST 1999 Subject: Re: Shell command from an app. Status: O the vxWorks Users Group Exploder wrote: > I would like to give a running application the ability to execute > a command using the VxWorks shell. I understand that it is not > reentrant... > > I have tried to use shellOrigStdSet to change the STDIN of the > shell to a pipe and a file. And then sent a printf command to > the pipe (or had it in a file) but never saw the output. I did > not redirect STDOUT or STDERR. > The shell can accept input from a pipe but it appears to expect the far end to write input one character at time. If you just do: write (pipeFd, "memShow\n", 8) the shell will only see the m, whereas if you send the 8 chars individually the shell is happy. I think this is because the shell does an isatty() on its OrigFd - being a pipe isatty() will return FALSE, in which case I suspect the shell assumes you are trying to talk to it with a 1960s teletype or something!. It also echos each character you send it back to whatever you have defined STDOOUT as (this is not the normal behaviour when interacting with a proper tty/pty device). Regards, Mike From owner-vxwexplo-process Thu Mar 25 07:17:55 1999 From: "Legoas, Marco A (FL51)" Date: Thu Mar 25 07:17:59 PST 1999 Subject: fft algorithms for PowerPC Status: O VxWorks, Tornado Does anyone know of a "DSP like" optimized math library (FFT algorithms, etc.) for the PowerPC family of CPU's? Any leads will be appreciated Marco A. Legoas malegoas@space.honeywell.com From owner-vxwexplo-process Thu Mar 25 11:56:02 1999 From: David Laight Date: Thu Mar 25 11:56:06 PST 1999 Subject: Re: Amtel 40400 and 'interrupt driver' Status: O I'm not sure what's in the ARM7T BSP, but the SA1100 strongarm (brutus BSP) seems similar. intVecChk is expected to return the vector number (used to index the array of function pointers populated by calls to intConnect()). You have free choice of the vector number - provided intVecChk and the intConnect caller agree (and the number is smaller than the declared interrupt vector table size). The strongarm interrupt code was, if not exactly broken, not a desirable piece of code. The strongarm fails to meet the model of a 68000 on a VmeBus so completely that I decided to redefine what interrupt 'number', 'vector' and 'level' meant. I have some 128 separately maskable possible interrupts (without considering the separate interrupt sources in some peripherals like the uarts). There are 53 possible general purpose io ports any of which could have an IRQ line connected. I don't want the uart driver (for example) to know HOW its interrupt is connected - it just wants to enable it. Some interrupts are inherantly higher priority than others, but two or three true interrupt 'levels' is more than enough. I can't allow 128 levels - the system would blow its interrupt stack if they all occurred in the 'correct' sequence. Also many of the interrupt sources are edge detects with software selectable edges - the driver knows which edge is required but doesn't want to know how to select it. This led me to define the argument to intEnable as: EDGE_RISING << 20 | EDGE_FALLING << 21 | priority << 16 | vector the system guesses an edge if one isn't supplied. priority of 0 is the lowest, 1-3 also exist but 3 is illegal for an isr as it would not be maskable. The only difficultly left is cascaded interrupts. GPIO 11-27 are cascaded through irq 11. The SA1101 cascaded 64 interrupts through a single SA1100 pin (and it is one of GPIO 11-27 on the MESA development board). The UCB1200 has 15 interrupt on one line and a horrid method of finding the active one. I didn't want to complicate intVecChk - so it causes an interrupt cascade routine to be called. Unfortunately the vector table base (it's malloc()ed) isn't in a global variable making the soft call of an ISR a little tricky! (Actually it is useful to be able to soft call an ISR (by number) anyway.) Enough rambling, back to getting the processor out of sleep... David > Subject: Atmel 40400 and 'interrupt driver' > From: Bill Pringlemeir > > I am attempting to get an 'interrupt driver' working with vxWorks and > the Atmel 40400 ARM processor. I am recieving control to the 'intVecChk' > function. However, I do not know what vector to pass back to vxWorks. > Should I hook the following function, > > _func_intConnectRtn - Connect the system call intConnect to our connect routine > _func_intEnableRtn, _func_intDisableRtn - Connect interrupt enable and disable > > These hook functions are undocumented. I don't know if it is wise to use > them. I used the ARM7T BSP amba interrupt controller code to try and create > something that will work for the atmel 40400. I think that I should be > returning the routine that was sent in intConnect(). > > The AMBA code just mentions that you should return a 'vector'. I don't > think they mean the interrupt code @0, but the routine to handle the > hardware (written in 'C'). > > Thanks for any illumination, > Bill Pringlemeir ---------------------------------------------------------------- David Laight email: dsl@tadpole.co.uk Tadpole Technology plc phone: +44 1223 278 256 Cambridge, UK fax: +44 1223 278 201 From owner-vxwexplo-process Fri Mar 26 04:00:14 1999 From: daemon@csg.lbl.gov Date: Fri Mar 26 04:00:19 PST 1999 Subject: comp.os.vxworks newsdigest Status: O Comp.Os.Vxworks Daily Digest Fri Mar 26 04:00:11 PST 1999 Subject: Select and Sockets Subject: Raster Graphics RG-101 PMC Color Graphics Board ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Select and Sockets Date: Thu, 25 Mar 1999 14:56:11 GMT From: eezekiel@my-dejanews.com Organization: Deja News - The Leader in Internet Discussion Message-ID: <7ddiq5$g7q$1@nnrp1.dejanews.com> We recently found out that the VxWorks select() cannot handle a file descriptor with a value >255. This is also confirmed on the WRS web site. We are planning on using >1000 sockets, and each socket consumes a file handle. I believe that this could cause applications (such as ftp) that do use select() to fail. Does anyone out there use more than 256 sockets/file descriptors and also use select()? If so, how did you get around this problem? The only way I can see of overcoming this problem is to avoid using select() completely. Thanks, Ephraim Ezekiel eezekiel@quarrytech.com - -----------== Posted via Deja News, The Discussion Network ==---------- http://www.dejanews.com/ Search, Read, Discuss, or Start Your Own --------------------------- Newsgroups: comp.os.vxworks Subject: Raster Graphics RG-101 PMC Color Graphics Board Date: Tue, 23 Mar 1999 08:42:52 -0800 From: "Leonard J. Reder" Organization: Jet Propulsion Laboratory Message-ID: <36F7C48C.F89CE85A@huey.jpl.nasa.gov> This is a multi-part message in MIME format. - --------------4FA4FC81B341F0294DCCD19C Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Hi, Is anyone using a Raster Graphics RG-101 PMC Color Graphics Board with a Motorola MVME-2307 PPC 604 board running VxWorks version 5.3.1. I am trying to run the raster graphics demo programs and the VGA monitor just lights up with a noisy lines pattern. I think something in the kernel may not be configured correctly. If someone has made this configuration work could you send me instructions how to make this work. Possibly send me a ~/mv2304.531.shell/config/mv2304/mv2600.h file with the proper configuration. I modified this file per the manufacturer and it does not seem to work. The BSP I am using for the MVME-2307 is #define BSP_VER_1_1 1 #define BSP_VERSION "1.1" #define BSP_REV "/4" Also if anyone has any suggestions for low cost frame buffer/graphics display boards that have VxWorks software support I would be interested. What I am trying to do is grab gray scale images from an NTSC camera and possibly overlay a measurement scale on top of the image. The Raster Graphics PMC card seems ideal for this and does not use a VME slot but rather is a PCI card which fits on the MVME-2307. Any replies would be very much appreciated. Thanks, Len - --------------4FA4FC81B341F0294DCCD19C Content-Type: text/x-vcard; charset=us-ascii; name="reder.vcf" Content-Transfer-Encoding: 7bit Content-Description: Card for Leonard J. Reder Content-Disposition: attachment; filename="reder.vcf" begin:vcard n:Reder;Leonard J. tel;fax:818-393-4357 tel;work:818-354-3639 x-mozilla-html:FALSE url:http://huey.jpl.nasa.gov/~reder org:Jet Propulsion Laboratory;Interferometry Systems and Technology Section 383 version:2.1 email;internet:reder@huey.jpl.nasa.gov title:Sr. Software Engineer adr;quoted-printable:;;4800 Oak Grove Dr.=0D=0A;Pasadena;CA.;91109;USA fn:Leonard J. Reder end:vcard - --------------4FA4FC81B341F0294DCCD19C-- --------------------------- End of New-News digest ********************** From owner-vxwexplo-process Fri Mar 26 08:44:43 1999 From: "Thomas Keith Buchanan" Date: Fri Mar 26 08:44:46 PST 1999 Subject: IDE problems Status: O I'm having problems with a solid-state IDE drive and VxWorks 5.1.1. Older versions of this drive have worked in the field for years. A newere version has intemittent failures loading the operating system. I suspect there is a timing issue with the IDE driver or an operating mode problem. The newer drive is actually an ATA drive operating in PIO mode 1. Unfortunately the ATA driver was not available way back when. I also cannot change the PIO mode from the BIOS on SBC running VxWorks. Any ideas, suggestions, relevant experiences? TIA, Keith Buchanan SPARTA, Inc. From owner-vxwexplo-process Fri Mar 26 10:10:19 1999 From: Ian Willlats Date: Fri Mar 26 10:10:23 PST 1999 Subject: Select and Sockets Status: O Ephraim Ezekiel wrote: > We recently found out that the VxWorks select() cannot handle a file > descriptor with a value >255. This is also confirmed on the WRS web site. > This is correct. It depends on the value of FD_SETSIZE (256) which is compiled into several parts of the operating system, most obviously selectLib.c. > Does anyone out there use more than 256 sockets/file descriptors and also use > select()? If so, how did you get around this problem? > > The only way I can see of overcoming this problem is to avoid using select() > completely. Well, that's certainly not a bad idea if you're not committed to select(). The VxWorks implementation of select() has some problems, not least of which is that it will tend to fragment memory if you use it with any frequency. However, there are some alternatives. Here are a couple I've used, on a project which needed up to 4096 file descriptors: 1) Use special versions of the relevant OS object files, built for us by WRS with our specified values of FD_SETSIZE. 2) Obtain source for selectLib.c and modify it to work with an unbounded FD_SETSIZE - in fact, defined for each call by the argument. I wouldn't recommend option (1) as it is a nuisance (maintenance-wise) for all concerned. I doubt that WRS would want to do this anyway. Option (2) works for us but it does require you to buy the selectLib.c source code. It's a very small amount of work (less than a day) once you get the source. Both these options preserve the standard calling interface to select(). Another alternative I considered was to write a special version of select() with a different API, not based on the fd_set data structure, which was better-suited to selecting on a small number of high-valued fds. One thing that has always concerned me about allowing more than 256 file descriptors is what happens if one of VxWorks' built-in tasks somehow gets an fd > 255 and tries to use it with select()? This would in principle cause problems with option (2), because the other parts of VxWorks apart from selectLib would still have a fixed FD_SETSIZE of 256, so it would still not be possible for them to select on fd's outside the range [0,255]. (It would not be a problem with option (1)). However, in practice this problem does not seem to arise - presumably because any such fds are opened when the system starts up and therefore tend to be low-numbered. I hope this ramble is of some use - by all means contact me directly if I can offer any more help. Regards, Ian --------------------------------------------------------------- Ian Willats e-mail: mailto:ihw@rtp.co.uk Real-Time Products Ltd. direct: +44 (0) 121 234 6637 Chancery House, tel: +44 (0) 121 234 6600 8 Edward Street, Birmingham. fax: +44 (0) 121 234 6611 B1 2RX. England. web: http://www.rtp.co.uk --------------------------------------------------------------- From owner-vxwexplo-process Fri Mar 26 17:11:41 1999 From: piyush agarwal Date: Fri Mar 26 17:11:45 PST 1999 Subject: Interface failure indication Status: O I am writing OSPF code for VxWorks. It shall run on cards connected by Ethernet. OSPF needs to know when its ethernet interface goes down. RFC says that this information is given by lower layer. Is there any way to hook up the callback function to ethernet driver or IP so that it is invoked on interface state changes. Or is there some other way to get this information (preferably) asynchronously. Thanks, Piyush From owner-vxwexplo-process Sat Mar 27 04:01:18 1999 From: daemon@csg.lbl.gov Date: Sat Mar 27 04:01:22 PST 1999 Subject: comp.os.vxworks newsdigest Status: O Comp.Os.Vxworks Daily Digest Sat Mar 27 04:01:15 PST 1999 Subject: Re: Disk Size Limitation for dosFs Question Subject: Systemclock Subject: Re: x86 and PCI Ethernet Subject: Re: Sockets in serial interrupt handler Subject: Re: How to draw dynamic graphic using Wind Web Server Subject: Re: OSE vs VxWorks ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: Disk Size Limitation for dosFs Question Date: Thu, 25 Mar 1999 08:37:43 GMT From: bobarmstrong@home.net (Robert Armstrong) Organization: @Home Network Message-ID: References: <36F7457E.25340BBD@siemens.at> <36F7BBB3.2A1E@dip3.ee.uct.ac.za> I believe the current limit is 8Gb. I believe lseek() is still limited to 2Gb for a single file though. Bob In article <36F7BBB3.2A1E@dip3.ee.uct.ac.za>, Trevor Pope wrote: >Andreas Lugbauer wrote: >> >> Hi, >> >> What is the limitation of the disk size for vxWorks dosFs file system? >> >> We realised that the ioctl function with function identifier FIONFREE >> returns the number of free bytes as 32 bit integer variable. >> This means the maximum number is 2 147 483 647. Is this only a >> limitation of the ioctl function or is there more behind? >> >> many thanks in advance >> >> Andreas >The maximum size is 4 gigs - the type of the size returned is a >32 bit unsigned int. You can handle larger discs, but then you will >have to code an alternative to the ioctl() to get the disc space (as >of about 1997 - this may have changed subsequently). We hit this limit >in about 1994, so that is how I know the answers. We haven't >continued with development in Tornado as we are looking at alternative >platforms for our next generation. Wind River may have mapped a >future for dosFs that I am unaware of, it certainly needs improvement. --------------------------- Newsgroups: comp.os.vxworks Subject: Systemclock Date: Fri, 26 Mar 1999 19:00:28 +0100 From: Armin Schmidle Organization: personal Message-ID: <36FBCB3B.8D6813EB@swol.de> How can I set the system clock on a vxWorks target ? A small code example could be helpful. Thanks in advance !!! Armin --------------------------- Newsgroups: comp.os.vxworks Subject: Re: x86 and PCI Ethernet Date: Thu, 25 Mar 1999 09:05:02 GMT From: bobarmstrong@home.net (Robert Armstrong) Organization: @Home Network Message-ID: <2%mK2.25171$FZ5.7922@news.rdc1.sfba.home.com> References: <36F8C09D.27BAFBAD@de.bosch.com> The latest I hear from WRS and Intel is that the 82559 will work if you treat it just like an 82557. In addition I have seen the dec21x4xEnd used. I hear there will be more PCI cards supported in Tornado 2 but I don't know any more than that. Bob In article <36F8C09D.27BAFBAD@de.bosch.com>, Joachim Zeller wrote: >Hi, >I start to look for any solution available to combine a x86 (Pentium >Board) with any PCI >Ethernet controller (all best fit would be a 10/100 card). The only >thing I found was the fei network interface. This interface is used with >82557 from Intel, this chip is not longer >produced, but changed to 82558 which should be available. Did anyone >know anything >about other PCI chips available for VxWorks ? Has anyone experience with >the integration of PCI Ethernet controllers ? I would be happy for any >response. >Thanks a lot! > > > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Sockets in serial interrupt handler Date: Thu, 25 Mar 1999 09:11:24 GMT From: bobarmstrong@home.net (Robert Armstrong) Organization: @Home Network Keywords: sockets, serial, interrupt Message-ID: <05nK2.25180$FZ5.7922@news.rdc1.sfba.home.com> References: <7d6095$gar$1@news.csita.unige.it> Just to add my weight to the ring buffer method, with a single source and single consumer you just don't need the complexity of the message queue. Just receive your characters, add them to the ring buffer and give a semaphore to a task which sends your data. I know the documentation covers this in a table but the general rule is that the ISRs cannot call anything which is or may potentially be blocking... an ISR can't wait on a task. Bob In article <7d6095$gar$1@news.csita.unige.it>, gabry@ian.ge.cnr.it wrote: >Hi all, > I'm writing a serial "repeater", i.e. a program having the task to >read a serial input and send it through a socket. I though to call the >socket's write(...) function directly in the serial interrupt handler every >time I receive the received char's interrupt. In your opinion, could this be a >dangerous approach? >Another approach could be the following: the interrupt handler when it >receives a received char's interrupt it before fills a queue and then it >unblocks a semaphore which signals that there are characters availables in the >buffer. Then, there is another task blocked on the same semaphore which, when >it unblocks, it reads the characters in the queue and sends them through the >socket. But I think this second solution could be very inefficient. Do you >have any suggestions? > >Many thanks in advance for your attention >Gabriele Bruzzone > >+---------------------------------------------------------------+ >| Gabriele BRUZZONE | >| Research Scientist of Robotics Department | >| C.N.R. - Istituto per l'Automazione Navale | >| Via De Marini, 6 Direct +(39)-010-6475657 | >| 16149-GENOVA-ITALY Fax +(39)-010-6475600 | >| mailto:gabry@ian.ge.cnr.it http://www.ian.ge.cnr.it/robotics | >+---------------------------------------------------------------+ --------------------------- Newsgroups: comp.os.vxworks Subject: Re: How to draw dynamic graphic using Wind Web Server Date: Fri, 26 Mar 1999 22:49:43 -0800 From: Chuck Hardy Organization: SBC Internet Services Message-ID: <36FC7F87.B4EA1DF3@pacbell.net> References: <36FB5373.F355A580@ms.chttl.com.tw> Reply-To: crhardy@pacbell.net You might want to consider using Java's RMI mechanism. It provides a simple, although not particularly fast, means to connect the device to an applet or application that can display the status information. Using RMI, you can easily send periodic updates to the display app. Wind River does have a Java virtual machine as part of their "Tornado for Java" product. A stripped down version called "Personal Java" is due out soon, and there will be an even lighter "Embedded Java" version a little later. "¼B¤_æÐ" wrote: > I have to design web-based management program. > It's purpose is to monitor the status of a device and > needs to be real-time updated by the web server > according to the status of the device. > Can any body give me some advices? > And, is any virtual java machine which can added > into the vxworks exist? > > Thank you!! > > yu-chen Liu --------------------------- Newsgroups: comp.realtime,comp.arch.embedded,comp.os.vxworks Subject: Re: OSE vs VxWorks Date: Tue, 23 Mar 1999 13:21:23 -0500 From: "J. Marshall" Organization: Posted via RemarQ Communities, Inc. Message-ID: <922213261.573.31@news.remarQ.com> References: I would recommend looking at LynxOS as well. Just my two cents. JM Johan Borkhuis wrote in message ... >Hello > >We are in the process of selecting a new operating system, and at the moment >we are focussing on OSE and VxWorks. We already have some VxWorks knowledge >but no OSE experience. >Does anyone have experience with OSE: how is the support, with respect to >problems, new (or older) device-drivers, new hardware and BSP's. >And how does OSE compare to VxWorks, based on the kernel itself (configuring, >runtime, stability) and on the development environment (compilers, debugging, >availability of tools, etc). > >Thanks for your time. > >Groeten, > Johan > >-- > o o o o o o o . . . _____J_o_h_a_n___B_o_r_k_h_u_i_s___ > o _____ || http://www.xs4all.nl/~borkhuis | > .][__n_n_|DD[ ====_____ | johan@borksoft.xs4all.nl | > >(________|__|_[_________]_|________________________________| > _/oo OOOOO oo` ooo ooo 'o!o!o o!o!o` > --------------------------- End of New-News digest ********************** From owner-vxwexplo-process Sat Mar 27 10:15:35 1999 From: David Beberman Date: Sat Mar 27 10:15:39 PST 1999 Subject: Looking for C++ class libraries for VxWorks Status: O I've been reading this mail list for a while, and checked the WindRiver website. Is there a listing of C++ class libraries for VxWorks, beyond the template libraries and the Vx wrapper classes offered by WindRiver? Our application is written strictly in C++ using GCC. (and for those "in the know" about GCC and templates, yes, we had our instantiation problems also. GCC can be very tricky about both templates, and inline'd methods) If there is a FAQ, or other listing that I've missed, a pointer would be appreciated. Regards David Beberman Logic Solutions, Inc. 104 Otis. St. Ste.29 Northboro, MA. 01532 508 351 9492 lsi@world.std.com From owner-vxwexplo-process Sat Mar 27 22:48:17 1999 From: Leonid Rosenboim Date: Sat Mar 27 22:48:22 PST 1999 Subject: Re: Select and Sockets Status: O Ephraim Ezekiel wrote: > We recently found out that the VxWorks select() cannot handle a file > descriptor with a value >255. This is also confirmed on the WRS web site. > Ephraim, it is been known for a long time that select has its limitation of handling a large number of file descriptors, and it has some other problems too. I recommend that you forget about select() altogether, and the file descriptor number limitation shall not be the only reason for that... You can in many cases just have a task block on a socket, thereby having as many tasks on your system as there are active connections. This does cost you memory, but is positively the most efficient mechanism to handle socket traffic, and does not employ any proprietary API. The other alternative is to use an undocumented "socket wakeup hook" which can activate your own callback when data arrives at a socket, or when socket buffer becomes available for write(). Unfortunately, I am tied by non-disclosure agreements with WRS, thus I can not elaborate on this. I have however left a considerable amount of information on this issue on the WRS support knowledge base, so if you call support they can easily get you the details. Leonid ex-WRS-er From owner-vxwexplo-process Sun Mar 28 02:55:35 1999 From: "G. Sudarshan" Date: Sun Mar 28 02:55:39 PST 1999 Subject: Re: Interface failure indication Status: O > I am writing OSPF code for VxWorks. It shall run on cards connected by > Ethernet. OSPF needs to know when its ethernet interface goes down. RFC > says that this information is given by lower layer. Is there any way to > hook up the callback function to ethernet driver or IP so that it is > invoked on interface state changes. Or is there some other way to get > this information (preferably) asynchronously. > In BSD OS we used routing socket messages to learn interface and kernel forwarding table changes. If VxWorks stack is similar to BSD like implementation then you may be able to have routing socket to receive such asynchronous information. Thanks, Sudarshan From owner-vxwexplo-process Sun Mar 28 21:00:47 1999 From: vasu@teil.soft.net (Vasudev Mulchandani) Date: Sun Mar 28 21:00:51 PST 1999 Subject: Problem with ULIP Status: O Hello ! I have a Solaris 2.6 installation and had to move it to Solaris 2.6 as our Solaris 2.5 crashed and we cannot locate the CD. When I installed VxWorks on it.. and then did a installUlipSolaris... I rebooted the system then.. but ULIP services fail to come up and the erro message comes up as SIOCSIFDSTADDR: IFF_POINTOPOINT not set Any idea as to why this happens ?? Thanx & Regards Vasu From owner-vxwexplo-process Sun Mar 28 23:13:41 1999 From: vasu Date: Sun Mar 28 23:13:45 PST 1999 Subject: ULIP related query Status: O Hello Everybody Ran into a problem while installing ULIP with my VxWorks. I installed VxWorks on Solaris 2.6 and then rebooted the system. ULIP services didnot comeup and it gave an error SIOCSIFDSTADDR: IFF_POINTOPOINT not set Couldnot decipher as to why this happens.. because of this launch doesnot give me VxSim.. HELP !! --vasu From owner-vxwexplo-process Mon Mar 29 01:19:33 1999 From: Francis Sreeves Date: Mon Mar 29 01:19:37 PST 1999 Subject: RE: Problem with ULIP Status: O On Monday, March 29, 1999 6:01 AM, the vxWorks Users Group Exploder [SMTP:vxwexplo@lbl.gov] wrote: > Submitted-by owner-vxwexplo-process Sun Mar 28 21:00:47 1999 > Submitted-by: vasu@teil.soft.net (Vasudev Mulchandani) > > Hello ! > > > I have a Solaris 2.6 installation and had to move it to Solaris 2.6 as > our Solaris 2.5 crashed and we cannot locate the CD. > > When I installed VxWorks on it.. and then did a installUlipSolaris... > I rebooted the system then.. but ULIP services fail to come up and > the erro message comes up as > > SIOCSIFDSTADDR: IFF_POINTOPOINT not set > > Any idea as to why this happens ?? > > Thanx & Regards > > Vasu For what it's worth, I've also seen the above messages (one for each service) when installing WRS' ULIP implementation on both Solaris 2.4 and 2.6. However, the ULIP services start and operate as expected apart from that. You could check whether this is the case for you. Regards, Francis -- Francis Sreeves tel: +44 (0)121 234 6600 Real Time Products Ltd. fax: +44 (0)121 234 6611 Chancery House, 8 Edward Street, email: mailto:frs@rtp.co.uk Birmingham B1 2RX website: http://www.rtp.co.uk From owner-vxwexplo-process Mon Mar 29 04:00:56 1999 From: daemon@csg.lbl.gov Date: Mon Mar 29 04:00:59 PST 1999 Subject: comp.os.vxworks newsdigest Status: O Comp.Os.Vxworks Daily Digest Mon Mar 29 04:00:51 PST 1999 Subject: Re: Simple Database Software For VxWorks Subject: Re: Shell command from an app. ------------------------------------------------------- Newsgroups: comp.os.vxworks,comp.arch.embedded Subject: Re: Simple Database Software For VxWorks Date: Wed, 24 Mar 1999 17:49:19 -0700 From: "Gene Gajewski" Organization: Rt66.COM, New Mexico's #1 ISP Message-ID: <7dc17e$o7o$1@news.rt66.com> References: <36F8F9B5.287DEB53@kronos.com> Mix Software has a product called the C Database ToolChest. I used this product many years ago. It's been updated - but I remember the architecture very well. It should be embeddable with little work, and the price is $49, no more than a book. Sneak it in on your expense report...... http://www.mixsoftware.com/ It implements a variable length record system, but uses an underlying fixed block pool manager. Essentially, it place as many records as can fit into a block, the records still being of variable length themselves. Same for the indexing btrees Very fast ISAM system. It will attempt to reuse holes within blocks created by record deletions. Should be able to make it lazy write too - further optimizing it. Dave Leney wrote in message news:36F8F9B5.287DEB53@kronos.com... > We are looking for a simple, flash-friendly, inexpensive database or > other type of indexed storage software which is available for VxWorks. > Our project is using an MPC860T with an M-Systems disk-on-a-chip flash > device for persistant storage. The VxWorks dosFS will be thr primary > interface to the flash device. > > We need to store both fixed and variable length data records in flash, > allowing for fast retrieval and for updating which includes insertion, > replacement and deletion. However, we don't want soething which will > cause un-necessary wear on the flash device. > > Thanks in advance for any responses > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Shell command from an app. Date: Thu, 25 Mar 1999 08:50:58 +0100 From: Wolfgang Meisl Organization: VIC - VLT Instrument Consortium Message-ID: <36F9EAE2.F3432620@bigbang.usm.uni-muenchen.de> References: <36F8FD4F.A83066D0@tekelec.com> Kevin Sapp wrote: > > I would like to give a running application the ability to execute > a command using the VxWorks shell. I understand that it is not > reentrant... > > I have tried to use shellOrigStdSet to change the STDIN of the > shell to a pipe and a file. And then sent a printf command to > the pipe (or had it in a file) but never saw the output. I did > not redirect STDOUT or STDERR. > > Any Ideas ? > Hi Kevin, are you sure that you need the VxWorks shell? You may call anything directly from a task like from the shell. The VxWorks is not comparable to UNIX shells. E.g., there are not shell-specific environment settings. It is only a simple C-Interpreter. - -------------------------------------------------------------------------- Wolfgang Meisl Universitaets-Sternwarte Muenchen wmeisl@bigbang.usm.uni-muenchen.de Tel ++49-+89-2180-5969 http://bigbang.usm.uni-muenchen.de:8002/~wmeisl/ - -------------------------------------------------------------------------- --------------------------- End of New-News digest ********************** From owner-vxwexplo-process Mon Mar 29 05:10:21 1999 From: Rajesh Gautam Date: Mon Mar 29 05:10:25 PST 1999 Subject: CMU's SNMP port Status: O Hello, Has anybody used the CMU SNMP library's port on VxWorks, which release of CMU was used, which versions of SNMP ( RFCs) does it support (v2c etc.). Mail to the authors mail address has bounced. Thanks , gautam. -- -------------------------------- Network Machines, 311, Ganga Shopping Complex, Noida, INDIA, 201301. Ph# 91-575-544783, 548517, 548518. Fax# 91-575-543721. From owner-vxwexplo-process Tue Mar 30 04:00:28 1999 From: daemon@csg.lbl.gov Date: Tue Mar 30 04:00:31 PST 1999 Subject: comp.os.vxworks newsdigest Status: O Comp.Os.Vxworks Daily Digest Tue Mar 30 04:00:24 PST 1999 Subject: Re: StrongARM 1100 (BRUTUS) Subject: MAXDOUBLE for vx works Subject: Execute host shell script ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: StrongARM 1100 (BRUTUS) Date: 29 Mar 1999 16:34:31 GMT From: gee@pcgee.dciem.dnd.ca (Tom Gee) Organization: DCIEM Message-ID: References: On Mon, 29 Mar 1999 07:45:29 GMT, Robert Armstrong wrote: >In article , "Peter Smith" wrote: >>Hi, >> >>Can anyone help with the answers to the following questions ? >> >>1) What is the BRUTUS board i.e. what's the Intel name for it ? > >It's still called Brutus as far as I know, they were showing the originals at >ESC west last year and repaired one for me in February. Hmm, I've searched the Intel developer website (developer.intel.com), and can't find any reference to "Brutus". They do have an SA1100 evaluation board, which I think is the same thing. Does anyone know for sure? Thanks, Tom - --- gee@dciem.dnd.ca SAILSS Life Support Development / DCIEM / Canada --------------------------- Newsgroups: comp.os.vxworks Subject: MAXDOUBLE for vx works Date: Mon, 29 Mar 1999 16:35:36 -0500 From: "Michael J. Abowd" Organization: CMU Software Engineering Message-ID: <7dornk$b2t$1@goldenapple.srv.cs.cmu.edu> Anyone know where MAXDOUBLE and MINDOUBLE are #define'ed? They are not in limits.h Thanks, Michael Abowd --------------------------- Newsgroups: comp.os.vxworks Subject: Execute host shell script Date: Tue, 30 Mar 1999 10:08:12 +0900 From: Aaron Jang Organization: Integra Telecom Co., Ltd. Seoul KOREA Message-ID: <370023FB.D1A7C729@integra.co.kr> Hello there? I want to do somthing on the host system. My target system using RSH protocol to access host file, and create a unix shell script file on the host. But I dont't know how to execute the host shell script file. The 'go' trigger is generated in my target system Any vxWorkers give me some advice. Thank you. - -- Aaron Jang (aaron@integra.co.kr,"ÀåÀÎÁø") Integra Telecom Co., Ltd. http://www.integra.co.kr Hwayoung Bdg., 112-2, Garak-Dong, Tel : +82-2-449-6851 Songpa-Gu, Seoul, 138-160, Korea Fax : +82-2-431-3741 --------------------------- End of New-News digest ********************** From owner-vxwexplo-process Tue Mar 30 08:18:15 1999 From: "Rasty Slutsker (STX)" Date: Tue Mar 30 08:18:18 PST 1999 Subject: x86 FPU Exceptions in VxWorks Status: O 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_01BE7AC8.B99CD0E4 Content-Type: text/plain; charset="ISO-8859-1" Hi, WRS does not supply FPU exception handler for x86 architecture. We've designed write simple one. The purpose was to stop a task caused the FPU fault and clear exception. Everything was fine until we tested how sscanf() handles NaNs. It has been discovered that sscanf() reads with FPU registers to detect invalid operation, moreover it writes something into Control Registers thus reprogramming FPU! Does somebody have experience with FPU exception handler for x86? Have anybody seen something similar? Regards, Rasty. ------_=_NextPart_001_01BE7AC8.B99CD0E4 Content-Type: text/html; charset="ISO-8859-1" Content-Transfer-Encoding: quoted-printable x86 FPU Exceptions in VxWorks

Hi,

WRS does not = supply FPU exception = handler for x86 = architecture.

We've designed = write simple one. The = purpose was to stop a task caused the FPU fault and = clear exception. Everything was = fine until we tested how sscanf() handles NaNs.  = It has been discovered that sscanf() reads with = FPU registers to detect invalid operation, moreover it writes something into Control Registers thus reprogramming = FPU!

Does somebody have experience = with FPU exception handler for = x86? Have anybody seen something similar?


Regards,

Rasty.

------_=_NextPart_001_01BE7AC8.B99CD0E4-- From owner-vxwexplo-process Tue Mar 30 09:03:15 1999 From: David Laight Date: Tue Mar 30 09:03:19 PST 1999 Subject: RE: StrongARM SA1100 (BRUTUS) Status: O The SA1100 evaluation board is indeed 'BRUTUS'. Intel also have a 'StrongARM SA-1100 Multimedia Development board with Companion SA-1101 development board' (MESA in some places). The brutus BSP runs on the former, the latter is different enough to require a BSP port. In particular the PCMCIA and display devices are different - so its hard working out what is going on. We now have prototypes of our SA-1100/SA-1101 board working under vxWorks (about 8000 lines of source for device drivers!) David ---------------------------------------------------------------- David Laight email: dsl@tadpole.co.uk Tadpole Technology plc phone: +44 1223 278 256 Cambridge, UK fax: +44 1223 278 201 From owner-vxwexplo-process Wed Mar 31 04:00:10 1999 From: daemon@csg.lbl.gov Date: Wed Mar 31 04:00:13 PST 1999 Subject: comp.os.vxworks newsdigest Status: O Comp.Os.Vxworks Daily Digest Wed Mar 31 04:00:07 PST 1999 Subject: Help - Port TCL to VxWorks ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Help - Port TCL to VxWorks Date: Tue, 30 Mar 1999 13:47:32 -0500 From: Edward Lee Organization: Nortel Networks Message-ID: <37011C44.AE317B10@nortelnetworks.com> Hi All, I'm looking to port TCL 8.1 to vxWorks. Has anyone here done this before? I'd appreciate some help. Thanks, Edward --------------------------- End of New-News digest ********************** From owner-vxwexplo-process Wed Mar 31 06:15:30 1999 From: Kevin Sapp Date: Wed Mar 31 06:15:34 PST 1999 Subject: Symbol Tables and loading. Status: O Hello, When using the VxWorks console and virtual consoles, different symbol tables get updated. Is there a way to stop this. For example, if I load a library type module at the console, and then later I open a vio using Tornado, and load a module that uses the libray module code I loaded, I get unresolves. Is there anyway to ask VxWorks loader to always put the symbols into the same table? Thanks Kevin From owner-vxwexplo-process Wed Mar 31 07:30:37 1999 From: "John R. Moore III" Date: Wed Mar 31 07:30:41 PST 1999 Subject: Re: Symbol Tables and loading. Status: O Make sure that you have the INCLUDE_SYM_TBL_SYNC feature included when you build the kernel, and when you launch your target server select the Target/Host symbol tables synchronization box when creating the target server. This setup will still have two different symbol tables, but a task is spawned that intermittently checks to see if the two symbol tables differ and will make them the same if they do. HTH John the vxWorks Users Group Exploder wrote: > > Submitted-by owner-vxwexplo-process Wed Mar 31 06:15:30 1999 > Submitted-by: Kevin Sapp > > Hello, > > When using the VxWorks console and virtual consoles, different > symbol tables get updated. Is there a way to stop this. > > For example, if I load a library type module at the console, and > then later I open a vio using Tornado, and load a module that uses > the libray module code I loaded, I get unresolves. > > Is there anyway to ask VxWorks loader to always put the symbols into > the same table? > > Thanks > Kevin > -- ------------------------------------------------------------ |John R. Moore | Voice -> 703 448 1683 x229 | |Principal Engineer | FAX -> 703 893 5494 | |EMAIL: | On the WEB! | |jrm@mclean.sparta.com | http://www.mclean.sparta.com | |___________________________|______________________________| | | | | /\\ SPARTA, Inc. | | | ///\\ 7926 Jones Branch Drive | | | /////\\ Suite 900 | | | ///// \\ McLean, VA 22102 | | | / \\\\\ \\ 703-448-0120 | | | \ \\\\\ // | | | \ ////// | | | \////// SPARTA | | | \//// Pride In Performance | | | \// -------------------- | | ------------------------------------------------------------ From owner-vxwexplo-process Wed Mar 31 17:15:15 1999 From: Debi Dawson-Lujan Date: Wed Mar 31 17:15:19 PST 1999 Subject: New Products from Intoto Inc. Status: O Hello, Previously you had contacted Technology Rendezvous Inc. for embedded system software solutions. Therefore, we would like to take this opportunity to inform you of our new spin-off called INTOTO INC. and the products we offer. Intoto Inc. provides embedded software solutions for Networking and Interconnectivity market. - IEEE-1394, IKE, IPSEC, NAT, USB, PCMCIA and Flash Software solutions. Below is the information regarding the new company Intoto Inc. and the solutions we offer. Should you like to receive some additional information please do not hesitate to contact me or visit our web site. http://www.intototinc.com Debi Dawson-Lujan Marketing & Sales Coordinator 3160 De La Cruz Blvd. Suite # 100 Santa Clara CA - 95054 USA T: 408 844 0480 x 303 F: 408 844 0488 ==================================================== Corporate Overview ---------------------------- Intoto Inc. develops, markets and manufactures reliable, off-the-shelf, and easy-to-use customized embedded system solutions that accelerate the customer's product development in the Networking and Interconnectivity market. Interconectivity Solutions ------------------------------------ - eFireStack - IEEE-1394 software and Protocols for Embedded Systems - LINCOLN - IEEE-1394 Link Layer and other related Cores - ePCMCIA - PCMCIA software with Card and Socket Services - eFMM - Flash software supporting on-board and PCMCIA Flash Devices - eUSBStack - Universal Serial Bus (USB) software with Class Drivers (Q3) Networking Solutions ------------------------------ - eNAT - Network Address Translation (NAT) Software - eIKE & eIPSEC - IPSEC software for Embedded Systems (Q2/Q3) - HTTP Server, DHCP and SNMP V1 Supported CPUs : x86, StrongARM, ARM, MIPS, SH, 68k, DSP, etc. Supported RTOS : VxWorks, pSOS, Nucleus, ThreadX, OSE, OS/9 Windows CE, QNX, etc. Markets and Associations -------------------------------------- Intoto's solutions address the needs of the Embedded market for digital multimedia, consumer electronics such as HDTV, Settop boxes and DVD players; for computer peripherals such as Printers, Scanners, Disk Drivers, MFPs; and for Networking, Virtual Private Networks such as VPN router, Switch-Router, Load Balancer, etc. Intoto is an active member of 1394TA, HAVi, PWG-C, USB, OpenCable, VESA. MS SerifINTOTO, INC. THE Solution Provider for Embedded Systems: IEEE-1394 PCMCIA Flash IPSEC USB sIKE+IPSEC Debi Dawson-Lujan Marketing & Sales Coordinator 3160 De La Cruz Blvd. Suite # 100 Santa Clara CA 95054 T: 408 844 0480 x 303 F: 408 844 0488